ViewVC logotype

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log

Revision 1639 - (show annotations)
Sun Feb 28 14:56:33 2016 UTC (4 years, 10 months ago) by ph10
File size: 288732 byte(s)
Fix typo in pcre_study().
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.
76 Version 8.38 23-November-2015
77 -----------------------------
79 1. If a group that contained a recursive back reference also contained a
80 forward reference subroutine call followed by a non-forward-reference
81 subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
82 compile correct code, leading to undefined behaviour or an internally
83 detected error. This bug was discovered by the LLVM fuzzer.
85 2. Quantification of certain items (e.g. atomic back references) could cause
86 incorrect code to be compiled when recursive forward references were
87 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
88 This bug was discovered by the LLVM fuzzer.
90 3. A repeated conditional group whose condition was a reference by name caused
91 a buffer overflow if there was more than one group with the given name.
92 This bug was discovered by the LLVM fuzzer.
94 4. A recursive back reference by name within a group that had the same name as
95 another group caused a buffer overflow. For example:
96 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
98 5. A forward reference by name to a group whose number is the same as the
99 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
100 a buffer overflow at compile time. This bug was discovered by the LLVM
101 fuzzer.
103 6. A lookbehind assertion within a set of mutually recursive subpatterns could
104 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
106 7. Another buffer overflow bug involved duplicate named groups with a
107 reference between their definition, with a group that reset capture
108 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
109 fixed by always allowing for more memory, even if not needed. (A proper fix
110 is implemented in PCRE2, but it involves more refactoring.)
112 8. There was no check for integer overflow in subroutine calls such as (?123).
114 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
115 being treated as a literal 'l' instead of causing an error.
117 10. There was a buffer overflow if pcre_exec() was called with an ovector of
118 size 1. This bug was found by american fuzzy lop.
120 11. If a non-capturing group containing a conditional group that could match
121 an empty string was repeated, it was not identified as matching an empty
122 string itself. For example: /^(?:(?(1)x|)+)+$()/.
124 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
125 \a and \e in test subject lines.
127 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
128 instead of the EBCDIC value.
130 14. The handling of \c in an EBCDIC environment has been revised so that it is
131 now compatible with the specification in Perl's perlebcdic page.
133 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
134 ASCII/Unicode. This has now been added to the list of characters that are
135 recognized as white space in EBCDIC.
137 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
138 error (correctly) when used outside a class, but did not give an error
139 within a class.
141 17. \h within a class was incorrectly compiled in EBCDIC environments.
143 18. A pattern with an unmatched closing parenthesis that contained a backward
144 assertion which itself contained a forward reference caused buffer
145 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
147 19. JIT should return with error when the compiled pattern requires more stack
148 space than the maximum.
150 20. A possessively repeated conditional group that could match an empty string,
151 for example, /(?(R))*+/, was incorrectly compiled.
153 21. Fix infinite recursion in the JIT compiler when certain patterns such as
154 /(?:|a|){100}x/ are analysed.
156 22. Some patterns with character classes involving [: and \\ were incorrectly
157 compiled and could cause reading from uninitialized memory or an incorrect
158 error diagnosis.
160 23. Pathological patterns containing many nested occurrences of [: caused
161 pcre_compile() to run for a very long time.
163 24. A conditional group with only one branch has an implicit empty alternative
164 branch and must therefore be treated as potentially matching an empty
165 string.
167 25. If (?R was followed by - or + incorrect behaviour happened instead of a
168 diagnostic.
170 26. Arrange to give up on finding the minimum matching length for overly
171 complex patterns.
173 27. Similar to (4) above: in a pattern with duplicated named groups and an
174 occurrence of (?| it is possible for an apparently non-recursive back
175 reference to become recursive if a later named group with the relevant
176 number is encountered. This could lead to a buffer overflow. Wen Guanxing
177 from Venustech ADLAB discovered this bug.
179 28. If pcregrep was given the -q option with -c or -l, or when handling a
180 binary file, it incorrectly wrote output to stdout.
182 29. The JIT compiler did not restore the control verb head in case of *THEN
183 control verbs. This issue was found by Karl Skomski with a custom LLVM
184 fuzzer.
186 30. Error messages for syntax errors following \g and \k were giving inaccurate
187 offsets in the pattern.
189 31. Added a check for integer overflow in conditions (?(<digits>) and
190 (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
191 fuzzer.
193 32. Handling recursive references such as (?2) when the reference is to a group
194 later in the pattern uses code that is very hacked about and error-prone.
195 It has been re-written for PCRE2. Here in PCRE1, a check has been added to
196 give an internal error if it is obvious that compiling has gone wrong.
198 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
199 This issue was found by Karl Skomski with a custom LLVM fuzzer.
201 34. The JIT compiler should restore the control chain for empty possessive
202 repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
204 35. Match limit check added to JIT recursion. This issue was found by Karl
205 Skomski with a custom LLVM fuzzer.
207 36. Yet another case similar to 27 above has been circumvented by an
208 unconditional allocation of extra memory. This issue is fixed "properly" in
209 PCRE2 by refactoring the way references are handled. Wen Guanxing
210 from Venustech ADLAB discovered this bug.
212 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
213 with a custom LLVM fuzzer.
215 38. Fixed a corner case of range optimization in JIT.
217 39. An incorrect error "overran compiling workspace" was given if there were
218 exactly enough group forward references such that the last one extended
219 into the workspace safety margin. The next one would have expanded the
220 workspace. The test for overflow was not including the safety margin.
222 40. A match limit issue is fixed in JIT which was found by Karl Skomski
223 with a custom LLVM fuzzer.
225 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
226 work under Windows. (Why has it taken so long for anyone to notice?)
228 42. In a character class such as [\W\p{Any}] where both a negative-type escape
229 ("not a word character") and a property escape were present, the property
230 escape was being ignored.
232 43. Fix crash caused by very long (*MARK) or (*THEN) names.
234 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
235 by a single ASCII character in a class item, was incorrectly compiled in
236 UCP mode. The POSIX class got lost, but only if the single character
237 followed it.
239 45. [:punct:] in UCP mode was matching some characters in the range 128-255
240 that should not have been matched.
242 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
243 class, all characters with code points greater than 255 are in the class.
244 When a Unicode property was also in the class (if PCRE_UCP is set, escapes
245 such as \w are turned into Unicode properties), wide characters were not
246 correctly handled, and could fail to match.
249 Version 8.37 28-April-2015
250 --------------------------
252 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
253 for those parentheses to be closed with whatever has been captured so far.
254 However, it was failing to mark any other groups between the hightest
255 capture so far and the currrent group as "unset". Thus, the ovector for
256 those groups contained whatever was previously there. An example is the
257 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
259 2. If an assertion condition was quantified with a minimum of zero (an odd
260 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
262 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
263 unrecognized modifier, a crash could occur.
265 4. An attempt to do global matching in pcretest with a zero-length ovector
266 caused a crash.
268 5. Fixed a memory leak during matching that could occur for a subpattern
269 subroutine call (recursive or otherwise) if the number of captured groups
270 that had to be saved was greater than ten.
272 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
273 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
274 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
276 7. A UTF pattern containing a "not" match of a non-ASCII character and a
277 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
279 8. When a pattern is compiled, it remembers the highest back reference so that
280 when matching, if the ovector is too small, extra memory can be obtained to
281 use instead. A conditional subpattern whose condition is a check on a
282 capture having happened, such as, for example in the pattern
283 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
284 setting the highest backreference number. This mattered only if pcre_exec()
285 was called with an ovector that was too small to hold the capture, and there
286 was no other kind of back reference (a situation which is probably quite
287 rare). The effect of the bug was that the condition was always treated as
288 FALSE when the capture could not be consulted, leading to a incorrect
289 behaviour by pcre_exec(). This bug has been fixed.
291 9. A reference to a duplicated named group (either a back reference or a test
292 for being set in a conditional) that occurred in a part of the pattern where
293 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
294 to be incorrectly calculated, leading to overwriting.
296 10. A mutually recursive set of back references such as (\2)(\1) caused a
297 segfault at study time (while trying to find the minimum matching length).
298 The infinite loop is now broken (with the minimum length unset, that is,
299 zero).
301 11. If an assertion that was used as a condition was quantified with a minimum
302 of zero, matching went wrong. In particular, if the whole group had
303 unlimited repetition and could match an empty string, a segfault was
304 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
305 assertions to be quantified, but not if they are being used as conditions,
306 so the above pattern is faulted by Perl. PCRE has now been changed so that
307 it also rejects such patterns.
309 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
310 failed to allow the zero-repeat case if pcre2_exec() was called with an
311 ovector too small to capture the group.
313 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
314 Red Hat Product Security:
316 (a) A crash if /K and /F were both set with the option to save the compiled
317 pattern.
319 (b) Another crash if the option to print captured substrings in a callout
320 was combined with setting a null ovector, for example \O\C+ as a subject
321 string.
323 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
324 forward reference repeated a large (but limited) number of times within a
325 repeated outer group that has a zero minimum quantifier, caused incorrect
326 code to be compiled, leading to the error "internal error:
327 previously-checked referenced subpattern not found" when an incorrect
328 memory address was read. This bug was reported as "heap overflow",
329 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
330 CVE-2015-2325.
332 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
333 call within a group that also contained a recursive back reference caused
334 incorrect code to be compiled. This bug was reported as "heap overflow",
335 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
336 number CVE-2015-2326.
338 24. Computing the size of the JIT read-only data in advance has been a source
339 of various issues, and new ones are still appear unfortunately. To fix
340 existing and future issues, size computation is eliminated from the code,
341 and replaced by on-demand memory allocation.
343 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
344 adjacent to the end of the range, and the range contained characters with
345 more than one other case, caused incorrect behaviour when compiled in UTF
346 mode. In that example, the range a-j was left out of the class.
348 26. Fix JIT compilation of conditional blocks, which assertion
349 is converted to (*FAIL). E.g: /(?(?!))/.
351 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
352 discovered by the LLVM fuzzer.
354 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
355 when this assertion was used as a condition, for example (?(?!)a|b). In
356 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
357 error about an unsupported item.
359 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
360 possessification code could take exponential time to complete. A recursion
361 depth limit of 1000 has been imposed to limit the resources used by this
362 optimization.
364 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
365 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
366 because \S ensures they are all in the class. The code for doing this was
367 interacting badly with the code for computing the amount of space needed to
368 compile the pattern, leading to a buffer overflow. This bug was discovered
369 by the LLVM fuzzer.
371 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
372 other kinds of group caused stack overflow at compile time. This bug was
373 discovered by the LLVM fuzzer.
375 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
376 between a subroutine call and its quantifier was incorrectly compiled,
377 leading to buffer overflow or other errors. This bug was discovered by the
378 LLVM fuzzer.
380 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
381 assertion after (?(. The code was failing to check the character after
382 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
383 was discovered by the LLVM fuzzer.
385 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
386 a fixed maximum following a group that contains a subroutine reference was
387 incorrectly compiled and could trigger buffer overflow. This bug was
388 discovered by the LLVM fuzzer.
390 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
391 caused a stack overflow instead of the diagnosis of a non-fixed length
392 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
394 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
395 (e.g. /(?<=\Ka)/) could make pcregrep loop.
397 37. There was a similar problem to 36 in pcretest for global matches.
399 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
400 and a subsequent item in the pattern caused a non-match, backtracking over
401 the repeated \X did not stop, but carried on past the start of the subject,
402 causing reference to random memory and/or a segfault. There were also some
403 other cases where backtracking after \C could crash. This set of bugs was
404 discovered by the LLVM fuzzer.
406 39. The function for finding the minimum length of a matching string could take
407 a very long time if mutual recursion was present many times in a pattern,
408 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
409 method has been implemented. This infelicity was discovered by the LLVM
410 fuzzer.
412 40. Static linking against the PCRE library using the pkg-config module was
413 failing on missing pthread symbols.
416 Version 8.36 26-September-2014
417 ------------------------------
419 1. Got rid of some compiler warnings in the C++ modules that were shown up by
420 -Wmissing-field-initializers and -Wunused-parameter.
422 2. The tests for quantifiers being too big (greater than 65535) were being
423 applied after reading the number, and stupidly assuming that integer
424 overflow would give a negative number. The tests are now applied as the
425 numbers are read.
427 3. Tidy code in pcre_exec.c where two branches that used to be different are
428 now the same.
430 4. The JIT compiler did not generate match limit checks for certain
431 bracketed expressions with quantifiers. This may lead to exponential
432 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
433 issue should be resolved now.
435 5. Fixed an issue, which occures when nested alternatives are optimized
436 with table jumps.
438 6. Inserted two casts and changed some ints to size_t in the light of some
439 reported 64-bit compiler warnings (Bugzilla 1477).
441 7. Fixed a bug concerned with zero-minimum possessive groups that could match
442 an empty string, which sometimes were behaving incorrectly in the
443 interpreter (though correctly in the JIT matcher). This pcretest input is
444 an example:
446 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
449 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
450 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
451 for an empty string was breaking the inner loop and carrying on at a lower
452 level, when possessive repeated groups should always return to a higher
453 level as they have no backtrack points in them. The empty string test now
454 occurs at the outer level.
456 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
457 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
459 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
460 doesn't).
462 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
463 to hold a pattern to be incorrectly computed (too small) when there were
464 named back references to duplicated names. This could cause "internal
465 error: code overflow" or "double free or corruption" or other memory
466 handling errors.
468 11. When named subpatterns had the same prefixes, back references could be
469 confused. For example, in this pattern:
471 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
473 the reference to 'Name' was incorrectly treated as a reference to a
474 duplicate name.
476 12. A pattern such as /^s?c/mi8 where the optional character has more than
477 one "other case" was incorrectly compiled such that it would only try to
478 match starting at "c".
480 13. When a pattern starting with \s was studied, VT was not included in the
481 list of possible starting characters; this should have been part of the
482 8.34/18 patch.
484 14. If a character class started [\Qx]... where x is any character, the class
485 was incorrectly terminated at the ].
487 15. If a pattern that started with a caseless match for a character with more
488 than one "other case" was studied, PCRE did not set up the starting code
489 unit bit map for the list of possible characters. Now it does. This is an
490 optimization improvement, not a bug fix.
492 16. The Unicode data tables have been updated to Unicode 7.0.0.
494 17. Fixed a number of memory leaks in pcregrep.
496 18. Avoid a compiler warning (from some compilers) for a function call with
497 a cast that removes "const" from an lvalue by using an intermediate
498 variable (to which the compiler does not object).
500 19. Incorrect code was compiled if a group that contained an internal recursive
501 back reference was optional (had quantifier with a minimum of zero). This
502 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
503 caused segmentation faults because of stack overflows at compile time.
505 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
506 group that is quantified with an indefinite repeat, caused a compile-time
507 loop which used up all the system stack and provoked a segmentation fault.
508 This was not the same bug as 19 above.
510 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
511 Patch by Mike Frysinger.
514 Version 8.35 04-April-2014
515 --------------------------
517 1. A new flag is set, when property checks are present in an XCLASS.
518 When this flag is not set, PCRE can perform certain optimizations
519 such as studying these XCLASS-es.
521 2. The auto-possessification of character sets were improved: a normal
522 and an extended character set can be compared now. Furthermore
523 the JIT compiler optimizes more character set checks.
525 3. Got rid of some compiler warnings for potentially uninitialized variables
526 that show up only when compiled with -O2.
528 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
529 of a match later then the end of the match. The pcretest program was not
530 handling the case sensibly - it was outputting from the start to the next
531 binary zero. It now reports this situation in a message, and outputs the
532 text from the end to the start.
534 5. Fast forward search is improved in JIT. Instead of the first three
535 characters, any three characters with fixed position can be searched.
536 Search order: first, last, middle.
538 6. Improve character range checks in JIT. Characters are read by an inprecise
539 function now, which returns with an unknown value if the character code is
540 above a certain threshold (e.g: 256). The only limitation is that the value
541 must be bigger than the threshold as well. This function is useful when
542 the characters above the threshold are handled in the same way.
544 7. The macros whose names start with RAWUCHAR are placeholders for a future
545 mode in which only the bottom 21 bits of 32-bit data items are used. To
546 make this more memorable for those maintaining the code, the names have
547 been changed to start with UCHAR21, and an extensive comment has been added
548 to their definition.
550 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
551 to the export list in Makefile.am (they were accidentally omitted from the
552 8.34 tarball).
554 9. The informational output from pcretest used the phrase "starting byte set"
555 which is inappropriate for the 16-bit and 32-bit libraries. As the output
556 for "first char" and "need char" really means "non-UTF-char", I've changed
557 "byte" to "char", and slightly reworded the output. The documentation about
558 these values has also been (I hope) clarified.
560 10. Another JIT related optimization: use table jumps for selecting the correct
561 backtracking path, when more than four alternatives are present inside a
562 bracket.
564 11. Empty match is not possible, when the minimum length is greater than zero,
565 and there is no \K in the pattern. JIT should avoid empty match checks in
566 such cases.
568 12. In a caseless character class with UCP support, when a character with more
569 than one alternative case was not the first character of a range, not all
570 the alternative cases were added to the class. For example, s and \x{17f}
571 are both alternative cases for S: the class [RST] was handled correctly,
572 but [R-T] was not.
574 13. The configure.ac file always checked for pthread support when JIT was
575 enabled. This is not used in Windows, so I have put this test inside a
576 check for the presence of windows.h (which was already tested for).
578 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
579 The algorithm provides a way to skip certain starting offsets, and usually
580 faster than linear prefix searches.
582 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
583 as for 'fr_FR' and 'french'. For some reason, however, it then used the
584 Windows-specific input and output files, which have 'french' screwed in.
585 So this could never have worked. One of the problems with locales is that
586 they aren't always the same. I have now updated RunTest so that it checks
587 the output of the locale test (test 3) against three different output
588 files, and it allows the test to pass if any one of them matches. With luck
589 this should make the test pass on some versions of Solaris where it was
590 failing. Because of the uncertainty, the script did not used to stop if
591 test 3 failed; it now does. If further versions of a French locale ever
592 come to light, they can now easily be added.
594 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
595 there was a message during ./configure, but it did not stop. This now
596 provokes an error. The invalid example in README has been corrected.
597 If a value less than the minimum is given, the minimum value has always
598 been used, but now a warning is given.
600 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
601 was a bug in the test system, which is now fixed. Also, the list of various
602 configurations that are tested for each release did not have one with both
603 16/32 bits and --enable-bar-anycrlf. It now does.
605 18. pcretest was missing "-C bsr" for displaying the \R default setting.
607 19. Little endian PowerPC systems are supported now by the JIT compiler.
609 20. The fast forward newline mechanism could enter to an infinite loop on
610 certain invalid UTF-8 input. Although we don't support these cases
611 this issue can be fixed by a performance optimization.
613 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
614 not take account if existing stack usage. There is now a new global
615 variable called pcre_stack_guard that can be set to point to an external
616 function to check stack availability. It is called at the start of
617 processing every parenthesized group.
619 22. A typo in the code meant that in ungreedy mode the max/min qualifier
620 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
621 not match "ab".
623 23. When UTF was disabled, the JIT program reported some incorrect compile
624 errors. These messages are silenced now.
626 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
627 compiler.
629 25. Change all the temporary files used in RunGrepTest to be different to those
630 used by RunTest so that the tests can be run simultaneously, for example by
631 "make -j check".
634 Version 8.34 15-December-2013
635 -----------------------------
637 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
638 executable memory. Patch inspired by Carsten Klein.
640 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
641 this macro is never tested and has no effect, because the work to support
642 coverage involves only compiling and linking options and special targets in
643 the Makefile. The comment in config.h implied that defining the macro would
644 enable coverage support, which is totally false. There was also support for
645 setting this macro in the CMake files (my fault, I just copied it from
646 configure). SUPPORT_GCOV has now been removed.
648 3. Make a small performance improvement in strlen16() and strlen32() in
649 pcretest.
651 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
652 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
653 these cases). There was also one in pcretest.c.
655 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
657 6. In UTF mode, the code for checking whether a group could match an empty
658 string (which is used for indefinitely repeated groups to allow for
659 breaking an infinite loop) was broken when the group contained a repeated
660 negated single-character class with a character that occupied more than one
661 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
662 UTF-8 mode). The effect was undefined: the group might or might not be
663 deemed as matching an empty string, or the program might have crashed.
665 7. The code for checking whether a group could match an empty string was not
666 recognizing that \h, \H, \v, \V, and \R must match a character.
668 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
669 an empty string. If it can, pcretest shows this in its information output.
671 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
672 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
673 matched by pcre_exec() without using JIT:
675 (a) If the rest of the pattern did not match after a maximal run of
676 grapheme clusters, the code for backing up to try with fewer of them
677 did not always back up over a full grapheme when characters that do not
678 have the modifier quality were involved, e.g. Hangul syllables.
680 (b) If the match point in a subject started with modifier character, and
681 there was no match, the code could incorrectly back up beyond the match
682 point, and potentially beyond the first character in the subject,
683 leading to a segfault or an incorrect match result.
685 10. A conditional group with an assertion condition could lead to PCRE
686 recording an incorrect first data item for a match if no other first data
687 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
688 first data item, and therefore matched "ca" after "c" instead of at the
689 start.
691 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
692 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
694 12. The source of pcregrep now includes z/OS-specific code so that it can be
695 compiled for z/OS as part of the special z/OS distribution.
697 13. Added the -T and -TM options to pcretest.
699 14. The code in pcre_compile.c for creating the table of named capturing groups
700 has been refactored. Instead of creating the table dynamically during the
701 actual compiling pass, the information is remembered during the pre-compile
702 pass (on the stack unless there are more than 20 named groups, in which
703 case malloc() is used) and the whole table is created before the actual
704 compile happens. This has simplified the code (it is now nearly 150 lines
705 shorter) and prepared the way for better handling of references to groups
706 with duplicate names.
708 15. A back reference to a named subpattern when there is more than one of the
709 same name now checks them in the order in which they appear in the pattern.
710 The first one that is set is used for the reference. Previously only the
711 first one was inspected. This change makes PCRE more compatible with Perl.
713 16. Unicode character properties were updated from Unicode 6.3.0.
715 17. The compile-time code for auto-possessification has been refactored, based
716 on a patch by Zoltan Herczeg. It now happens after instead of during
717 compilation. The code is cleaner, and more cases are handled. The option
718 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
719 options in pcretest are provided to set it. It can also be set by
720 (*NO_AUTO_POSSESS) at the start of a pattern.
722 18. The character VT has been added to the default ("C" locale) set of
723 characters that match \s and are generally treated as white space,
724 following this same change in Perl 5.18. There is now no difference between
725 "Perl space" and "POSIX space". Whether VT is treated as white space in
726 other locales depends on the locale.
728 19. The code for checking named groups as conditions, either for being set or
729 for being recursed, has been refactored (this is related to 14 and 15
730 above). Processing unduplicated named groups should now be as fast at
731 numerical groups, and processing duplicated groups should be faster than
732 before.
734 20. Two patches to the CMake build system, by Alexander Barkov:
736 (1) Replace the "source" command by "." in CMakeLists.txt because
737 "source" is a bash-ism.
739 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
740 without these the CMake build does not work on Solaris.
742 21. Perl has changed its handling of \8 and \9. If there is no previously
743 encountered capturing group of those numbers, they are treated as the
744 literal characters 8 and 9 instead of a binary zero followed by the
745 literals. PCRE now does the same.
747 22. Following Perl, added \o{} to specify codepoints in octal, making it
748 possible to specify values greater than 0777 and also making them
749 unambiguous.
751 23. Perl now gives an error for missing closing braces after \x{... instead of
752 treating the string as literal. PCRE now does the same.
754 24. RunTest used to grumble if an inappropriate test was selected explicitly,
755 but just skip it when running all tests. This make it awkward to run ranges
756 of tests when one of them was inappropriate. Now it just skips any
757 inappropriate tests, as it always did when running all tests.
759 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
760 character types such as \d or \w, too many callouts were inserted, and the
761 data that they returned was rubbish.
763 26. In UCP mode, \s was not matching two of the characters that Perl matches,
764 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
765 were matched by \h. The code has now been refactored so that the lists of
766 the horizontal and vertical whitespace characters used for \h and \v (which
767 are defined only in one place) are now also used for \s.
769 27. Add JIT support for the 64 bit TileGX architecture.
770 Patch by Jiong Wang (Tilera Corporation).
772 28. Possessive quantifiers for classes (both explicit and automatically
773 generated) now use special opcodes instead of wrapping in ONCE brackets.
775 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
776 (because it's meaningless), this was not happening when PCRE_CASELESS was
777 set. Not wrong, but inefficient.
779 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
780 properties for \w, \d, etc) is present in a test regex. Otherwise if the
781 test contains no characters greater than 255, Perl doesn't realise it
782 should be using Unicode semantics.
784 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
785 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
786 does in Unicode mode.
788 32. Added the "forbid" facility to pcretest so that putting tests into the
789 wrong test files can sometimes be quickly detected.
791 33. There is now a limit (default 250) on the depth of nesting of parentheses.
792 This limit is imposed to control the amount of system stack used at compile
793 time. It can be changed at build time by --with-parens-nest-limit=xxx or
794 the equivalent in CMake.
796 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
797 errors. Perl warns for these when in warning mode, but PCRE has no facility
798 for giving warnings.
800 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
801 However, this was not working for (?!) because it is optimized to (*FAIL),
802 for which PCRE does not allow quantifiers. The optimization is now disabled
803 when a quantifier follows (?!). I can't see any use for this, but it makes
804 things uniform.
806 36. Perl no longer allows group names to start with digits, so I have made this
807 change also in PCRE. It simplifies the code a bit.
809 37. In extended mode, Perl ignores spaces before a + that indicates a
810 possessive quantifier. PCRE allowed a space before the quantifier, but not
811 before the possessive +. It now does.
813 38. The use of \K (reset reported match start) within a repeated possessive
814 group such as (a\Kb)*+ was not working.
816 40. Document that the same character tables must be used at compile time and
817 run time, and that the facility to pass tables to pcre_exec() and
818 pcre_dfa_exec() is for use only with saved/restored patterns.
820 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
821 features for Builds with MSVC:
823 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
824 builds). The .rc files can be used to set FileDescription and many other
825 attributes.
827 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
828 This allows higher-level build scripts which want .pdb files to avoid
829 hard-coding the exact files needed."
831 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
832 mean "start of word" and "end of word", respectively, as a transition aid.
834 43. A minimizing repeat of a class containing codepoints greater than 255 in
835 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
836 compiled to use the heap for recursion.
838 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
839 is configured.
842 Version 8.33 28-May-2013
843 ------------------------
845 1. Added 'U' to some constants that are compared to unsigned integers, to
846 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
847 variables that are added to signed variables, to ensure the result is
848 signed and can be negated.
850 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
851 CMake config files.
853 3. Revise the creation of config.h.generic so that all boolean macros are
854 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
855 overriding via -D on the command line possible.
857 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
858 to unsigned int is reported to make a quite noticeable speed difference in
859 a specific Windows environment. Testing on Linux did also appear to show
860 some benefit (and it is clearly not harmful). Also fixed the definition of
861 Xop which should be unsigned.
863 5. Related to (4), changing the definition of the intermediate variable cc
864 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
865 improvements.
867 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
868 unnecessary spaces.
870 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
871 and later.
873 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
875 9. Optimizing fast_forward_start_bits in JIT.
877 10. Adding support for callouts in JIT, and fixing some issues revealed
878 during this work. Namely:
880 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
882 (b) Minimum length was not checked before the matching is started.
884 11. The value of capture_last that is passed to callouts was incorrect in some
885 cases when there was a capture on one path that was subsequently abandoned
886 after a backtrack. Also, the capture_last value is now reset after a
887 recursion, since all captures are also reset in this case.
889 12. The interpreter no longer returns the "too many substrings" error in the
890 case when an overflowing capture is in a branch that is subsequently
891 abandoned after a backtrack.
893 13. In the pathological case when an offset vector of size 2 is used, pcretest
894 now prints out the matched string after a yield of 0 or 1.
896 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
897 Only supported by the JIT compiler at the moment.
899 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
901 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
902 offset of the starting point of the matching process, provided the offsets
903 vector is large enough.
905 17. The \A escape now records a lookbehind value of 1, though its execution
906 does not actually inspect the previous character. This is to ensure that,
907 in partial multi-segment matching, at least one character from the old
908 segment is retained when a new segment is processed. Otherwise, if there
909 are no lookbehinds in the pattern, \A might match incorrectly at the start
910 of a new segment.
912 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
914 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
915 gives some modest performance improvement in 8-bit mode.
917 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
918 be expressed in certain programming languages using Universal Character
919 Names.
921 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
922 which points out that "non characters" are not "characters that may not
923 appear in Unicode strings" but rather "characters that are reserved for
924 internal use and have only local meaning".
926 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
927 there was a conditional group that depended on an assertion, if the
928 assertion was false, the callout that immediately followed the alternation
929 in the condition was skipped when pcre_exec() was used for matching.
931 23. Allow an explicit callout to be inserted before an assertion that is the
932 condition for a conditional group, for compatibility with automatic
933 callouts, which always insert a callout at this point.
935 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
936 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
938 25. (*PRUNE) is now supported by the JIT compiler.
940 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
942 27. Fix the case where there are two or more SKIPs with arguments that may be
943 ignored.
945 28. (*SKIP) is now supported by the JIT compiler.
947 29. (*THEN) is now supported by the JIT compiler.
949 30. Update RunTest with additional test selector options.
951 31. The way PCRE handles backtracking verbs has been changed in two ways.
953 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
954 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
955 backtracking. In some cases this makes it more Perl-compatible, but Perl's
956 rather obscure rules do not always do the same thing.
958 (2) Previously, backtracking verbs were confined within assertions. This is
959 no longer the case for positive assertions, except for (*ACCEPT). Again,
960 this sometimes improves Perl compatibility, and sometimes does not.
962 32. A number of tests that were in test 2 because Perl did things differently
963 have been moved to test 1, because either Perl or PCRE has changed, and
964 these tests are now compatible.
966 32. Backtracking control verbs are now handled in the same way in JIT and
967 interpreter.
969 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
970 contained a forward subroutine reference caused a compile error.
972 34. Auto-detect and optimize limited repetitions in JIT.
974 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
975 blocking (*UTF) etc.
977 36. In the interpreter, maximizing pattern repetitions for characters and
978 character types now use tail recursion, which reduces stack usage.
980 37. The value of the max lookbehind was not correctly preserved if a compiled
981 and saved regex was reloaded on a host of different endianness.
983 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
984 of the compiled pattern block, expand the flags field from 16 to 32 bits
985 because it was almost full.
987 39. Try madvise first before posix_madvise.
989 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
990 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
991 original change locked out all matches of empty strings. This has been
992 changed so that one match of an empty string per line is recognized.
993 Subsequent searches on the same line (for colouring or for --only-matching,
994 for example) do not recognize empty strings.
996 41. Applied a user patch to fix a number of spelling mistakes in comments.
998 42. Data lines longer than 65536 caused pcretest to crash.
1000 43. Clarified the data type for length and startoffset arguments for pcre_exec
1001 and pcre_dfa_exec in the function-specific man pages, where they were
1002 explicitly stated to be in bytes, never having been updated. I also added
1003 some clarification to the pcreapi man page.
1005 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
1006 a segmentation fault.
1009 Version 8.32 30-November-2012
1010 -----------------------------
1012 1. Improved JIT compiler optimizations for first character search and single
1013 character iterators.
1015 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
1016 Patch by Daniel Richard G.
1018 3. Single character iterator optimizations in the JIT compiler.
1020 4. Improved JIT compiler optimizations for character ranges.
1022 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1023 Reported by Giuseppe D'Angelo.
1025 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1026 of a line, was being set incorrectly in cases where .* appeared inside
1027 atomic brackets at the start of a pattern, or where there was a subsequent
1028 *PRUNE or *SKIP.
1030 7. Improved instruction cache flush for POWER/PowerPC.
1031 Patch by Daniel Richard G.
1033 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
1034 grep:
1036 (a) There is now no limit to the number of patterns to be matched.
1038 (b) An error is given if a pattern is too long.
1040 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1041 are now supported.
1043 (d) --exclude-from and --include-from (multiple use) have been added.
1045 (e) Exclusions and inclusions now apply to all files and directories, not
1046 just to those obtained from scanning a directory recursively.
1048 (f) Multiple uses of -f and --file-list are now supported.
1050 (g) In a Windows environment, the default for -d has been changed from
1051 "read" (the GNU grep default) to "skip", because otherwise the presence
1052 of a directory in the file list provokes an error.
1054 (h) The documentation has been revised and clarified in places.
1056 9. Improve the matching speed of capturing brackets.
1058 10. Changed the meaning of \X so that it now matches a Unicode extended
1059 grapheme cluster.
1061 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1062 out POSIX threads when JIT support is configured.
1064 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1066 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1067 structure could go wrong in environments where size_t is not the same size
1068 as int.
1070 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1071 set.
1073 15. The EBCDIC support had decayed; later updates to the code had included
1074 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1075 general tidy up of EBCDIC-related issues, and the documentation was also
1076 not quite right. There is now a test that can be run on ASCII systems to
1077 check some of the EBCDIC-related things (but is it not a full test).
1079 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1080 in a small tidy to the code.
1082 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1084 18. If the --only-matching (-o) option in pcregrep is specified multiple
1085 times, each one causes appropriate output. For example, -o1 -o2 outputs the
1086 substrings matched by the 1st and 2nd capturing parentheses. A separating
1087 string can be specified by --om-separator (default empty).
1089 19. Improving the first n character searches.
1091 20. Turn case lists for horizontal and vertical white space into macros so that
1092 they are defined only once.
1094 21. This set of changes together give more compatible Unicode case-folding
1095 behaviour for characters that have more than one other case when UCP
1096 support is available.
1098 (a) The Unicode property table now has offsets into a new table of sets of
1099 three or more characters that are case-equivalent. The MultiStage2.py
1100 script that generates these tables (the pcre_ucd.c file) now scans
1101 CaseFolding.txt instead of UnicodeData.txt for character case
1102 information.
1104 (b) The code for adding characters or ranges of characters to a character
1105 class has been abstracted into a generalized function that also handles
1106 case-independence. In UTF-mode with UCP support, this uses the new data
1107 to handle characters with more than one other case.
1109 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1110 whose other case is greater than 256 are now correctly matched
1111 caselessly. Previously, the high codepoint matched the low one, but not
1112 vice versa.
1114 (d) The processing of \h, \H, \v, and \ in character classes now makes use
1115 of the new class addition function, using character lists defined as
1116 macros alongside the case definitions of 20 above.
1118 (e) Caseless back references now work with characters that have more than
1119 one other case.
1121 (f) General caseless matching of characters with more than one other case
1122 is supported.
1124 22. Unicode character properties were updated from Unicode 6.2.0
1126 23. Improved CMake support under Windows. Patch by Daniel Richard G.
1128 24. Add support for 32-bit character strings, and UTF-32
1130 25. Major JIT compiler update (code refactoring and bugfixing).
1131 Experimental Sparc 32 support is added.
1133 26. Applied a modified version of Daniel Richard G's patch to create
1134 pcre.h.generic and config.h.generic by "make" instead of in the
1135 PrepareRelease script.
1137 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1138 pcre_compile.c when checking for a zero character.
1140 28. Introducing a native interface for JIT. Through this interface, the compiled
1141 machine code can be directly executed. The purpose of this interface is to
1142 provide fast pattern matching, so several sanity checks are not performed.
1143 However, feature tests are still performed. The new interface provides
1144 1.4x speedup compared to the old one.
1146 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1147 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1148 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1150 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1151 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1152 the "old" RFC 2279). Instead, it ended up passing a negative length to
1153 pcre_exec().
1155 31. Add support for GCC's visibility feature to hide internal functions.
1157 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1158 "unknown -C option" after outputting 0 or 1.
1160 33. There is now support for generating a code coverage report for the test
1161 suite in environments where gcc is the compiler and lcov is installed. This
1162 is mainly for the benefit of the developers.
1164 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1165 unaddressable using valgrind annotations, allowing valgrind to detect
1166 invalid memory accesses. This is mainly for the benefit of the developers.
1168 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1170 26. Give configure error if --enable-cpp but no C++ compiler found.
1173 Version 8.31 06-July-2012
1174 -------------------------
1176 1. Fixing a wrong JIT test case and some compiler warnings.
1178 2. Removed a bashism from the RunTest script.
1180 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1181 to unsigned type, result still unsigned" that was given by an MS compiler
1182 on encountering the code "-sizeof(xxx)".
1184 4. Partial matching support is added to the JIT compiler.
1186 5. Fixed several bugs concerned with partial matching of items that consist
1187 of more than one character:
1189 (a) /^(..)\1/ did not partially match "aba" because checking references was
1190 done on an "all or nothing" basis. This also applied to repeated
1191 references.
1193 (b) \R did not give a hard partial match if \r was found at the end of the
1194 subject.
1196 (c) \X did not give a hard partial match after matching one or more
1197 characters at the end of the subject.
1199 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1200 a partial match for the string "\r".
1202 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1203 a partial match for a CR character at the end of the subject string.
1205 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1206 running pcretest, the text "(JIT)" added to the output whenever JIT is
1207 actually used to run the match.
1209 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1210 or /S+[+] with a digit between 1 and 7.
1212 8. OP_NOT now supports any UTF character not just single-byte ones.
1214 9. (*MARK) control verb is now supported by the JIT compiler.
1216 10. The command "./RunTest list" lists the available tests without actually
1217 running any of them. (Because I keep forgetting what they all are.)
1221 12. Applied a (slightly modified) user-supplied patch that improves performance
1222 when the heap is used for recursion (compiled with --disable-stack-for-
1223 recursion). Instead of malloc and free for each heap frame each time a
1224 logical recursion happens, frames are retained on a chain and re-used where
1225 possible. This sometimes gives as much as 30% improvement.
1227 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1228 call.
1230 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1232 15. It is now possible to link pcretest with libedit as an alternative to
1233 libreadline.
1235 16. (*COMMIT) control verb is now supported by the JIT compiler.
1237 17. The Unicode data tables have been updated to Unicode 6.1.0.
1239 18. Added --file-list option to pcregrep.
1241 19. Added binary file support to pcregrep, including the -a, --binary-files,
1242 -I, and --text options.
1244 20. The madvise function is renamed for posix_madvise for QNX compatibility
1245 reasons. Fixed by Giuseppe D'Angelo.
1247 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1249 22. Changed the option for creating symbolic links for 16-bit man pages from
1250 -s to -sf so that re-installing does not cause issues.
1252 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1254 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1255 matches in certain environments (the workspace was not being correctly
1256 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1257 some of the workspace data at the beginning of a restart.
1259 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1260 was not doing so when it should - probably a typo introduced by SVN 528
1261 (change 8.10/14).
1263 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1264 \w+ when the character tables indicated that \x{c4} was a word character.
1265 There were several related cases, all because the tests for doing a table
1266 lookup were testing for characters less than 127 instead of 255.
1268 27. If a pattern contains capturing parentheses that are not used in a match,
1269 their slots in the ovector are set to -1. For those that are higher than
1270 any matched groups, this happens at the end of processing. In the case when
1271 there were back references that the ovector was too small to contain
1272 (causing temporary malloc'd memory to be used during matching), and the
1273 highest capturing number was not used, memory off the end of the ovector
1274 was incorrectly being set to -1. (It was using the size of the temporary
1275 memory instead of the true size.)
1277 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1278 ovector size, it uses memory at the end of the block that it has got.
1280 29. Check for an overlong MARK name and give an error at compile time. The
1281 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1283 30. JIT compiler update.
1285 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1286 Rill for the patch.
1288 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1290 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1292 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1293 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1294 was enabled.
1296 35. Improve JIT code generation for greedy plus quantifier.
1298 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1299 1 to "aa" instead of to an empty string. The bug affected repeated groups
1300 that could potentially match an empty string.
1302 37. Optimizing single character iterators in JIT.
1304 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1305 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1306 codepoints that are too big for the mode are faulted, and in a UTF mode,
1307 disallowed codepoints are also faulted.
1309 39. If PCRE was compiled with UTF support, in three places in the DFA
1310 matcher there was code that should only have been obeyed in UTF mode, but
1311 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1312 processing when bytes with values greater than 127 were present. In 16-bit
1313 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1314 both cases the values are those that cannot be the first data item in a UTF
1315 character. The three items that might have provoked this were recursions,
1316 possessively repeated groups, and atomic groups.
1318 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1319 and pcregrep, because some OS require shared objects to be explicitly
1320 passed to ld, causing the link step to fail if they are not.
1322 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1323 mode, patterns that started with \h* or \R* might be incorrectly matched.
1326 Version 8.30 04-February-2012
1327 -----------------------------
1329 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1330 name is defined in ctype.h.
1332 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1333 only in quite long subpatterns.
1335 3. Removed the function pcre_info(), which has been obsolete and deprecated
1336 since it was replaced by pcre_fullinfo() in February 2000.
1338 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1339 match a (*MARK), and the match failed at the start of the subject, a
1340 reference to memory before the start of the subject could occur. This bug
1341 was introduced by fix 17 of release 8.21.
1343 5. A reference to an unset group with zero minimum repetition was giving
1344 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1345 /(another)?(\1?)test/ matched against "hello world test". This bug was
1346 introduced in release 8.13.
1348 6. Add support for 16-bit character strings (a large amount of work involving
1349 many changes and refactorings).
1351 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1352 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1353 from a file.
1355 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1356 rounding is not applied in this particular case).
1358 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1359 if they appear, or are escaped, in patterns.
1361 10. Get rid of a number of -Wunused-but-set-variable warnings.
1363 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1364 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1365 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1366 also returns the mark "x". This bug applied to capturing parentheses,
1367 non-capturing parentheses, and atomic parentheses. It also applied to some
1368 assertions.
1370 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1371 information out of configure.ac instead of relying on pcre.h.generic, which
1372 is not stored in the repository.
1374 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1375 -lreadline.
1377 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1379 15. Applied Graycode's patch to put the top-level frame on the stack rather
1380 than the heap when not using the stack for recursion. This gives a
1381 performance improvement in many cases when recursion is not deep.
1383 16. Experimental code added to "pcretest -C" to output the stack frame size.
1386 Version 8.21 12-Dec-2011
1387 ------------------------
1389 1. Updating the JIT compiler.
1391 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1392 are added as well.
1394 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1395 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1396 calling _pcre_jit_exec. Some extra comments are added.
1398 4. (*MARK) settings inside atomic groups that do not contain any capturing
1399 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1400 was introduced by change 18 for 8.20.
1402 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1403 ECMA-262 standard.
1405 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1406 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1407 This bug was probably introduced by change 9 of 8.13.
1409 7. While fixing 6 above, I noticed that a number of other items were being
1410 incorrectly rejected as "not fixed length". This arose partly because newer
1411 opcodes had not been added to the fixed-length checking code. I have (a)
1412 corrected the bug and added tests for these items, and (b) arranged for an
1413 error to occur if an unknown opcode is encountered while checking for fixed
1414 length instead of just assuming "not fixed length". The items that were
1415 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1416 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1417 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1419 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1420 being incorrectly compiled and would have given unpredicatble results.
1422 9. A possessively repeated subpattern with minimum repeat count greater than
1423 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1424 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1425 the first (A) could occur when it should not.
1427 10. Add a cast and remove a redundant test from the code.
1429 11. JIT should use pcre_malloc/pcre_free for allocation.
1431 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1432 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1433 is anything other than /usr, -L is still shown).
1435 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1437 14. Perl does not support \N without a following name in a [] class; PCRE now
1438 also gives an error.
1440 15. If a forward reference was repeated with an upper limit of around 2000,
1441 it caused the error "internal error: overran compiling workspace". The
1442 maximum number of forward references (including repeats) was limited by the
1443 internal workspace, and dependent on the LINK_SIZE. The code has been
1444 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1445 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1446 of around 200,000 forward references. While doing this, I also speeded up
1447 the filling in of repeated forward references.
1449 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1450 incorrectly expecting the subject to contain another "a" after the start.
1452 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1453 in the match, the SKIP should be ignored. This was not happening; instead
1454 the SKIP was being treated as NOMATCH. For patterns such as
1455 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1456 tested.
1458 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1459 now much more compatible with Perl, in particular in cases where the result
1460 is a non-match for a non-anchored pattern. For example, if
1461 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1462 "m", where previously it did not return a name. A side effect of this
1463 change is that for partial matches, the last encountered mark name is
1464 returned, as for non matches. A number of tests that were previously not
1465 Perl-compatible have been moved into the Perl-compatible test files. The
1466 refactoring has had the pleasing side effect of removing one argument from
1467 the match() function, thus reducing its stack requirements.
1469 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1470 subsequent uses of /S (without +) incorrectly behaved like /S+.
1472 21. Retrieve executable code size support for the JIT compiler and fixing
1473 some warnings.
1475 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1476 not work when the shorter character appeared right at the end of the
1477 subject string.
1479 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1480 systems.
1482 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1483 output it when the /M option is used in pcretest.
1485 25. The CheckMan script was not being included in the distribution. Also, added
1486 an explicit "perl" to run Perl scripts from the PrepareRelease script
1487 because this is reportedly needed in Windows.
1489 26. If study data was being save in a file and studying had not found a set of
1490 "starts with" bytes for the pattern, the data written to the file (though
1491 never used) was taken from uninitialized memory and so caused valgrind to
1492 complain.
1494 27. Updated RunTest.bat as provided by Sheri Pierce.
1496 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1498 29. Computation of memory usage for the table of capturing group names was
1499 giving an unnecessarily large value.
1502 Version 8.20 21-Oct-2011
1503 ------------------------
1505 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1506 a POSIX class. After further experiments with Perl, which convinced me that
1507 Perl has bugs and confusions, a closing square bracket is no longer allowed
1508 in a POSIX name. This bug also affected patterns with classes that started
1509 with full stops.
1511 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1512 captured substring, but while checking the failing first alternative,
1513 substring 1 is temporarily captured. If the output vector supplied to
1514 pcre_exec() was not big enough for this capture, the yield of the function
1515 was still zero ("insufficient space for captured substrings"). This cannot
1516 be totally fixed without adding another stack variable, which seems a lot
1517 of expense for a edge case. However, I have improved the situation in cases
1518 such as /(a)(b)x|abc/ matched against "abc", where the return code
1519 indicates that fewer than the maximum number of slots in the ovector have
1520 been set.
1522 3. Related to (2) above: when there are more back references in a pattern than
1523 slots in the output vector, pcre_exec() uses temporary memory during
1524 matching, and copies in the captures as far as possible afterwards. It was
1525 using the entire output vector, but this conflicts with the specification
1526 that only 2/3 is used for passing back captured substrings. Now it uses
1527 only the first 2/3, for compatibility. This is, of course, another edge
1528 case.
1530 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1531 main code base, and can be used by building with --enable-jit. When this is
1532 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1533 runtime --no-jit option is given.
1535 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1536 ovector, the return from the function was zero, implying that there were
1537 other matches that did not fit. The correct "exactly full" value is now
1538 returned.
1540 6. If a subpattern that was called recursively or as a subroutine contained
1541 (*PRUNE) or any other control that caused it to give a non-standard return,
1542 invalid errors such as "Error -26 (nested recursion at the same subject
1543 position)" or even infinite loops could occur.
1545 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1546 computing the minimum length on reaching *ACCEPT, and so ended up with the
1547 wrong value of 1 rather than 0. Further investigation indicates that
1548 computing a minimum subject length in the presence of *ACCEPT is difficult
1549 (think back references, subroutine calls), and so I have changed the code
1550 so that no minimum is registered for a pattern that contains *ACCEPT.
1552 8. If (*THEN) was present in the first (true) branch of a conditional group,
1553 it was not handled as intended. [But see 16 below.]
1555 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1556 Sheri Pierce.
1558 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1559 the first byte in a match must be "a".
1561 11. Change 17 for 8.13 increased the recursion depth for patterns like
1562 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1563 pattern contains any instances of (*THEN). If it does not, the old
1564 optimizations are restored. It would be nice to do this on a per-group
1565 basis, but at the moment that is not feasible.
1567 12. In some environments, the output of pcretest -C is CRLF terminated. This
1568 broke RunTest's code that checks for the link size. A single white space
1569 character after the value is now allowed for.
1571 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1572 For "fr", it uses the Windows-specific input and output files.
1574 14. If (*THEN) appeared in a group that was called recursively or as a
1575 subroutine, it did not work as intended. [But see next item.]
1577 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1578 pattern fragments (but not containing any | characters). If A and B are
1579 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1580 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1581 D. In other words, Perl considers parentheses that do not contain any |
1582 characters to be part of a surrounding alternative, whereas PCRE was
1583 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1584 differently. PCRE now behaves in the same way as Perl, except in the case
1585 of subroutine/recursion calls such as (?1) which have in any case always
1586 been different (but PCRE had them first :-).
1588 16. Related to 15 above: Perl does not treat the | in a conditional group as
1589 creating alternatives. Such a group is treated in the same way as an
1590 ordinary group without any | characters when processing (*THEN). PCRE has
1591 been changed to match Perl's behaviour.
1593 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1594 RunGrepTest script failed.
1596 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1597 inevitable for groups that contain captures, but it can lead to a lot of
1598 stack use in large patterns. The old behaviour has been restored for atomic
1599 groups that do not contain any capturing parentheses.
1601 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1602 suppress the check for a minimum subject length at run time. (If it was
1603 given to pcre_exec() or pcre_dfa_exec() it did work.)
1605 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1606 fail to work when decoding hex characters in data strings in EBCDIC
1607 environments.
1609 21. It appears that in at least one Mac OS environment, the isxdigit() function
1610 is implemented as a macro that evaluates to its argument more than once,
1611 contravening the C 90 Standard (I haven't checked a later standard). There
1612 was an instance in pcretest which caused it to go wrong when processing
1613 \x{...} escapes in subject strings. The has been rewritten to avoid using
1614 things like p++ in the argument of isxdigit().
1617 Version 8.13 16-Aug-2011
1618 ------------------------
1620 1. The Unicode data tables have been updated to Unicode 6.0.0.
1622 2. Two minor typos in pcre_internal.h have been fixed.
1624 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1625 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1626 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1628 4. There were a number of related bugs in the code for matching backrefences
1629 caselessly in UTF-8 mode when codes for the characters concerned were
1630 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1631 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1632 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1633 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1634 2-byte code at the end of the subject (it thought there wasn't enough data
1635 left).
1637 5. Comprehensive information about what went wrong is now returned by
1638 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1639 as the output vector has at least 2 elements. The offset of the start of
1640 the failing character and a reason code are placed in the vector.
1642 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1643 now returned is for the first byte of the failing character, instead of the
1644 last byte inspected. This is an incompatible change, but I hope it is small
1645 enough not to be a problem. It makes the returned offset consistent with
1646 pcre_exec() and pcre_dfa_exec().
1648 7. pcretest now gives a text phrase as well as the error number when
1649 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1650 failure, the offset and reason code are output.
1652 8. When \R was used with a maximizing quantifier it failed to skip backwards
1653 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1654 back over a single character (\n). This seems wrong (because it treated the
1655 two characters as a single entity when going forwards), conflicts with the
1656 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1657 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1658 has been changed.
1660 9. Some internal refactoring has changed the processing so that the handling
1661 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1662 time (the PCRE_DOTALL option was changed this way some time ago: version
1663 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1664 which was always a bit of a fudge. It also means that there is one less
1665 argument for the match() function, which reduces its stack requirements
1666 slightly. This change also fixes an incompatibility with Perl: the pattern
1667 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1669 10. More internal refactoring has drastically reduced the number of recursive
1670 calls to match() for possessively repeated groups such as (abc)++ when
1671 using pcre_exec().
1673 11. While implementing 10, a number of bugs in the handling of groups were
1674 discovered and fixed:
1676 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1677 (a|)*(?1) gave a compile-time internal error.
1678 ((a|)+)+ did not notice that the outer group could match an empty string.
1679 (^a|^)+ was not marked as anchored.
1680 (.*a|.*)+ was not marked as matching at start or after a newline.
1682 12. Yet more internal refactoring has removed another argument from the match()
1683 function. Special calls to this function are now indicated by setting a
1684 value in a variable in the "match data" data block.
1686 13. Be more explicit in pcre_study() instead of relying on "default" for
1687 opcodes that mean there is no starting character; this means that when new
1688 ones are added and accidentally left out of pcre_study(), testing should
1689 pick them up.
1691 14. The -s option of pcretest has been documented for ages as being an old
1692 synonym of -m (show memory usage). I have changed it to mean "force study
1693 for every regex", that is, assume /S for every regex. This is similar to -i
1694 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1695 using it. It makes it easier to run collections of tests with and without
1696 study enabled, and thereby test pcre_study() more easily. All the standard
1697 tests are now run with and without -s (but some patterns can be marked as
1698 "never study" - see 20 below).
1700 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1701 restoration of the capturing data to the outer values was not happening
1702 correctly.
1704 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1705 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1706 pattern had matched an empty string, and so incorrectly returned a no
1707 match.
1709 17. There was optimizing code for the last branch of non-capturing parentheses,
1710 and also for the obeyed branch of a conditional subexpression, which used
1711 tail recursion to cut down on stack usage. Unfortunately, now that there is
1712 the possibility of (*THEN) occurring in these branches, tail recursion is
1713 no longer possible because the return has to be checked for (*THEN). These
1714 two optimizations have therefore been removed. [But see 8.20/11 above.]
1716 18. If a pattern containing \R was studied, it was assumed that \R always
1717 matched two bytes, thus causing the minimum subject length to be
1718 incorrectly computed because \R can also match just one byte.
1720 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1721 was incorrectly computed.
1723 20. If /S is present twice on a test pattern in pcretest input, it now
1724 *disables* studying, thereby overriding the use of -s on the command line
1725 (see 14 above). This is necessary for one or two tests to keep the output
1726 identical in both cases.
1728 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1729 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1731 22. When an atomic group that contained a capturing parenthesis was
1732 successfully matched, but the branch in which it appeared failed, the
1733 capturing was not being forgotten if a higher numbered group was later
1734 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1735 group 1 to "a", when in fact it should be unset. This applied to multi-
1736 branched capturing and non-capturing groups, repeated or not, and also to
1737 positive assertions (capturing in negative assertions does not happen
1738 in PCRE) and also to nested atomic groups.
1740 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1741 subject after a captured substring, to make it easier to tell which of a
1742 number of identical substrings has been captured.
1744 24. The way atomic groups are processed by pcre_exec() has been changed so that
1745 if they are repeated, backtracking one repetition now resets captured
1746 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1747 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1748 "aaa". Previously, it would have been "a". As part of this code
1749 refactoring, the way recursive calls are handled has also been changed.
1751 25. If an assertion condition captured any substrings, they were not passed
1752 back unless some other capturing happened later. For example, if
1753 (?(?=(a))a) was matched against "a", no capturing was returned.
1755 26. When studying a pattern that contained subroutine calls or assertions,
1756 the code for finding the minimum length of a possible match was handling
1757 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1758 group 1 called group 2 while simultaneously a separate group 2 called group
1759 1). A stack overflow occurred in this case. I have fixed this by limiting
1760 the recursion depth to 10.
1762 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1763 Fortmann. This supports explicit test numbers on the command line, and has
1764 argument validation and error reporting.
1766 28. An instance of \X with an unlimited repeat could fail if at any point the
1767 first character it looked at was a mark character.
1769 29. Some minor code refactoring concerning Unicode properties and scripts
1770 should reduce the stack requirement of match() slightly.
1772 30. Added the '=' option to pcretest to check the setting of unused capturing
1773 slots at the end of the pattern, which are documented as being -1, but are
1774 not included in the return count.
1776 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1777 compiled something random. Now it gives a compile-time error (as does
1778 Perl).
1780 32. A *MARK encountered during the processing of a positive assertion is now
1781 recorded and passed back (compatible with Perl).
1783 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1784 had alternative anchored branches, the search for a second match in a line
1785 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1786 matched the line "0102" twice. The same bug affected patterns that started
1787 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1788 twice.
1790 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1791 does, and because of capturing effects, quantifying parenthesized
1792 assertions may at times be useful. Quantifiers are now allowed for
1793 parenthesized assertions.
1795 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1797 36. \g was being checked for fancy things in a character class, when it should
1798 just be a literal "g".
1800 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1801 appearance of a nested POSIX class supersedes an apparent external class.
1802 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1803 unescaped square brackets may also appear as part of class names. For
1804 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1805 more like Perl. (But see 8.20/1 above.)
1807 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1808 was because it thought it was \N{name}, which is not supported).
1810 39. Add minix to OS list not supporting the -S option in pcretest.
1812 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1813 cannot analyze patterns in sufficient detail to catch mutual recursions
1814 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1815 subgroup is called recursively as a subpattern for a second time at the
1816 same position in the subject string. In previous releases this might have
1817 been caught by the recursion limit, or it might have run out of stack.
1819 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1820 happen only once. PCRE was, however incorrectly giving a compile time error
1821 "recursive call could loop indefinitely" because it cannot analyze the
1822 pattern in sufficient detail. The compile time test no longer happens when
1823 PCRE is compiling a conditional subpattern, but actual runaway loops are
1824 now caught at runtime (see 40 above).
1826 42. It seems that Perl allows any characters other than a closing parenthesis
1827 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1828 has been changed to be the same.
1830 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1831 as not to get warnings when autogen.sh is called. Also changed
1832 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1834 44. To help people who use pcregrep to scan files containing exceedingly long
1835 lines, the following changes have been made:
1837 (a) The default value of the buffer size parameter has been increased from
1838 8K to 20K. (The actual buffer used is three times this size.)
1840 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1841 PCRE is built.
1843 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1844 to be set at run time.
1846 (d) Numerical values in pcregrep options can be followed by K or M, for
1847 example --buffer-size=50K.
1849 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1850 given and the return code is set to 2.
1852 45. Add a pointer to the latest mark to the callout data block.
1854 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1855 partial match of an empty string instead of no match. This was specific to
1856 the use of ".".
1858 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1859 complete match instead of a partial match. This bug was dependent on both
1860 the PCRE_UTF8 and PCRE_DOTALL options being set.
1862 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1863 starting byte set, because \b was not being ignored.
1866 Version 8.12 15-Jan-2011
1867 ------------------------
1869 1. Fixed some typos in the markup of the man pages, and wrote a script that
1870 checks for such things as part of the documentation building process.
1872 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1873 --match-limit and --recursion-limit options (added for 8.11). In
1874 particular, this made one of the standard tests fail. (The integer value
1875 went into the wrong half of a long int.)
1877 3. If the --colour option was given to pcregrep with -v (invert match), it
1878 did strange things, either producing crazy output, or crashing. It should,
1879 of course, ignore a request for colour when reporting lines that do not
1880 match.
1882 4. Another pcregrep bug caused similar problems if --colour was specified with
1883 -M (multiline) and the pattern match finished with a line ending.
1885 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1886 matched in multiline mode, the following line was shown as part of the
1887 match. This seems wrong, so I have changed it.
1889 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1890 the check for further matches in the same line (so they could be coloured)
1891 to overrun the end of the current line. If another match was found, it was
1892 incorrectly shown (and then shown again when found in the next line).
1894 7. If pcregrep was compiled under Windows, there was a reference to the
1895 function pcregrep_exit() before it was defined. I am assuming this was
1896 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1897 reported by a user. I've moved the definition above the reference.
1900 Version 8.11 10-Dec-2010
1901 ------------------------
1903 1. (*THEN) was not working properly if there were untried alternatives prior
1904 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1905 backtracked to try for "b" instead of moving to the next alternative branch
1906 at the same level (in this case, to look for "c"). The Perl documentation
1907 is clear that when (*THEN) is backtracked onto, it goes to the "next
1908 alternative in the innermost enclosing group".
1910 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1911 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1912 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1913 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1914 (*THEN).
1916 3. If \s appeared in a character class, it removed the VT character from
1917 the class, even if it had been included by some previous item, for example
1918 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1919 of \s, but is part of the POSIX "space" class.)
1921 4. A partial match never returns an empty string (because you can always
1922 match an empty string at the end of the subject); however the checking for
1923 an empty string was starting at the "start of match" point. This has been
1924 changed to the "earliest inspected character" point, because the returned
1925 data for a partial match starts at this character. This means that, for
1926 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1927 (previously it gave "no match").
1929 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1930 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1931 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1932 has an implication that the given string is incomplete (because a partial
1933 match is preferred over a full match). For this reason, these items now
1934 give a partial match in this situation. [Aside: previously, the one case
1935 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1936 match rather than a full match, which was wrong by the old rules, but is
1937 now correct.]
1939 6. There was a bug in the handling of #-introduced comments, recognized when
1940 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1941 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1942 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1943 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1944 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1945 places in pcre_compile().
1947 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1948 comments when looking ahead for named forward references to subpatterns,
1949 the only newline sequence it recognized was NL. It now handles newlines
1950 according to the set newline convention.
1952 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1953 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1954 cater for a lack of strerror(). These oversights have been fixed.
1956 9. Added --match-limit and --recursion-limit to pcregrep.
1958 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1960 11. When the -o option was used, pcregrep was setting a return code of 1, even
1961 when matches were found, and --line-buffered was not being honoured.
1963 12. Added an optional parentheses number to the -o and --only-matching options
1964 of pcregrep.
1966 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1967 can match an empty string. The code to do it in pcretest and pcredemo
1968 needed fixing:
1970 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1971 only one byte after an empty string match just before CRLF (this case
1972 just got forgotten; "any" and "anycrlf" were OK).
1974 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1975 mode when an empty string match preceded an ASCII character followed by
1976 a non-ASCII character. (The code for advancing by one character rather
1977 than one byte was nonsense.)
1979 (c) The pcredemo.c sample program did not have any code at all to handle
1980 the cases when CRLF is a valid newline sequence.
1982 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1983 as a starting offset was within the subject string. There is now a new
1984 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1985 negative or greater than the length of the string. In order to test this,
1986 pcretest is extended to allow the setting of negative starting offsets.
1988 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1989 starting offset points to the beginning of a UTF-8 character was
1990 unnecessarily clumsy. I tidied it up.
1992 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1993 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1995 17. Nobody had reported that the --include_dir option, which was added in
1996 release 7.7 should have been called --include-dir (hyphen, not underscore)
1997 for compatibility with GNU grep. I have changed it to --include-dir, but
1998 left --include_dir as an undocumented synonym, and the same for
1999 --exclude-dir, though that is not available in GNU grep, at least as of
2000 release 2.5.4.
2002 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
2003 characters from a string of bytes) have been redefined so as not to use
2004 loops, in order to improve performance in some environments. At the same
2005 time, I abstracted some of the common code into auxiliary macros to save
2006 repetition (this should not affect the compiled code).
2008 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
2009 compile-time error is now given if \c is not followed by an ASCII
2010 character, that is, a byte less than 128. (In EBCDIC mode, the code is
2011 different, and any byte value is allowed.)
2013 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
2014 START_OPTIMIZE option, which is now allowed at compile time - but just
2015 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
2016 to pcregrep and other applications that have no direct access to PCRE
2017 options. The new /Y option in pcretest sets this option when calling
2018 pcre_compile().
2020 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
2021 back references. Groups containing recursive back references were forced to
2022 be atomic by that change, but in the case of named groups, the amount of
2023 memory required was incorrectly computed, leading to "Failed: internal
2024 error: code overflow". This has been fixed.
2026 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
2027 pcretest.c, to avoid build problems in some Borland environments.
2030 Version 8.10 25-Jun-2010
2031 ------------------------
2033 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2034 THEN.
2036 2. (*ACCEPT) was not working when inside an atomic group.
2038 3. Inside a character class, \B is treated as a literal by default, but
2039 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2040 causes the error). The code is unchanged, but I tidied the documentation.
2042 4. Inside a character class, PCRE always treated \R and \X as literals,
2043 whereas Perl faults them if its -w option is set. I have changed PCRE so
2044 that it faults them when PCRE_EXTRA is set.
2046 5. Added support for \N, which always matches any character other than
2047 newline. (It is the same as "." when PCRE_DOTALL is not set.)
2049 6. When compiling pcregrep with newer versions of gcc which may have
2050 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2051 declared with attribute warn_unused_result" were given. Just casting the
2052 result to (void) does not stop the warnings; a more elaborate fudge is
2053 needed. I've used a macro to implement this.
2055 7. Minor change to pcretest.c to avoid a compiler warning.
2057 8. Added four artifical Unicode properties to help with an option to make
2058 \s etc use properties (see next item). The new properties are: Xan
2059 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2061 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2062 use Unicode properties. (*UCP) at the start of a pattern can be used to set
2063 this option. Modified pcretest to add /W to test this facility. Added
2064 REG_UCP to make it available via the POSIX interface.
2066 10. Added --line-buffered to pcregrep.
2068 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2069 studied, and the match started with a letter with a code point greater than
2070 127 whose first byte was different to the first byte of the other case of
2071 the letter, the other case of this starting letter was not recognized
2072 (#976).
2074 12. If a pattern that was studied started with a repeated Unicode property
2075 test, for example, \p{Nd}+, there was the theoretical possibility of
2076 setting up an incorrect bitmap of starting bytes, but fortunately it could
2077 not have actually happened in practice until change 8 above was made (it
2078 added property types that matched character-matching opcodes).
2080 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2081 possible starting bytes for non-anchored patterns.
2083 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2084 \R, and also a number of cases that involve Unicode properties, both
2085 explicit and implicit when PCRE_UCP is set.
2087 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2088 input, it could crash or give wrong results if characters with values
2089 greater than 0xc0 were present in the subject string. (Detail: it assumed
2090 UTF-8 input when processing these items.)
2092 16. Added a lot of (int) casts to avoid compiler warnings in systems where
2093 size_t is 64-bit (#991).
2095 17. Added a check for running out of memory when PCRE is compiled with
2096 --disable-stack-for-recursion (#990).
2098 18. If the last data line in a file for pcretest does not have a newline on
2099 the end, a newline was missing in the output.
2101 19. The default pcre_chartables.c file recognizes only ASCII characters (values
2102 less than 128) in its various bitmaps. However, there is a facility for
2103 generating tables according to the current locale when PCRE is compiled. It
2104 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2105 characters, are recognized by isspace() and therefore were getting set in
2106 these tables, and indeed these tables seem to approximate to ISO 8859. This
2107 caused a problem in UTF-8 mode when pcre_study() was used to create a list
2108 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2109 which of course cannot start UTF-8 characters. I have changed the code so
2110 that only real ASCII characters (less than 128) and the correct starting
2111 bytes for UTF-8 encodings are set for characters greater than 127 when in
2112 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2113 altogether.)
2115 20. Added the /T option to pcretest so as to be able to run tests with non-
2116 standard character tables, thus making it possible to include the tests
2117 used for 19 above in the standard set of tests.
2119 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2120 reference to a subpattern the other side of a comment that contains an
2121 opening parenthesis caused either an internal compiling error, or a
2122 reference to the wrong subpattern.
2125 Version 8.02 19-Mar-2010
2126 ------------------------
2128 1. The Unicode data tables have been updated to Unicode 5.2.0.
2130 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2131 configured.
2133 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2134 original author of that file, following a query about its status.
2136 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2137 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2139 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2140 quantifier applied to a forward-referencing subroutine call, could compile
2141 incorrect code or give the error "internal error: previously-checked
2142 referenced subpattern not found".
2144 6. Both MS Visual Studio and Symbian OS have problems with initializing
2145 variables to point to external functions. For these systems, therefore,
2146 pcre_malloc etc. are now initialized to local functions that call the
2147 relevant global functions.
2149 7. There were two entries missing in the vectors called coptable and poptable
2150 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2151 I've fixed the data, and added a kludgy way of testing at compile time that
2152 the lengths are correct (equal to the number of opcodes).
2154 8. Following on from 7, I added a similar kludge to check the length of the
2155 eint vector in pcreposix.c.
2157 9. Error texts for pcre_compile() are held as one long string to avoid too
2158 much relocation at load time. To find a text, the string is searched,
2159 counting zeros. There was no check for running off the end of the string,
2160 which could happen if a new error number was added without updating the
2161 string.
2163 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2165 11. \K was not working if it appeared in an atomic group or in a group that
2166 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2167 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2168 the assertion is positive, but not if it is negative.
2170 12. Change 11 fortuitously reduced the size of the stack frame used in the
2171 "match()" function of pcre_exec.c by one pointer. Forthcoming
2172 implementation of support for (*MARK) will need an extra pointer on the
2173 stack; I have reserved it now, so that the stack frame size does not
2174 decrease.
2176 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2177 item in branch that calls a recursion is a subroutine call - as in the
2178 second branch in the above example - was incorrectly given the compile-
2179 time error "recursive call could loop indefinitely" because pcre_compile()
2180 was not correctly checking the subroutine for matching a non-empty string.
2182 14. The checks for overrunning compiling workspace could trigger after an
2183 overrun had occurred. This is a "should never occur" error, but it can be
2184 triggered by pathological patterns such as hundreds of nested parentheses.
2185 The checks now trigger 100 bytes before the end of the workspace.
2187 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2190 Version 8.01 19-Jan-2010
2191 ------------------------
2193 1. If a pattern contained a conditional subpattern with only one branch (in
2194 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2195 computed the wrong minimum data length (which is of course zero for such
2196 subpatterns). This could cause incorrect "no match" results.
2198 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2199 the pattern is reset in the first branch, pcre_compile() failed with
2200 "internal error: code overflow at offset...". This happened only when
2201 the reset was to the original external option setting. (An optimization
2202 abstracts leading options settings into an external setting, which was the
2203 cause of this.)
2205 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2206 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2207 assertion pattern did not match (meaning that the assertion was true), it
2208 was incorrectly treated as false if the SKIP had been reached during the
2209 matching. This also applied to assertions used as conditions.
2211 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2212 assertion subpattern, including such a pattern used as a condition,
2213 unpredictable results occurred, instead of the error return
2216 5. The C++ GlobalReplace function was not working like Perl for the special
2217 situation when an empty string is matched. It now does the fancy magic
2218 stuff that is necessary.
2220 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2221 removed. (These were left over from very, very early versions of PCRE.)
2223 7. Some cosmetic changes to the code to make life easier when compiling it
2224 as part of something else:
2226 (a) Change DEBUG to PCRE_DEBUG.
2228 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2229 called "current" as "current_branch", to prevent a collision with the
2230 Linux macro when compiled as a kernel module.
2232 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2233 prevent a collision with the Linux macro when compiled as a kernel
2234 module.
2236 8. In pcre_compile() there are some checks for integer overflows that used to
2237 cast potentially large values to (double). This has been changed to that
2238 when building, a check for int64_t is made, and if it is found, it is used
2239 instead, thus avoiding the use of floating point arithmetic. (There is no
2240 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2241 double.
2243 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2244 2005 (difference between two addresses compared to an unsigned value).
2246 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2247 custom one, because of the following reported problem in Windows:
2249 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2250 under Win32.
2251 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2252 therefore missing the function definition.
2253 - The compiler thus generates a "C" signature for the test function.
2254 - The linker fails to find the "C" function.
2255 - PCRE fails to configure if asked to do so against libbz2.
2257 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2258 messages were output:
2260 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2261 rerunning libtoolize, to keep the correct libtool macros in-tree.
2262 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2264 I have done both of these things.
2266 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2267 most of the time, it *can* run out if it is given a pattern that contains a
2268 runaway infinite recursion. I updated the discussion in the pcrestack man
2269 page.
2271 13. Now that we have gone to the x.xx style of version numbers, the minor
2272 version may start with zero. Using 08 or 09 is a bad idea because users
2273 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2274 interpreted as invalid octal numbers. I've updated the previous comment in
2275 configure.ac, and also added a check that gives an error if 08 or 09 are
2276 used.
2278 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2279 causing partial matching to fail when the end of the subject matched \W
2280 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2282 15. There were some discrepancies between the declarations in pcre_internal.h
2283 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2284 their definitions. The declarations used "const uschar *" and the
2285 definitions used USPTR. Even though USPTR is normally defined as "const
2286 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2287 reported that: "This difference in casting confuses some C++ compilers, for
2288 example, SunCC recognizes above declarations as different functions and
2289 generates broken code for hbpcre." I have changed the declarations to use
2290 USPTR.
2292 16. GNU libtool is named differently on some systems. The autogen.sh script now
2293 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2294 (FreeBSD).
2296 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2297 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2298 comment: "Figure out how to create a longlong from a string: strtoll and
2299 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2300 instance, but it only takes 2 args instead of 3!"
2302 18. A subtle bug concerned with back references has been fixed by a change of
2303 specification, with a corresponding code fix. A pattern such as
2304 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2305 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2306 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2307 same bug. Such groups have to be quantified to be useful, or contained
2308 inside another quantified group. (If there's no repetition, the reference
2309 can never match.) The problem arises because, having left the group and
2310 moved on to the rest of the pattern, a later failure that backtracks into
2311 the group uses the captured value from the final iteration of the group
2312 rather than the correct earlier one. I have fixed this in PCRE by forcing
2313 any group that contains a reference to itself to be an atomic group; that
2314 is, there cannot be any backtracking into it once it has completed. This is
2315 similar to recursive and subroutine calls.
2318 Version 8.00 19-Oct-09
2319 ----------------------
2321 1. The table for translating pcre_compile() error codes into POSIX error codes
2322 was out-of-date, and there was no check on the pcre_compile() error code
2323 being within the table. This could lead to an OK return being given in
2324 error.
2326 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2327 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2328 in a Windows environment.
2330 3. The pcregrep --count option prints the count for each file even when it is
2331 zero, as does GNU grep. However, pcregrep was also printing all files when
2332 --files-with-matches was added. Now, when both options are given, it prints
2333 counts only for those files that have at least one match. (GNU grep just
2334 prints the file name in this circumstance, but including the count seems
2335 more useful - otherwise, why use --count?) Also ensured that the
2336 combination -clh just lists non-zero counts, with no names.
2338 4. The long form of the pcregrep -F option was incorrectly implemented as
2339 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2340 but it seems right to fix it, and I didn't think it was worth preserving
2341 the old behaviour.
2343 5. The command line items --regex=pattern and --regexp=pattern were not
2344 recognized by pcregrep, which required --regex pattern or --regexp pattern
2345 (with a space rather than an '='). The man page documented the '=' forms,
2346 which are compatible with GNU grep; these now work.
2348 6. No libpcreposix.pc file was created for pkg-config; there was just
2349 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2351 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2352 when UCP support is not needed, by modifying the Python script that
2353 generates it from Unicode data files. This should not matter if the module
2354 is correctly used as a library, but I received one complaint about 50K of
2355 unwanted data. My guess is that the person linked everything into his
2356 program rather than using a library. Anyway, it does no harm.
2358 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2359 was a minimum greater than 1 for a wide character in a possessive
2360 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2361 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2362 character. Chaos in the form of incorrect output or a compiling loop could
2363 result.
2365 9. The restrictions on what a pattern can contain when partial matching is
2366 requested for pcre_exec() have been removed. All patterns can now be
2367 partially matched by this function. In addition, if there are at least two
2368 slots in the offset vector, the offset of the earliest inspected character
2369 for the match and the offset of the end of the subject are set in them when
2370 PCRE_ERROR_PARTIAL is returned.
2372 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2373 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2374 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2375 and may be more useful for multi-segment matching.
2377 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2378 used to be given if ever the end of the subject was reached; now it is
2379 given only if matching could not proceed because another character was
2380 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2381 string "Z", which now yields "no match" instead of "partial match". In the
2382 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2383 final character ended with (*FAIL).
2385 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2386 if the pattern had a "must contain" character that was already found in the
2387 earlier partial match, unless partial matching was again requested. For
2388 example, with the pattern /dog.(body)?/, the "must contain" character is
2389 "g". If the first part-match was for the string "dog", restarting with
2390 "sbody" failed. This bug has been fixed.
2392 13. The string returned by pcre_dfa_exec() after a partial match has been
2393 changed so that it starts at the first inspected character rather than the
2394 first character of the match. This makes a difference only if the pattern
2395 starts with a lookbehind assertion or \b or \B (\K is not supported by
2396 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2397 matching functions compatible, and I think it's the right thing to do.
2399 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2400 so that the demonstration program is easily available in environments where
2401 PCRE has not been installed from source.
2403 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2404 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2405 library.
2407 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2408 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2409 is not the first non-POSIX option to be added. Clearly some people find
2410 these options useful.
2412 17. If a caller to the POSIX matching function regexec() passes a non-zero
2413 value for nmatch with a NULL value for pmatch, the value of
2414 nmatch is forced to zero.
2416 18. RunGrepTest did not have a test for the availability of the -u option of
2417 the diff command, as RunTest does. It now checks in the same way as
2418 RunTest, and also checks for the -b option.
2420 19. If an odd number of negated classes containing just a single character
2421 interposed, within parentheses, between a forward reference to a named
2422 subpattern and the definition of the subpattern, compilation crashed with
2423 an internal error, complaining that it could not find the referenced
2424 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2425 [The bug was that it was starting one character too far in when skipping
2426 over the character class, thus treating the ] as data rather than
2427 terminating the class. This meant it could skip too much.]
2429 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2430 /g option in pcretest when the pattern contains \K, which makes it possible
2431 to have an empty string match not at the start, even when the pattern is
2432 anchored. Updated pcretest and pcredemo to use this option.
2434 21. If the maximum number of capturing subpatterns in a recursion was greater
2435 than the maximum at the outer level, the higher number was returned, but
2436 with unset values at the outer level. The correct (outer level) value is
2437 now given.
2439 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2440 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2441 make it do so. The string so far is captured, making this feature
2442 compatible with Perl.
2444 23. The tests have been re-organized, adding tests 11 and 12, to make it
2445 possible to check the Perl 5.10 features against Perl 5.10.
2447 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2448 pattern matches a fixed length string. PCRE did not allow this; now it
2449 does. Neither allows recursion.
2451 25. I finally figured out how to implement a request to provide the minimum
2452 length of subject string that was needed in order to match a given pattern.
2453 (It was back references and recursion that I had previously got hung up
2454 on.) This code has now been added to pcre_study(); it finds a lower bound
2455 to the length of subject needed. It is not necessarily the greatest lower
2456 bound, but using it to avoid searching strings that are too short does give
2457 some useful speed-ups. The value is available to calling programs via
2458 pcre_fullinfo().
2460 26. While implementing 25, I discovered to my embarrassment that pcretest had
2461 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2462 study optimizations had never been tested with that matching function.
2463 Oops. What is worse, even when it was passed study data, there was a bug in
2464 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2465 were also very few tests of studied patterns with pcre_dfa_exec().
2467 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2468 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2469 on the other side of the coin, they are no longer allowed to have different
2470 names, because these cannot be distinguished in PCRE, and this has caused
2471 confusion. (This is a difference from Perl.)
2473 28. When duplicate subpattern names are present (necessarily with different
2474 numbers, as required by 27 above), and a test is made by name in a
2475 conditional pattern, either for a subpattern having been matched, or for
2476 recursion in such a pattern, all the associated numbered subpatterns are
2477 tested, and the overall condition is true if the condition is true for any
2478 one of them. This is the way Perl works, and is also more like the way
2479 testing by number works.
2482 Version 7.9 11-Apr-09
2483 ---------------------
2485 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2486 (pcretest), all targets were linked against these libraries. This included
2487 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2488 libraries. This caused unwanted dependencies to be created. This problem
2489 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2490 pcretest is linked with readline.
2492 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2493 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2494 moved outside it again, because FALSE and TRUE are already defined in AIX,
2495 but BOOL is not.
2497 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2498 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2500 4. The pcregrep documentation said spaces were inserted as well as colons (or
2501 hyphens) following file names and line numbers when outputting matching
2502 lines. This is not true; no spaces are inserted. I have also clarified the
2503 wording for the --colour (or --color) option.
2505 5. In pcregrep, when --colour was used with -o, the list of matching strings
2506 was not coloured; this is different to GNU grep, so I have changed it to be
2507 the same.
2509 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2510 each matching line was coloured. Now it goes on to look for further matches
2511 of any of the test patterns, which is the same behaviour as GNU grep.
2513 7. A pattern that could match an empty string could cause pcregrep to loop; it
2514 doesn't make sense to accept an empty string match in pcregrep, so I have
2515 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2516 seems to be how GNU grep behaves. [But see later change 40 for release
2517 8.33.]
2519 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2520 start or after a newline", because the conditional assertion was not being
2521 correctly handled. The rule now is that both the assertion and what follows
2522 in the first alternative must satisfy the test.
2524 9. If auto-callout was enabled in a pattern with a conditional group whose
2525 condition was an assertion, PCRE could crash during matching, both with
2526 pcre_exec() and pcre_dfa_exec().
2528 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2529 used for matching.
2531 11. Unicode property support in character classes was not working for
2532 characters (bytes) greater than 127 when not in UTF-8 mode.
2534 12. Added the -M command line option to pcretest.
2536 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2538 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2540 16. Added comments and documentation about mis-use of no_arg in the C++
2541 wrapper.
2543 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2544 from Martin Jerabek that uses macro names for all relevant character and
2545 string constants.
2547 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2548 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2549 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2550 these, but not everybody uses configure.
2552 19. A conditional group that had only one branch was not being correctly
2553 recognized as an item that could match an empty string. This meant that an
2554 enclosing group might also not be so recognized, causing infinite looping
2555 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2556 with the subject "ab", where knowledge that the repeated group can match
2557 nothing is needed in order to break the loop.
2559 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2560 exec(), but without supplying a callout function, matching went wrong.
2562 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2563 leak if the size of the offset vector was greater than 30. When the vector
2564 is smaller, the saved offsets during recursion go onto a local stack
2565 vector, but for larger vectors malloc() is used. It was failing to free
2566 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2567 error, in fact).
2569 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2570 heapframe that is used only when UTF-8 support is enabled. This caused no
2571 problem, but was untidy.
2573 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2574 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2575 included within another project.
2577 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2578 slightly modified by me:
2580 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2581 not building pcregrep.
2583 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2584 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2586 25. Forward references, both numeric and by name, in patterns that made use of
2587 duplicate group numbers, could behave incorrectly or give incorrect errors,
2588 because when scanning forward to find the reference group, PCRE was not
2589 taking into account the duplicate group numbers. A pattern such as
2590 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2592 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2593 the feature of a custom pointer more persuasive (as requested by a user).
2595 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2596 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2597 pre-defined.
2599 28. Added support for (*UTF8) at the start of a pattern.
2601 29. Arrange for flags added by the "release type" setting in CMake to be shown
2602 in the configuration summary.
2605 Version 7.8 05-Sep-08
2606 ---------------------
2608 1. Replaced UCP searching code with optimized version as implemented for Ad
2609 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2610 stage table and inline lookup instead of a function, giving speed ups of 2
2611 to 5 times on some simple patterns that I tested. Permission was given to
2612 distribute the MultiStage2.py script that generates the tables (it's not in
2613 the tarball, but is in the Subversion repository).
2615 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2616 scripts.
2618 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2619 a group with a zero qualifier. The result of the study could be incorrect,
2620 or the function might crash, depending on the pattern.
2622 4. Caseless matching was not working for non-ASCII characters in back
2623 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2624 It now works when Unicode Property Support is available.
2626 5. In pcretest, an escape such as \x{de} in the data was always generating
2627 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2628 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2629 truncation.
2631 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2633 7. Added two (int) casts to pcregrep when printing the difference of two
2634 pointers, in case they are 64-bit values.
2636 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2637 test 2 if it fails.
2639 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2640 and a #define of that name to empty if it is not externally set. This is to
2641 allow users of MSVC to set it if necessary.
2643 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2644 the convenience functions in the pcre_get.c source file.
2646 11. An option change at the start of a pattern that had top-level alternatives
2647 could cause overwriting and/or a crash. This command provoked a crash in
2648 some environments:
2650 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2652 This potential security problem was recorded as CVE-2008-2371.
2654 12. For a pattern where the match had to start at the beginning or immediately
2655 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2656 pcre_dfa_exec() could read past the end of the passed subject if there was
2657 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2658 pcretest so that it places the subject at the end of its malloc-ed buffer.
2660 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2661 exec() might read past the end of the data buffer in UTF-8 mode.
2663 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2664 the data contained the byte 0x85 as part of a UTF-8 character within its
2665 first line. This applied both to normal and DFA matching.
2667 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2668 /^[^d]*?$/8 failed to match "abc".
2670 16. Added a missing copyright notice to pcrecpp_internal.h.
2672 17. Make it more clear in the documentation that values returned from
2673 pcre_exec() in ovector are byte offsets, not character counts.
2675 18. Tidied a few places to stop certain compilers from issuing warnings.
2677 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2678 supplied by Stefan Weber. I made a further small update for 7.8 because
2679 there is a change of source arrangements: the pcre_searchfuncs.c module is
2680 replaced by pcre_ucd.c.
2683 Version 7.7 07-May-08
2684 ---------------------
2686 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2687 a string to a long long, pretend we don't even have a long long." This is
2688 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2690 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2691 pre-7.6 versions, which defined a global no_arg variable instead of putting
2692 it in the RE class. (See also #8 below.)
2694 3. Remove a line of dead code, identified by coverity and reported by Nuno
2695 Lopes.
2697 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2699 (1) The include/exclude patterns were being applied to the whole pathnames
2700 of files, instead of just to the final components.
2702 (2) If there was more than one level of directory, the subdirectories were
2703 skipped unless they satisfied the include/exclude conditions. This is
2704 inconsistent with GNU grep (and could even be seen as contrary to the
2705 pcregrep specification - which I improved to make it absolutely clear).
2706 The action now is always to scan all levels of directory, and just
2707 apply the include/exclude patterns to regular files.
2709 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2710 --exclude_dir in the tests to avoid scanning .svn directories.
2712 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2713 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2714 doesn't support NULs in patterns.
2716 7. Added some missing "const"s to declarations of static tables in
2717 pcre_compile.c and pcre_dfa_exec.c.
2719 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2720 caused by fix #2 above. (Subsequently also a second patch to fix the
2721 first patch. And a third patch - this was a messy problem.)
2723 9. Applied Craig's patch to remove the use of push_back().
2725 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2726 matching function regexec().
2728 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2729 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2730 references. PCRE supports relative numbers with this syntax (I don't think
2731 Oniguruma does).
2733 12. Previously, a group with a zero repeat such as (...){0} was completely
2734 omitted from the compiled regex. However, this means that if the group
2735 was called as a subroutine from elsewhere in the pattern, things went wrong
2736 (an internal error was given). Such groups are now left in the compiled
2737 pattern, with a new opcode that causes them to be skipped at execution
2738 time.
2740 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2741 to the way PCRE behaves:
2743 (a) A lone ] character is dis-allowed (Perl treats it as data).
2745 (b) A back reference to an unmatched subpattern matches an empty string
2746 (Perl fails the current match path).
2748 (c) A data ] in a character class must be notated as \] because if the
2749 first data character in a class is ], it defines an empty class. (In
2750 Perl it is not possible to have an empty class.) The empty class []
2751 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2752 The negative empty class [^] matches any one character, independently
2753 of the DOTALL setting.
2755 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2756 non-existent subpattern following a character class starting with ']' and
2757 containing () gave an internal compiling error instead of "reference to
2758 non-existent subpattern". Fortunately, when the pattern did exist, the
2759 compiled code was correct. (When scanning forwards to check for the
2760 existence of the subpattern, it was treating the data ']' as terminating
2761 the class, so got the count wrong. When actually compiling, the reference
2762 was subsequently set up correctly.)
2764 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2765 it was being rejected as not supported by pcre_dfa_exec(), even though
2766 other assertions are supported. I have made pcre_dfa_exec() support
2767 (*FAIL).
2769 16. The implementation of 13c above involved the invention of a new opcode,
2770 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2771 cannot be changed at match time, I realized I could make a small
2772 improvement to matching performance by compiling OP_ALLANY instead of
2773 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2774 on the OP_ANY path.
2776 17. Compiling pcretest on Windows with readline support failed without the
2777 following two fixes: (1) Make the unistd.h include conditional on
2778 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2780 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2781 ncurses library to be included for pcretest when ReadLine support is
2782 requested, but also to allow for it to be overridden. This patch came from
2783 Daniel Bergström.
2785 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2786 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2787 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2788 spotting this.
2791 Version 7.6 28-Jan-08
2792 ---------------------
2794 1. A character class containing a very large number of characters with
2795 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2796 overflow.
2798 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2799 HAVE_LONG_LONG is not defined.
2801 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2802 bring them up to date and include new features. This patch includes:
2804 - Fixed PH's badly added libz and libbz2 support.
2805 - Fixed a problem with static linking.
2806 - Added pcredemo. [But later removed - see 7 below.]
2807 - Fixed dftables problem and added an option.
2808 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2810 - Added readline support for pcretest.
2811 - Added an listing of the option settings after cmake has run.
2813 4. A user submitted a patch to Makefile that makes it easy to create
2814 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2815 Makefile.am that cause it to include this special target, without
2816 affecting anything else. Note that the same mingw target plus all
2817 the other distribution libraries and programs are now supported
2818 when configuring with CMake (see 6 below) instead of with
2819 Configure/Make.
2821 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2822 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2823 exported in the Windows port". It has not yet been confirmed that the patch
2824 solves the problem, but it does no harm.
2826 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2827 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2828 with CMake, and also correct the comment about stack recursion.
2830 7. Remove the automatic building of pcredemo from the ./configure system and
2831 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2832 of a program that users should build themselves after PCRE is installed, so
2833 building it automatically is not really right. What is more, it gave
2834 trouble in some build environments.
2836 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2839 Version 7.5 10-Jan-08
2840 ---------------------
2842 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2843 values in parens when parsing an RE using the C++ wrapper."
2845 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2846 Characters greater than 255 were excluded from the class instead of being
2847 included.
2849 3. The same bug as (2) above applied to negated POSIX classes such as
2850 [:^space:].
2852 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2853 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2854 I have changed it.
2856 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2857 first named subpattern) and a construct such as (?&a) would reference the
2858 first named subpattern whose name started with "a" (in other words, the
2859 length check was missing). Both these problems are fixed. "Subpattern name
2860 expected" is now given for (?&) (a zero-length name), and this patch also
2861 makes it give the same error for \k'' (previously it complained that that
2862 was a reference to a non-existent subpattern).
2864 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2865 this is right because (?- can be followed by option settings as well as by
2866 digits. I have, however, made the messages clearer.
2868 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2869 than the number used in the conditional) now cause a compile-time error.
2870 This is actually not compatible with Perl, which accepts such patterns, but
2871 treats the conditional as always being FALSE (as PCRE used to), but it
2872 seems to me that giving a diagnostic is better.
2874 8. Change "alphameric" to the more common word "alphanumeric" in comments
2875 and messages.
2877 9. Fix two occurrences of "backslash" in comments that should have been
2878 "backspace".
2880 10. Remove two redundant lines of code that can never be obeyed (their function
2881 was moved elsewhere).
2883 11. The program that makes PCRE's Unicode character property table had a bug
2884 which caused it to generate incorrect table entries for sequences of
2885 characters that have the same character type, but are in different scripts.
2886 It amalgamated them into a single range, with the script of the first of
2887 them. In other words, some characters were in the wrong script. There were
2888 thirteen such cases, affecting characters in the following ranges:
2890 U+002b0 - U+002c1
2891 U+0060c - U+0060d
2892 U+0061e - U+00612
2893 U+0064b - U+0065e
2894 U+0074d - U+0076d
2895 U+01800 - U+01805
2896 U+01d00 - U+01d77
2897 U+01d9b - U+01dbf
2898 U+0200b - U+0200f
2899 U+030fc - U+030fe
2900 U+03260 - U+0327f
2901 U+0fb46 - U+0fbb1
2902 U+10450 - U+1049d
2904 12. The -o option (show only the matching part of a line) for pcregrep was not
2905 compatible with GNU grep in that, if there was more than one match in a
2906 line, it showed only the first of them. It now behaves in the same way as
2907 GNU grep.
2909 13. If the -o and -v options were combined for pcregrep, it printed a blank
2910 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2911 does the same. The return code can be used to tell if there were any
2912 non-matching lines.
2914 14. Added --file-offsets and --line-offsets to pcregrep.
2916 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2917 infinitely looping recursion. The bug was that positive lookaheads were not
2918 being skipped when checking for a possible empty match (negative lookaheads
2919 and both kinds of lookbehind were skipped).
2921 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2922 inclusion of <windows.h> to before rather than after the definition of
2923 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2925 17. Specifying a possessive quantifier with a specific limit for a Unicode
2926 character property caused pcre_compile() to compile bad code, which led at
2927 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2928 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2929 caused the error; without that there was no problem.
2931 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2933 19. Added --enable-pcretest-libreadline.
2935 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2936 RE::GlobalReplace(). As a result, the number of replacements returned was
2937 double what it should be. I removed one of the increments, but Craig sent a
2938 later patch that removed the other one (the right fix) and added unit tests
2939 that check the return values (which was not done before).
2941 21. Several CMake things:
2943 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2944 the names libpcre and libpcreposix, not just pcre and pcreposix.
2946 (2) The above change means that pcretest and pcregrep are now correctly
2947 linked with the newly-built libraries, not previously installed ones.
2951 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2952 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2953 UTF-8 newline character). The key issue is that the pattern starts .*;
2954 this means that the match must be either at the beginning, or after a
2955 newline. The bug was in the code for advancing after a failed match and
2956 checking that the new position followed a newline. It was not taking
2957 account of UTF-8 characters correctly.
2959 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2960 character classes. PCRE was not treating the sequence [:...:] as a
2961 character class unless the ... were all letters. Perl, however, seems to
2962 allow any characters between [: and :], though of course it rejects as
2963 unknown any "names" that contain non-letters, because all the known class
2964 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2965 for example, whereas PCRE did not - it did not recognize a POSIX character
2966 class. This seemed a bit dangerous, so the code has been changed to be
2967 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2968 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2969 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2970 Perl does, and where it didn't before.
2972 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2973 Windows environments %n is disabled by default.
2976 Version 7.4 21-Sep-07
2977 ---------------------
2979 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2980 means that a class such as [\s] counted as "explicit reference to CR or
2981 LF". That isn't really right - the whole point of the change was to try to
2982 help when there was an actual mention of one of the two characters. So now
2983 the change happens only if \r or \n (or a literal CR or LF) character is
2984 encountered.
2986 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2987 of both had grown to the point where there were only 3 bits left.
2988 Fortunately, there was spare space in the data structure, and so I have
2989 moved the internal flags into a new 16-bit field to free up more option
2990 bits.
2992 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2993 but did not set the internal JCHANGED flag - either of these is enough to
2994 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2995 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2996 start sets both bits.
2998 4. Added options (at build time, compile time, exec time) to change \R from
2999 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
3001 5. doc/pcresyntax.html was missing from the distribution.
3003 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
3004 compatibility, even though it is no longer used.
3006 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
3007 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
3008 windows.h file is present (where different names are used). [This was
3009 reversed later after testing - see 16 below.]
3011 8. Changed all #include <config.h> to #include "config.h". There were also
3012 some further <pcre.h> cases that I changed to "pcre.h".
3014 9. When pcregrep was used with the --colour option, it missed the line ending
3015 sequence off the lines that it output.
3017 10. It was pointed out to me that arrays of string pointers cause lots of
3018 relocations when a shared library is dynamically loaded. A technique of
3019 using a single long string with a table of offsets can drastically reduce
3020 these. I have refactored PCRE in four places to do this. The result is
3021 dramatic:
3023 Originally: 290
3024 After changing UCP table: 187
3025 After changing error message table: 43
3026 After changing table of "verbs" 36
3027 After changing table of Posix names 22
3029 Thanks to the folks working on Gregex for glib for this insight.
3031 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3032 unicode-properties was also set.
3034 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3036 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3037 checked only for CRLF.
3039 14. Added casts to pcretest.c to avoid compiler warnings.
3041 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3043 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3044 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3045 entirely. This removes changes made in 7 above.
3047 17. The CMake files have been updated, and there is now more information about
3048 building with CMake in the NON-UNIX-USE document.
3051 Version 7.3 28-Aug-07
3052 ---------------------
3054 1. In the rejigging of the build system that eventually resulted in 7.1, the
3055 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3056 brackets there is not right, since it causes compilers to look for an
3057 installed pcre.h, not the version that is in the source that is being
3058 compiled (which of course may be different). I have changed it back to:
3060 #include "pcre.h"
3062 I have a vague recollection that the change was concerned with compiling in
3063 different directories, but in the new build system, that is taken care of
3064 by the VPATH setting the Makefile.
3066 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3067 when the subject happened to end in the byte 0x85 (e.g. if the last
3068 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3069 characters but of course it shouldn't be taken as a newline when it is part
3070 of another character. The bug was that, for an unlimited repeat of . in
3071 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3072 characters when looking for a newline.
3074 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3076 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3077 in debug output.
3079 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3080 long printing in the pcrecpp unittest when running under MinGW.
3082 6. ESC_K was left out of the EBCDIC table.
3084 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3085 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3086 limit also applies to "virtual nesting" when a pattern is recursive, and in
3087 this case 1000 isn't so big. I have been able to remove this limit at the
3088 expense of backing off one optimization in certain circumstances. Normally,
3089 when pcre_exec() would call its internal match() function recursively and
3090 immediately return the result unconditionally, it uses a "tail recursion"
3091 feature to save stack. However, when a subpattern that can match an empty
3092 string has an unlimited repetition quantifier, it no longer makes this
3093 optimization. That gives it a stack frame in which to save the data for
3094 checking that an empty string has been matched. Previously this was taken
3095 from the 1000-entry workspace that had been reserved. So now there is no
3096 explicit limit, but more stack is used.
3098 8. Applied Daniel's patches to solve problems with the import/export magic
3099 syntax that is required for Windows, and which was going wrong for the
3100 pcreposix and pcrecpp parts of the library. These were overlooked when this
3101 problem was solved for the main library.
3103 9. There were some crude static tests to avoid integer overflow when computing
3104 the size of patterns that contain repeated groups with explicit upper
3105 limits. As the maximum quantifier is 65535, the maximum group length was
3106 set at 30,000 so that the product of these two numbers did not overflow a
3107 32-bit integer. However, it turns out that people want to use groups that
3108 are longer than 30,000 bytes (though not repeat them that many times).
3109 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3110 made it possible to implement the integer overflow checks in a much more
3111 dynamic way, which I have now done. The artificial limitation on group
3112 length has been removed - we now have only the limit on the total length of
3113 the compiled pattern, which depends on the LINK_SIZE setting.
3115 10. Fixed a bug in the documentation for get/copy named substring when
3116 duplicate names are permitted. If none of the named substrings are set, the
3117 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3118 empty string.
3120 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3121 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3122 because the ] is interpreted as the first data character and the
3123 terminating ] is not found. PCRE has been made compatible with Perl in this
3124 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3125 cause memory overwriting.
3127 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3128 string has been matched (to stop an infinite loop). It was not recognizing
3129 a conditional subpattern that could match an empty string if that
3130 subpattern was within another subpattern. For example, it looped when
3131 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3132 condition was not nested. This bug has been fixed.
3134 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3135 past the start of the subject in the presence of bytes with the top bit
3136 set, for example "\x8aBCD".
3138 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3139 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3141 14. Optimized (?!) to (*FAIL).
3143 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3144 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3145 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3146 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3147 does: it's just the validity check that is more restrictive.
3149 16. Inserted checks for integer overflows during escape sequence (backslash)
3150 processing, and also fixed erroneous offset values for syntax errors during
3151 backslash processing.
3153 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3154 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3156 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3157 caused an overrun.
3159 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3160 something other than just ASCII characters) inside a group that had an
3161 unlimited repeat caused a loop at compile time (while checking to see
3162 whether the group could match an empty string).
3164 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3165 [\P{Any}] did so. (Error in the code for printing property names.)
3167 21. An orphan \E inside a character class could cause a crash.
3169 22. A repeated capturing bracket such as (A)? could cause a wild memory
3170 reference during compilation.
3172 23. There are several functions in pcre_compile() that scan along a compiled
3173 expression for various reasons (e.g. to see if it's fixed length for look
3174 behind). There were bugs in these functions when a repeated \p or \P was
3175 present in the pattern. These operators have additional parameters compared
3176 with \d, etc, and these were not being taken into account when moving along
3177 the compiled data. Specifically:
3179 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3180 length.
3182 (b) An item such as \pL+ within a repeated group could cause crashes or
3183 loops.
3185 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3186 "reference to non-existent subpattern" error.
3188 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3190 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3191 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3193 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3195 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3196 character were causing crashes (broken optimization).
3198 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3199 \p or \P) caused a compile-time loop.
3201 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3202 break. For example, the unstudied pattern [\r\n]A does not match the string
3203 "\r\nA" because change 7.0/46 below moves the current point on by two
3204 characters after failing to match at the start. However, the pattern \nA
3205 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3206 the same is true. There doesn't seem any very clean way out of this, but
3207 what I have chosen to do makes the common cases work: PCRE now takes note
3208 of whether there can be an explicit match for \r or \n anywhere in the
3209 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3210 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3211 pattern has explicit CR or LF references.
3213 29. Added (*CR) etc for changing newline setting at start of pattern.
3216 Version 7.2 19-Jun-07
3217 ---------------------
3219 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3220 which is apparently normally available under Windows.
3222 2. Re-jig the pcregrep tests with different newline settings in an attempt
3223 to make them independent of the local environment's newline setting.
3225 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3227 4. Some of the "internals" tests were previously cut out when the link size
3228 was not 2, because the output contained actual offsets. The recent new
3229 "Z" feature of pcretest means that these can be cut out, making the tests
3230 usable with all link sizes.
3232 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3233 stack recursion. This gives a massive performance boost under BSD, but just
3234 a small improvement under Linux. However, it saves one field in the frame
3235 in all cases.
3237 6. Added more features from the forthcoming Perl 5.10:
3239 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3240 recursion call. It refers to the nth most recently opened parentheses.
3242 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3243 to be opened parentheses.
3245 (c) Conditions that refer to capturing parentheses can be specified
3246 relatively, for example, (?(-2)... or (?(+3)...
3248 (d) \K resets the start of the current match so that everything before
3249 is not part of it.
3251 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3253 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3254 reference syntax.
3256 (g) (?| introduces a group in which the numbering of parentheses in each
3257 alternative starts with the same number.
3259 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3261 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3264 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3265 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3266 for detecting groups that can match an empty string.
3268 9. A pattern with a very large number of alternatives (more than several
3269 hundred) was running out of internal workspace during the pre-compile
3270 phase, where pcre_compile() figures out how much memory will be needed. A
3271 bit of new cunning has reduced the workspace needed for groups with
3272 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3273 workspace instead of running out of the 4096 that are available.
3275 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3277 11. Applied patch from Google to remove an optimization that didn't quite work.
3278 The report of the bug said:
3280 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3281 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3282 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3284 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3285 it matched the wrong number of bytes.
3288 Version 7.1 24-Apr-07
3289 ---------------------
3291 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3292 that is more "standard", making use of automake and other Autotools. There
3293 is some re-arrangement of the files and adjustment of comments consequent
3294 on this.
3296 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3297 for recursive directory scanning broke on some systems because the files
3298 are not scanned in any specific order and on different systems the order
3299 was different. A call to "sort" has been inserted into RunGrepTest for the
3300 approprate test as a short-term fix. In the longer term there may be an
3301 alternative.
3303 3. I had an email from Eric Raymond about problems translating some of PCRE's
3304 man pages to HTML (despite the fact that I distribute HTML pages, some
3305 people do their own conversions for various reasons). The problems
3306 concerned the use of low-level troff macros .br and .in. I have therefore
3307 removed all such uses from the man pages (some were redundant, some could
3308 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3309 HTML has been updated to handle .nf/.fi and to complain if it encounters
3310 .br or .in.
3312 4. Updated comments in configure.ac that get placed in config.h.in and also
3313 arranged for config.h to be included in the distribution, with the name
3314 config.h.generic, for the benefit of those who have to compile without
3315 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3317 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3318 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3319 makevp.bat for the current PCRE, using the additional files
3320 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3322 6. A Windows user reported a minor discrepancy with test 2, which turned out
3323 to be caused by a trailing space on an input line that had got lost in his
3324 copy. The trailing space was an accident, so I've just removed it.
3326 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3327 that is needed.
3329 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3330 as "const" (a) because they are and (b) because it helps the PHP
3331 maintainers who have recently made a script to detect big data structures
3332 in the php code that should be moved to the .rodata section. I remembered
3333 to update Builducptable as well, so it won't revert if ucptable.h is ever
3334 re-created.
3336 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3337 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3338 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3339 support is not required. This saves 1.5-2K of code, which is important in
3340 some applications.
3342 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3343 so as not to refer to the tables, even though these functions will never be
3344 called when UTF-8 support is disabled. Otherwise there are problems with a
3345 shared library.
3347 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3349 (a) It was defining its arguments as char * instead of void *.
3351 (b) It was assuming that all moves were upwards in memory; this was true
3352 a long time ago when I wrote it, but is no longer the case.
3354 The emulated memove() is provided for those environments that have neither
3355 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3356 is clearly not the case, as these two bugs were recently reported.
3358 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3359 and Detrail to create the HTML documentation, the .txt form of the man
3360 pages, and it removes trailing spaces from listed files. It also creates
3361 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3362 case, it wraps all the #defines with #ifndefs. This script should be run
3363 before "make dist".
3365 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3366 with Unicode property support.
3368 (a) For a maximizing quantifier, if the two different cases of the
3369 character were of different lengths in their UTF-8 codings (there are
3370 some cases like this - I found 11), and the matching function had to
3371 back up over a mixture of the two cases, it incorrectly assumed they
3372 were both the same length.
3374 (b) When PCRE was configured to use the heap rather than the stack for
3375 recursion during matching, it was not correctly preserving the data for
3376 the other case of a UTF-8 character when checking ahead for a match
3377 while processing a minimizing repeat. If the check also involved
3378 matching a wide character, but failed, corruption could cause an
3379 erroneous result when trying to check for a repeat of the original
3380 character.
3382 13. Some tidying changes to the testing mechanism:
3384 (a) The RunTest script now detects the internal link size and whether there
3385 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3386 values substituted by "configure". (The RunGrepTest script already did
3387 this for UTF-8.) The configure.ac script no longer substitutes the
3388 relevant variables.
3390 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3391 with length and offset values. This means that the output is different
3392 for different internal link sizes. Test 2 is skipped for link sizes
3393 other than 2 because of this, bypassing the problem. Unfortunately,
3394 there was also a test in test 3 (the locale tests) that used /B and
3395 failed for link sizes other than 2. Rather than cut the whole test out,
3396 I have added a new /Z option to pcretest that replaces the length and
3397 offset values with spaces. This is now used to make test 3 independent
3398 of link size. (Test 2 will be tidied up later.)
3400 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3401 segmentation fault instead of returning the appropriate error message.
3403 15. In multiline mode when the newline sequence was set to "any", the pattern
3404 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3405 This doesn't seem right; it now treats the CRLF combination as the line
3406 ending, and so does not match in that case. It's only a pattern such as ^$
3407 that would hit this one: something like ^ABC$ would have failed after \r
3408 and then tried again after \r\n.
3410 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3411 in an attempt to make files that differ only in their line terminators
3412 compare equal. This works on Linux.
3414 17. Under certain error circumstances pcregrep might try to free random memory
3415 as it exited. This is now fixed, thanks to valgrind.
3417 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3418 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3419 was because its rules for how to advance for /g after matching an empty
3420 string at the end of a line did not allow for this case. They now check for
3421 it specially.
3423 20. pcretest is supposed to handle patterns and data of any length, by
3424 extending its buffers when necessary. It was getting this wrong when the
3425 buffer for a data line had to be extended.
3427 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3428 CRLF as a newline sequence.
3430 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3431 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3432 I have nevertheless tidied it up.
3434 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3436 24. Added a man page for pcre-config.
3439 Version 7.0 19-Dec-06
3440 ---------------------
3442 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3443 moving to gcc 4.1.1.
3445 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3446 sys/time.h, which is documented as needed for this function. It doesn't
3447 seem to matter on Linux, but it showed up on some releases of OS X.
3449 3. It seems that there are systems where bytes whose values are greater than
3450 127 match isprint() in the "C" locale. The "C" locale should be the
3451 default when a C program starts up. In most systems, only ASCII printing
3452 characters match isprint(). This difference caused the output from pcretest
3453 to vary, making some of the tests fail. I have changed pcretest so that:
3455 (a) When it is outputting text in the compiled version of a pattern, bytes
3456 other than 32-126 are always shown as hex escapes.
3458 (b) When it is outputting text that is a matched part of a subject string,
3459 it does the same, unless a different locale has been set for the match
3460 (using the /L modifier). In this case, it uses isprint() to decide.
3462 4. Fixed a major bug that caused incorrect computation of the amount of memory
3463 required for a compiled pattern when options that changed within the
3464 pattern affected the logic of the preliminary scan that determines the
3465 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3466 that the computed length was too small. The symptoms of this bug were
3467 either the PCRE error "internal error: code overflow" from pcre_compile(),
3468 or a glibc crash with a message such as "pcretest: free(): invalid next
3469 size (fast)". Examples of patterns that provoked this bug (shown in
3470 pcretest format) are:
3472 /(?-x: )/x
3473 /(?x)(?-x: \s*#\s*)/
3474 /((?i)[\x{c0}])/8
3475 /(?i:[\x{c0}])/8
3477 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3478 is now done differently.
3480 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3481 wrapper classes; (b) implement a new function in the C++ scanner that is
3482 more efficient than the old way of doing things because it avoids levels of
3483 recursion in the regex matching; (c) add a paragraph to the documentation
3484 for the FullMatch() function.
3486 6. The escape sequence \n was being treated as whatever was defined as
3487 "newline". Not only was this contrary to the documentation, which states
3488 that \n is character 10 (hex 0A), but it also went horribly wrong when
3489 "newline" was defined as CRLF. This has been fixed.
3491 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3492 was being set to -1 for the "end of line" case (supposedly a value that no
3493 character can have). Though this value is never used (the check for end of
3494 line is "zero bytes in current character"), it caused compiler complaints.
3495 I've changed it to 0xffffffff.
3497 8. In pcre_version.c, the version string was being built by a sequence of
3498 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3499 string (as it is for production releases) called a macro with an empty
3500 argument. The C standard says the result of this is undefined. The gcc
3501 compiler treats it as an empty string (which was what was wanted) but it is
3502 reported that Visual C gives an error. The source has been hacked around to
3503 avoid this problem.
3505 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3506 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3507 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3508 of them did).
3510 10. Originally, pcretest opened its input and output without "b"; then I was
3511 told that "b" was needed in some environments, so it was added for release
3512 5.0 to both the input and output. (It makes no difference on Unix-like
3513 systems.) Later I was told that it is wrong for the input on Windows. I've
3514 now abstracted the modes into two macros, to make it easier to fiddle with
3515 them, and removed "b" from the input mode under Windows.
3517 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3519 12. Added -help and --help to pcretest as an official way of being reminded
3520 of the options.
3522 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3523 and pcrecpp.cc because they annoy compilers at high warning levels.
3525 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3527 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3528 scripts are not C programs :-) and which was not noticed because it works
3529 on Linux.
3531 16. pcretest is supposed to handle any length of pattern and data line (as one
3532 line or as a continued sequence of lines) by extending its input buffer if
3533 necessary. This feature was broken for very long pattern lines, leading to
3534 a string of junk being passed to pcre_compile() if the pattern was longer
3535 than about 50K.
3537 17. I have done a major re-factoring of the way pcre_compile() computes the
3538 amount of memory needed for a compiled pattern. Previously, there was code
3539 that made a preliminary scan of the pattern in order to do this. That was
3540 OK when PCRE was new, but as the facilities have expanded, it has become
3541 harder and harder to keep it in step with the real compile phase, and there
3542 have been a number of bugs (see for example, 4 above). I have now found a
3543 cunning way of running the real compile function in a "fake" mode that
3544 enables it to compute how much memory it would need, while actually only
3545 ever using a few hundred bytes of working memory and without too many
3546 tests of the mode. This should make future maintenance and development
3547 easier. A side effect of this work is that the limit of 200 on the nesting
3548 depth of parentheses has been removed (though this was never a serious
3549 limitation, I suspect). However, there is a downside: pcre_compile() now
3550 runs more slowly than before (30% or more, depending on the pattern). I
3551 hope this isn't a big issue. There is no effect on runtime performance.
3553 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3554 newline (only possible for the last line of a file) and it was a
3555 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3557 19. Added additional timing features to pcretest. (1) The -tm option now times
3558 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3559 separate command line item, by a number that specifies the number of
3560 repeats to use when timing. The default is 50000; this gives better
3561 precision, but takes uncomfortably long for very large patterns.
3563 20. Extended pcre_study() to be more clever in cases where a branch of a
3564 subpattern has no definite first character. For example, (a*|b*)[cd] would
3565 previously give no result from pcre_study(). Now it recognizes that the
3566 first character must be a, b, c, or d.
3568 21. There was an incorrect error "recursive call could loop indefinitely" if
3569 a subpattern (or the entire pattern) that was being tested for matching an
3570 empty string contained only one non-empty item after a nested subpattern.
3571 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3572 incorrectly, because the \d was being skipped in the check.
3574 22. The pcretest program now has a new pattern option /B and a command line
3575 option -b, which is equivalent to adding /B to every pattern. This causes
3576 it to show the compiled bytecode, without the additional information that
3577 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3578 is the same as /B/I).
3580 23. A new optimization is now able automatically to treat some sequences such
3581 as a*b as a*+b. More specifically, if something simple (such as a character
3582 or a simple class like \d) has an unlimited quantifier, and is followed by
3583 something that cannot possibly match the quantified thing, the quantifier
3584 is automatically "possessified".
3586 24. A recursive reference to a subpattern whose number was greater than 39
3587 went wrong under certain circumstances in UTF-8 mode. This bug could also
3588 have affected the operation of pcre_study().
3590 25. Realized that a little bit of performance could be had by replacing
3591 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3593 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3595 27. Possessive quantifiers such as a++ were previously implemented by turning
3596 them into atomic groups such as ($>a+). Now they have their own opcodes,
3597 which improves performance. This includes the automatically created ones
3598 from 23 above.
3600 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3601 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3602 the first matched character to be a colon. This applied both to named and
3603 numbered groups.
3605 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3607 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3608 building PCRE on Windows easier, so I have included it in the distribution.
3610 31. There is now a check in pcretest against a ridiculously large number being
3611 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3612 loop, the loop is abandoned.
3614 32. Forward references to subpatterns in conditions such as (?(2)...) where
3615 subpattern 2 is defined later cause pcre_compile() to search forwards in
3616 the pattern for the relevant set of parentheses. This search went wrong
3617 when there were unescaped parentheses in a character class, parentheses
3618 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3620 33. "Subroutine" calls and backreferences were previously restricted to
3621 referencing subpatterns earlier in the regex. This restriction has now
3622 been removed.
3624 34. Added a number of extra features that are going to be in Perl 5.10. On the
3625 whole, these are just syntactic alternatives for features that PCRE had
3626 previously implemented using the Python syntax or my own invention. The
3627 other formats are all retained for compatibility.
3629 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3630 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3631 also .NET compatible.
3633 (b) A recursion or subroutine call to a named group can now be defined as
3634 (?&name) as well as (?P>name).
3636 (c) A backreference to a named group can now be defined as \k<name> or
3637 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3638 5.10, are also .NET compatible.
3640 (d) A conditional reference to a named group can now use the syntax
3641 (?(<name>) or (?('name') as well as (?(name).
3643 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3644 groups (named and numbered) that are never evaluated inline, but can be
3645 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3646 is always false. There may be only one alternative in such a group.
3648 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3649 as the simple (?(R). The condition is true only if the most recent
3650 recursion is that of the given number or name. It does not search out
3651 through the entire recursion stack.
3653 (g) The escape \gN or \g{N} has been added, where N is a positive or
3654 negative number, specifying an absolute or relative reference.
3656 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3657 some "unreachable code" warnings.
3659 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3660 things, this adds five new scripts.
3662 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3663 There were also incompatibilities regarding the handling of \Q..\E inside
3664 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3665 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3667 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3668 matches an empty string, and forcibly breaks the loop. There were bugs in
3669 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3670 against aaaa the result was just "a" rather than "aaaa", for example. Two
3671 separate and independent bugs (that affected different cases) have been
3672 fixed.
3674 39. Refactored the code to abolish the use of different opcodes for small
3675 capturing bracket numbers. This is a tidy that I avoided doing when I
3676 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3677 The new approach is not only tidier, it makes it possible to reduce the
3678 memory needed to fix the previous bug (38).
3680 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3681 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3682 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3683 mode.
3685 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3686 report.
3688 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3689 copy construction and assignment in the C++ wrapper.
3691 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3692 couple of bugs that could have given wrong results in the "--newline=crlf"
3693 case.
3695 44. Added a number of casts and did some reorganization of signed/unsigned int
3696 variables following suggestions from Dair Grant. Also renamed the variable
3697 "this" as "item" because it is a C++ keyword.
3699 45. Arranged for dftables to add
3701 #include "pcre_internal.h"
3703 to pcre_chartables.c because without it, gcc 4.x may remove the array
3704 definition from the final binary if PCRE is built into a static library and
3705 dead code stripping is activated.
3707 46. For an unanchored pattern, if a match attempt fails at the start of a
3708 newline sequence, and the newline setting is CRLF or ANY, and the next two
3709 characters are CRLF, advance by two characters instead of one.
3712 Version 6.7 04-Jul-06
3713 ---------------------
3715 1. In order to handle tests when input lines are enormously long, pcretest has
3716 been re-factored so that it automatically extends its buffers when
3717 necessary. The code is crude, but this _is_ just a test program. The
3718 default size has been increased from 32K to 50K.
3720 2. The code in pcre_study() was using the value of the re argument before
3721 testing it for NULL. (Of course, in any sensible call of the function, it
3722 won't be NULL.)
3724 3. The memmove() emulation function in pcre_internal.h, which is used on
3725 systems that lack both memmove() and bcopy() - that is, hardly ever -
3726 was missing a "static" storage class specifier.
3728 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3729 containing an extended class (one that cannot be represented by a bitmap
3730 because it contains high-valued characters or Unicode property items, e.g.
3731 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3732 pattern, but PCRE should not loop if you do not (it no longer does).
3733 [Detail: two cases were found: (a) a repeated subpattern containing an
3734 extended class; (b) a recursive reference to a subpattern that followed a
3735 previous extended class. It wasn't skipping over the extended class
3736 correctly when UTF-8 mode was not set.]
3738 5. A negated single-character class was not being recognized as fixed-length
3739 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3740 compile error "lookbehind assertion is not fixed length".
3742 6. The RunPerlTest auxiliary script was showing an unexpected difference
3743 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3744 write a Perl script that can interpret lines of an input file either as
3745 byte characters or as UTF-8, which is what "perltest" was being required to
3746 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3747 can't do is switch easily at run time between having the "use utf8;" pragma
3748 or not. In the end, I fudged it by using the RunPerlTest script to insert
3749 "use utf8;" explicitly for the UTF-8 tests.
3751 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3752 the end of the subject string, contrary to the documentation and to what
3753 Perl does. This was true of both matching functions. Now it matches only at
3754 the start of the subject and immediately after *internal* newlines.
3756 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3757 a pointer to an int instead of a pointer to an unsigned long int. This
3758 caused problems on 64-bit systems.
3760 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3761 instance of the 'standard' template library not being so standard".
3763 10. There was no check on the number of named subpatterns nor the maximum
3764 length of a subpattern name. The product of these values is used to compute
3765 the size of the memory block for a compiled pattern. By supplying a very
3766 long subpattern name and a large number of named subpatterns, the size
3767 computation could be caused to overflow. This is now prevented by limiting
3768 the length of names to 32 characters, and the number of named subpatterns
3769 to 10,000.
3771 11. Subpatterns that are repeated with specific counts have to be replicated in
3772 the compiled pattern. The size of memory for this was computed from the
3773 length of the subpattern and the repeat count. The latter is limited to
3774 65535, but there was no limit on the former, meaning that integer overflow
3775 could in principle occur. The compiled length of a repeated subpattern is
3776 now limited to 30,000 bytes in order to prevent this.
3778 12. Added the optional facility to have named substrings with the same name.
3780 13. Added the ability to use a named substring as a condition, using the
3781 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3782 are numbers (not recommended). Forward references are permitted.
3784 14. Added forward references in named backreferences (if you see what I mean).
3786 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3787 pattern could run off the end of the subject. For example, the pattern
3788 "(?s)(.{1,5})"8 did this with the subject "ab".
3790 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3791 PCRE_CASELESS was set when matching characters that were quantified with ?
3792 or *.
3794 17. A character class other than a single negated character that had a minimum
3795 but no maximum quantifier - for example [ab]{6,} - was not handled
3796 correctly by pce_dfa_exec(). It would match only one character.
3798 18. A valid (though odd) pattern that looked like a POSIX character
3799 class but used an invalid character after [ (for example [[,abc,]]) caused
3800 pcre_compile() to give the error "Failed: internal error: code overflow" or
3801 in some cases to crash with a glibc free() error. This could even happen if
3802 the pattern terminated after [[ but there just happened to be a sequence of
3803 letters, a binary zero, and a closing ] in the memory that followed.
3805 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3806 over the years. Originally (before any Unicode support), just the bottom 8
3807 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3808 output from "man perlunicode" includes this:
3810 The regular expression compiler produces polymorphic opcodes. That
3811 is, the pattern adapts to the data and automatically switches to
3812 the Unicode character scheme when presented with Unicode data--or
3813 instead uses a traditional byte scheme when presented with byte
3814 data.
3816 Sadly, a wide octal escape does not cause a switch, and in a string with
3817 no other multibyte characters, these octal escapes are treated as before.
3818 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3819 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3820 Unicode string.
3822 I have not perpetrated such confusion in PCRE. Up till now, it took just
3823 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3824 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3825 translate to the appropriate multibyte character.
3827 29. Applied some refactoring to reduce the number of warnings from Microsoft
3828 and Borland compilers. This has included removing the fudge introduced
3829 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3830 a warning about an unused variable.
3832 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3833 characters since release 4.0, because Perl (from release 5.004) does not.
3834 [Or at least, is documented not to: some releases seem to be in conflict
3835 with the documentation.] However, when a pattern was studied with
3836 pcre_study() and all its branches started with \s, PCRE still included VT
3837 as a possible starting character. Of course, this did no harm; it just
3838 caused an unnecessary match attempt.
3840 22. Removed a now-redundant internal flag bit that recorded the fact that case
3841 dependency changed within the pattern. This was once needed for "required
3842 byte" processing, but is no longer used. This recovers a now-scarce options
3843 bit. Also moved the least significant internal flag bit to the most-
3844 significant bit of the word, which was not previously used (hangover from
3845 the days when it was an int rather than a uint) to free up another bit for
3846 the future.
3848 23. Added support for CRLF line endings as well as CR and LF. As well as the
3849 default being selectable at build time, it can now be changed at runtime
3850 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3851 specify that it is scanning data with non-default line endings.
3853 24. Changed the definition of CXXLINK to make it agree with the definition of
3854 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3856 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3857 recursions. This makes a big different to stack usage for some patterns.
3859 26. If a subpattern containing a named recursion or subroutine reference such
3860 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3861 the space required for the compiled pattern went wrong and gave too small a
3862 value. Depending on the environment, this could lead to "Failed: internal
3863 error: code overflow at offset 49" or "glibc detected double free or
3864 corruption" errors.
3866 27. Applied patches from Google (a) to support the new newline modes and (b) to
3867 advance over multibyte UTF-8 characters in GlobalReplace.
3869 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3870 difference for some implementation of PCRE in some Windows version.
3872 29. Added some extra testing facilities to pcretest:
3874 \q<number> in a data line sets the "match limit" value
3875 \Q<number> in a data line sets the "match recursion limt" value
3876 -S <number> sets the stack size, where <number> is in megabytes
3878 The -S option isn't available for Windows.
3881 Version 6.6 06-Feb-06
3882 ---------------------
3884 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3885 in pcreposix.h. I have copied the definition from pcre.h.
3887 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3888 because pcre.h is no longer a built file.
3890 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3891 not normally included in the compiled code.
3894 Version 6.5 01-Feb-06
3895 ---------------------
3897 1. When using the partial match feature with pcre_dfa_exec(), it was not
3898 anchoring the second and subsequent partial matches at the new starting
3899 point. This could lead to incorrect results. For example, with the pattern
3900 /1234/, partially matching against "123" and then "a4" gave a match.
3902 2. Changes to pcregrep:
3904 (a) All non-match returns from pcre_exec() were being treated as failures
3905 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3906 error message is output. Some extra information is given for the
3908 probably the only errors that are likely to be caused by users (by
3909 specifying a regex that has nested indefinite repeats, for instance).
3910 If there are more than 20 of these errors, pcregrep is abandoned.
3912 (b) A binary zero was treated as data while matching, but terminated the
3913 output line if it was written out. This has been fixed: binary zeroes
3914 are now no different to any other data bytes.
3916 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3917 used to set a locale for matching. The --locale=xxxx long option has
3918 been added (no short equivalent) to specify a locale explicitly on the
3919 pcregrep command, overriding the environment variables.
3921 (d) When -B was used with -n, some line numbers in the output were one less
3922 than they should have been.
3924 (e) Added the -o (--only-matching) option.
3926 (f) If -A or -C was used with -c (count only), some lines of context were
3927 accidentally printed for the final match.
3929 (g) Added the -H (--with-filename) option.
3931 (h) The combination of options -rh failed to suppress file names for files
3932 that were found from directory arguments.
3934 (i) Added the -D (--devices) and -d (--directories) options.
3936 (j) Added the -F (--fixed-strings) option.
3938 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3940 (l) Added the --colo(u)r option.
3942 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3943 is not present by default.
3945 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3946 items such as (?R) or (?1), when the recursion could match a number of
3947 alternatives. If it matched one of the alternatives, but subsequently,
3948 outside the recursion, there was a failure, the code tried to back up into
3949 the recursion. However, because of the way PCRE is implemented, this is not
3950 possible, and the result was an incorrect result from the match.
3952 In order to prevent this happening, the specification of recursion has
3953 been changed so that all such subpatterns are automatically treated as
3954 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3956 4. I had overlooked the fact that, in some locales, there are characters for
3957 which isalpha() is true but neither isupper() nor islower() are true. In
3958 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3959 and ordfeminine) are like this. This affected the treatment of \w and \W
3960 when they appeared in character classes, but not when they appeared outside
3961 a character class. The bit map for "word" characters is now created
3962 separately from the results of isalnum() instead of just taking it from the
3963 upper, lower, and digit maps. (Plus the underscore character, of course.)
3965 5. The above bug also affected the handling of POSIX character classes such as
3966 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3967 permanent tables. Instead, the bit maps for such a class were previously
3968 created as the appropriate unions of the upper, lower, and digit bitmaps.
3969 Now they are created by subtraction from the [[:word:]] class, which has
3970 its own bitmap.
3972 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3973 It is created by subtracting the vertical space characters (\x09, \x0a,
3974 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3975 subtraction was done in the overall bitmap for a character class, meaning
3976 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3977 be recognized. This bug has been fixed.
3979 7. Patches from the folks at Google:
3981 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3982 real life, but is still worth protecting against".
3984 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3985 regular expressions".
3987 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3988 have it.
3990 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3991 "configure" and the latter not, in order to fix a problem somebody had
3992 with compiling the Arg class on HP-UX.
3994 (e) Improve the error-handling of the C++ wrapper a little bit.
3996 (f) New tests for checking recursion limiting.
3998 8. The pcre_memmove() function, which is used only if the environment does not
3999 have a standard memmove() function (and is therefore rarely compiled),
4000 contained two bugs: (a) use of int instead of size_t, and (b) it was not
4001 returning a result (though PCRE never actually uses the result).
4003 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
4004 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
4005 returned instead of calling malloc() with an overflowing number that would
4006 most likely cause subsequent chaos.
4008 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
4010 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
4011 with this option is matched, the nmatch and pmatch options of regexec() are
4012 ignored.
4014 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
4015 provided in case anyone wants to the the POSIX interface with UTF-8
4016 strings.
4018 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
4019 C++ linking (needed for some HP-UX environments).
4021 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
4022 (unused parameter) and in the pcre_printint() function (omitted "default"
4023 switch label when the default is to do nothing).
4025 15. Added some code to make it possible, when PCRE is compiled as a C++
4026 library, to replace subject pointers for pcre_exec() with a smart pointer
4027 class, thus making it possible to process discontinuous strings.
4029 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
4030 much the same function. They were added by different people who were trying
4031 to make PCRE easy to compile on non-Unix systems. It has been suggested
4032 that PCRE_EXPORT be abolished now that there is more automatic apparatus
4033 for compiling on Windows systems. I have therefore replaced it with
4034 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
4035 defaults to "extern" for C or "extern C" for C++, which works fine on
4036 Unix-like systems. It is now possible to override the value of PCRE_DATA_
4037 SCOPE with something explicit in config.h. In addition:
4039 (a) pcreposix.h still had just "extern" instead of either of these macros;
4040 I have replaced it with PCRE_DATA_SCOPE.
4042 (b) Functions such as _pcre_xclass(), which are internal to the library,
4043 but external in the C sense, all had PCRE_EXPORT in their definitions.
4044 This is apparently wrong for the Windows case, so I have removed it.
4045 (It makes no difference on Unix-like systems.)
4047 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
4048 of recursive calls to match(). This is different to MATCH_LIMIT because
4049 that limits the total number of calls to match(), not all of which increase
4050 the depth of recursion. Limiting the recursion depth limits the amount of
4051 stack (or heap if NO_RECURSE is set) that is used. The default can be set
4052 when PCRE is compiled, and changed at run time. A patch from Google adds
4053 this functionality to the C++ interface.
4055 18. Changes to the handling of Unicode character properties:
4057 (a) Updated the table to Unicode 4.1.0.
4059 (b) Recognize characters that are not in the table as "Cn" (undefined).
4061 (c) I revised the way the table is implemented to a much improved format
4062 which includes recognition of ranges. It now supports the ranges that
4063 are defined in UnicodeData.txt, and it also amalgamates other
4064 characters into ranges. This has reduced the number of entries in the
4065 table from around 16,000 to around 3,000, thus reducing its size
4066 considerably. I realized I did not need to use a tree structure after
4067 all - a binary chop search is just as efficient. Having reduced the
4068 number of entries, I extended their size from 6 bytes to 8 bytes to
4069 allow for more data.
4071 (d) Added support for Unicode script names via properties such as \p{Han}.
4073 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
4074 matching that character.
4076 20. When matching a repeated Unicode property with a minimum greater than zero,
4077 (for example \pL{2,}), PCRE could look past the end of the subject if it
4078 reached it while seeking the minimum number of characters. This could
4079 happen only if some of the characters were more than one byte long, because
4080 there is a check for at least the minimum number of bytes.
4082 21. Refactored the implementation of \p and \P so as to be more general, to
4083 allow for more different types of property in future. This has changed the
4084 compiled form incompatibly. Anybody with saved compiled patterns that use
4085 \p or \P will have to recompile them.
4087 22. Added "Any" and "L&" to the supported property types.
4089 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
4090 but give a compile time error if the value is greater than 0xff.
4092 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
4093 accidentally not being installed or uninstalled.
4095 25. The pcre.h file was built from pcre.h.in, but the only changes that were
4096 made were to insert the current release number. This seemed silly, because
4097 it made things harder for people building PCRE on systems that don't run
4098 "configure". I have turned pcre.h into a distributed file, no longer built
4099 by "configure", with the version identification directly included. There is
4100 no longer a pcre.h.in file.
4102 However, this change necessitated a change to the pcre-config script as
4103 well. It is built from pcre-config.in, and one of the substitutions was the
4104 release number. I have updated configure.ac so that ./configure now finds
4105 the release number by grepping pcre.h.
4107 26. Added the ability to run the tests under valgrind.
4110 Version 6.4 05-Sep-05
4111 ---------------------
4113 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
4114 "--" to be printed when multiple files were scanned, even when none of the
4115 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
4116 consider it to be a bug, and have restored the previous behaviour.
4118 2. A couple of code tidies to get rid of compiler warnings.
4120 3. The pcretest program used to cheat by referring to symbols in the library
4121 whose names begin with _pcre_. These are internal symbols that are not
4122 really supposed to be visible externally, and in some environments it is
4123 possible to suppress them. The cheating is now confined to including
4124 certain files from the library's source, which is a bit cleaner.
4126 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
4127 file's purpose clearer.
4129 5. Reorganized pcre_ucp_findchar().
4132 Version 6.3 15-Aug-05
4133 ---------------------
4135 1. The file libpcre.pc.in did not have general read permission in the tarball.
4137 2. There were some problems when building without C++ support:
4139 (a) If C++ support was not built, "make install" and "make test" still
4140 tried to test it.
4142 (b) There were problems when the value of CXX was explicitly set. Some
4143 changes have been made to try to fix these, and ...
4145 (c) --disable-cpp can now be used to explicitly disable C++ support.
4147 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4148 backslash in a target when C++ was disabled. This confuses some
4149 versions of "make", apparently. Using an intermediate variable solves
4150 this. (Same for CPP_LOBJ.)
4152 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4153 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4154 necessary on certain architectures.
4156 4. Added a setting of -export-symbols-regex to the link command to remove
4157 those symbols that are exported in the C sense, but actually are local
4158 within the library, and not documented. Their names all begin with
4159 "_pcre_". This is not a perfect job, because (a) we have to except some
4160 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4161 available (and never for static libraries). I have made a note to try to
4162 find a way round (a) in the future.
4165 Version 6.2 01-Aug-05
4166 ---------------------
4168 1. There was no test for integer overflow of quantifier values. A construction
4169 such as {1111111111111111} would give undefined results. What is worse, if
4170 a minimum quantifier for a parenthesized subpattern overflowed and became
4171 negative, the calculation of the memory size went wrong. This could have
4172 led to memory overwriting.
4174 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4176 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4177 operating environments where this matters.
4179 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4180 PCRE options from within the C++ wrapper.
4182 5. Named capturing subpatterns were not being correctly counted when a pattern
4183 was compiled. This caused two problems: (a) If there were more than 100
4184 such subpatterns, the calculation of the memory needed for the whole
4185 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4186 back references of the form \12, where the number was greater than 9, were
4187 not recognized as back references, even though there were sufficient
4188 previous subpatterns.
4190 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4191 versions of gcc, e.g. 2.95.4.
4194 Version 6.1 21-Jun-05
4195 ---------------------
4197 1. There was one reference to the variable "posix" in pcretest.c that was not
4198 surrounded by "#if !defined NOPOSIX".
4200 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4201 the cross-check on the old pcre_info() function, for the benefit of the
4202 cut-down version of PCRE that is currently imported into Exim.
4204 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4205 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4206 patterns that start (?i) but I don't think that matters. The use of (?i) is
4207 just an example; this all applies to the other options as well.
4209 4. Since libtool seems to echo the compile commands it is issuing, the output
4210 from "make" can be reduced a bit by putting "@" in front of each libtool
4211 compile command.
4213 5. Patch from the folks at Google for configure.in to be a bit more thorough
4214 in checking for a suitable C++ installation before trying to compile the
4215 C++ stuff. This should fix a reported problem when a compiler was present,
4216 but no suitable headers.
4218 6. The man pages all had just "PCRE" as their title. I have changed them to
4219 be the relevant file name. I have also arranged that these names are
4220 retained in the file doc/pcre.txt, which is a concatenation in text format
4221 of all the man pages except the little individual ones for each function.
4223 7. The NON-UNIX-USE file had not been updated for the different set of source
4224 files that come with release 6. I also added a few comments about the C++
4225 wrapper.
4228 Version 6.0 07-Jun-05
4229 ---------------------
4231 1. Some minor internal re-organization to help with my DFA experiments.
4233 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4234 didn't matter for the library itself when fully configured, but did matter
4235 when compiling without UCP support, or within Exim, where the ucp files are
4236 not imported.
4238 3. Refactoring of the library code to split up the various functions into
4239 different source modules. The addition of the new DFA matching code (see
4240 below) to a single monolithic source would have made it really too
4241 unwieldy, quite apart from causing all the code to be include in a
4242 statically linked application, when only some functions are used. This is
4243 relevant even without the DFA addition now that patterns can be compiled in
4244 one application and matched in another.
4246 The downside of splitting up is that there have to be some external
4247 functions and data tables that are used internally in different modules of
4248 the library but which are not part of the API. These have all had their
4249 names changed to start with "_pcre_" so that they are unlikely to clash
4250 with other external names.
4252 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4253 a different (DFA) algorithm. Although it is slower than the original
4254 function, it does have some advantages for certain types of matching
4255 problem.
4257 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4258 including restarting after a partial match.
4260 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4261 defined when compiling for Windows was sent to me. I have put it into the
4262 code, though I have no means of testing or verifying it.
4264 7. Added the pcre_refcount() auxiliary function.
4266 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4267 match before or at the first newline in the subject string. In pcretest,
4268 the /f option on a pattern can be used to set this.
4270 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4271 would behave wrongly. This has been present in PCRE since release 4.0.
4273 10. A number of changes to the pcregrep command:
4275 (a) Refactored how -x works; insert ^(...)$ instead of setting
4276 PCRE_ANCHORED and checking the length, in preparation for adding
4277 something similar for -w.
4279 (b) Added the -w (match as a word) option.
4281 (c) Refactored the way lines are read and buffered so as to have more
4282 than one at a time available.
4284 (d) Implemented a pcregrep test script.
4286 (e) Added the -M (multiline match) option. This allows patterns to match
4287 over several lines of the subject. The buffering ensures that at least
4288 8K, or the rest of the document (whichever is the shorter) is available
4289 for matching (and similarly the previous 8K for lookbehind assertions).
4291 (f) Changed the --help output so that it now says
4293 -w, --word-regex(p)
4295 instead of two lines, one with "regex" and the other with "regexp"
4296 because that confused at least one person since the short forms are the
4297 same. (This required a bit of code, as the output is generated
4298 automatically from a table. It wasn't just a text change.)
4300 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4301 option but starts with a hyphen. Could be a pattern or a path name
4302 starting with a hyphen, for instance.
4304 (h) "-" can be given as a file name to represent stdin.
4306 (i) When file names are being printed, "(standard input)" is used for
4307 the standard input, for compatibility with GNU grep. Previously
4308 "<stdin>" was used.
4310 (j) The option --label=xxx can be used to supply a name to be used for
4311 stdin when file names are being printed. There is no short form.
4313 (k) Re-factored the options decoding logic because we are going to add
4314 two more options that take data. Such options can now be given in four
4315 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4317 (l) Added the -A, -B, and -C options for requesting that lines of context
4318 around matches be printed.
4320 (m) Added the -L option to print the names of files that do not contain
4321 any matching lines, that is, the complement of -l.
4323 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4324 continue to scan other files.
4326 (o) The -s option was incorrectly implemented. For compatibility with other
4327 greps, it now suppresses the error message for a non-existent or non-
4328 accessible file (but not the return code). There is a new option called
4329 -q that suppresses the output of matching lines, which was what -s was
4330 previously doing.
4332 (p) Added --include and --exclude options to specify files for inclusion
4333 and exclusion when recursing.
4335 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4336 Hopefully, it now does.
4338 12. Missing cast in pcre_study().
4340 13. Added an "uninstall" target to the makefile.
4342 14. Replaced "extern" in the function prototypes in Makefile.in with
4343 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4344 world, but is set differently for Windows.
4346 15. Added a second compiling function called pcre_compile2(). The only
4347 difference is that it has an extra argument, which is a pointer to an
4348 integer error code. When there is a compile-time failure, this is set
4349 non-zero, in addition to the error test pointer being set to point to an
4350 error message. The new argument may be NULL if no error number is required
4351 (but then you may as well call pcre_compile(), which is now just a
4352 wrapper). This facility is provided because some applications need a
4353 numeric error indication, but it has also enabled me to tidy up the way
4354 compile-time errors are handled in the POSIX wrapper.
4356 16. Added VPATH=.libs to the makefile; this should help when building with one
4357 prefix path and installing with another. (Or so I'm told by someone who
4358 knows more about this stuff than I do.)
4360 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4361 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4362 match everything, including newlines. This is not POSIX-compatible, but
4363 somebody wanted the feature. From pcretest it can be activated by using
4364 both the P and the s flags.
4366 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4368 19. libpcre.pc was being incorrectly installed as executable.
4370 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4371 it now also looks for '\r' so that it will work unmodified on Windows.
4373 21. Added Google's contributed C++ wrapper to the distribution.
4375 22. Added some untidy missing memory free() calls in pcretest, to keep
4376 Electric Fence happy when testing.
4380 Version 5.0 13-Sep-04
4381 ---------------------
4383 1. Internal change: literal characters are no longer packed up into items
4384 containing multiple characters in a single byte-string. Each character
4385 is now matched using a separate opcode. However, there may be more than one
4386 byte in the character in UTF-8 mode.
4388 2. The pcre_callout_block structure has two new fields: pattern_position and
4389 next_item_length. These contain the offset in the pattern to the next match
4390 item, and its length, respectively.
4392 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4393 insertion of callouts before each pattern item. Added the /C option to
4394 pcretest to make use of this.
4396 4. On the advice of a Windows user, the lines
4398 #if defined(_WIN32) || defined(WIN32)
4399 _setmode( _fileno( stdout ), 0x8000 );
4400 #endif /* defined(_WIN32) || defined(WIN32) */
4402 have been added to the source of pcretest. This apparently does useful
4403 magic in relation to line terminators.
4405 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4406 for the benefit of those environments where the "b" makes a difference.
4408 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4409 to know about it. I have put a hack into configure.in that adds in code
4410 to set GCC=yes if CC=icc. This seems to end up at a point in the
4411 generated configure script that is early enough to affect the setting of
4412 compiler options, which is what is needed, but I have no means of testing
4413 whether it really works. (The user who reported this had patched the
4414 generated configure script, which of course I cannot do.)
4416 LATER: After change 22 below (new libtool files), the configure script
4417 seems to know about icc (and also ecc). Therefore, I have commented out
4418 this hack in configure.in.
4420 7. Added support for pkg-config (2 patches were sent in).
4422 8. Negated POSIX character classes that used a combination of internal tables
4423 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4424 [[:^ascii]]. Typically, they would match almost any characters. The other
4425 POSIX classes were not broken in this way.
4427 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4428 to find the match, as PCRE was deluded into thinking that the match had to
4429 start at the start point or following a newline. The same bug applied to
4430 patterns with negative forward assertions or any backward assertions
4431 preceding ".*" at the start, unless the pattern required a fixed first
4432 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4434 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4435 starting at the last subject character, bytes beyond the end of the subject
4436 string were read.
4438 11. Renamed the variable "class" as "classbits" to make life easier for C++
4439 users. (Previously there was a macro definition, but it apparently wasn't
4440 enough.)
4442 12. Added the new field "tables" to the extra data so that tables can be passed
4443 in at exec time, or the internal tables can be re-selected. This allows
4444 a compiled regex to be saved and re-used at a later time by a different
4445 program that might have everything at different addresses.
4447 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4448 -R library as well as a -L library.
4450 14. The debugging options of pcretest (-d on the command line or D on a
4451 pattern) showed incorrect output for anything following an extended class
4452 that contained multibyte characters and which was followed by a quantifier.
4454 15. Added optional support for general category Unicode character properties
4455 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4456 support. It adds about 90K to the size of the library. The meanings of the
4457 inbuilt class escapes such as \d and \s have NOT been changed.
4459 16. Updated pcredemo.c to include calls to free() to release the memory for the
4460 compiled pattern.
4462 17. The generated file chartables.c was being created in the source directory
4463 instead of in the building directory. This caused the build to fail if the
4464 source directory was different from the building directory, and was
4465 read-only.
4467 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4468 file. No doubt somebody will tell me if they don't make sense... Also added
4469 Dan Mooney's comments about building on OpenVMS.
4471 19. Added support for partial matching via the PCRE_PARTIAL option for
4472 pcre_exec() and the \P data escape in pcretest.
4474 20. Extended pcretest with 3 new pattern features:
4476 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4477 write the compiled pattern to the file whose name is "rest-of-line".
4478 This is a straight binary dump of the data, with the saved pointer to
4479 the character tables forced to be NULL. The study data, if any, is
4480 written too. After writing, pcretest reads a new pattern.
4482 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4483 compiled pattern from the given file. There must not be any
4484 occurrences of "<" in the file name (pretty unlikely); if there are,
4485 pcretest will instead treat the initial "<" as a pattern delimiter.
4486 After reading in the pattern, pcretest goes on to read data lines as
4487 usual.
4489 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4490 and 16-bit fields in a compiled pattern, to simulate a pattern that
4491 was compiled on a host of opposite endianness.
4493 21. The pcre-exec() function can now cope with patterns that were compiled on
4494 hosts of opposite endianness, with this restriction:
4496 As for any compiled expression that is saved and used later, the tables
4497 pointer field cannot be preserved; the extra_data field in the arguments
4498 to pcre_exec() should be used to pass in a tables address if a value
4499 other than the default internal tables were used at compile time.
4501 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4502 now diagnosed as an error. Previously, most of the time, a negative number
4503 would have been treated as zero, but if in addition "ovector" was passed as
4504 NULL, a crash could occur.
4506 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4507 new versions from the libtool 1.5 distribution (the last one is a copy of
4508 a file called libtool.m4). This seems to have fixed the need to patch
4509 "configure" to support Darwin 1.3 (which I used to do). However, I still
4510 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4511 workstation).
4513 24. Changed the PCRE licence to be the more standard "BSD" licence.
4516 Version 4.5 01-Dec-03
4517 ---------------------
4519 1. There has been some re-arrangement of the code for the match() function so
4520 that it can be compiled in a version that does not call itself recursively.
4521 Instead, it keeps those local variables that need separate instances for
4522 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4523 needs to "recurse". Keeping track of where control must go is done by means
4524 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4525 hide most of the details from the main code, and operates only if
4526 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4527 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4528 operating.
4530 To make it easier for callers to provide specially tailored get/free
4531 functions for this usage, two new functions, pcre_stack_malloc, and
4532 pcre_stack_free, are used. They are always called in strict stacking order,
4533 and the size of block requested is always the same.
4535 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4536 PCRE has been compiled to use the stack or the heap for recursion. The
4537 -C option of pcretest uses this to show which version is compiled.
4539 A new data escape \S, is added to pcretest; it causes the amounts of store
4540 obtained and freed by both kinds of malloc/free at match time to be added
4541 to the output.
4543 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4544 what's available on my current Linux desktop machine.
4546 3. When matching a UTF-8 string, the test for a valid string at the start has
4547 been extended. If start_offset is not zero, PCRE now checks that it points
4548 to a byte that is the start of a UTF-8 character. If not, it returns
4549 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4550 this is necessary because there may be backward assertions in the pattern.
4551 When matching the same subject several times, it may save resources to use
4552 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4554 4. The code for checking the validity of UTF-8 strings has been tightened so
4555 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4556 containing "overlong sequences".
4558 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4559 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4560 should have been "&", but it just so happened that all the cases this let
4561 through by mistake were picked up later in the function.
4563 6. I had used a variable called "isblank" - this is a C99 function, causing
4564 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4566 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4567 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4568 so the version is shown; (c) stop "make test" echoing "./RunTest".
4570 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4572 9. The prototype for memmove() for systems that don't have it was using
4573 size_t, but the inclusion of the header that defines size_t was later. I've
4574 moved the #includes for the C headers earlier to avoid this.
4576 10. Added some adjustments to the code to make it easier to compiler on certain
4577 special systems:
4579 (a) Some "const" qualifiers were missing.
4580 (b) Added the macro EXPORT before all exported functions; by default this
4581 is defined to be empty.
4582 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4583 that it reads its output file name as an argument instead of writing
4584 to the standard output and assuming this can be redirected.
4586 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4587 class containing characters with values greater than 255, PCRE compilation
4588 went into a loop.
4590 12. A recursive reference to a subpattern that was within another subpattern
4591 that had a minimum quantifier of zero caused PCRE to crash. For example,
4592 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4593 recursion. If the recursively-called subpattern itself had a zero repeat,
4594 that was OK.
4596 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4597 buffer into which it was copied (for escape processing) was still set at
4598 1024, so long lines caused crashes.
4600 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4601 "internal error: code overflow...". This applied to any character class
4602 that was followed by a possessive quantifier.
4604 15. Modified the Makefile to add libpcre.la as a prerequisite for
4605 libpcreposix.la because I was told this is needed for a parallel build to
4606 work.
4608 16. If a pattern that contained .* following optional items at the start was
4609 studied, the wrong optimizing data was generated, leading to matching
4610 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4611 matching string must start with a or b or c. The correct conclusion for
4612 this pattern is that a match can start with any character.
4615 Version 4.4 13-Aug-03
4616 ---------------------
4618 1. In UTF-8 mode, a character class containing characters with values between
4619 127 and 255 was not handled correctly if the compiled pattern was studied.
4620 In fixing this, I have also improved the studying algorithm for such
4621 classes (slightly).
4623 2. Three internal functions had redundant arguments passed to them. Removal
4624 might give a very teeny performance improvement.
4626 3. Documentation bug: the value of the capture_top field in a callout is *one
4627 more than* the number of the hightest numbered captured substring.
4629 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4630 in incorrectly linking with a previously installed version. They now link
4631 explicitly with libpcre.la.
4633 5. configure.in no longer needs to recognize Cygwin specially.
4635 6. A problem in pcre.in for Windows platforms is fixed.
4637 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4638 pcretest, it used to include the size of the study block as part of its
4639 output. Unfortunately, the structure contains a field that has a different
4640 size on different hardware architectures. This meant that the tests that
4641 showed this size failed. As the block is currently always of a fixed size,
4642 this information isn't actually particularly useful in pcretest output, so
4643 I have just removed it.
4645 8. Three pre-processor statements accidentally did not start in column 1.
4646 Sadly, there are *still* compilers around that complain, even though
4647 standard C has not required this for well over a decade. Sigh.
4649 9. In pcretest, the code for checking callouts passed small integers in the
4650 callout_data field, which is a void * field. However, some picky compilers
4651 complained about the casts involved for this on 64-bit systems. Now
4652 pcretest passes the address of the small integer instead, which should get
4653 rid of the warnings.
4655 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4656 both compile and run time, and gives an error if an invalid UTF-8 sequence
4657 is found. There is a option for disabling this check in cases where the
4658 string is known to be correct and/or the maximum performance is wanted.
4660 11. In response to a bug report, I changed one line in Makefile.in from
4662 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4663 to
4664 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4666 to look similar to other lines, but I have no way of telling whether this
4667 is the right thing to do, as I do not use Windows. No doubt I'll get told
4668 if it's wrong...
4671 Version 4.3 21-May-03
4672 ---------------------
4674 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4675 Makefile.
4677 2. Some refactoring to improve the quality of the code:
4679 (i) The utf8_table... variables are now declared "const".
4681 (ii) The code for \cx, which used the "case flipping" table to upper case
4682 lower case letters, now just substracts 32. This is ASCII-specific,
4683 but the whole concept of \cx is ASCII-specific, so it seems
4684 reasonable.
4686 (iii) PCRE was using its character types table to recognize decimal and
4687 hexadecimal digits in the pattern. This is silly, because it handles
4688 only 0-9, a-f, and A-F, but the character types table is locale-
4689 specific, which means strange things might happen. A private
4690 table is now used for this - though it costs 256 bytes, a table is
4691 much faster than multiple explicit tests. Of course, the standard
4692 character types table is still used for matching digits in subject
4693 strings against \d.
4695 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4696 ending in _t are). So I've renamed it as ESC_tee.
4698 3. The first argument for regexec() in the POSIX wrapper should have been
4699 defined as "const".
4701 4. Changed pcretest to use malloc() for its buffers so that they can be
4702 Electric Fenced for debugging.
4704 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4705 to read one or more bytes before the start of the subject string. Often this
4706 had no effect on PCRE's behaviour, but in some circumstances it could
4707 provoke a segmentation fault.
4709 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4710 to try to read one or more bytes before the start of the subject string.
4712 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4713 UTF-8 support could misbehave in various ways if the subject string
4714 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4715 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4716 back over UTF-8 characters.)
4719 Version 4.2 14-Apr-03
4720 ---------------------
4722 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4724 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4725 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4726 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4727 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4728 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4730 Note: automatic setting of the BUILD variables is not yet working
4731 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4732 compile-time but not at link-time
4733 [LINK]: use for linking executables only
4734 make different versions for Windows and non-Windows
4735 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4736 libraries
4737 [LINK_FOR_BUILD]: new variable
4738 [OBJEXT]: use throughout
4739 [EXEEXT]: use throughout
4740 <winshared>: new target
4741 <wininstall>: new target
4742 <dftables.o>: use native compiler
4743 <dftables>: use native linker
4744 <install>: handle Windows platform correctly
4745 <clean>: ditto
4746 <check>: ditto
4747 copy DLL to top builddir before testing
4749 As part of these changes, -no-undefined was removed again. This was reported
4750 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4751 in any case.
4753 3. Some tidies to get rid of compiler warnings:
4755 . In the match_data structure, match_limit was an unsigned long int, whereas
4756 match_call_count was an int. I've made them both unsigned long ints.
4758 . In pcretest the fact that a const uschar * doesn't automatically cast to
4759 a void * provoked a warning.
4761 . Turning on some more compiler warnings threw up some "shadow" variables
4762 and a few more missing casts.
4764 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4765 option, a class that contained a single character with a value between 128
4766 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4768 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4769 option, a class that contained several characters, but with at least one
4770 whose value was between 128 and 255 caused PCRE to crash.
4773 Version 4.1 12-Mar-03
4774 ---------------------
4776 1. Compiling with gcc -pedantic found a couple of places where casts were
4777 needed, and a string in dftables.c that was longer than standard compilers are
4778 required to support.
4780 2. Compiling with Sun's compiler found a few more places where the code could
4781 be tidied up in order to avoid warnings.
4783 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4784 first of these names is deprecated in the latest Autoconf in favour of the name
4785 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4786 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4787 analogy I have changed it to CFLAGS_FOR_BUILD.
4789 4. Added -no-undefined to the linking command in the Makefile, because this is
4790 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4791 linking step for the pcreposix library.
4793 5. PCRE was failing to diagnose the case of two named groups with the same
4794 name.
4796 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4797 literal character that is needed in the subject for a match, and scans along to
4798 ensure that it is present before embarking on the full matching process. This
4799 saves time in cases of nested unlimited repeats that are never going to match.
4800 Problem: the scan can take a lot of time if the subject is very long (e.g.
4801 megabytes), thus penalizing straightforward matches. It is now done only if the
4802 amount of subject to be scanned is less than 1000 bytes.
4804 7. A lesser problem with the same optimization is that it was recording the
4805 first character of an anchored pattern as "needed", thus provoking a search
4806 right along the subject, even when the first match of the pattern was going to
4807 fail. The "needed" character is now not set for anchored patterns, unless it
4808 follows something in the pattern that is of non-fixed length. Thus, it still
4809 fulfils its original purpose of finding quick non-matches in cases of nested
4810 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4813 Version 4.0 17-Feb-03
4814 ---------------------
4816 1. If a comment in an extended regex that started immediately after a meta-item
4817 extended to the end of string, PCRE compiled incorrect data. This could lead to
4818 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4820 2. Moved to autoconf 2.53 and libtool 1.4.2.
4822 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4823 the special perltest8 script is no longer needed - all the tests can be run
4824 from a single perltest script.
4826 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4827 by \s. It has now been removed in PCRE. This means it isn't recognized as
4828 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4829 class [:space:] *does* include VT, thereby creating a mess.
4831 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4832 space and tab.
4834 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4835 its new features into the main test script, reducing the number of scripts.
4837 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4838 were backward compatible, and made the (?i) apply to the whole pattern, as if
4839 /i were given. Now it behaves more logically, and applies the option setting
4840 only to what follows. PCRE has been changed to follow suit. However, if it
4841 finds options settings right at the start of the pattern, it extracts them into
4842 the global options, as before. Thus, they show up in the info data.
4844 8. Added support for the \Q...\E escape sequence. Characters in between are
4845 treated as literals. This is slightly different from Perl in that $ and @ are
4846 also handled as literals inside the quotes. In Perl, they will cause variable
4847 interpolation. Note the following examples:
4849 Pattern PCRE matches Perl matches
4851 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4852 \Qabc\$xyz\E abc\$xyz abc\$xyz
4853 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4855 For compatibility with Perl, \Q...\E sequences are recognized inside character
4856 classes as well as outside them.
4858 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4859 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4860 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4861 signed/unsigned warnings.
4863 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4864 option for pcretest, so I've replaced it by a simple function that does just
4865 that job.
4867 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4868 "pcregrep -".
4870 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4871 Java package. This provides some syntactic sugar for simple cases of what my
4872 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4873 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4874 item, you can use this simplified notation. Note that only makes sense with
4875 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4876 greediness, whatever the setting of the PCRE_UNGREEDY option.
4878 13. A change of greediness default within a pattern was not taking effect at
4879 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4880 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4881 was abstracted outside.
4883 14. PCRE now supports the \G assertion. It is true when the current matching
4884 position is at the start point of the match. This differs from \A when the
4885 starting offset is non-zero. Used with the /g option of pcretest (or similar
4886 code), it works in the same way as it does for Perl's /g option. If all
4887 alternatives of a regex begin with \G, the expression is anchored to the start
4888 match position, and the "anchored" flag is set in the compiled expression.
4890 15. Some bugs concerning the handling of certain option changes within patterns
4891 have been fixed. These applied to options other than (?ims). For example,
4892 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4893 been the other way round. Some of this was related to change 7 above.
4895 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4896 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4897 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4898 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4900 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4901 mode. Unlike ".", it always matches newline, whatever the setting of
4902 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4903 assertions. Perl allows it, but it doesn't (in general) work because it can't
4904 calculate the length of the lookbehind. At least, that's the case for Perl
4905 5.8.0 - I've been told they are going to document that it doesn't work in
4906 future.
4908 18. Added an error diagnosis for escapes that PCRE does not support: these are
4909 \L, \l, \N, \P, \p, \U, \u, and \X.
4911 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4912 reading past the end of the pattern in cases such as /[abcd/.
4914 20. PCRE was getting more memory than necessary for patterns with classes that
4915 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4917 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4918 compiling PCRE for use with Virtual Pascal.
4920 22. Small fix to the Makefile to make it work properly if the build is done
4921 outside the source tree.
4923 23. Added a new extension: a condition to go with recursion. If a conditional
4924 subpattern starts with (?(R) the "true" branch is used if recursion has
4925 happened, whereas the "false" branch is used only at the top level.
4927 24. When there was a very long string of literal characters (over 255 bytes
4928 without UTF support, over 250 bytes with UTF support), the computation of how
4929 much memory was required could be incorrect, leading to segfaults or other
4930 strange effects.
4932 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4933 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4934 there was a subsequent back reference to those brackets. This meant that, for
4935 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4936 possible to check for precisely this case. All we can do is abandon the
4937 optimization if .* occurs inside capturing brackets when there are any back
4938 references whatsoever. (See below for a better fix that came later.)
4940 26. The handling of the optimization for finding the first character of a
4941 non-anchored pattern, and for finding a character that is required later in the
4942 match were failing in some cases. This didn't break the matching; it just
4943 failed to optimize when it could. The way this is done has been re-implemented.
4945 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4947 28. Added a new feature that provides some of the functionality that Perl
4948 provides with (?{...}). The facility is termed a "callout". The way it is done
4949 in PCRE is for the caller to provide an optional function, by setting
4950 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4951 global variable. By default it is unset, which disables all calling out. To get
4952 the function called, the regex must include (?C) at appropriate points. This
4953 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4954 This provides a means of identifying different callout points. When PCRE
4955 reaches such a point in the regex, if pcre_callout has been set, the external
4956 function is called. It is provided with data in a structure called
4957 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4958 matching continues; if it returns a non-zero value, the match at the current
4959 point fails. However, backtracking will occur if possible. [This was changed
4960 later and other features added - see item 49 below.]
4962 29. pcretest is upgraded to test the callout functionality. It provides a
4963 callout function that displays information. By default, it shows the start of
4964 the match and the current position in the text. There are some new data escapes
4965 to vary what happens:
4967 \C+ in addition, show current contents of captured substrings
4968 \C- do not supply a callout function
4969 \C!n return 1 when callout number n is reached
4970 \C!n!m return 1 when callout number n is reached for the mth time
4972 30. If pcregrep was called with the -l option and just a single file name, it
4973 output "<stdin>" if a match was found, instead of the file name.
4975 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4976 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4977 pcre_exec(). This saves a malloc/free per call. The default value of
4978 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4979 when configuring.
4981 32. The default maximum size of a compiled pattern is 64K. There have been a
4982 few cases of people hitting this limit. The code now uses macros to handle the
4983 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4984 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4985 configuring. Tests 2 and 5 work only with 2-byte links because they output
4986 debugging information about compiled patterns.
4988 33. Internal code re-arrangements:
4990 (a) Moved the debugging function for printing out a compiled regex into
4991 its own source file (printint.c) and used #include to pull it into
4992 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4993 separate copies.
4995 (b) Defined the list of op-code names for debugging as a macro in
4996 internal.h so that it is next to the definition of the opcodes.
4998 (c) Defined a table of op-code lengths for simpler skipping along compiled
4999 code. This is again a macro in internal.h so that it is next to the
5000 definition of the opcodes.
5002 34. Added support for recursive calls to individual subpatterns, along the
5003 lines of Robin Houston's patch (but implemented somewhat differently).
5005 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
5006 allow it to read and process whole directories in Win32. This code was
5007 contributed by Lionel Fourquaux; it has not been tested by me.
5009 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
5010 used to name a group. Names consist of alphanumerics and underscores, and must
5011 be unique. Back references use the syntax (?P=name) and recursive calls use
5012 (?P>name) which is a PCRE extension to the Python extension. Groups still have
5013 numbers. The function pcre_fullinfo() can be used after compilation to extract
5014 a name/number map. There are three relevant calls:
5016 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
5017 PCRE_INFO_NAMECOUNT yields the number of entries
5018 PCRE_INFO_NAMETABLE yields a pointer to the map.
5020 The map is a vector of fixed-size entries. The size of each entry depends on
5021 the length of the longest name used. The first two bytes of each entry are the
5022 group number, most significant byte first. There follows the corresponding
5023 name, zero terminated. The names are in alphabetical order.
5025 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
5026 case instead of 255. Making it the same both with and without UTF-8 support
5027 means that the same test output works with both.
5029 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
5030 calling malloc() with a zero argument.
5032 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
5033 optimization. I've improved things by keeping a bitmap of backreferences with
5034 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
5035 fact referenced, the optimization can be applied. It is unlikely that a
5036 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
5037 the match to follow \n) will appear inside brackets with a number greater than
5038 31, but if it does, any back reference > 31 suppresses the optimization.
5040 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
5041 of disabling numbered capturing parentheses. Any opening parenthesis that is
5042 not followed by ? behaves as if it were followed by ?: but named parentheses
5043 can still be used for capturing (and they will acquire numbers in the usual
5044 way).
5046 41. Redesigned the return codes from the match() function into yes/no/error so
5047 that errors can be passed back from deep inside the nested calls. A malloc
5048 failure while inside a recursive subpattern call now causes the
5049 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
5051 42. It is now possible to set a limit on the number of times the match()
5052 function is called in a call to pcre_exec(). This facility makes it possible to
5053 limit the amount of recursion and backtracking, though not in a directly
5054 obvious way, because the match() function is used in a number of different
5055 circumstances. The count starts from zero for each position in the subject
5056 string (for non-anchored patterns). The default limit is, for compatibility, a
5057 large number, namely 10 000 000. You can change this in two ways:
5059 (a) When configuring PCRE before making, you can use --with-match-limit=n
5060 to set a default value for the compiled library.
5062 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
5063 a different value is set. See 45 below.
5065 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
5067 43. Added a new function pcre_config(int, void *) to enable run-time extraction
5068 of things that can be changed at compile time. The first argument specifies
5069 what is wanted and the second points to where the information is to be placed.
5070 The current list of available information is:
5074 The output is an integer that is set to one if UTF-8 support is available;
5075 otherwise it is set to zero.
5079 The output is an integer that it set to the value of the code that is used for
5080 newline. It is either LF (10) or CR (13).
5084 The output is an integer that contains the number of bytes used for internal
5085 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
5089 The output is an integer that contains the threshold above which the POSIX
5090 interface uses malloc() for output vectors. See item 31 above.
5094 The output is an unsigned integer that contains the default limit of the number
5095 of match() calls in a pcre_exec() execution. See 42 above.
5097 44. pcretest has been upgraded by the addition of the -C option. This causes it
5098 to extract all the available output from the new pcre_config() function, and to
5099 output it. The program then exits immediately.
5101 45. A need has arisen to pass over additional data with calls to pcre_exec() in
5102 order to support additional features. One way would have been to define
5103 pcre_exec2() (for example) with extra arguments, but this would not have been
5104 extensible, and would also have required all calls to the original function to
5105 be mapped to the new one. Instead, I have chosen to extend the mechanism that
5106 is used for passing in "extra" data from pcre_study().
5108 The pcre_extra structure is now exposed and defined in pcre.h. It currently
5109 contains the following fields:
5111 flags a bitmap indicating which of the following fields are set
5112 study_data opaque data from pcre_study()
5113 match_limit a way of specifying a limit on match() calls for a specific
5114 call to pcre_exec()
5115 callout_data data for callouts (see 49 below)
5117 The flag bits are also defined in pcre.h, and are
5123 The pcre_study() function now returns one of these new pcre_extra blocks, with
5124 the actual study data pointed to by the study_data field, and the
5125 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
5126 before. That is, this change is entirely upwards-compatible and requires no
5127 change to existing code.
5129 If you want to pass in additional data to pcre_exec(), you can either place it
5130 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5131 block.
5133 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5134 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5135 times with different match limits, until it finds the minimum value needed for
5136 pcre_exec() to complete. The value is then output. This can be instructive; for
5137 most simple matches the number is quite small, but for pathological cases it
5138 gets very large very quickly.
5140 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5141 returns the size of the data block pointed to by the study_data field in a
5142 pcre_extra block, that is, the value that was passed as the argument to
5143 pcre_malloc() when PCRE was getting memory in which to place the information
5144 created by pcre_study(). The fourth argument should point to a size_t variable.
5145 pcretest has been extended so that this information is shown after a successful
5146 pcre_study() call when information about the compiled regex is being displayed.
5148 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5149 because what follows is always an absolute path. (Later: it turns out that this
5150 is more than cosmetic for MinGW, because it doesn't like empty path
5151 components.)
5153 49. Some changes have been made to the callout feature (see 28 above):
5155 (i) A callout function now has three choices for what it returns:
5157 0 => success, carry on matching
5158 > 0 => failure at this point, but backtrack if possible
5159 < 0 => serious error, return this value from pcre_exec()
5161 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5162 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5163 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5164 use by callout functions. It will never be used by PCRE itself.
5166 (ii) The pcre_extra structure (see 45 above) has a void * field called
5167 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5168 pcre_callout_block structure has a field of the same name. The contents of
5169 the field passed in the pcre_extra structure are passed to the callout
5170 function in the corresponding field in the callout block. This makes it
5171 easier to use the same callout-containing regex from multiple threads. For
5172 testing, the pcretest program has a new data escape
5174 \C*n pass the number n (may be negative) as callout_data
5176 If the callout function in pcretest receives a non-zero value as
5177 callout_data, it returns that value.
5179 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5180 there were some redundant $(CFLAGS) in commands that are now specified as
5181 $(LINK), which already includes $(CFLAGS).
5183 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5184 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5185 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5186 one-byte characters throughout. Note that case-insensitive matching applies
5187 only to characters whose values are less than 256. PCRE doesn't support the
5188 notion of cases for higher-valued characters.
5190 (i) A character class whose characters are all within 0-255 is handled as
5191 a bit map, and the map is inverted for negative classes. Previously, a
5192 character > 255 always failed to match such a class; however it should
5193 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5195 (ii) A negated character class with a single character < 255 is coded as
5196 "not this character" (OP_NOT). This wasn't working properly when the test
5197 character was multibyte, either singly or repeated.
5199 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5200 mode, for example: \x{100}{2,3}.
5202 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5203 singly or repeated) now correctly test multibyte characters. However,
5204 PCRE doesn't recognize any characters with values greater than 255 as
5205 digits, spaces, or word characters. Such characters always match \D, \S,
5206 and \W, and never match \d, \s, or \w.
5208 (v) Classes may now contain characters and character ranges with values
5209 greater than 255. For example: [ab\x{100}-\x{400}].
5211 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5212 PCRE in UTF-8 mode.
5214 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5215 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5216 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5217 value.)
5219 53. The single man page has become too large. I have therefore split it up into
5220 a number of separate man pages. These also give rise to individual HTML pages;
5221 these are now put in a separate directory, and there is an index.html page that
5222 lists them all. Some hyperlinking between the pages has been installed.
5224 54. Added convenience functions for handling named capturing parentheses.
5226 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5227 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5228 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5229 are faulted.
5231 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5232 calling configure. These values are used when compiling the dftables.c program
5233 which is run to generate the source of the default character tables. They
5234 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5235 you will need to set these values.
5237 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5240 Version 3.9 02-Jan-02
5241 ---------------------
5243 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5245 2. If --disable-static was given, the building process failed when trying to
5246 build pcretest and pcregrep. (For some reason it was using libtool to compile
5247 them, which is not right, as they aren't part of the library.)
5250 Version 3.8 18-Dec-01
5251 ---------------------
5253 1. The experimental UTF-8 code was completely screwed up. It was packing the
5254 bytes in the wrong order. How dumb can you get?
5257 Version 3.7 29-Oct-01
5258 ---------------------
5260 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5261 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5262 this didn't happen under Solaris 8, where I normally test things.
5264 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5265 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5266 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5267 this fix an hour or so after the initial 3.7 release.)
5270 Version 3.6 23-Oct-01
5271 ---------------------
5273 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5274 offsets passed as NULL with zero offset count.
5276 2. The config.guess and config.sub files had not been updated when I moved to
5277 the latest autoconf.
5280 Version 3.5 15-Aug-01
5281 ---------------------
5283 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5284 had been forgotten.
5286 2. By using declared but undefined structures, we can avoid using "void"
5287 definitions in pcre.h while keeping the internal definitions of the structures
5288 private.
5290 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5291 user point of view, this means that both static and shared libraries are built
5292 by default, but this can be individually controlled. More of the work of
5293 handling this static/shared cases is now inside libtool instead of PCRE's make
5294 file.
5296 4. The pcretest utility is now installed along with pcregrep because it is
5297 useful for users (to test regexs) and by doing this, it automatically gets
5298 relinked by libtool. The documentation has been turned into a man page, so
5299 there are now .1, .txt, and .html versions in /doc.
5301 5. Upgrades to pcregrep:
5302 (i) Added long-form option names like gnu grep.
5303 (ii) Added --help to list all options with an explanatory phrase.
5304 (iii) Added -r, --recursive to recurse into sub-directories.
5305 (iv) Added -f, --file to read patterns from a file.
5307 6. pcre_exec() was referring to its "code" argument before testing that
5308 argument for NULL (and giving an error if it was NULL).
5310 7. Upgraded Makefile.in to allow for compiling in a different directory from
5311 the source directory.
5313 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5314 options bits, the pointer it was passed was to an int instead of to an unsigned
5315 long int. This mattered only on 64-bit systems.
5317 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5318 generated) instead of pcre.in, which it its source. Also made the same change
5319 in several of the .c files.
5321 10. A new release of gcc defines printf() as a macro, which broke pcretest
5322 because it had an ifdef in the middle of a string argument for printf(). Fixed
5323 by using separate calls to printf().
5325 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5326 script, to force use of CR or LF instead of \n in the source. On non-Unix
5327 systems, the value can be set in config.h.
5329 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5330 absolute limit. Changed the text of the error message to make this clear, and
5331 likewise updated the man page.
5333 13. The limit of 99 on the number of capturing subpatterns has been removed.
5334 The new limit is 65535, which I hope will not be a "real" limit.
5337 Version 3.4 22-Aug-00
5338 ---------------------
5340 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5342 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5345 Version 3.3 01-Aug-00
5346 ---------------------
5348 1. If an octal character was given, but the value was greater than \377, it
5349 was not getting masked to the least significant bits, as documented. This could
5350 lead to crashes in some systems.
5352 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5353 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5355 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5356 These just pass their arguments on to (pcre_free)(), but they are provided
5357 because some uses of PCRE bind it to non-C systems that can call its functions,
5358 but cannot call free() or pcre_free() directly.
5360 4. Add "make test" as a synonym for "make check". Corrected some comments in
5361 the Makefile.
5363 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5364 Makefile.
5366 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5367 command called pgrep for grepping around the active processes.
5369 7. Added the beginnings of support for UTF-8 character strings.
5371 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5372 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5373 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5374 out for the ar command.)
5377 Version 3.2 12-May-00
5378 ---------------------
5380 This is purely a bug fixing release.
5382 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5383 of ZA. This was just one example of several cases that could provoke this bug,
5384 which was introduced by change 9 of version 2.00. The code for breaking
5385 infinite loops after an iteration that matches an empty string was't working
5386 correctly.
5388 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5389 when matched against abbab (for example). After matching an empty string, it
5390 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5391 caused it to match further down the string than it should.
5393 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5394 was there because it doesn't seem to be needed, and it causes trouble on some
5395 systems, as it is not a Standard C header. It has been removed.
5397 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5398 were reported on the Macintosh. The changes were from
5400 while ((c = *(++ptr)) != 0 && c != '\n');
5401 to
5402 while ((c = *(++ptr)) != 0 && c != '\n') ;
5404 Totally extraordinary, but if that's what it takes...
5406 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5407 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5408 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5409 assumes the way PCRE uses memmove() (always moving upwards).
5411 6. PCRE is being used in one environment where strchr() is not available. There
5412 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5413 faster code anyway.
5416 Version 3.1 09-Feb-00
5417 ---------------------
5419 The only change in this release is the fixing of some bugs in Makefile.in for
5420 the "install" target:
5422 (1) It was failing to install pcreposix.h.
5424 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5427 Version 3.0 01-Feb-00
5428 ---------------------
5430 1. Add support for the /+ modifier to perltest (to output $` like it does in
5431 pcretest).
5433 2. Add support for the /g modifier to perltest.
5435 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5436 matches null strings.
5438 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5439 pattern. Perl treats empty patterns specially - it reuses the most recent
5440 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5441 effect.
5443 5. The POSIX interface was broken in that it was just handing over the POSIX
5444 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5445 required a bigger vector, with some working space on the end. This means that
5446 the POSIX wrapper now has to get and free some memory, and copy the results.
5448 6. Added some simple autoconf support, placing the test data and the
5449 documentation in separate directories, re-organizing some of the
5450 information files, and making it build pcre-config (a GNU standard). Also added
5451 libtool support for building PCRE as a shared library, which is now the
5452 default.
5454 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5455 09 are not valid octal constants. Single digits will be used for minor values
5456 less than 10.
5458 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5459 existing programs that set these in the POSIX interface can use PCRE without
5460 modification.
5462 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5463 return all that pcre_info() returns, plus additional data. The pcre_info()
5464 function is retained for compatibility, but is considered to be obsolete.
5466 10. Added experimental recursion feature (?R) to handle one common case that
5467 Perl 5.6 will be able to do with (?p{...}).
5469 11. Added support for POSIX character classes like [:alpha:], which Perl is
5470 adopting.
5473 Version 2.08 31-Aug-99
5474 ----------------------
5476 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5477 trying to match at the startoffset position, but instead was moving forward to
5478 the next newline as if a previous match had failed.
5480 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5481 and could get into a loop if a null string was matched other than at the start
5482 of the subject.
5484 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5485 be distinguished at compile time, and for completeness also added PCRE_DATE.
5487 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5488 in GnuWin32 environments.
5491 Version 2.07 29-Jul-99
5492 ----------------------
5494 1. The documentation is now supplied in plain text form and HTML as well as in
5495 the form of man page sources.
5497 2. C++ compilers don't like assigning (void *) values to other pointer types.
5498 In particular this affects malloc(). Although there is no problem in Standard
5499 C, I've put in casts to keep C++ compilers happy.
5501 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5502 should be (const char *).
5504 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5505 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5506 However, I haven't made this a standard facility. The documentation doesn't
5507 mention it, and the Makefile doesn't support it.
5509 5. The Makefile now contains an "install" target, with editable destinations at
5510 the top of the file. The pcretest program is not installed.
5512 6. pgrep -V now gives the PCRE version number and date.
5514 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5515 causing the entire string to be ignored, instead of just the last character.
5517 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5518 non-matching string, it can take a very, very long time, even for strings of
5519 quite modest length, because of the nested recursion. PCRE now does better in
5520 some of these cases. It does this by remembering the last required literal
5521 character in the pattern, and pre-searching the subject to ensure it is present
5522 before running the real match. In other words, it applies a heuristic to detect
5523 some types of certain failure quickly, and in the above example, if presented
5524 with a string that has no trailing " it gives "no match" very quickly.
5526 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5527 other alternatives are tried instead.
5530 Version 2.06 09-Jun-99
5531 ----------------------
5533 1. Change pcretest's output for amount of store used to show just the code
5534 space, because the remainder (the data block) varies in size between 32-bit and
5535 64-bit systems.
5537 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5538 start matching at. This allows lookbehinds to work when searching for multiple
5539 occurrences in a string.
5541 3. Added additional options to pcretest for testing multiple occurrences:
5543 /+ outputs the rest of the string that follows a match
5544 /g loops for multiple occurrences, using the new startoffset argument
5545 /G loops for multiple occurrences by passing an incremented pointer
5547 4. PCRE wasn't doing the "first character" optimization for patterns starting
5548 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5549 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5550 the letter 'x'. On long subject strings, this gives a significant speed-up.
5553 Version 2.05 21-Apr-99
5554 ----------------------
5556 1. Changed the type of magic_number from int to long int so that it works
5557 properly on 16-bit systems.
5559 2. Fixed a bug which caused patterns starting with .* not to work correctly
5560 when the subject string contained newline characters. PCRE was assuming
5561 anchoring for such patterns in all cases, which is not correct because .* will
5562 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5563 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5564 must be retried after every newline in the subject.
5567 Version 2.04 18-Feb-99
5568 ----------------------
5570 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5571 computation of the store needed to hold the pattern was incorrect (too large).
5572 If such patterns were nested a few deep, this could multiply and become a real
5573 problem.
5575 2. Added /M option to pcretest to show the memory requirement of a specific
5576 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5578 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5579 compiled in such a way that the backtracking after subsequent failure was
5580 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5581 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5584 Version 2.03 02-Feb-99
5585 ----------------------
5587 1. Fixed typo and small mistake in man page.
5589 2. Added 4th condition (GPL supersedes if conflict) and created separate
5590 LICENCE file containing the conditions.
5592 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5593 Perl, that is the internal \ allows the delimiter to be included in the
5594 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5595 the final delimiter, add \ to the end of the pattern (to test the error).
5597 4. Added the convenience functions for extracting substrings after a successful
5598 match. Updated pcretest to make it able to test these functions.
5601 Version 2.02 14-Jan-99
5602 ----------------------
5604 1. Initialized the working variables associated with each extraction so that
5605 their saving and restoring doesn't refer to uninitialized store.
5607 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5608 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5609 fix the problem.
5611 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5612 calls, and wasn't printing the correct value for compiling calls. Increased the
5613 default value of LOOPREPEAT, and the number of significant figures in the
5614 times.
5616 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5618 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5619 a building problem on Windows NT with a FAT file system.
5622 Version 2.01 21-Oct-98
5623 ----------------------
5625 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5626 to character tables built by pcre_maketables() in the current locale. If NULL
5627 is passed, the default tables are used.
5630 Version 2.00 24-Sep-98
5631 ----------------------
5633 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5634 it any more.
5636 2. Allow quantification of (?>) groups, and make it work correctly.
5638 3. The first character computation wasn't working for (?>) groups.
5640 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5641 end of the subject) and add 5.005's \z, which really does match only at the
5642 very end of the subject.
5644 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5646 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5647 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5648 localized options. All options to pcre_study() were also removed.
5650 7. Add other new features from 5.005:
5652 $(?<= positive lookbehind
5653 $(?<! negative lookbehind
5654 (?imsx-imsx) added the unsetting capability
5655 such a setting is global if at outer level; local otherwise
5656 (?imsx-imsx:) non-capturing groups with option setting
5657 (?(cond)re|re) conditional pattern matching
5659 A backreference to itself in a repeated group matches the previous
5660 captured string.
5662 8. General tidying up of studying (both automatic and via "study")
5663 consequential on the addition of new assertions.
5665 9. As in 5.005, unlimited repeated groups that could match an empty substring
5666 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5667 runtime if any iteration does actually match an empty substring.
5669 10. Include the RunTest script in the distribution.
5671 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5672 discrepancies, some of which were old and were also with respect to 5.004. They
5673 have now been fixed.
5676 Version 1.09 28-Apr-98
5677 ----------------------
5679 1. A negated single character class followed by a quantifier with a minimum
5680 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5681 program crashes, or just wrong answers. This did not apply to negated classes
5682 containing more than one character, or to minima other than one.
5685 Version 1.08 27-Mar-98
5686 ----------------------
5688 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5690 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5691 latter must appear before anything that relies on it in the pattern.
5694 Version 1.07 16-Feb-98
5695 ----------------------
5697 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5698 repeat of a potentially empty string).
5701 Version 1.06 23-Jan-98
5702 ----------------------
5704 1. Added Markus Oberhumer's little patches for C++.
5706 2. Literal strings longer than 255 characters were broken.
5709 Version 1.05 23-Dec-97
5710 ----------------------
5712 1. Negated character classes containing more than one character were failing if
5713 PCRE_CASELESS was set at run time.
5716 Version 1.04 19-Dec-97
5717 ----------------------
5719 1. Corrected the man page, where some "const" qualifiers had been omitted.
5721 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5722 input syntax.
5724 3. Fixed memory leak which occurred when a regex with back references was
5725 matched with an offsets vector that wasn't big enough. The temporary memory
5726 that is used in this case wasn't being freed if the match failed.
5728 4. Tidied pcretest to ensure it frees memory that it gets.
5730 5. Temporary memory was being obtained in the case where the passed offsets
5731 vector was exactly big enough.
5733 6. Corrected definition of offsetof() from change 5 below.
5735 7. I had screwed up change 6 below and broken the rules for the use of
5736 setjmp(). Now fixed.
5739 Version 1.03 18-Dec-97
5740 ----------------------
5742 1. A erroneous regex with a missing opening parenthesis was correctly
5743 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5744 on some systems.
5746 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5747 it was reported that one broken compiler failed on the former because "code" is
5748 also an independent variable.
5750 3. The erroneous regex a[]b caused an array overrun reference.
5752 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5753 fail on data ending with that character. (It was going on too far, and checking
5754 the next character, typically a binary zero.) This was specific to the
5755 optimized code for single-character negative classes.
5757 5. Added a contributed patch from the TIN world which does the following:
5759 + Add an undef for memmove, in case the the system defines a macro for it.
5761 + Add a definition of offsetof(), in case there isn't one. (I don't know
5762 the reason behind this - offsetof() is part of the ANSI standard - but
5763 it does no harm).
5765 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5766 most of the places where whitespace preceded '#'. I have given up and
5767 allowed the remaining 2 cases to be at the margin.
5769 + Rename some variables in pcre to eliminate shadowing. This seems very
5770 pedantic, but does no harm, of course.
5772 6. Moved the call to setjmp() into its own function, to get rid of warnings
5773 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5775 7. Constructs such as \d{8,} were compiling into the equivalent of
5776 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5777 outcome, but in this particular case used more store than had been allocated,
5778 which caused the bug to be discovered because it threw up an internal error.
5780 8. The debugging code in both pcre and pcretest for outputting the compiled
5781 form of a regex was going wrong in the case of back references followed by
5782 curly-bracketed repeats.
5785 Version 1.02 12-Dec-97
5786 ----------------------
5788 1. Typos in pcre.3 and comments in the source fixed.
5790 2. Applied a contributed patch to get rid of places where it used to remove
5791 'const' from variables, and fixed some signed/unsigned and uninitialized
5792 variable warnings.
5794 3. Added the "runtest" target to Makefile.
5796 4. Set default compiler flag to -O2 rather than just -O.
5799 Version 1.01 19-Nov-97
5800 ----------------------
5802 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5803 like /([ab]*)*/, that is, for classes with more than one character in them.
5805 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5806 as /((?>a*))*/ (a PCRE_EXTRA facility).
5809 Version 1.00 18-Nov-97
5810 ----------------------
5812 1. Added compile-time macros to support systems such as SunOS4 which don't have
5813 memmove() or strerror() but have other things that can be used instead.
5815 2. Arranged that "make clean" removes the executables.
5818 Version 0.99 27-Oct-97
5819 ----------------------
5821 1. Fixed bug in code for optimizing classes with only one character. It was
5822 initializing a 32-byte map regardless, which could cause it to run off the end
5823 of the memory it had got.
5825 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5828 Version 0.98 22-Oct-97
5829 ----------------------
5831 1. Fixed bug in code for handling temporary memory usage when there are more
5832 back references than supplied space in the ovector. This could cause segfaults.
5835 Version 0.97 21-Oct-97
5836 ----------------------
5838 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5840 2. Optimized negated single characters not to use a bit map.
5842 3. Brought error texts together as macro definitions; clarified some of them;
5843 fixed one that was wrong - it said "range out of order" when it meant "invalid
5844 escape sequence".
5846 4. Changed some char * arguments to const char *.
5848 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5850 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5851 pcretest.
5854 Version 0.96 16-Oct-97
5855 ----------------------
5857 1. Added a simple "pgrep" utility to the distribution.
5859 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5860 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5861 where "ddd" means "one or more decimal digits".
5863 3. Fixed serious bug. If a pattern had a back reference, but the call to
5864 pcre_exec() didn't supply a large enough ovector to record the related
5865 identifying subpattern, the match always failed. PCRE now remembers the number
5866 of the largest back reference, and gets some temporary memory in which to save
5867 the offsets during matching if necessary, in order to ensure that
5868 backreferences always work.
5870 4. Increased the compatibility with Perl in a number of ways:
5872 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5873 to request this handling. The option can be set at compile or exec time.
5875 (b) $ matches before a terminating newline by default; an option
5876 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5877 mode). The option can be set at compile or exec time.
5879 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5880 the same as Perl's. If the decimal number it represents is less than 10
5881 or there aren't that many previous left capturing parentheses, an octal
5882 escape is read. Inside a character class, it's always an octal escape,
5883 even if it is a single digit.
5885 (d) An escaped but undefined alphabetic character is taken as a literal,
5886 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5887 escapes.
5889 (e) {0} is now permitted. (The previous item is removed from the compiled
5890 pattern).
5892 5. Changed all the names of code files so that the basic parts are no longer
5893 than 10 characters, and abolished the teeny "globals.c" file.
5895 6. Changed the handling of character classes; they are now done with a 32-byte
5896 bit map always.
5898 7. Added the -d and /D options to pcretest to make it possible to look at the
5899 internals of compilation without having to recompile pcre.
5902 Version 0.95 23-Sep-97
5903 ----------------------
5905 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5906 \x20 at the start of a run of normal characters. These were being treated as
5907 real characters, instead of the source characters being re-checked.
5910 Version 0.94 18-Sep-97
5911 ----------------------
5913 1. The functions are now thread-safe, with the caveat that the global variables
5914 containing pointers to malloc() and free() or alternative functions are the
5915 same for all threads.
5917 2. Get pcre_study() to generate a bitmap of initial characters for non-
5918 anchored patterns when this is possible, and use it if passed to pcre_exec().
5921 Version 0.93 15-Sep-97
5922 ----------------------
5924 1. /(b)|(:+)/ was computing an incorrect first character.
5926 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5927 but not actually doing anything yet.
5929 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5930 as Perl does (e.g. [-az] or [az-]).
5932 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5933 all possible positions.
5935 5. Split up into different modules to avoid including unneeded functions in a
5936 compiled binary. However, compile and exec are still in one module. The "study"
5937 function is split off.
5939 6. The character tables are now in a separate module whose source is generated
5940 by an auxiliary program - but can then be edited by hand if required. There are
5941 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5942 toupper() in the code.
5944 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5945 make them global. Abolish the function for setting them, as the caller can now
5946 set them directly.
5949 Version 0.92 11-Sep-97
5950 ----------------------
5952 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5953 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5955 2. Caseless matching was not working in character classes if the characters in
5956 the pattern were in upper case.
5958 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5960 4. Make PCRE_ANCHORED public and accept as a compile option.
5962 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5963 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5964 pass them.
5966 6. Give an error if bad option bits passed at compile or run time.
5968 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5969 pcretest to cause it to pass that flag.
5971 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5972 options, and the first character, if set.
5974 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5977 Version 0.91 10-Sep-97
5978 ----------------------
5980 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5981 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5983 2. PCRE was looping on encountering an indefinitely repeated back reference to
5984 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5985 Perl does - treats the match as successful.
5987 ****


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

  ViewVC Help
Powered by ViewVC 1.1.5