ViewVC logotype

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log

Revision 1579 - (show annotations)
Tue Jul 21 13:47:22 2015 UTC (4 years, 4 months ago) by ph10
File size: 281560 byte(s)
Fix "run for ever" bug for deeply nested [: sequences.
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 8.38 xx-xxx-xxxx
8 ------------------------
10 1. If a group that contained a recursive back reference also contained a
11 forward reference subroutine call followed by a non-forward-reference
12 subroutine call, for example /.((?2)(?R)\1)()/, pcre2_compile() failed to
13 compile correct code, leading to undefined behaviour or an internally
14 detected error. This bug was discovered by the LLVM fuzzer.
16 2. Quantification of certain items (e.g. atomic back references) could cause
17 incorrect code to be compiled when recursive forward references were
18 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
19 This bug was discovered by the LLVM fuzzer.
21 3. A repeated conditional group whose condition was a reference by name caused
22 a buffer overflow if there was more than one group with the given name.
23 This bug was discovered by the LLVM fuzzer.
25 4. A recursive back reference by name within a group that had the same name as
26 another group caused a buffer overflow. For example:
27 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
29 5. A forward reference by name to a group whose number is the same as the
30 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
31 a buffer overflow at compile time. This bug was discovered by the LLVM
32 fuzzer.
34 6. A lookbehind assertion within a set of mutually recursive subpatterns could
35 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
37 7. Another buffer overflow bug involved duplicate named groups with a
38 reference between their definition, with a group that reset capture
39 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
40 fixed by always allowing for more memory, even if not needed. (A proper fix
41 is implemented in PCRE2, but it involves more refactoring.)
43 8. There was no check for integer overflow in subroutine calls such as (?123).
45 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
46 being treated as a literal 'l' instead of causing an error.
48 10. There was a buffer overflow if pcre_exec() was called with an ovector of
49 size 1. This bug was found by american fuzzy lop.
51 11. If a non-capturing group containing a conditional group that could match
52 an empty string was repeated, it was not identified as matching an empty
53 string itself. For example: /^(?:(?(1)x|)+)+$()/.
55 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
56 \a and \e in test subject lines.
58 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
59 instead of the EBCDIC value.
61 14. The handling of \c in an EBCDIC environment has been revised so that it is
62 now compatible with the specification in Perl's perlebcdic page.
64 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
65 ASCII/Unicode. This has now been added to the list of characters that are
66 recognized as white space in EBCDIC.
68 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
69 error (correctly) when used outside a class, but did not give an error
70 within a class.
72 17. \h within a class was incorrectly compiled in EBCDIC environments.
74 18. A pattern with an unmatched closing parenthesis that contained a backward
75 assertion which itself contained a forward reference caused buffer
76 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
78 19. JIT should return with error when the compiled pattern requires more stack
79 space than the maximum.
81 20. A possessively repeated conditional group that could match an empty string,
82 for example, /(?(R))*+/, was incorrectly compiled.
84 21. Fix infinite recursion in the JIT compiler when certain patterns such as
85 /(?:|a|){100}x/ are analysed.
87 22. Some patterns with character classes involving [: and \\ were incorrectly
88 compiled and could cause reading from uninitialized memory or an incorrect
89 error diagnosis.
91 16. Pathological patterns containing many nested occurrences of [: caused
92 pcre_compile() to run for a very long time.
95 Version 8.37 28-April-2015
96 --------------------------
98 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
99 for those parentheses to be closed with whatever has been captured so far.
100 However, it was failing to mark any other groups between the hightest
101 capture so far and the currrent group as "unset". Thus, the ovector for
102 those groups contained whatever was previously there. An example is the
103 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
105 2. If an assertion condition was quantified with a minimum of zero (an odd
106 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
108 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
109 unrecognized modifier, a crash could occur.
111 4. An attempt to do global matching in pcretest with a zero-length ovector
112 caused a crash.
114 5. Fixed a memory leak during matching that could occur for a subpattern
115 subroutine call (recursive or otherwise) if the number of captured groups
116 that had to be saved was greater than ten.
118 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
119 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
120 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
122 7. A UTF pattern containing a "not" match of a non-ASCII character and a
123 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
125 8. When a pattern is compiled, it remembers the highest back reference so that
126 when matching, if the ovector is too small, extra memory can be obtained to
127 use instead. A conditional subpattern whose condition is a check on a
128 capture having happened, such as, for example in the pattern
129 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
130 setting the highest backreference number. This mattered only if pcre_exec()
131 was called with an ovector that was too small to hold the capture, and there
132 was no other kind of back reference (a situation which is probably quite
133 rare). The effect of the bug was that the condition was always treated as
134 FALSE when the capture could not be consulted, leading to a incorrect
135 behaviour by pcre_exec(). This bug has been fixed.
137 9. A reference to a duplicated named group (either a back reference or a test
138 for being set in a conditional) that occurred in a part of the pattern where
139 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
140 to be incorrectly calculated, leading to overwriting.
142 10. A mutually recursive set of back references such as (\2)(\1) caused a
143 segfault at study time (while trying to find the minimum matching length).
144 The infinite loop is now broken (with the minimum length unset, that is,
145 zero).
147 11. If an assertion that was used as a condition was quantified with a minimum
148 of zero, matching went wrong. In particular, if the whole group had
149 unlimited repetition and could match an empty string, a segfault was
150 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
151 assertions to be quantified, but not if they are being used as conditions,
152 so the above pattern is faulted by Perl. PCRE has now been changed so that
153 it also rejects such patterns.
155 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
156 failed to allow the zero-repeat case if pcre2_exec() was called with an
157 ovector too small to capture the group.
159 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
160 Red Hat Product Security:
162 (a) A crash if /K and /F were both set with the option to save the compiled
163 pattern.
165 (b) Another crash if the option to print captured substrings in a callout
166 was combined with setting a null ovector, for example \O\C+ as a subject
167 string.
169 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
170 forward reference repeated a large (but limited) number of times within a
171 repeated outer group that has a zero minimum quantifier, caused incorrect
172 code to be compiled, leading to the error "internal error:
173 previously-checked referenced subpattern not found" when an incorrect
174 memory address was read. This bug was reported as "heap overflow",
175 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
176 CVE-2015-2325.
178 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
179 call within a group that also contained a recursive back reference caused
180 incorrect code to be compiled. This bug was reported as "heap overflow",
181 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
182 number CVE-2015-2326.
184 24. Computing the size of the JIT read-only data in advance has been a source
185 of various issues, and new ones are still appear unfortunately. To fix
186 existing and future issues, size computation is eliminated from the code,
187 and replaced by on-demand memory allocation.
189 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
190 adjacent to the end of the range, and the range contained characters with
191 more than one other case, caused incorrect behaviour when compiled in UTF
192 mode. In that example, the range a-j was left out of the class.
194 26. Fix JIT compilation of conditional blocks, which assertion
195 is converted to (*FAIL). E.g: /(?(?!))/.
197 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
198 discovered by the LLVM fuzzer.
200 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
201 when this assertion was used as a condition, for example (?(?!)a|b). In
202 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
203 error about an unsupported item.
205 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
206 possessification code could take exponential time to complete. A recursion
207 depth limit of 1000 has been imposed to limit the resources used by this
208 optimization.
210 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
211 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
212 because \S ensures they are all in the class. The code for doing this was
213 interacting badly with the code for computing the amount of space needed to
214 compile the pattern, leading to a buffer overflow. This bug was discovered
215 by the LLVM fuzzer.
217 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
218 other kinds of group caused stack overflow at compile time. This bug was
219 discovered by the LLVM fuzzer.
221 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
222 between a subroutine call and its quantifier was incorrectly compiled,
223 leading to buffer overflow or other errors. This bug was discovered by the
224 LLVM fuzzer.
226 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
227 assertion after (?(. The code was failing to check the character after
228 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
229 was discovered by the LLVM fuzzer.
231 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
232 a fixed maximum following a group that contains a subroutine reference was
233 incorrectly compiled and could trigger buffer overflow. This bug was
234 discovered by the LLVM fuzzer.
236 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
237 caused a stack overflow instead of the diagnosis of a non-fixed length
238 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
240 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
241 (e.g. /(?<=\Ka)/) could make pcregrep loop.
243 37. There was a similar problem to 36 in pcretest for global matches.
245 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
246 and a subsequent item in the pattern caused a non-match, backtracking over
247 the repeated \X did not stop, but carried on past the start of the subject,
248 causing reference to random memory and/or a segfault. There were also some
249 other cases where backtracking after \C could crash. This set of bugs was
250 discovered by the LLVM fuzzer.
252 39. The function for finding the minimum length of a matching string could take
253 a very long time if mutual recursion was present many times in a pattern,
254 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
255 method has been implemented. This infelicity was discovered by the LLVM
256 fuzzer.
258 40. Static linking against the PCRE library using the pkg-config module was
259 failing on missing pthread symbols.
262 Version 8.36 26-September-2014
263 ------------------------------
265 1. Got rid of some compiler warnings in the C++ modules that were shown up by
266 -Wmissing-field-initializers and -Wunused-parameter.
268 2. The tests for quantifiers being too big (greater than 65535) were being
269 applied after reading the number, and stupidly assuming that integer
270 overflow would give a negative number. The tests are now applied as the
271 numbers are read.
273 3. Tidy code in pcre_exec.c where two branches that used to be different are
274 now the same.
276 4. The JIT compiler did not generate match limit checks for certain
277 bracketed expressions with quantifiers. This may lead to exponential
278 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
279 issue should be resolved now.
281 5. Fixed an issue, which occures when nested alternatives are optimized
282 with table jumps.
284 6. Inserted two casts and changed some ints to size_t in the light of some
285 reported 64-bit compiler warnings (Bugzilla 1477).
287 7. Fixed a bug concerned with zero-minimum possessive groups that could match
288 an empty string, which sometimes were behaving incorrectly in the
289 interpreter (though correctly in the JIT matcher). This pcretest input is
290 an example:
292 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
295 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
296 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
297 for an empty string was breaking the inner loop and carrying on at a lower
298 level, when possessive repeated groups should always return to a higher
299 level as they have no backtrack points in them. The empty string test now
300 occurs at the outer level.
302 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
303 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
305 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
306 doesn't).
308 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
309 to hold a pattern to be incorrectly computed (too small) when there were
310 named back references to duplicated names. This could cause "internal
311 error: code overflow" or "double free or corruption" or other memory
312 handling errors.
314 11. When named subpatterns had the same prefixes, back references could be
315 confused. For example, in this pattern:
317 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
319 the reference to 'Name' was incorrectly treated as a reference to a
320 duplicate name.
322 12. A pattern such as /^s?c/mi8 where the optional character has more than
323 one "other case" was incorrectly compiled such that it would only try to
324 match starting at "c".
326 13. When a pattern starting with \s was studied, VT was not included in the
327 list of possible starting characters; this should have been part of the
328 8.34/18 patch.
330 14. If a character class started [\Qx]... where x is any character, the class
331 was incorrectly terminated at the ].
333 15. If a pattern that started with a caseless match for a character with more
334 than one "other case" was studied, PCRE did not set up the starting code
335 unit bit map for the list of possible characters. Now it does. This is an
336 optimization improvement, not a bug fix.
338 16. The Unicode data tables have been updated to Unicode 7.0.0.
340 17. Fixed a number of memory leaks in pcregrep.
342 18. Avoid a compiler warning (from some compilers) for a function call with
343 a cast that removes "const" from an lvalue by using an intermediate
344 variable (to which the compiler does not object).
346 19. Incorrect code was compiled if a group that contained an internal recursive
347 back reference was optional (had quantifier with a minimum of zero). This
348 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
349 caused segmentation faults because of stack overflows at compile time.
351 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
352 group that is quantified with an indefinite repeat, caused a compile-time
353 loop which used up all the system stack and provoked a segmentation fault.
354 This was not the same bug as 19 above.
356 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
357 Patch by Mike Frysinger.
360 Version 8.35 04-April-2014
361 --------------------------
363 1. A new flag is set, when property checks are present in an XCLASS.
364 When this flag is not set, PCRE can perform certain optimizations
365 such as studying these XCLASS-es.
367 2. The auto-possessification of character sets were improved: a normal
368 and an extended character set can be compared now. Furthermore
369 the JIT compiler optimizes more character set checks.
371 3. Got rid of some compiler warnings for potentially uninitialized variables
372 that show up only when compiled with -O2.
374 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
375 of a match later then the end of the match. The pcretest program was not
376 handling the case sensibly - it was outputting from the start to the next
377 binary zero. It now reports this situation in a message, and outputs the
378 text from the end to the start.
380 5. Fast forward search is improved in JIT. Instead of the first three
381 characters, any three characters with fixed position can be searched.
382 Search order: first, last, middle.
384 6. Improve character range checks in JIT. Characters are read by an inprecise
385 function now, which returns with an unknown value if the character code is
386 above a certain threshold (e.g: 256). The only limitation is that the value
387 must be bigger than the threshold as well. This function is useful when
388 the characters above the threshold are handled in the same way.
390 7. The macros whose names start with RAWUCHAR are placeholders for a future
391 mode in which only the bottom 21 bits of 32-bit data items are used. To
392 make this more memorable for those maintaining the code, the names have
393 been changed to start with UCHAR21, and an extensive comment has been added
394 to their definition.
396 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
397 to the export list in Makefile.am (they were accidentally omitted from the
398 8.34 tarball).
400 9. The informational output from pcretest used the phrase "starting byte set"
401 which is inappropriate for the 16-bit and 32-bit libraries. As the output
402 for "first char" and "need char" really means "non-UTF-char", I've changed
403 "byte" to "char", and slightly reworded the output. The documentation about
404 these values has also been (I hope) clarified.
406 10. Another JIT related optimization: use table jumps for selecting the correct
407 backtracking path, when more than four alternatives are present inside a
408 bracket.
410 11. Empty match is not possible, when the minimum length is greater than zero,
411 and there is no \K in the pattern. JIT should avoid empty match checks in
412 such cases.
414 12. In a caseless character class with UCP support, when a character with more
415 than one alternative case was not the first character of a range, not all
416 the alternative cases were added to the class. For example, s and \x{17f}
417 are both alternative cases for S: the class [RST] was handled correctly,
418 but [R-T] was not.
420 13. The configure.ac file always checked for pthread support when JIT was
421 enabled. This is not used in Windows, so I have put this test inside a
422 check for the presence of windows.h (which was already tested for).
424 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
425 The algorithm provides a way to skip certain starting offsets, and usually
426 faster than linear prefix searches.
428 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
429 as for 'fr_FR' and 'french'. For some reason, however, it then used the
430 Windows-specific input and output files, which have 'french' screwed in.
431 So this could never have worked. One of the problems with locales is that
432 they aren't always the same. I have now updated RunTest so that it checks
433 the output of the locale test (test 3) against three different output
434 files, and it allows the test to pass if any one of them matches. With luck
435 this should make the test pass on some versions of Solaris where it was
436 failing. Because of the uncertainty, the script did not used to stop if
437 test 3 failed; it now does. If further versions of a French locale ever
438 come to light, they can now easily be added.
440 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
441 there was a message during ./configure, but it did not stop. This now
442 provokes an error. The invalid example in README has been corrected.
443 If a value less than the minimum is given, the minimum value has always
444 been used, but now a warning is given.
446 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
447 was a bug in the test system, which is now fixed. Also, the list of various
448 configurations that are tested for each release did not have one with both
449 16/32 bits and --enable-bar-anycrlf. It now does.
451 18. pcretest was missing "-C bsr" for displaying the \R default setting.
453 19. Little endian PowerPC systems are supported now by the JIT compiler.
455 20. The fast forward newline mechanism could enter to an infinite loop on
456 certain invalid UTF-8 input. Although we don't support these cases
457 this issue can be fixed by a performance optimization.
459 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
460 not take account if existing stack usage. There is now a new global
461 variable called pcre_stack_guard that can be set to point to an external
462 function to check stack availability. It is called at the start of
463 processing every parenthesized group.
465 22. A typo in the code meant that in ungreedy mode the max/min qualifier
466 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
467 not match "ab".
469 23. When UTF was disabled, the JIT program reported some incorrect compile
470 errors. These messages are silenced now.
472 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
473 compiler.
475 25. Change all the temporary files used in RunGrepTest to be different to those
476 used by RunTest so that the tests can be run simultaneously, for example by
477 "make -j check".
480 Version 8.34 15-December-2013
481 -----------------------------
483 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
484 executable memory. Patch inspired by Carsten Klein.
486 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
487 this macro is never tested and has no effect, because the work to support
488 coverage involves only compiling and linking options and special targets in
489 the Makefile. The comment in config.h implied that defining the macro would
490 enable coverage support, which is totally false. There was also support for
491 setting this macro in the CMake files (my fault, I just copied it from
492 configure). SUPPORT_GCOV has now been removed.
494 3. Make a small performance improvement in strlen16() and strlen32() in
495 pcretest.
497 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
498 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
499 these cases). There was also one in pcretest.c.
501 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
503 6. In UTF mode, the code for checking whether a group could match an empty
504 string (which is used for indefinitely repeated groups to allow for
505 breaking an infinite loop) was broken when the group contained a repeated
506 negated single-character class with a character that occupied more than one
507 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
508 UTF-8 mode). The effect was undefined: the group might or might not be
509 deemed as matching an empty string, or the program might have crashed.
511 7. The code for checking whether a group could match an empty string was not
512 recognizing that \h, \H, \v, \V, and \R must match a character.
514 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
515 an empty string. If it can, pcretest shows this in its information output.
517 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
518 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
519 matched by pcre_exec() without using JIT:
521 (a) If the rest of the pattern did not match after a maximal run of
522 grapheme clusters, the code for backing up to try with fewer of them
523 did not always back up over a full grapheme when characters that do not
524 have the modifier quality were involved, e.g. Hangul syllables.
526 (b) If the match point in a subject started with modifier character, and
527 there was no match, the code could incorrectly back up beyond the match
528 point, and potentially beyond the first character in the subject,
529 leading to a segfault or an incorrect match result.
531 10. A conditional group with an assertion condition could lead to PCRE
532 recording an incorrect first data item for a match if no other first data
533 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
534 first data item, and therefore matched "ca" after "c" instead of at the
535 start.
537 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
538 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
540 12. The source of pcregrep now includes z/OS-specific code so that it can be
541 compiled for z/OS as part of the special z/OS distribution.
543 13. Added the -T and -TM options to pcretest.
545 14. The code in pcre_compile.c for creating the table of named capturing groups
546 has been refactored. Instead of creating the table dynamically during the
547 actual compiling pass, the information is remembered during the pre-compile
548 pass (on the stack unless there are more than 20 named groups, in which
549 case malloc() is used) and the whole table is created before the actual
550 compile happens. This has simplified the code (it is now nearly 150 lines
551 shorter) and prepared the way for better handling of references to groups
552 with duplicate names.
554 15. A back reference to a named subpattern when there is more than one of the
555 same name now checks them in the order in which they appear in the pattern.
556 The first one that is set is used for the reference. Previously only the
557 first one was inspected. This change makes PCRE more compatible with Perl.
559 16. Unicode character properties were updated from Unicode 6.3.0.
561 17. The compile-time code for auto-possessification has been refactored, based
562 on a patch by Zoltan Herczeg. It now happens after instead of during
563 compilation. The code is cleaner, and more cases are handled. The option
564 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
565 options in pcretest are provided to set it. It can also be set by
566 (*NO_AUTO_POSSESS) at the start of a pattern.
568 18. The character VT has been added to the default ("C" locale) set of
569 characters that match \s and are generally treated as white space,
570 following this same change in Perl 5.18. There is now no difference between
571 "Perl space" and "POSIX space". Whether VT is treated as white space in
572 other locales depends on the locale.
574 19. The code for checking named groups as conditions, either for being set or
575 for being recursed, has been refactored (this is related to 14 and 15
576 above). Processing unduplicated named groups should now be as fast at
577 numerical groups, and processing duplicated groups should be faster than
578 before.
580 20. Two patches to the CMake build system, by Alexander Barkov:
582 (1) Replace the "source" command by "." in CMakeLists.txt because
583 "source" is a bash-ism.
585 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
586 without these the CMake build does not work on Solaris.
588 21. Perl has changed its handling of \8 and \9. If there is no previously
589 encountered capturing group of those numbers, they are treated as the
590 literal characters 8 and 9 instead of a binary zero followed by the
591 literals. PCRE now does the same.
593 22. Following Perl, added \o{} to specify codepoints in octal, making it
594 possible to specify values greater than 0777 and also making them
595 unambiguous.
597 23. Perl now gives an error for missing closing braces after \x{... instead of
598 treating the string as literal. PCRE now does the same.
600 24. RunTest used to grumble if an inappropriate test was selected explicitly,
601 but just skip it when running all tests. This make it awkward to run ranges
602 of tests when one of them was inappropriate. Now it just skips any
603 inappropriate tests, as it always did when running all tests.
605 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
606 character types such as \d or \w, too many callouts were inserted, and the
607 data that they returned was rubbish.
609 26. In UCP mode, \s was not matching two of the characters that Perl matches,
610 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
611 were matched by \h. The code has now been refactored so that the lists of
612 the horizontal and vertical whitespace characters used for \h and \v (which
613 are defined only in one place) are now also used for \s.
615 27. Add JIT support for the 64 bit TileGX architecture.
616 Patch by Jiong Wang (Tilera Corporation).
618 28. Possessive quantifiers for classes (both explicit and automatically
619 generated) now use special opcodes instead of wrapping in ONCE brackets.
621 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
622 (because it's meaningless), this was not happening when PCRE_CASELESS was
623 set. Not wrong, but inefficient.
625 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
626 properties for \w, \d, etc) is present in a test regex. Otherwise if the
627 test contains no characters greater than 255, Perl doesn't realise it
628 should be using Unicode semantics.
630 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
631 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
632 does in Unicode mode.
634 32. Added the "forbid" facility to pcretest so that putting tests into the
635 wrong test files can sometimes be quickly detected.
637 33. There is now a limit (default 250) on the depth of nesting of parentheses.
638 This limit is imposed to control the amount of system stack used at compile
639 time. It can be changed at build time by --with-parens-nest-limit=xxx or
640 the equivalent in CMake.
642 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
643 errors. Perl warns for these when in warning mode, but PCRE has no facility
644 for giving warnings.
646 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
647 However, this was not working for (?!) because it is optimized to (*FAIL),
648 for which PCRE does not allow quantifiers. The optimization is now disabled
649 when a quantifier follows (?!). I can't see any use for this, but it makes
650 things uniform.
652 36. Perl no longer allows group names to start with digits, so I have made this
653 change also in PCRE. It simplifies the code a bit.
655 37. In extended mode, Perl ignores spaces before a + that indicates a
656 possessive quantifier. PCRE allowed a space before the quantifier, but not
657 before the possessive +. It now does.
659 38. The use of \K (reset reported match start) within a repeated possessive
660 group such as (a\Kb)*+ was not working.
662 40. Document that the same character tables must be used at compile time and
663 run time, and that the facility to pass tables to pcre_exec() and
664 pcre_dfa_exec() is for use only with saved/restored patterns.
666 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
667 features for Builds with MSVC:
669 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
670 builds). The .rc files can be used to set FileDescription and many other
671 attributes.
673 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
674 This allows higher-level build scripts which want .pdb files to avoid
675 hard-coding the exact files needed."
677 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
678 mean "start of word" and "end of word", respectively, as a transition aid.
680 43. A minimizing repeat of a class containing codepoints greater than 255 in
681 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
682 compiled to use the heap for recursion.
684 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
685 is configured.
688 Version 8.33 28-May-2013
689 ------------------------
691 1. Added 'U' to some constants that are compared to unsigned integers, to
692 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
693 variables that are added to signed variables, to ensure the result is
694 signed and can be negated.
696 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
697 CMake config files.
699 3. Revise the creation of config.h.generic so that all boolean macros are
700 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
701 overriding via -D on the command line possible.
703 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
704 to unsigned int is reported to make a quite noticeable speed difference in
705 a specific Windows environment. Testing on Linux did also appear to show
706 some benefit (and it is clearly not harmful). Also fixed the definition of
707 Xop which should be unsigned.
709 5. Related to (4), changing the definition of the intermediate variable cc
710 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
711 improvements.
713 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
714 unnecessary spaces.
716 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
717 and later.
719 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
721 9. Optimizing fast_forward_start_bits in JIT.
723 10. Adding support for callouts in JIT, and fixing some issues revealed
724 during this work. Namely:
726 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
728 (b) Minimum length was not checked before the matching is started.
730 11. The value of capture_last that is passed to callouts was incorrect in some
731 cases when there was a capture on one path that was subsequently abandoned
732 after a backtrack. Also, the capture_last value is now reset after a
733 recursion, since all captures are also reset in this case.
735 12. The interpreter no longer returns the "too many substrings" error in the
736 case when an overflowing capture is in a branch that is subsequently
737 abandoned after a backtrack.
739 13. In the pathological case when an offset vector of size 2 is used, pcretest
740 now prints out the matched string after a yield of 0 or 1.
742 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
743 Only supported by the JIT compiler at the moment.
745 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
747 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
748 offset of the starting point of the matching process, provided the offsets
749 vector is large enough.
751 17. The \A escape now records a lookbehind value of 1, though its execution
752 does not actually inspect the previous character. This is to ensure that,
753 in partial multi-segment matching, at least one character from the old
754 segment is retained when a new segment is processed. Otherwise, if there
755 are no lookbehinds in the pattern, \A might match incorrectly at the start
756 of a new segment.
758 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
760 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
761 gives some modest performance improvement in 8-bit mode.
763 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
764 be expressed in certain programming languages using Universal Character
765 Names.
767 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
768 which points out that "non characters" are not "characters that may not
769 appear in Unicode strings" but rather "characters that are reserved for
770 internal use and have only local meaning".
772 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
773 there was a conditional group that depended on an assertion, if the
774 assertion was false, the callout that immediately followed the alternation
775 in the condition was skipped when pcre_exec() was used for matching.
777 23. Allow an explicit callout to be inserted before an assertion that is the
778 condition for a conditional group, for compatibility with automatic
779 callouts, which always insert a callout at this point.
781 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
782 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
784 25. (*PRUNE) is now supported by the JIT compiler.
786 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
788 27. Fix the case where there are two or more SKIPs with arguments that may be
789 ignored.
791 28. (*SKIP) is now supported by the JIT compiler.
793 29. (*THEN) is now supported by the JIT compiler.
795 30. Update RunTest with additional test selector options.
797 31. The way PCRE handles backtracking verbs has been changed in two ways.
799 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
800 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
801 backtracking. In some cases this makes it more Perl-compatible, but Perl's
802 rather obscure rules do not always do the same thing.
804 (2) Previously, backtracking verbs were confined within assertions. This is
805 no longer the case for positive assertions, except for (*ACCEPT). Again,
806 this sometimes improves Perl compatibility, and sometimes does not.
808 32. A number of tests that were in test 2 because Perl did things differently
809 have been moved to test 1, because either Perl or PCRE has changed, and
810 these tests are now compatible.
812 32. Backtracking control verbs are now handled in the same way in JIT and
813 interpreter.
815 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
816 contained a forward subroutine reference caused a compile error.
818 34. Auto-detect and optimize limited repetitions in JIT.
820 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
821 blocking (*UTF) etc.
823 36. In the interpreter, maximizing pattern repetitions for characters and
824 character types now use tail recursion, which reduces stack usage.
826 37. The value of the max lookbehind was not correctly preserved if a compiled
827 and saved regex was reloaded on a host of different endianness.
829 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
830 of the compiled pattern block, expand the flags field from 16 to 32 bits
831 because it was almost full.
833 39. Try madvise first before posix_madvise.
835 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
836 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
837 original change locked out all matches of empty strings. This has been
838 changed so that one match of an empty string per line is recognized.
839 Subsequent searches on the same line (for colouring or for --only-matching,
840 for example) do not recognize empty strings.
842 41. Applied a user patch to fix a number of spelling mistakes in comments.
844 42. Data lines longer than 65536 caused pcretest to crash.
846 43. Clarified the data type for length and startoffset arguments for pcre_exec
847 and pcre_dfa_exec in the function-specific man pages, where they were
848 explicitly stated to be in bytes, never having been updated. I also added
849 some clarification to the pcreapi man page.
851 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
852 a segmentation fault.
855 Version 8.32 30-November-2012
856 -----------------------------
858 1. Improved JIT compiler optimizations for first character search and single
859 character iterators.
861 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
862 Patch by Daniel Richard G.
864 3. Single character iterator optimizations in the JIT compiler.
866 4. Improved JIT compiler optimizations for character ranges.
868 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
869 Reported by Giuseppe D'Angelo.
871 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
872 of a line, was being set incorrectly in cases where .* appeared inside
873 atomic brackets at the start of a pattern, or where there was a subsequent
874 *PRUNE or *SKIP.
876 7. Improved instruction cache flush for POWER/PowerPC.
877 Patch by Daniel Richard G.
879 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
880 grep:
882 (a) There is now no limit to the number of patterns to be matched.
884 (b) An error is given if a pattern is too long.
886 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
887 are now supported.
889 (d) --exclude-from and --include-from (multiple use) have been added.
891 (e) Exclusions and inclusions now apply to all files and directories, not
892 just to those obtained from scanning a directory recursively.
894 (f) Multiple uses of -f and --file-list are now supported.
896 (g) In a Windows environment, the default for -d has been changed from
897 "read" (the GNU grep default) to "skip", because otherwise the presence
898 of a directory in the file list provokes an error.
900 (h) The documentation has been revised and clarified in places.
902 9. Improve the matching speed of capturing brackets.
904 10. Changed the meaning of \X so that it now matches a Unicode extended
905 grapheme cluster.
907 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
908 out POSIX threads when JIT support is configured.
910 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
912 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
913 structure could go wrong in environments where size_t is not the same size
914 as int.
916 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
917 set.
919 15. The EBCDIC support had decayed; later updates to the code had included
920 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
921 general tidy up of EBCDIC-related issues, and the documentation was also
922 not quite right. There is now a test that can be run on ASCII systems to
923 check some of the EBCDIC-related things (but is it not a full test).
925 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
926 in a small tidy to the code.
928 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
930 18. If the --only-matching (-o) option in pcregrep is specified multiple
931 times, each one causes appropriate output. For example, -o1 -o2 outputs the
932 substrings matched by the 1st and 2nd capturing parentheses. A separating
933 string can be specified by --om-separator (default empty).
935 19. Improving the first n character searches.
937 20. Turn case lists for horizontal and vertical white space into macros so that
938 they are defined only once.
940 21. This set of changes together give more compatible Unicode case-folding
941 behaviour for characters that have more than one other case when UCP
942 support is available.
944 (a) The Unicode property table now has offsets into a new table of sets of
945 three or more characters that are case-equivalent. The MultiStage2.py
946 script that generates these tables (the pcre_ucd.c file) now scans
947 CaseFolding.txt instead of UnicodeData.txt for character case
948 information.
950 (b) The code for adding characters or ranges of characters to a character
951 class has been abstracted into a generalized function that also handles
952 case-independence. In UTF-mode with UCP support, this uses the new data
953 to handle characters with more than one other case.
955 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
956 whose other case is greater than 256 are now correctly matched
957 caselessly. Previously, the high codepoint matched the low one, but not
958 vice versa.
960 (d) The processing of \h, \H, \v, and \ in character classes now makes use
961 of the new class addition function, using character lists defined as
962 macros alongside the case definitions of 20 above.
964 (e) Caseless back references now work with characters that have more than
965 one other case.
967 (f) General caseless matching of characters with more than one other case
968 is supported.
970 22. Unicode character properties were updated from Unicode 6.2.0
972 23. Improved CMake support under Windows. Patch by Daniel Richard G.
974 24. Add support for 32-bit character strings, and UTF-32
976 25. Major JIT compiler update (code refactoring and bugfixing).
977 Experimental Sparc 32 support is added.
979 26. Applied a modified version of Daniel Richard G's patch to create
980 pcre.h.generic and config.h.generic by "make" instead of in the
981 PrepareRelease script.
983 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
984 pcre_compile.c when checking for a zero character.
986 28. Introducing a native interface for JIT. Through this interface, the compiled
987 machine code can be directly executed. The purpose of this interface is to
988 provide fast pattern matching, so several sanity checks are not performed.
989 However, feature tests are still performed. The new interface provides
990 1.4x speedup compared to the old one.
992 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
993 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
994 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
996 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
997 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
998 the "old" RFC 2279). Instead, it ended up passing a negative length to
999 pcre_exec().
1001 31. Add support for GCC's visibility feature to hide internal functions.
1003 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1004 "unknown -C option" after outputting 0 or 1.
1006 33. There is now support for generating a code coverage report for the test
1007 suite in environments where gcc is the compiler and lcov is installed. This
1008 is mainly for the benefit of the developers.
1010 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1011 unaddressable using valgrind annotations, allowing valgrind to detect
1012 invalid memory accesses. This is mainly for the benefit of the developers.
1014 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1016 26. Give configure error if --enable-cpp but no C++ compiler found.
1019 Version 8.31 06-July-2012
1020 -------------------------
1022 1. Fixing a wrong JIT test case and some compiler warnings.
1024 2. Removed a bashism from the RunTest script.
1026 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1027 to unsigned type, result still unsigned" that was given by an MS compiler
1028 on encountering the code "-sizeof(xxx)".
1030 4. Partial matching support is added to the JIT compiler.
1032 5. Fixed several bugs concerned with partial matching of items that consist
1033 of more than one character:
1035 (a) /^(..)\1/ did not partially match "aba" because checking references was
1036 done on an "all or nothing" basis. This also applied to repeated
1037 references.
1039 (b) \R did not give a hard partial match if \r was found at the end of the
1040 subject.
1042 (c) \X did not give a hard partial match after matching one or more
1043 characters at the end of the subject.
1045 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1046 a partial match for the string "\r".
1048 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1049 a partial match for a CR character at the end of the subject string.
1051 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1052 running pcretest, the text "(JIT)" added to the output whenever JIT is
1053 actually used to run the match.
1055 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1056 or /S+[+] with a digit between 1 and 7.
1058 8. OP_NOT now supports any UTF character not just single-byte ones.
1060 9. (*MARK) control verb is now supported by the JIT compiler.
1062 10. The command "./RunTest list" lists the available tests without actually
1063 running any of them. (Because I keep forgetting what they all are.)
1067 12. Applied a (slightly modified) user-supplied patch that improves performance
1068 when the heap is used for recursion (compiled with --disable-stack-for-
1069 recursion). Instead of malloc and free for each heap frame each time a
1070 logical recursion happens, frames are retained on a chain and re-used where
1071 possible. This sometimes gives as much as 30% improvement.
1073 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1074 call.
1076 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1078 15. It is now possible to link pcretest with libedit as an alternative to
1079 libreadline.
1081 16. (*COMMIT) control verb is now supported by the JIT compiler.
1083 17. The Unicode data tables have been updated to Unicode 6.1.0.
1085 18. Added --file-list option to pcregrep.
1087 19. Added binary file support to pcregrep, including the -a, --binary-files,
1088 -I, and --text options.
1090 20. The madvise function is renamed for posix_madvise for QNX compatibility
1091 reasons. Fixed by Giuseppe D'Angelo.
1093 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1095 22. Changed the option for creating symbolic links for 16-bit man pages from
1096 -s to -sf so that re-installing does not cause issues.
1098 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1100 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1101 matches in certain environments (the workspace was not being correctly
1102 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1103 some of the workspace data at the beginning of a restart.
1105 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1106 was not doing so when it should - probably a typo introduced by SVN 528
1107 (change 8.10/14).
1109 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1110 \w+ when the character tables indicated that \x{c4} was a word character.
1111 There were several related cases, all because the tests for doing a table
1112 lookup were testing for characters less than 127 instead of 255.
1114 27. If a pattern contains capturing parentheses that are not used in a match,
1115 their slots in the ovector are set to -1. For those that are higher than
1116 any matched groups, this happens at the end of processing. In the case when
1117 there were back references that the ovector was too small to contain
1118 (causing temporary malloc'd memory to be used during matching), and the
1119 highest capturing number was not used, memory off the end of the ovector
1120 was incorrectly being set to -1. (It was using the size of the temporary
1121 memory instead of the true size.)
1123 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1124 ovector size, it uses memory at the end of the block that it has got.
1126 29. Check for an overlong MARK name and give an error at compile time. The
1127 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1129 30. JIT compiler update.
1131 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1132 Rill for the patch.
1134 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1136 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1138 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1139 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1140 was enabled.
1142 35. Improve JIT code generation for greedy plus quantifier.
1144 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1145 1 to "aa" instead of to an empty string. The bug affected repeated groups
1146 that could potentially match an empty string.
1148 37. Optimizing single character iterators in JIT.
1150 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1151 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1152 codepoints that are too big for the mode are faulted, and in a UTF mode,
1153 disallowed codepoints are also faulted.
1155 39. If PCRE was compiled with UTF support, in three places in the DFA
1156 matcher there was code that should only have been obeyed in UTF mode, but
1157 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1158 processing when bytes with values greater than 127 were present. In 16-bit
1159 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1160 both cases the values are those that cannot be the first data item in a UTF
1161 character. The three items that might have provoked this were recursions,
1162 possessively repeated groups, and atomic groups.
1164 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1165 and pcregrep, because some OS require shared objects to be explicitly
1166 passed to ld, causing the link step to fail if they are not.
1168 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1169 mode, patterns that started with \h* or \R* might be incorrectly matched.
1172 Version 8.30 04-February-2012
1173 -----------------------------
1175 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1176 name is defined in ctype.h.
1178 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1179 only in quite long subpatterns.
1181 3. Removed the function pcre_info(), which has been obsolete and deprecated
1182 since it was replaced by pcre_fullinfo() in February 2000.
1184 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1185 match a (*MARK), and the match failed at the start of the subject, a
1186 reference to memory before the start of the subject could occur. This bug
1187 was introduced by fix 17 of release 8.21.
1189 5. A reference to an unset group with zero minimum repetition was giving
1190 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1191 /(another)?(\1?)test/ matched against "hello world test". This bug was
1192 introduced in release 8.13.
1194 6. Add support for 16-bit character strings (a large amount of work involving
1195 many changes and refactorings).
1197 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1198 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1199 from a file.
1201 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1202 rounding is not applied in this particular case).
1204 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1205 if they appear, or are escaped, in patterns.
1207 10. Get rid of a number of -Wunused-but-set-variable warnings.
1209 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1210 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1211 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1212 also returns the mark "x". This bug applied to capturing parentheses,
1213 non-capturing parentheses, and atomic parentheses. It also applied to some
1214 assertions.
1216 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1217 information out of configure.ac instead of relying on pcre.h.generic, which
1218 is not stored in the repository.
1220 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1221 -lreadline.
1223 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1225 15. Applied Graycode's patch to put the top-level frame on the stack rather
1226 than the heap when not using the stack for recursion. This gives a
1227 performance improvement in many cases when recursion is not deep.
1229 16. Experimental code added to "pcretest -C" to output the stack frame size.
1232 Version 8.21 12-Dec-2011
1233 ------------------------
1235 1. Updating the JIT compiler.
1237 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1238 are added as well.
1240 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1241 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1242 calling _pcre_jit_exec. Some extra comments are added.
1244 4. (*MARK) settings inside atomic groups that do not contain any capturing
1245 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1246 was introduced by change 18 for 8.20.
1248 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1249 ECMA-262 standard.
1251 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1252 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1253 This bug was probably introduced by change 9 of 8.13.
1255 7. While fixing 6 above, I noticed that a number of other items were being
1256 incorrectly rejected as "not fixed length". This arose partly because newer
1257 opcodes had not been added to the fixed-length checking code. I have (a)
1258 corrected the bug and added tests for these items, and (b) arranged for an
1259 error to occur if an unknown opcode is encountered while checking for fixed
1260 length instead of just assuming "not fixed length". The items that were
1261 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1262 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1263 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1265 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1266 being incorrectly compiled and would have given unpredicatble results.
1268 9. A possessively repeated subpattern with minimum repeat count greater than
1269 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1270 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1271 the first (A) could occur when it should not.
1273 10. Add a cast and remove a redundant test from the code.
1275 11. JIT should use pcre_malloc/pcre_free for allocation.
1277 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1278 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1279 is anything other than /usr, -L is still shown).
1281 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1283 14. Perl does not support \N without a following name in a [] class; PCRE now
1284 also gives an error.
1286 15. If a forward reference was repeated with an upper limit of around 2000,
1287 it caused the error "internal error: overran compiling workspace". The
1288 maximum number of forward references (including repeats) was limited by the
1289 internal workspace, and dependent on the LINK_SIZE. The code has been
1290 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1291 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1292 of around 200,000 forward references. While doing this, I also speeded up
1293 the filling in of repeated forward references.
1295 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1296 incorrectly expecting the subject to contain another "a" after the start.
1298 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1299 in the match, the SKIP should be ignored. This was not happening; instead
1300 the SKIP was being treated as NOMATCH. For patterns such as
1301 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1302 tested.
1304 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1305 now much more compatible with Perl, in particular in cases where the result
1306 is a non-match for a non-anchored pattern. For example, if
1307 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1308 "m", where previously it did not return a name. A side effect of this
1309 change is that for partial matches, the last encountered mark name is
1310 returned, as for non matches. A number of tests that were previously not
1311 Perl-compatible have been moved into the Perl-compatible test files. The
1312 refactoring has had the pleasing side effect of removing one argument from
1313 the match() function, thus reducing its stack requirements.
1315 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1316 subsequent uses of /S (without +) incorrectly behaved like /S+.
1318 21. Retrieve executable code size support for the JIT compiler and fixing
1319 some warnings.
1321 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1322 not work when the shorter character appeared right at the end of the
1323 subject string.
1325 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1326 systems.
1328 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1329 output it when the /M option is used in pcretest.
1331 25. The CheckMan script was not being included in the distribution. Also, added
1332 an explicit "perl" to run Perl scripts from the PrepareRelease script
1333 because this is reportedly needed in Windows.
1335 26. If study data was being save in a file and studying had not found a set of
1336 "starts with" bytes for the pattern, the data written to the file (though
1337 never used) was taken from uninitialized memory and so caused valgrind to
1338 complain.
1340 27. Updated RunTest.bat as provided by Sheri Pierce.
1342 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1344 29. Computation of memory usage for the table of capturing group names was
1345 giving an unnecessarily large value.
1348 Version 8.20 21-Oct-2011
1349 ------------------------
1351 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1352 a POSIX class. After further experiments with Perl, which convinced me that
1353 Perl has bugs and confusions, a closing square bracket is no longer allowed
1354 in a POSIX name. This bug also affected patterns with classes that started
1355 with full stops.
1357 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1358 captured substring, but while checking the failing first alternative,
1359 substring 1 is temporarily captured. If the output vector supplied to
1360 pcre_exec() was not big enough for this capture, the yield of the function
1361 was still zero ("insufficient space for captured substrings"). This cannot
1362 be totally fixed without adding another stack variable, which seems a lot
1363 of expense for a edge case. However, I have improved the situation in cases
1364 such as /(a)(b)x|abc/ matched against "abc", where the return code
1365 indicates that fewer than the maximum number of slots in the ovector have
1366 been set.
1368 3. Related to (2) above: when there are more back references in a pattern than
1369 slots in the output vector, pcre_exec() uses temporary memory during
1370 matching, and copies in the captures as far as possible afterwards. It was
1371 using the entire output vector, but this conflicts with the specification
1372 that only 2/3 is used for passing back captured substrings. Now it uses
1373 only the first 2/3, for compatibility. This is, of course, another edge
1374 case.
1376 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1377 main code base, and can be used by building with --enable-jit. When this is
1378 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1379 runtime --no-jit option is given.
1381 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1382 ovector, the return from the function was zero, implying that there were
1383 other matches that did not fit. The correct "exactly full" value is now
1384 returned.
1386 6. If a subpattern that was called recursively or as a subroutine contained
1387 (*PRUNE) or any other control that caused it to give a non-standard return,
1388 invalid errors such as "Error -26 (nested recursion at the same subject
1389 position)" or even infinite loops could occur.
1391 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1392 computing the minimum length on reaching *ACCEPT, and so ended up with the
1393 wrong value of 1 rather than 0. Further investigation indicates that
1394 computing a minimum subject length in the presence of *ACCEPT is difficult
1395 (think back references, subroutine calls), and so I have changed the code
1396 so that no minimum is registered for a pattern that contains *ACCEPT.
1398 8. If (*THEN) was present in the first (true) branch of a conditional group,
1399 it was not handled as intended. [But see 16 below.]
1401 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1402 Sheri Pierce.
1404 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1405 the first byte in a match must be "a".
1407 11. Change 17 for 8.13 increased the recursion depth for patterns like
1408 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1409 pattern contains any instances of (*THEN). If it does not, the old
1410 optimizations are restored. It would be nice to do this on a per-group
1411 basis, but at the moment that is not feasible.
1413 12. In some environments, the output of pcretest -C is CRLF terminated. This
1414 broke RunTest's code that checks for the link size. A single white space
1415 character after the value is now allowed for.
1417 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1418 For "fr", it uses the Windows-specific input and output files.
1420 14. If (*THEN) appeared in a group that was called recursively or as a
1421 subroutine, it did not work as intended. [But see next item.]
1423 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1424 pattern fragments (but not containing any | characters). If A and B are
1425 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1426 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1427 D. In other words, Perl considers parentheses that do not contain any |
1428 characters to be part of a surrounding alternative, whereas PCRE was
1429 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1430 differently. PCRE now behaves in the same way as Perl, except in the case
1431 of subroutine/recursion calls such as (?1) which have in any case always
1432 been different (but PCRE had them first :-).
1434 16. Related to 15 above: Perl does not treat the | in a conditional group as
1435 creating alternatives. Such a group is treated in the same way as an
1436 ordinary group without any | characters when processing (*THEN). PCRE has
1437 been changed to match Perl's behaviour.
1439 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1440 RunGrepTest script failed.
1442 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1443 inevitable for groups that contain captures, but it can lead to a lot of
1444 stack use in large patterns. The old behaviour has been restored for atomic
1445 groups that do not contain any capturing parentheses.
1447 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1448 suppress the check for a minimum subject length at run time. (If it was
1449 given to pcre_exec() or pcre_dfa_exec() it did work.)
1451 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1452 fail to work when decoding hex characters in data strings in EBCDIC
1453 environments.
1455 21. It appears that in at least one Mac OS environment, the isxdigit() function
1456 is implemented as a macro that evaluates to its argument more than once,
1457 contravening the C 90 Standard (I haven't checked a later standard). There
1458 was an instance in pcretest which caused it to go wrong when processing
1459 \x{...} escapes in subject strings. The has been rewritten to avoid using
1460 things like p++ in the argument of isxdigit().
1463 Version 8.13 16-Aug-2011
1464 ------------------------
1466 1. The Unicode data tables have been updated to Unicode 6.0.0.
1468 2. Two minor typos in pcre_internal.h have been fixed.
1470 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1471 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1472 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1474 4. There were a number of related bugs in the code for matching backrefences
1475 caselessly in UTF-8 mode when codes for the characters concerned were
1476 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1477 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1478 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1479 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1480 2-byte code at the end of the subject (it thought there wasn't enough data
1481 left).
1483 5. Comprehensive information about what went wrong is now returned by
1484 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1485 as the output vector has at least 2 elements. The offset of the start of
1486 the failing character and a reason code are placed in the vector.
1488 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1489 now returned is for the first byte of the failing character, instead of the
1490 last byte inspected. This is an incompatible change, but I hope it is small
1491 enough not to be a problem. It makes the returned offset consistent with
1492 pcre_exec() and pcre_dfa_exec().
1494 7. pcretest now gives a text phrase as well as the error number when
1495 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1496 failure, the offset and reason code are output.
1498 8. When \R was used with a maximizing quantifier it failed to skip backwards
1499 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1500 back over a single character (\n). This seems wrong (because it treated the
1501 two characters as a single entity when going forwards), conflicts with the
1502 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1503 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1504 has been changed.
1506 9. Some internal refactoring has changed the processing so that the handling
1507 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1508 time (the PCRE_DOTALL option was changed this way some time ago: version
1509 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1510 which was always a bit of a fudge. It also means that there is one less
1511 argument for the match() function, which reduces its stack requirements
1512 slightly. This change also fixes an incompatibility with Perl: the pattern
1513 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1515 10. More internal refactoring has drastically reduced the number of recursive
1516 calls to match() for possessively repeated groups such as (abc)++ when
1517 using pcre_exec().
1519 11. While implementing 10, a number of bugs in the handling of groups were
1520 discovered and fixed:
1522 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1523 (a|)*(?1) gave a compile-time internal error.
1524 ((a|)+)+ did not notice that the outer group could match an empty string.
1525 (^a|^)+ was not marked as anchored.
1526 (.*a|.*)+ was not marked as matching at start or after a newline.
1528 12. Yet more internal refactoring has removed another argument from the match()
1529 function. Special calls to this function are now indicated by setting a
1530 value in a variable in the "match data" data block.
1532 13. Be more explicit in pcre_study() instead of relying on "default" for
1533 opcodes that mean there is no starting character; this means that when new
1534 ones are added and accidentally left out of pcre_study(), testing should
1535 pick them up.
1537 14. The -s option of pcretest has been documented for ages as being an old
1538 synonym of -m (show memory usage). I have changed it to mean "force study
1539 for every regex", that is, assume /S for every regex. This is similar to -i
1540 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1541 using it. It makes it easier to run collections of tests with and without
1542 study enabled, and thereby test pcre_study() more easily. All the standard
1543 tests are now run with and without -s (but some patterns can be marked as
1544 "never study" - see 20 below).
1546 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1547 restoration of the capturing data to the outer values was not happening
1548 correctly.
1550 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1551 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1552 pattern had matched an empty string, and so incorrectly returned a no
1553 match.
1555 17. There was optimizing code for the last branch of non-capturing parentheses,
1556 and also for the obeyed branch of a conditional subexpression, which used
1557 tail recursion to cut down on stack usage. Unfortunately, now that there is
1558 the possibility of (*THEN) occurring in these branches, tail recursion is
1559 no longer possible because the return has to be checked for (*THEN). These
1560 two optimizations have therefore been removed. [But see 8.20/11 above.]
1562 18. If a pattern containing \R was studied, it was assumed that \R always
1563 matched two bytes, thus causing the minimum subject length to be
1564 incorrectly computed because \R can also match just one byte.
1566 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1567 was incorrectly computed.
1569 20. If /S is present twice on a test pattern in pcretest input, it now
1570 *disables* studying, thereby overriding the use of -s on the command line
1571 (see 14 above). This is necessary for one or two tests to keep the output
1572 identical in both cases.
1574 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1575 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1577 22. When an atomic group that contained a capturing parenthesis was
1578 successfully matched, but the branch in which it appeared failed, the
1579 capturing was not being forgotten if a higher numbered group was later
1580 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1581 group 1 to "a", when in fact it should be unset. This applied to multi-
1582 branched capturing and non-capturing groups, repeated or not, and also to
1583 positive assertions (capturing in negative assertions does not happen
1584 in PCRE) and also to nested atomic groups.
1586 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1587 subject after a captured substring, to make it easier to tell which of a
1588 number of identical substrings has been captured.
1590 24. The way atomic groups are processed by pcre_exec() has been changed so that
1591 if they are repeated, backtracking one repetition now resets captured
1592 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1593 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1594 "aaa". Previously, it would have been "a". As part of this code
1595 refactoring, the way recursive calls are handled has also been changed.
1597 25. If an assertion condition captured any substrings, they were not passed
1598 back unless some other capturing happened later. For example, if
1599 (?(?=(a))a) was matched against "a", no capturing was returned.
1601 26. When studying a pattern that contained subroutine calls or assertions,
1602 the code for finding the minimum length of a possible match was handling
1603 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1604 group 1 called group 2 while simultaneously a separate group 2 called group
1605 1). A stack overflow occurred in this case. I have fixed this by limiting
1606 the recursion depth to 10.
1608 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1609 Fortmann. This supports explicit test numbers on the command line, and has
1610 argument validation and error reporting.
1612 28. An instance of \X with an unlimited repeat could fail if at any point the
1613 first character it looked at was a mark character.
1615 29. Some minor code refactoring concerning Unicode properties and scripts
1616 should reduce the stack requirement of match() slightly.
1618 30. Added the '=' option to pcretest to check the setting of unused capturing
1619 slots at the end of the pattern, which are documented as being -1, but are
1620 not included in the return count.
1622 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1623 compiled something random. Now it gives a compile-time error (as does
1624 Perl).
1626 32. A *MARK encountered during the processing of a positive assertion is now
1627 recorded and passed back (compatible with Perl).
1629 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1630 had alternative anchored branches, the search for a second match in a line
1631 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1632 matched the line "0102" twice. The same bug affected patterns that started
1633 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1634 twice.
1636 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1637 does, and because of capturing effects, quantifying parenthesized
1638 assertions may at times be useful. Quantifiers are now allowed for
1639 parenthesized assertions.
1641 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1643 36. \g was being checked for fancy things in a character class, when it should
1644 just be a literal "g".
1646 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1647 appearance of a nested POSIX class supersedes an apparent external class.
1648 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1649 unescaped square brackets may also appear as part of class names. For
1650 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1651 more like Perl. (But see 8.20/1 above.)
1653 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1654 was because it thought it was \N{name}, which is not supported).
1656 39. Add minix to OS list not supporting the -S option in pcretest.
1658 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1659 cannot analyze patterns in sufficient detail to catch mutual recursions
1660 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1661 subgroup is called recursively as a subpattern for a second time at the
1662 same position in the subject string. In previous releases this might have
1663 been caught by the recursion limit, or it might have run out of stack.
1665 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1666 happen only once. PCRE was, however incorrectly giving a compile time error
1667 "recursive call could loop indefinitely" because it cannot analyze the
1668 pattern in sufficient detail. The compile time test no longer happens when
1669 PCRE is compiling a conditional subpattern, but actual runaway loops are
1670 now caught at runtime (see 40 above).
1672 42. It seems that Perl allows any characters other than a closing parenthesis
1673 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1674 has been changed to be the same.
1676 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1677 as not to get warnings when autogen.sh is called. Also changed
1678 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1680 44. To help people who use pcregrep to scan files containing exceedingly long
1681 lines, the following changes have been made:
1683 (a) The default value of the buffer size parameter has been increased from
1684 8K to 20K. (The actual buffer used is three times this size.)
1686 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1687 PCRE is built.
1689 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1690 to be set at run time.
1692 (d) Numerical values in pcregrep options can be followed by K or M, for
1693 example --buffer-size=50K.
1695 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1696 given and the return code is set to 2.
1698 45. Add a pointer to the latest mark to the callout data block.
1700 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1701 partial match of an empty string instead of no match. This was specific to
1702 the use of ".".
1704 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1705 complete match instead of a partial match. This bug was dependent on both
1706 the PCRE_UTF8 and PCRE_DOTALL options being set.
1708 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1709 starting byte set, because \b was not being ignored.
1712 Version 8.12 15-Jan-2011
1713 ------------------------
1715 1. Fixed some typos in the markup of the man pages, and wrote a script that
1716 checks for such things as part of the documentation building process.
1718 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1719 --match-limit and --recursion-limit options (added for 8.11). In
1720 particular, this made one of the standard tests fail. (The integer value
1721 went into the wrong half of a long int.)
1723 3. If the --colour option was given to pcregrep with -v (invert match), it
1724 did strange things, either producing crazy output, or crashing. It should,
1725 of course, ignore a request for colour when reporting lines that do not
1726 match.
1728 4. Another pcregrep bug caused similar problems if --colour was specified with
1729 -M (multiline) and the pattern match finished with a line ending.
1731 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1732 matched in multiline mode, the following line was shown as part of the
1733 match. This seems wrong, so I have changed it.
1735 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1736 the check for further matches in the same line (so they could be coloured)
1737 to overrun the end of the current line. If another match was found, it was
1738 incorrectly shown (and then shown again when found in the next line).
1740 7. If pcregrep was compiled under Windows, there was a reference to the
1741 function pcregrep_exit() before it was defined. I am assuming this was
1742 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1743 reported by a user. I've moved the definition above the reference.
1746 Version 8.11 10-Dec-2010
1747 ------------------------
1749 1. (*THEN) was not working properly if there were untried alternatives prior
1750 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1751 backtracked to try for "b" instead of moving to the next alternative branch
1752 at the same level (in this case, to look for "c"). The Perl documentation
1753 is clear that when (*THEN) is backtracked onto, it goes to the "next
1754 alternative in the innermost enclosing group".
1756 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1757 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1758 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1759 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1760 (*THEN).
1762 3. If \s appeared in a character class, it removed the VT character from
1763 the class, even if it had been included by some previous item, for example
1764 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1765 of \s, but is part of the POSIX "space" class.)
1767 4. A partial match never returns an empty string (because you can always
1768 match an empty string at the end of the subject); however the checking for
1769 an empty string was starting at the "start of match" point. This has been
1770 changed to the "earliest inspected character" point, because the returned
1771 data for a partial match starts at this character. This means that, for
1772 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1773 (previously it gave "no match").
1775 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1776 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1777 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1778 has an implication that the given string is incomplete (because a partial
1779 match is preferred over a full match). For this reason, these items now
1780 give a partial match in this situation. [Aside: previously, the one case
1781 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1782 match rather than a full match, which was wrong by the old rules, but is
1783 now correct.]
1785 6. There was a bug in the handling of #-introduced comments, recognized when
1786 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1787 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1788 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1789 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1790 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1791 places in pcre_compile().
1793 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1794 comments when looking ahead for named forward references to subpatterns,
1795 the only newline sequence it recognized was NL. It now handles newlines
1796 according to the set newline convention.
1798 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1799 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1800 cater for a lack of strerror(). These oversights have been fixed.
1802 9. Added --match-limit and --recursion-limit to pcregrep.
1804 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1806 11. When the -o option was used, pcregrep was setting a return code of 1, even
1807 when matches were found, and --line-buffered was not being honoured.
1809 12. Added an optional parentheses number to the -o and --only-matching options
1810 of pcregrep.
1812 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1813 can match an empty string. The code to do it in pcretest and pcredemo
1814 needed fixing:
1816 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1817 only one byte after an empty string match just before CRLF (this case
1818 just got forgotten; "any" and "anycrlf" were OK).
1820 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1821 mode when an empty string match preceded an ASCII character followed by
1822 a non-ASCII character. (The code for advancing by one character rather
1823 than one byte was nonsense.)
1825 (c) The pcredemo.c sample program did not have any code at all to handle
1826 the cases when CRLF is a valid newline sequence.
1828 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1829 as a starting offset was within the subject string. There is now a new
1830 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1831 negative or greater than the length of the string. In order to test this,
1832 pcretest is extended to allow the setting of negative starting offsets.
1834 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1835 starting offset points to the beginning of a UTF-8 character was
1836 unnecessarily clumsy. I tidied it up.
1838 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1839 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1841 17. Nobody had reported that the --include_dir option, which was added in
1842 release 7.7 should have been called --include-dir (hyphen, not underscore)
1843 for compatibility with GNU grep. I have changed it to --include-dir, but
1844 left --include_dir as an undocumented synonym, and the same for
1845 --exclude-dir, though that is not available in GNU grep, at least as of
1846 release 2.5.4.
1848 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1849 characters from a string of bytes) have been redefined so as not to use
1850 loops, in order to improve performance in some environments. At the same
1851 time, I abstracted some of the common code into auxiliary macros to save
1852 repetition (this should not affect the compiled code).
1854 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1855 compile-time error is now given if \c is not followed by an ASCII
1856 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1857 different, and any byte value is allowed.)
1859 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1860 START_OPTIMIZE option, which is now allowed at compile time - but just
1861 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1862 to pcregrep and other applications that have no direct access to PCRE
1863 options. The new /Y option in pcretest sets this option when calling
1864 pcre_compile().
1866 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1867 back references. Groups containing recursive back references were forced to
1868 be atomic by that change, but in the case of named groups, the amount of
1869 memory required was incorrectly computed, leading to "Failed: internal
1870 error: code overflow". This has been fixed.
1872 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1873 pcretest.c, to avoid build problems in some Borland environments.
1876 Version 8.10 25-Jun-2010
1877 ------------------------
1879 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1880 THEN.
1882 2. (*ACCEPT) was not working when inside an atomic group.
1884 3. Inside a character class, \B is treated as a literal by default, but
1885 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1886 causes the error). The code is unchanged, but I tidied the documentation.
1888 4. Inside a character class, PCRE always treated \R and \X as literals,
1889 whereas Perl faults them if its -w option is set. I have changed PCRE so
1890 that it faults them when PCRE_EXTRA is set.
1892 5. Added support for \N, which always matches any character other than
1893 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1895 6. When compiling pcregrep with newer versions of gcc which may have
1896 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1897 declared with attribute warn_unused_result" were given. Just casting the
1898 result to (void) does not stop the warnings; a more elaborate fudge is
1899 needed. I've used a macro to implement this.
1901 7. Minor change to pcretest.c to avoid a compiler warning.
1903 8. Added four artifical Unicode properties to help with an option to make
1904 \s etc use properties (see next item). The new properties are: Xan
1905 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1907 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1908 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1909 this option. Modified pcretest to add /W to test this facility. Added
1910 REG_UCP to make it available via the POSIX interface.
1912 10. Added --line-buffered to pcregrep.
1914 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1915 studied, and the match started with a letter with a code point greater than
1916 127 whose first byte was different to the first byte of the other case of
1917 the letter, the other case of this starting letter was not recognized
1918 (#976).
1920 12. If a pattern that was studied started with a repeated Unicode property
1921 test, for example, \p{Nd}+, there was the theoretical possibility of
1922 setting up an incorrect bitmap of starting bytes, but fortunately it could
1923 not have actually happened in practice until change 8 above was made (it
1924 added property types that matched character-matching opcodes).
1926 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1927 possible starting bytes for non-anchored patterns.
1929 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1930 \R, and also a number of cases that involve Unicode properties, both
1931 explicit and implicit when PCRE_UCP is set.
1933 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1934 input, it could crash or give wrong results if characters with values
1935 greater than 0xc0 were present in the subject string. (Detail: it assumed
1936 UTF-8 input when processing these items.)
1938 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1939 size_t is 64-bit (#991).
1941 17. Added a check for running out of memory when PCRE is compiled with
1942 --disable-stack-for-recursion (#990).
1944 18. If the last data line in a file for pcretest does not have a newline on
1945 the end, a newline was missing in the output.
1947 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1948 less than 128) in its various bitmaps. However, there is a facility for
1949 generating tables according to the current locale when PCRE is compiled. It
1950 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1951 characters, are recognized by isspace() and therefore were getting set in
1952 these tables, and indeed these tables seem to approximate to ISO 8859. This
1953 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1954 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1955 which of course cannot start UTF-8 characters. I have changed the code so
1956 that only real ASCII characters (less than 128) and the correct starting
1957 bytes for UTF-8 encodings are set for characters greater than 127 when in
1958 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1959 altogether.)
1961 20. Added the /T option to pcretest so as to be able to run tests with non-
1962 standard character tables, thus making it possible to include the tests
1963 used for 19 above in the standard set of tests.
1965 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1966 reference to a subpattern the other side of a comment that contains an
1967 opening parenthesis caused either an internal compiling error, or a
1968 reference to the wrong subpattern.
1971 Version 8.02 19-Mar-2010
1972 ------------------------
1974 1. The Unicode data tables have been updated to Unicode 5.2.0.
1976 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1977 configured.
1979 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1980 original author of that file, following a query about its status.
1982 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1983 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1985 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1986 quantifier applied to a forward-referencing subroutine call, could compile
1987 incorrect code or give the error "internal error: previously-checked
1988 referenced subpattern not found".
1990 6. Both MS Visual Studio and Symbian OS have problems with initializing
1991 variables to point to external functions. For these systems, therefore,
1992 pcre_malloc etc. are now initialized to local functions that call the
1993 relevant global functions.
1995 7. There were two entries missing in the vectors called coptable and poptable
1996 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1997 I've fixed the data, and added a kludgy way of testing at compile time that
1998 the lengths are correct (equal to the number of opcodes).
2000 8. Following on from 7, I added a similar kludge to check the length of the
2001 eint vector in pcreposix.c.
2003 9. Error texts for pcre_compile() are held as one long string to avoid too
2004 much relocation at load time. To find a text, the string is searched,
2005 counting zeros. There was no check for running off the end of the string,
2006 which could happen if a new error number was added without updating the
2007 string.
2009 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2011 11. \K was not working if it appeared in an atomic group or in a group that
2012 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2013 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2014 the assertion is positive, but not if it is negative.
2016 12. Change 11 fortuitously reduced the size of the stack frame used in the
2017 "match()" function of pcre_exec.c by one pointer. Forthcoming
2018 implementation of support for (*MARK) will need an extra pointer on the
2019 stack; I have reserved it now, so that the stack frame size does not
2020 decrease.
2022 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2023 item in branch that calls a recursion is a subroutine call - as in the
2024 second branch in the above example - was incorrectly given the compile-
2025 time error "recursive call could loop indefinitely" because pcre_compile()
2026 was not correctly checking the subroutine for matching a non-empty string.
2028 14. The checks for overrunning compiling workspace could trigger after an
2029 overrun had occurred. This is a "should never occur" error, but it can be
2030 triggered by pathological patterns such as hundreds of nested parentheses.
2031 The checks now trigger 100 bytes before the end of the workspace.
2033 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2036 Version 8.01 19-Jan-2010
2037 ------------------------
2039 1. If a pattern contained a conditional subpattern with only one branch (in
2040 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2041 computed the wrong minimum data length (which is of course zero for such
2042 subpatterns). This could cause incorrect "no match" results.
2044 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2045 the pattern is reset in the first branch, pcre_compile() failed with
2046 "internal error: code overflow at offset...". This happened only when
2047 the reset was to the original external option setting. (An optimization
2048 abstracts leading options settings into an external setting, which was the
2049 cause of this.)
2051 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2052 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2053 assertion pattern did not match (meaning that the assertion was true), it
2054 was incorrectly treated as false if the SKIP had been reached during the
2055 matching. This also applied to assertions used as conditions.
2057 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2058 assertion subpattern, including such a pattern used as a condition,
2059 unpredictable results occurred, instead of the error return
2062 5. The C++ GlobalReplace function was not working like Perl for the special
2063 situation when an empty string is matched. It now does the fancy magic
2064 stuff that is necessary.
2066 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2067 removed. (These were left over from very, very early versions of PCRE.)
2069 7. Some cosmetic changes to the code to make life easier when compiling it
2070 as part of something else:
2072 (a) Change DEBUG to PCRE_DEBUG.
2074 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2075 called "current" as "current_branch", to prevent a collision with the
2076 Linux macro when compiled as a kernel module.
2078 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2079 prevent a collision with the Linux macro when compiled as a kernel
2080 module.
2082 8. In pcre_compile() there are some checks for integer overflows that used to
2083 cast potentially large values to (double). This has been changed to that
2084 when building, a check for int64_t is made, and if it is found, it is used
2085 instead, thus avoiding the use of floating point arithmetic. (There is no
2086 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2087 double.
2089 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2090 2005 (difference between two addresses compared to an unsigned value).
2092 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2093 custom one, because of the following reported problem in Windows:
2095 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2096 under Win32.
2097 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2098 therefore missing the function definition.
2099 - The compiler thus generates a "C" signature for the test function.
2100 - The linker fails to find the "C" function.
2101 - PCRE fails to configure if asked to do so against libbz2.
2103 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2104 messages were output:
2106 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2107 rerunning libtoolize, to keep the correct libtool macros in-tree.
2108 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2110 I have done both of these things.
2112 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2113 most of the time, it *can* run out if it is given a pattern that contains a
2114 runaway infinite recursion. I updated the discussion in the pcrestack man
2115 page.
2117 13. Now that we have gone to the x.xx style of version numbers, the minor
2118 version may start with zero. Using 08 or 09 is a bad idea because users
2119 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2120 interpreted as invalid octal numbers. I've updated the previous comment in
2121 configure.ac, and also added a check that gives an error if 08 or 09 are
2122 used.
2124 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2125 causing partial matching to fail when the end of the subject matched \W
2126 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2128 15. There were some discrepancies between the declarations in pcre_internal.h
2129 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2130 their definitions. The declarations used "const uschar *" and the
2131 definitions used USPTR. Even though USPTR is normally defined as "const
2132 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2133 reported that: "This difference in casting confuses some C++ compilers, for
2134 example, SunCC recognizes above declarations as different functions and
2135 generates broken code for hbpcre." I have changed the declarations to use
2136 USPTR.
2138 16. GNU libtool is named differently on some systems. The autogen.sh script now
2139 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2140 (FreeBSD).
2142 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2143 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2144 comment: "Figure out how to create a longlong from a string: strtoll and
2145 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2146 instance, but it only takes 2 args instead of 3!"
2148 18. A subtle bug concerned with back references has been fixed by a change of
2149 specification, with a corresponding code fix. A pattern such as
2150 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2151 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2152 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2153 same bug. Such groups have to be quantified to be useful, or contained
2154 inside another quantified group. (If there's no repetition, the reference
2155 can never match.) The problem arises because, having left the group and
2156 moved on to the rest of the pattern, a later failure that backtracks into
2157 the group uses the captured value from the final iteration of the group
2158 rather than the correct earlier one. I have fixed this in PCRE by forcing
2159 any group that contains a reference to itself to be an atomic group; that
2160 is, there cannot be any backtracking into it once it has completed. This is
2161 similar to recursive and subroutine calls.
2164 Version 8.00 19-Oct-09
2165 ----------------------
2167 1. The table for translating pcre_compile() error codes into POSIX error codes
2168 was out-of-date, and there was no check on the pcre_compile() error code
2169 being within the table. This could lead to an OK return being given in
2170 error.
2172 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2173 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2174 in a Windows environment.
2176 3. The pcregrep --count option prints the count for each file even when it is
2177 zero, as does GNU grep. However, pcregrep was also printing all files when
2178 --files-with-matches was added. Now, when both options are given, it prints
2179 counts only for those files that have at least one match. (GNU grep just
2180 prints the file name in this circumstance, but including the count seems
2181 more useful - otherwise, why use --count?) Also ensured that the
2182 combination -clh just lists non-zero counts, with no names.
2184 4. The long form of the pcregrep -F option was incorrectly implemented as
2185 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2186 but it seems right to fix it, and I didn't think it was worth preserving
2187 the old behaviour.
2189 5. The command line items --regex=pattern and --regexp=pattern were not
2190 recognized by pcregrep, which required --regex pattern or --regexp pattern
2191 (with a space rather than an '='). The man page documented the '=' forms,
2192 which are compatible with GNU grep; these now work.
2194 6. No libpcreposix.pc file was created for pkg-config; there was just
2195 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2197 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2198 when UCP support is not needed, by modifying the Python script that
2199 generates it from Unicode data files. This should not matter if the module
2200 is correctly used as a library, but I received one complaint about 50K of
2201 unwanted data. My guess is that the person linked everything into his
2202 program rather than using a library. Anyway, it does no harm.
2204 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2205 was a minimum greater than 1 for a wide character in a possessive
2206 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2207 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2208 character. Chaos in the form of incorrect output or a compiling loop could
2209 result.
2211 9. The restrictions on what a pattern can contain when partial matching is
2212 requested for pcre_exec() have been removed. All patterns can now be
2213 partially matched by this function. In addition, if there are at least two
2214 slots in the offset vector, the offset of the earliest inspected character
2215 for the match and the offset of the end of the subject are set in them when
2216 PCRE_ERROR_PARTIAL is returned.
2218 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2219 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2220 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2221 and may be more useful for multi-segment matching.
2223 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2224 used to be given if ever the end of the subject was reached; now it is
2225 given only if matching could not proceed because another character was
2226 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2227 string "Z", which now yields "no match" instead of "partial match". In the
2228 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2229 final character ended with (*FAIL).
2231 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2232 if the pattern had a "must contain" character that was already found in the
2233 earlier partial match, unless partial matching was again requested. For
2234 example, with the pattern /dog.(body)?/, the "must contain" character is
2235 "g". If the first part-match was for the string "dog", restarting with
2236 "sbody" failed. This bug has been fixed.
2238 13. The string returned by pcre_dfa_exec() after a partial match has been
2239 changed so that it starts at the first inspected character rather than the
2240 first character of the match. This makes a difference only if the pattern
2241 starts with a lookbehind assertion or \b or \B (\K is not supported by
2242 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2243 matching functions compatible, and I think it's the right thing to do.
2245 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2246 so that the demonstration program is easily available in environments where
2247 PCRE has not been installed from source.
2249 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2250 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2251 library.
2253 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2254 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2255 is not the first non-POSIX option to be added. Clearly some people find
2256 these options useful.
2258 17. If a caller to the POSIX matching function regexec() passes a non-zero
2259 value for nmatch with a NULL value for pmatch, the value of
2260 nmatch is forced to zero.
2262 18. RunGrepTest did not have a test for the availability of the -u option of
2263 the diff command, as RunTest does. It now checks in the same way as
2264 RunTest, and also checks for the -b option.
2266 19. If an odd number of negated classes containing just a single character
2267 interposed, within parentheses, between a forward reference to a named
2268 subpattern and the definition of the subpattern, compilation crashed with
2269 an internal error, complaining that it could not find the referenced
2270 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2271 [The bug was that it was starting one character too far in when skipping
2272 over the character class, thus treating the ] as data rather than
2273 terminating the class. This meant it could skip too much.]
2275 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2276 /g option in pcretest when the pattern contains \K, which makes it possible
2277 to have an empty string match not at the start, even when the pattern is
2278 anchored. Updated pcretest and pcredemo to use this option.
2280 21. If the maximum number of capturing subpatterns in a recursion was greater
2281 than the maximum at the outer level, the higher number was returned, but
2282 with unset values at the outer level. The correct (outer level) value is
2283 now given.
2285 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2286 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2287 make it do so. The string so far is captured, making this feature
2288 compatible with Perl.
2290 23. The tests have been re-organized, adding tests 11 and 12, to make it
2291 possible to check the Perl 5.10 features against Perl 5.10.
2293 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2294 pattern matches a fixed length string. PCRE did not allow this; now it
2295 does. Neither allows recursion.
2297 25. I finally figured out how to implement a request to provide the minimum
2298 length of subject string that was needed in order to match a given pattern.
2299 (It was back references and recursion that I had previously got hung up
2300 on.) This code has now been added to pcre_study(); it finds a lower bound
2301 to the length of subject needed. It is not necessarily the greatest lower
2302 bound, but using it to avoid searching strings that are too short does give
2303 some useful speed-ups. The value is available to calling programs via
2304 pcre_fullinfo().
2306 26. While implementing 25, I discovered to my embarrassment that pcretest had
2307 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2308 study optimizations had never been tested with that matching function.
2309 Oops. What is worse, even when it was passed study data, there was a bug in
2310 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2311 were also very few tests of studied patterns with pcre_dfa_exec().
2313 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2314 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2315 on the other side of the coin, they are no longer allowed to have different
2316 names, because these cannot be distinguished in PCRE, and this has caused
2317 confusion. (This is a difference from Perl.)
2319 28. When duplicate subpattern names are present (necessarily with different
2320 numbers, as required by 27 above), and a test is made by name in a
2321 conditional pattern, either for a subpattern having been matched, or for
2322 recursion in such a pattern, all the associated numbered subpatterns are
2323 tested, and the overall condition is true if the condition is true for any
2324 one of them. This is the way Perl works, and is also more like the way
2325 testing by number works.
2328 Version 7.9 11-Apr-09
2329 ---------------------
2331 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2332 (pcretest), all targets were linked against these libraries. This included
2333 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2334 libraries. This caused unwanted dependencies to be created. This problem
2335 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2336 pcretest is linked with readline.
2338 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2339 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2340 moved outside it again, because FALSE and TRUE are already defined in AIX,
2341 but BOOL is not.
2343 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2344 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2346 4. The pcregrep documentation said spaces were inserted as well as colons (or
2347 hyphens) following file names and line numbers when outputting matching
2348 lines. This is not true; no spaces are inserted. I have also clarified the
2349 wording for the --colour (or --color) option.
2351 5. In pcregrep, when --colour was used with -o, the list of matching strings
2352 was not coloured; this is different to GNU grep, so I have changed it to be
2353 the same.
2355 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2356 each matching line was coloured. Now it goes on to look for further matches
2357 of any of the test patterns, which is the same behaviour as GNU grep.
2359 7. A pattern that could match an empty string could cause pcregrep to loop; it
2360 doesn't make sense to accept an empty string match in pcregrep, so I have
2361 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2362 seems to be how GNU grep behaves. [But see later change 40 for release
2363 8.33.]
2365 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2366 start or after a newline", because the conditional assertion was not being
2367 correctly handled. The rule now is that both the assertion and what follows
2368 in the first alternative must satisfy the test.
2370 9. If auto-callout was enabled in a pattern with a conditional group whose
2371 condition was an assertion, PCRE could crash during matching, both with
2372 pcre_exec() and pcre_dfa_exec().
2374 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2375 used for matching.
2377 11. Unicode property support in character classes was not working for
2378 characters (bytes) greater than 127 when not in UTF-8 mode.
2380 12. Added the -M command line option to pcretest.
2382 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2384 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2386 16. Added comments and documentation about mis-use of no_arg in the C++
2387 wrapper.
2389 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2390 from Martin Jerabek that uses macro names for all relevant character and
2391 string constants.
2393 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2394 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2395 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2396 these, but not everybody uses configure.
2398 19. A conditional group that had only one branch was not being correctly
2399 recognized as an item that could match an empty string. This meant that an
2400 enclosing group might also not be so recognized, causing infinite looping
2401 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2402 with the subject "ab", where knowledge that the repeated group can match
2403 nothing is needed in order to break the loop.
2405 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2406 exec(), but without supplying a callout function, matching went wrong.
2408 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2409 leak if the size of the offset vector was greater than 30. When the vector
2410 is smaller, the saved offsets during recursion go onto a local stack
2411 vector, but for larger vectors malloc() is used. It was failing to free
2412 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2413 error, in fact).
2415 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2416 heapframe that is used only when UTF-8 support is enabled. This caused no
2417 problem, but was untidy.
2419 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2420 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2421 included within another project.
2423 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2424 slightly modified by me:
2426 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2427 not building pcregrep.
2429 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2430 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2432 25. Forward references, both numeric and by name, in patterns that made use of
2433 duplicate group numbers, could behave incorrectly or give incorrect errors,
2434 because when scanning forward to find the reference group, PCRE was not
2435 taking into account the duplicate group numbers. A pattern such as
2436 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2438 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2439 the feature of a custom pointer more persuasive (as requested by a user).
2441 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2442 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2443 pre-defined.
2445 28. Added support for (*UTF8) at the start of a pattern.
2447 29. Arrange for flags added by the "release type" setting in CMake to be shown
2448 in the configuration summary.
2451 Version 7.8 05-Sep-08
2452 ---------------------
2454 1. Replaced UCP searching code with optimized version as implemented for Ad
2455 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2456 stage table and inline lookup instead of a function, giving speed ups of 2
2457 to 5 times on some simple patterns that I tested. Permission was given to
2458 distribute the MultiStage2.py script that generates the tables (it's not in
2459 the tarball, but is in the Subversion repository).
2461 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2462 scripts.
2464 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2465 a group with a zero qualifier. The result of the study could be incorrect,
2466 or the function might crash, depending on the pattern.
2468 4. Caseless matching was not working for non-ASCII characters in back
2469 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2470 It now works when Unicode Property Support is available.
2472 5. In pcretest, an escape such as \x{de} in the data was always generating
2473 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2474 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2475 truncation.
2477 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2479 7. Added two (int) casts to pcregrep when printing the difference of two
2480 pointers, in case they are 64-bit values.
2482 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2483 test 2 if it fails.
2485 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2486 and a #define of that name to empty if it is not externally set. This is to
2487 allow users of MSVC to set it if necessary.
2489 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2490 the convenience functions in the pcre_get.c source file.
2492 11. An option change at the start of a pattern that had top-level alternatives
2493 could cause overwriting and/or a crash. This command provoked a crash in
2494 some environments:
2496 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2498 This potential security problem was recorded as CVE-2008-2371.
2500 12. For a pattern where the match had to start at the beginning or immediately
2501 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2502 pcre_dfa_exec() could read past the end of the passed subject if there was
2503 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2504 pcretest so that it places the subject at the end of its malloc-ed buffer.
2506 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2507 exec() might read past the end of the data buffer in UTF-8 mode.
2509 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2510 the data contained the byte 0x85 as part of a UTF-8 character within its
2511 first line. This applied both to normal and DFA matching.
2513 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2514 /^[^d]*?$/8 failed to match "abc".
2516 16. Added a missing copyright notice to pcrecpp_internal.h.
2518 17. Make it more clear in the documentation that values returned from
2519 pcre_exec() in ovector are byte offsets, not character counts.
2521 18. Tidied a few places to stop certain compilers from issuing warnings.
2523 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2524 supplied by Stefan Weber. I made a further small update for 7.8 because
2525 there is a change of source arrangements: the pcre_searchfuncs.c module is
2526 replaced by pcre_ucd.c.
2529 Version 7.7 07-May-08
2530 ---------------------
2532 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2533 a string to a long long, pretend we don't even have a long long." This is
2534 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2536 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2537 pre-7.6 versions, which defined a global no_arg variable instead of putting
2538 it in the RE class. (See also #8 below.)
2540 3. Remove a line of dead code, identified by coverity and reported by Nuno
2541 Lopes.
2543 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2545 (1) The include/exclude patterns were being applied to the whole pathnames
2546 of files, instead of just to the final components.
2548 (2) If there was more than one level of directory, the subdirectories were
2549 skipped unless they satisfied the include/exclude conditions. This is
2550 inconsistent with GNU grep (and could even be seen as contrary to the
2551 pcregrep specification - which I improved to make it absolutely clear).
2552 The action now is always to scan all levels of directory, and just
2553 apply the include/exclude patterns to regular files.
2555 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2556 --exclude_dir in the tests to avoid scanning .svn directories.
2558 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2559 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2560 doesn't support NULs in patterns.
2562 7. Added some missing "const"s to declarations of static tables in
2563 pcre_compile.c and pcre_dfa_exec.c.
2565 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2566 caused by fix #2 above. (Subsequently also a second patch to fix the
2567 first patch. And a third patch - this was a messy problem.)
2569 9. Applied Craig's patch to remove the use of push_back().
2571 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2572 matching function regexec().
2574 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2575 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2576 references. PCRE supports relative numbers with this syntax (I don't think
2577 Oniguruma does).
2579 12. Previously, a group with a zero repeat such as (...){0} was completely
2580 omitted from the compiled regex. However, this means that if the group
2581 was called as a subroutine from elsewhere in the pattern, things went wrong
2582 (an internal error was given). Such groups are now left in the compiled
2583 pattern, with a new opcode that causes them to be skipped at execution
2584 time.
2586 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2587 to the way PCRE behaves:
2589 (a) A lone ] character is dis-allowed (Perl treats it as data).
2591 (b) A back reference to an unmatched subpattern matches an empty string
2592 (Perl fails the current match path).
2594 (c) A data ] in a character class must be notated as \] because if the
2595 first data character in a class is ], it defines an empty class. (In
2596 Perl it is not possible to have an empty class.) The empty class []
2597 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2598 The negative empty class [^] matches any one character, independently
2599 of the DOTALL setting.
2601 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2602 non-existent subpattern following a character class starting with ']' and
2603 containing () gave an internal compiling error instead of "reference to
2604 non-existent subpattern". Fortunately, when the pattern did exist, the
2605 compiled code was correct. (When scanning forwards to check for the
2606 existence of the subpattern, it was treating the data ']' as terminating
2607 the class, so got the count wrong. When actually compiling, the reference
2608 was subsequently set up correctly.)
2610 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2611 it was being rejected as not supported by pcre_dfa_exec(), even though
2612 other assertions are supported. I have made pcre_dfa_exec() support
2613 (*FAIL).
2615 16. The implementation of 13c above involved the invention of a new opcode,
2616 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2617 cannot be changed at match time, I realized I could make a small
2618 improvement to matching performance by compiling OP_ALLANY instead of
2619 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2620 on the OP_ANY path.
2622 17. Compiling pcretest on Windows with readline support failed without the
2623 following two fixes: (1) Make the unistd.h include conditional on
2624 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2626 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2627 ncurses library to be included for pcretest when ReadLine support is
2628 requested, but also to allow for it to be overridden. This patch came from
2629 Daniel Bergström.
2631 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2632 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2633 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2634 spotting this.
2637 Version 7.6 28-Jan-08
2638 ---------------------
2640 1. A character class containing a very large number of characters with
2641 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2642 overflow.
2644 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2645 HAVE_LONG_LONG is not defined.
2647 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2648 bring them up to date and include new features. This patch includes:
2650 - Fixed PH's badly added libz and libbz2 support.
2651 - Fixed a problem with static linking.
2652 - Added pcredemo. [But later removed - see 7 below.]
2653 - Fixed dftables problem and added an option.
2654 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2656 - Added readline support for pcretest.
2657 - Added an listing of the option settings after cmake has run.
2659 4. A user submitted a patch to Makefile that makes it easy to create
2660 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2661 Makefile.am that cause it to include this special target, without
2662 affecting anything else. Note that the same mingw target plus all
2663 the other distribution libraries and programs are now supported
2664 when configuring with CMake (see 6 below) instead of with
2665 Configure/Make.
2667 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2668 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2669 exported in the Windows port". It has not yet been confirmed that the patch
2670 solves the problem, but it does no harm.
2672 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2673 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2674 with CMake, and also correct the comment about stack recursion.
2676 7. Remove the automatic building of pcredemo from the ./configure system and
2677 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2678 of a program that users should build themselves after PCRE is installed, so
2679 building it automatically is not really right. What is more, it gave
2680 trouble in some build environments.
2682 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2685 Version 7.5 10-Jan-08
2686 ---------------------
2688 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2689 values in parens when parsing an RE using the C++ wrapper."
2691 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2692 Characters greater than 255 were excluded from the class instead of being
2693 included.
2695 3. The same bug as (2) above applied to negated POSIX classes such as
2696 [:^space:].
2698 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2699 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2700 I have changed it.
2702 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2703 first named subpattern) and a construct such as (?&a) would reference the
2704 first named subpattern whose name started with "a" (in other words, the
2705 length check was missing). Both these problems are fixed. "Subpattern name
2706 expected" is now given for (?&) (a zero-length name), and this patch also
2707 makes it give the same error for \k'' (previously it complained that that
2708 was a reference to a non-existent subpattern).
2710 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2711 this is right because (?- can be followed by option settings as well as by
2712 digits. I have, however, made the messages clearer.
2714 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2715 than the number used in the conditional) now cause a compile-time error.
2716 This is actually not compatible with Perl, which accepts such patterns, but
2717 treats the conditional as always being FALSE (as PCRE used to), but it
2718 seems to me that giving a diagnostic is better.
2720 8. Change "alphameric" to the more common word "alphanumeric" in comments
2721 and messages.
2723 9. Fix two occurrences of "backslash" in comments that should have been
2724 "backspace".
2726 10. Remove two redundant lines of code that can never be obeyed (their function
2727 was moved elsewhere).
2729 11. The program that makes PCRE's Unicode character property table had a bug
2730 which caused it to generate incorrect table entries for sequences of
2731 characters that have the same character type, but are in different scripts.
2732 It amalgamated them into a single range, with the script of the first of
2733 them. In other words, some characters were in the wrong script. There were
2734 thirteen such cases, affecting characters in the following ranges:
2736 U+002b0 - U+002c1
2737 U+0060c - U+0060d
2738 U+0061e - U+00612
2739 U+0064b - U+0065e
2740 U+0074d - U+0076d
2741 U+01800 - U+01805
2742 U+01d00 - U+01d77
2743 U+01d9b - U+01dbf
2744 U+0200b - U+0200f
2745 U+030fc - U+030fe
2746 U+03260 - U+0327f
2747 U+0fb46 - U+0fbb1
2748 U+10450 - U+1049d
2750 12. The -o option (show only the matching part of a line) for pcregrep was not
2751 compatible with GNU grep in that, if there was more than one match in a
2752 line, it showed only the first of them. It now behaves in the same way as
2753 GNU grep.
2755 13. If the -o and -v options were combined for pcregrep, it printed a blank
2756 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2757 does the same. The return code can be used to tell if there were any
2758 non-matching lines.
2760 14. Added --file-offsets and --line-offsets to pcregrep.
2762 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2763 infinitely looping recursion. The bug was that positive lookaheads were not
2764 being skipped when checking for a possible empty match (negative lookaheads
2765 and both kinds of lookbehind were skipped).
2767 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2768 inclusion of <windows.h> to before rather than after the definition of
2769 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2771 17. Specifying a possessive quantifier with a specific limit for a Unicode
2772 character property caused pcre_compile() to compile bad code, which led at
2773 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2774 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2775 caused the error; without that there was no problem.
2777 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2779 19. Added --enable-pcretest-libreadline.
2781 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2782 RE::GlobalReplace(). As a result, the number of replacements returned was
2783 double what it should be. I removed one of the increments, but Craig sent a
2784 later patch that removed the other one (the right fix) and added unit tests
2785 that check the return values (which was not done before).
2787 21. Several CMake things:
2789 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2790 the names libpcre and libpcreposix, not just pcre and pcreposix.
2792 (2) The above change means that pcretest and pcregrep are now correctly
2793 linked with the newly-built libraries, not previously installed ones.
2797 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2798 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2799 UTF-8 newline character). The key issue is that the pattern starts .*;
2800 this means that the match must be either at the beginning, or after a
2801 newline. The bug was in the code for advancing after a failed match and
2802 checking that the new position followed a newline. It was not taking
2803 account of UTF-8 characters correctly.
2805 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2806 character classes. PCRE was not treating the sequence [:...:] as a
2807 character class unless the ... were all letters. Perl, however, seems to
2808 allow any characters between [: and :], though of course it rejects as
2809 unknown any "names" that contain non-letters, because all the known class
2810 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2811 for example, whereas PCRE did not - it did not recognize a POSIX character
2812 class. This seemed a bit dangerous, so the code has been changed to be
2813 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2814 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2815 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2816 Perl does, and where it didn't before.
2818 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2819 Windows environments %n is disabled by default.
2822 Version 7.4 21-Sep-07
2823 ---------------------
2825 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2826 means that a class such as [\s] counted as "explicit reference to CR or
2827 LF". That isn't really right - the whole point of the change was to try to
2828 help when there was an actual mention of one of the two characters. So now
2829 the change happens only if \r or \n (or a literal CR or LF) character is
2830 encountered.
2832 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2833 of both had grown to the point where there were only 3 bits left.
2834 Fortunately, there was spare space in the data structure, and so I have
2835 moved the internal flags into a new 16-bit field to free up more option
2836 bits.
2838 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2839 but did not set the internal JCHANGED flag - either of these is enough to
2840 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2841 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2842 start sets both bits.
2844 4. Added options (at build time, compile time, exec time) to change \R from
2845 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2847 5. doc/pcresyntax.html was missing from the distribution.
2849 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2850 compatibility, even though it is no longer used.
2852 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2853 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2854 windows.h file is present (where different names are used). [This was
2855 reversed later after testing - see 16 below.]
2857 8. Changed all #include <config.h> to #include "config.h". There were also
2858 some further <pcre.h> cases that I changed to "pcre.h".
2860 9. When pcregrep was used with the --colour option, it missed the line ending
2861 sequence off the lines that it output.
2863 10. It was pointed out to me that arrays of string pointers cause lots of
2864 relocations when a shared library is dynamically loaded. A technique of
2865 using a single long string with a table of offsets can drastically reduce
2866 these. I have refactored PCRE in four places to do this. The result is
2867 dramatic:
2869 Originally: 290
2870 After changing UCP table: 187
2871 After changing error message table: 43
2872 After changing table of "verbs" 36
2873 After changing table of Posix names 22
2875 Thanks to the folks working on Gregex for glib for this insight.
2877 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2878 unicode-properties was also set.
2880 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2882 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2883 checked only for CRLF.
2885 14. Added casts to pcretest.c to avoid compiler warnings.
2887 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2889 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2890 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2891 entirely. This removes changes made in 7 above.
2893 17. The CMake files have been updated, and there is now more information about
2894 building with CMake in the NON-UNIX-USE document.
2897 Version 7.3 28-Aug-07
2898 ---------------------
2900 1. In the rejigging of the build system that eventually resulted in 7.1, the
2901 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2902 brackets there is not right, since it causes compilers to look for an
2903 installed pcre.h, not the version that is in the source that is being
2904 compiled (which of course may be different). I have changed it back to:
2906 #include "pcre.h"
2908 I have a vague recollection that the change was concerned with compiling in
2909 different directories, but in the new build system, that is taken care of
2910 by the VPATH setting the Makefile.
2912 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2913 when the subject happened to end in the byte 0x85 (e.g. if the last
2914 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2915 characters but of course it shouldn't be taken as a newline when it is part
2916 of another character. The bug was that, for an unlimited repeat of . in
2917 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2918 characters when looking for a newline.
2920 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2922 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2923 in debug output.
2925 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2926 long printing in the pcrecpp unittest when running under MinGW.
2928 6. ESC_K was left out of the EBCDIC table.
2930 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2931 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2932 limit also applies to "virtual nesting" when a pattern is recursive, and in
2933 this case 1000 isn't so big. I have been able to remove this limit at the
2934 expense of backing off one optimization in certain circumstances. Normally,
2935 when pcre_exec() would call its internal match() function recursively and
2936 immediately return the result unconditionally, it uses a "tail recursion"
2937 feature to save stack. However, when a subpattern that can match an empty
2938 string has an unlimited repetition quantifier, it no longer makes this
2939 optimization. That gives it a stack frame in which to save the data for
2940 checking that an empty string has been matched. Previously this was taken
2941 from the 1000-entry workspace that had been reserved. So now there is no
2942 explicit limit, but more stack is used.
2944 8. Applied Daniel's patches to solve problems with the import/export magic
2945 syntax that is required for Windows, and which was going wrong for the
2946 pcreposix and pcrecpp parts of the library. These were overlooked when this
2947 problem was solved for the main library.
2949 9. There were some crude static tests to avoid integer overflow when computing
2950 the size of patterns that contain repeated groups with explicit upper
2951 limits. As the maximum quantifier is 65535, the maximum group length was
2952 set at 30,000 so that the product of these two numbers did not overflow a
2953 32-bit integer. However, it turns out that people want to use groups that
2954 are longer than 30,000 bytes (though not repeat them that many times).
2955 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2956 made it possible to implement the integer overflow checks in a much more
2957 dynamic way, which I have now done. The artificial limitation on group
2958 length has been removed - we now have only the limit on the total length of
2959 the compiled pattern, which depends on the LINK_SIZE setting.
2961 10. Fixed a bug in the documentation for get/copy named substring when
2962 duplicate names are permitted. If none of the named substrings are set, the
2963 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2964 empty string.
2966 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2967 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2968 because the ] is interpreted as the first data character and the
2969 terminating ] is not found. PCRE has been made compatible with Perl in this
2970 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2971 cause memory overwriting.
2973 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2974 string has been matched (to stop an infinite loop). It was not recognizing
2975 a conditional subpattern that could match an empty string if that
2976 subpattern was within another subpattern. For example, it looped when
2977 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2978 condition was not nested. This bug has been fixed.
2980 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2981 past the start of the subject in the presence of bytes with the top bit
2982 set, for example "\x8aBCD".
2984 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2985 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2987 14. Optimized (?!) to (*FAIL).
2989 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2990 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2991 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2992 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2993 does: it's just the validity check that is more restrictive.
2995 16. Inserted checks for integer overflows during escape sequence (backslash)
2996 processing, and also fixed erroneous offset values for syntax errors during
2997 backslash processing.
2999 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3000 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3002 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3003 caused an overrun.
3005 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3006 something other than just ASCII characters) inside a group that had an
3007 unlimited repeat caused a loop at compile time (while checking to see
3008 whether the group could match an empty string).
3010 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3011 [\P{Any}] did so. (Error in the code for printing property names.)
3013 21. An orphan \E inside a character class could cause a crash.
3015 22. A repeated capturing bracket such as (A)? could cause a wild memory
3016 reference during compilation.
3018 23. There are several functions in pcre_compile() that scan along a compiled
3019 expression for various reasons (e.g. to see if it's fixed length for look
3020 behind). There were bugs in these functions when a repeated \p or \P was
3021 present in the pattern. These operators have additional parameters compared
3022 with \d, etc, and these were not being taken into account when moving along
3023 the compiled data. Specifically:
3025 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3026 length.
3028 (b) An item such as \pL+ within a repeated group could cause crashes or
3029 loops.
3031 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3032 "reference to non-existent subpattern" error.
3034 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3036 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3037 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3039 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3041 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3042 character were causing crashes (broken optimization).
3044 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3045 \p or \P) caused a compile-time loop.
3047 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3048 break. For example, the unstudied pattern [\r\n]A does not match the string
3049 "\r\nA" because change 7.0/46 below moves the current point on by two
3050 characters after failing to match at the start. However, the pattern \nA
3051 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3052 the same is true. There doesn't seem any very clean way out of this, but
3053 what I have chosen to do makes the common cases work: PCRE now takes note
3054 of whether there can be an explicit match for \r or \n anywhere in the
3055 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3056 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3057 pattern has explicit CR or LF references.
3059 29. Added (*CR) etc for changing newline setting at start of pattern.
3062 Version 7.2 19-Jun-07
3063 ---------------------
3065 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3066 which is apparently normally available under Windows.
3068 2. Re-jig the pcregrep tests with different newline settings in an attempt
3069 to make them independent of the local environment's newline setting.
3071 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3073 4. Some of the "internals" tests were previously cut out when the link size
3074 was not 2, because the output contained actual offsets. The recent new
3075 "Z" feature of pcretest means that these can be cut out, making the tests
3076 usable with all link sizes.
3078 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3079 stack recursion. This gives a massive performance boost under BSD, but just
3080 a small improvement under Linux. However, it saves one field in the frame
3081 in all cases.
3083 6. Added more features from the forthcoming Perl 5.10:
3085 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3086 recursion call. It refers to the nth most recently opened parentheses.
3088 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3089 to be opened parentheses.
3091 (c) Conditions that refer to capturing parentheses can be specified
3092 relatively, for example, (?(-2)... or (?(+3)...
3094 (d) \K resets the start of the current match so that everything before
3095 is not part of it.
3097 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3099 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3100 reference syntax.
3102 (g) (?| introduces a group in which the numbering of parentheses in each
3103 alternative starts with the same number.
3105 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3107 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3110 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3111 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3112 for detecting groups that can match an empty string.
3114 9. A pattern with a very large number of alternatives (more than several
3115 hundred) was running out of internal workspace during the pre-compile
3116 phase, where pcre_compile() figures out how much memory will be needed. A
3117 bit of new cunning has reduced the workspace needed for groups with
3118 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3119 workspace instead of running out of the 4096 that are available.
3121 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3123 11. Applied patch from Google to remove an optimization that didn't quite work.
3124 The report of the bug said:
3126 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3127 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3128 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3130 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3131 it matched the wrong number of bytes.
3134 Version 7.1 24-Apr-07
3135 ---------------------
3137 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3138 that is more "standard", making use of automake and other Autotools. There
3139 is some re-arrangement of the files and adjustment of comments consequent
3140 on this.
3142 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3143 for recursive directory scanning broke on some systems because the files
3144 are not scanned in any specific order and on different systems the order
3145 was different. A call to "sort" has been inserted into RunGrepTest for the
3146 approprate test as a short-term fix. In the longer term there may be an
3147 alternative.
3149 3. I had an email from Eric Raymond about problems translating some of PCRE's
3150 man pages to HTML (despite the fact that I distribute HTML pages, some
3151 people do their own conversions for various reasons). The problems
3152 concerned the use of low-level troff macros .br and .in. I have therefore
3153 removed all such uses from the man pages (some were redundant, some could
3154 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3155 HTML has been updated to handle .nf/.fi and to complain if it encounters
3156 .br or .in.
3158 4. Updated comments in configure.ac that get placed in config.h.in and also
3159 arranged for config.h to be included in the distribution, with the name
3160 config.h.generic, for the benefit of those who have to compile without
3161 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3163 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3164 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3165 makevp.bat for the current PCRE, using the additional files
3166 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3168 6. A Windows user reported a minor discrepancy with test 2, which turned out
3169 to be caused by a trailing space on an input line that had got lost in his
3170 copy. The trailing space was an accident, so I've just removed it.
3172 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3173 that is needed.
3175 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3176 as "const" (a) because they are and (b) because it helps the PHP
3177 maintainers who have recently made a script to detect big data structures
3178 in the php code that should be moved to the .rodata section. I remembered
3179 to update Builducptable as well, so it won't revert if ucptable.h is ever
3180 re-created.
3182 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3183 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3184 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3185 support is not required. This saves 1.5-2K of code, which is important in
3186 some applications.
3188 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3189 so as not to refer to the tables, even though these functions will never be
3190 called when UTF-8 support is disabled. Otherwise there are problems with a
3191 shared library.
3193 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3195 (a) It was defining its arguments as char * instead of void *.
3197 (b) It was assuming that all moves were upwards in memory; this was true
3198 a long time ago when I wrote it, but is no longer the case.
3200 The emulated memove() is provided for those environments that have neither
3201 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3202 is clearly not the case, as these two bugs were recently reported.
3204 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3205 and Detrail to create the HTML documentation, the .txt form of the man
3206 pages, and it removes trailing spaces from listed files. It also creates
3207 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3208 case, it wraps all the #defines with #ifndefs. This script should be run
3209 before "make dist".
3211 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3212 with Unicode property support.
3214 (a) For a maximizing quantifier, if the two different cases of the
3215 character were of different lengths in their UTF-8 codings (there are
3216 some cases like this - I found 11), and the matching function had to
3217 back up over a mixture of the two cases, it incorrectly assumed they
3218 were both the same length.
3220 (b) When PCRE was configured to use the heap rather than the stack for
3221 recursion during matching, it was not correctly preserving the data for
3222 the other case of a UTF-8 character when checking ahead for a match
3223 while processing a minimizing repeat. If the check also involved
3224 matching a wide character, but failed, corruption could cause an
3225 erroneous result when trying to check for a repeat of the original
3226 character.
3228 13. Some tidying changes to the testing mechanism:
3230 (a) The RunTest script now detects the internal link size and whether there
3231 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3232 values substituted by "configure". (The RunGrepTest script already did
3233 this for UTF-8.) The configure.ac script no longer substitutes the
3234 relevant variables.
3236 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3237 with length and offset values. This means that the output is different
3238 for different internal link sizes. Test 2 is skipped for link sizes
3239 other than 2 because of this, bypassing the problem. Unfortunately,
3240 there was also a test in test 3 (the locale tests) that used /B and
3241 failed for link sizes other than 2. Rather than cut the whole test out,
3242 I have added a new /Z option to pcretest that replaces the length and
3243 offset values with spaces. This is now used to make test 3 independent
3244 of link size. (Test 2 will be tidied up later.)
3246 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3247 segmentation fault instead of returning the appropriate error message.
3249 15. In multiline mode when the newline sequence was set to "any", the pattern
3250 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3251 This doesn't seem right; it now treats the CRLF combination as the line
3252 ending, and so does not match in that case. It's only a pattern such as ^$
3253 that would hit this one: something like ^ABC$ would have failed after \r
3254 and then tried again after \r\n.
3256 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3257 in an attempt to make files that differ only in their line terminators
3258 compare equal. This works on Linux.
3260 17. Under certain error circumstances pcregrep might try to free random memory
3261 as it exited. This is now fixed, thanks to valgrind.
3263 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3264 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3265 was because its rules for how to advance for /g after matching an empty
3266 string at the end of a line did not allow for this case. They now check for
3267 it specially.
3269 20. pcretest is supposed to handle patterns and data of any length, by
3270 extending its buffers when necessary. It was getting this wrong when the
3271 buffer for a data line had to be extended.
3273 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3274 CRLF as a newline sequence.
3276 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3277 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3278 I have nevertheless tidied it up.
3280 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3282 24. Added a man page for pcre-config.
3285 Version 7.0 19-Dec-06
3286 ---------------------
3288 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3289 moving to gcc 4.1.1.
3291 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3292 sys/time.h, which is documented as needed for this function. It doesn't
3293 seem to matter on Linux, but it showed up on some releases of OS X.
3295 3. It seems that there are systems where bytes whose values are greater than
3296 127 match isprint() in the "C" locale. The "C" locale should be the
3297 default when a C program starts up. In most systems, only ASCII printing
3298 characters match isprint(). This difference caused the output from pcretest
3299 to vary, making some of the tests fail. I have changed pcretest so that:
3301 (a) When it is outputting text in the compiled version of a pattern, bytes
3302 other than 32-126 are always shown as hex escapes.
3304 (b) When it is outputting text that is a matched part of a subject string,
3305 it does the same, unless a different locale has been set for the match
3306 (using the /L modifier). In this case, it uses isprint() to decide.
3308 4. Fixed a major bug that caused incorrect computation of the amount of memory
3309 required for a compiled pattern when options that changed within the
3310 pattern affected the logic of the preliminary scan that determines the
3311 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3312 that the computed length was too small. The symptoms of this bug were
3313 either the PCRE error "internal error: code overflow" from pcre_compile(),
3314 or a glibc crash with a message such as "pcretest: free(): invalid next
3315 size (fast)". Examples of patterns that provoked this bug (shown in
3316 pcretest format) are:
3318 /(?-x: )/x
3319 /(?x)(?-x: \s*#\s*)/
3320 /((?i)[\x{c0}])/8
3321 /(?i:[\x{c0}])/8
3323 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3324 is now done differently.
3326 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3327 wrapper classes; (b) implement a new function in the C++ scanner that is
3328 more efficient than the old way of doing things because it avoids levels of
3329 recursion in the regex matching; (c) add a paragraph to the documentation
3330 for the FullMatch() function.
3332 6. The escape sequence \n was being treated as whatever was defined as
3333 "newline". Not only was this contrary to the documentation, which states
3334 that \n is character 10 (hex 0A), but it also went horribly wrong when
3335 "newline" was defined as CRLF. This has been fixed.
3337 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3338 was being set to -1 for the "end of line" case (supposedly a value that no
3339 character can have). Though this value is never used (the check for end of
3340 line is "zero bytes in current character"), it caused compiler complaints.
3341 I've changed it to 0xffffffff.
3343 8. In pcre_version.c, the version string was being built by a sequence of
3344 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3345 string (as it is for production releases) called a macro with an empty
3346 argument. The C standard says the result of this is undefined. The gcc
3347 compiler treats it as an empty string (which was what was wanted) but it is
3348 reported that Visual C gives an error. The source has been hacked around to
3349 avoid this problem.
3351 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3352 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3353 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3354 of them did).
3356 10. Originally, pcretest opened its input and output without "b"; then I was
3357 told that "b" was needed in some environments, so it was added for release
3358 5.0 to both the input and output. (It makes no difference on Unix-like
3359 systems.) Later I was told that it is wrong for the input on Windows. I've
3360 now abstracted the modes into two macros, to make it easier to fiddle with
3361 them, and removed "b" from the input mode under Windows.
3363 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3365 12. Added -help and --help to pcretest as an official way of being reminded
3366 of the options.
3368 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3369 and pcrecpp.cc because they annoy compilers at high warning levels.
3371 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3373 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3374 scripts are not C programs :-) and which was not noticed because it works
3375 on Linux.
3377 16. pcretest is supposed to handle any length of pattern and data line (as one
3378 line or as a continued sequence of lines) by extending its input buffer if
3379 necessary. This feature was broken for very long pattern lines, leading to
3380 a string of junk being passed to pcre_compile() if the pattern was longer
3381 than about 50K.
3383 17. I have done a major re-factoring of the way pcre_compile() computes the
3384 amount of memory needed for a compiled pattern. Previously, there was code
3385 that made a preliminary scan of the pattern in order to do this. That was
3386 OK when PCRE was new, but as the facilities have expanded, it has become
3387 harder and harder to keep it in step with the real compile phase, and there
3388 have been a number of bugs (see for example, 4 above). I have now found a
3389 cunning way of running the real compile function in a "fake" mode that
3390 enables it to compute how much memory it would need, while actually only
3391 ever using a few hundred bytes of working memory and without too many
3392 tests of the mode. This should make future maintenance and development
3393 easier. A side effect of this work is that the limit of 200 on the nesting
3394 depth of parentheses has been removed (though this was never a serious
3395 limitation, I suspect). However, there is a downside: pcre_compile() now
3396 runs more slowly than before (30% or more, depending on the pattern). I
3397 hope this isn't a big issue. There is no effect on runtime performance.
3399 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3400 newline (only possible for the last line of a file) and it was a
3401 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3403 19. Added additional timing features to pcretest. (1) The -tm option now times
3404 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3405 separate command line item, by a number that specifies the number of
3406 repeats to use when timing. The default is 50000; this gives better
3407 precision, but takes uncomfortably long for very large patterns.
3409 20. Extended pcre_study() to be more clever in cases where a branch of a
3410 subpattern has no definite first character. For example, (a*|b*)[cd] would
3411 previously give no result from pcre_study(). Now it recognizes that the
3412 first character must be a, b, c, or d.
3414 21. There was an incorrect error "recursive call could loop indefinitely" if
3415 a subpattern (or the entire pattern) that was being tested for matching an
3416 empty string contained only one non-empty item after a nested subpattern.
3417 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3418 incorrectly, because the \d was being skipped in the check.
3420 22. The pcretest program now has a new pattern option /B and a command line
3421 option -b, which is equivalent to adding /B to every pattern. This causes
3422 it to show the compiled bytecode, without the additional information that
3423 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3424 is the same as /B/I).
3426 23. A new optimization is now able automatically to treat some sequences such
3427 as a*b as a*+b. More specifically, if something simple (such as a character
3428 or a simple class like \d) has an unlimited quantifier, and is followed by
3429 something that cannot possibly match the quantified thing, the quantifier
3430 is automatically "possessified".
3432 24. A recursive reference to a subpattern whose number was greater than 39
3433 went wrong under certain circumstances in UTF-8 mode. This bug could also
3434 have affected the operation of pcre_study().
3436 25. Realized that a little bit of performance could be had by replacing
3437 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3439 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3441 27. Possessive quantifiers such as a++ were previously implemented by turning
3442 them into atomic groups such as ($>a+). Now they have their own opcodes,
3443 which improves performance. This includes the automatically created ones
3444 from 23 above.
3446 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3447 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3448 the first matched character to be a colon. This applied both to named and
3449 numbered groups.
3451 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3453 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3454 building PCRE on Windows easier, so I have included it in the distribution.
3456 31. There is now a check in pcretest against a ridiculously large number being
3457 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3458 loop, the loop is abandoned.
3460 32. Forward references to subpatterns in conditions such as (?(2)...) where
3461 subpattern 2 is defined later cause pcre_compile() to search forwards in
3462 the pattern for the relevant set of parentheses. This search went wrong
3463 when there were unescaped parentheses in a character class, parentheses
3464 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3466 33. "Subroutine" calls and backreferences were previously restricted to
3467 referencing subpatterns earlier in the regex. This restriction has now
3468 been removed.
3470 34. Added a number of extra features that are going to be in Perl 5.10. On the
3471 whole, these are just syntactic alternatives for features that PCRE had
3472 previously implemented using the Python syntax or my own invention. The
3473 other formats are all retained for compatibility.
3475 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3476 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3477 also .NET compatible.
3479 (b) A recursion or subroutine call to a named group can now be defined as
3480 (?&name) as well as (?P>name).
3482 (c) A backreference to a named group can now be defined as \k<name> or
3483 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3484 5.10, are also .NET compatible.
3486 (d) A conditional reference to a named group can now use the syntax
3487 (?(<name>) or (?('name') as well as (?(name).
3489 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3490 groups (named and numbered) that are never evaluated inline, but can be
3491 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3492 is always false. There may be only one alternative in such a group.
3494 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3495 as the simple (?(R). The condition is true only if the most recent
3496 recursion is that of the given number or name. It does not search out
3497 through the entire recursion stack.
3499 (g) The escape \gN or \g{N} has been added, where N is a positive or
3500 negative number, specifying an absolute or relative reference.
3502 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3503 some "unreachable code" warnings.
3505 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3506 things, this adds five new scripts.
3508 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3509 There were also incompatibilities regarding the handling of \Q..\E inside
3510 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3511 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3513 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3514 matches an empty string, and forcibly breaks the loop. There were bugs in
3515 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3516 against aaaa the result was just "a" rather than "aaaa", for example. Two
3517 separate and independent bugs (that affected different cases) have been
3518 fixed.
3520 39. Refactored the code to abolish the use of different opcodes for small
3521 capturing bracket numbers. This is a tidy that I avoided doing when I
3522 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3523 The new approach is not only tidier, it makes it possible to reduce the
3524 memory needed to fix the previous bug (38).
3526 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3527 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3528 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3529 mode.
3531 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3532 report.
3534 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3535 copy construction and assignment in the C++ wrapper.
3537 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3538 couple of bugs that could have given wrong results in the "--newline=crlf"
3539 case.
3541 44. Added a number of casts and did some reorganization of signed/unsigned int
3542 variables following suggestions from Dair Grant. Also renamed the variable
3543 "this" as "item" because it is a C++ keyword.
3545 45. Arranged for dftables to add
3547 #include "pcre_internal.h"
3549 to pcre_chartables.c because without it, gcc 4.x may remove the array
3550 definition from the final binary if PCRE is built into a static library and
3551 dead code stripping is activated.
3553 46. For an unanchored pattern, if a match attempt fails at the start of a
3554 newline sequence, and the newline setting is CRLF or ANY, and the next two
3555 characters are CRLF, advance by two characters instead of one.
3558 Version 6.7 04-Jul-06
3559 ---------------------
3561 1. In order to handle tests when input lines are enormously long, pcretest has
3562 been re-factored so that it automatically extends its buffers when
3563 necessary. The code is crude, but this _is_ just a test program. The
3564 default size has been increased from 32K to 50K.
3566 2. The code in pcre_study() was using the value of the re argument before
3567 testing it for NULL. (Of course, in any sensible call of the function, it
3568 won't be NULL.)
3570 3. The memmove() emulation function in pcre_internal.h, which is used on
3571 systems that lack both memmove() and bcopy() - that is, hardly ever -
3572 was missing a "static" storage class specifier.
3574 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3575 containing an extended class (one that cannot be represented by a bitmap
3576 because it contains high-valued characters or Unicode property items, e.g.
3577 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3578 pattern, but PCRE should not loop if you do not (it no longer does).
3579 [Detail: two cases were found: (a) a repeated subpattern containing an
3580 extended class; (b) a recursive reference to a subpattern that followed a
3581 previous extended class. It wasn't skipping over the extended class
3582 correctly when UTF-8 mode was not set.]
3584 5. A negated single-character class was not being recognized as fixed-length
3585 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3586 compile error "lookbehind assertion is not fixed length".
3588 6. The RunPerlTest auxiliary script was showing an unexpected difference
3589 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3590 write a Perl script that can interpret lines of an input file either as
3591 byte characters or as UTF-8, which is what "perltest" was being required to
3592 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3593 can't do is switch easily at run time between having the "use utf8;" pragma
3594 or not. In the end, I fudged it by using the RunPerlTest script to insert
3595 "use utf8;" explicitly for the UTF-8 tests.
3597 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3598 the end of the subject string, contrary to the documentation and to what
3599 Perl does. This was true of both matching functions. Now it matches only at
3600 the start of the subject and immediately after *internal* newlines.
3602 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3603 a pointer to an int instead of a pointer to an unsigned long int. This
3604 caused problems on 64-bit systems.
3606 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3607 instance of the 'standard' template library not being so standard".
3609 10. There was no check on the number of named subpatterns nor the maximum
3610 length of a subpattern name. The product of these values is used to compute
3611 the size of the memory block for a compiled pattern. By supplying a very
3612 long subpattern name and a large number of named subpatterns, the size
3613 computation could be caused to overflow. This is now prevented by limiting
3614 the length of names to 32 characters, and the number of named subpatterns
3615 to 10,000.
3617 11. Subpatterns that are repeated with specific counts have to be replicated in
3618 the compiled pattern. The size of memory for this was computed from the
3619 length of the subpattern and the repeat count. The latter is limited to
3620 65535, but there was no limit on the former, meaning that integer overflow
3621 could in principle occur. The compiled length of a repeated subpattern is
3622 now limited to 30,000 bytes in order to prevent this.
3624 12. Added the optional facility to have named substrings with the same name.
3626 13. Added the ability to use a named substring as a condition, using the
3627 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3628 are numbers (not recommended). Forward references are permitted.
3630 14. Added forward references in named backreferences (if you see what I mean).
3632 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3633 pattern could run off the end of the subject. For example, the pattern
3634 "(?s)(.{1,5})"8 did this with the subject "ab".
3636 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3637 PCRE_CASELESS was set when matching characters that were quantified with ?
3638 or *.
3640 17. A character class other than a single negated character that had a minimum
3641 but no maximum quantifier - for example [ab]{6,} - was not handled
3642 correctly by pce_dfa_exec(). It would match only one character.
3644 18. A valid (though odd) pattern that looked like a POSIX character
3645 class but used an invalid character after [ (for example [[,abc,]]) caused
3646 pcre_compile() to give the error "Failed: internal error: code overflow" or
3647 in some cases to crash with a glibc free() error. This could even happen if
3648 the pattern terminated after [[ but there just happened to be a sequence of
3649 letters, a binary zero, and a closing ] in the memory that followed.
3651 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3652 over the years. Originally (before any Unicode support), just the bottom 8
3653 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3654 output from "man perlunicode" includes this:
3656 The regular expression compiler produces polymorphic opcodes. That
3657 is, the pattern adapts to the data and automatically switches to
3658 the Unicode character scheme when presented with Unicode data--or
3659 instead uses a traditional byte scheme when presented with byte
3660 data.
3662 Sadly, a wide octal escape does not cause a switch, and in a string with
3663 no other multibyte characters, these octal escapes are treated as before.
3664 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3665 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3666 Unicode string.
3668 I have not perpetrated such confusion in PCRE. Up till now, it took just
3669 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3670 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3671 translate to the appropriate multibyte character.
3673 29. Applied some refactoring to reduce the number of warnings from Microsoft
3674 and Borland compilers. This has included removing the fudge introduced
3675 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3676 a warning about an unused variable.
3678 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3679 characters since release 4.0, because Perl (from release 5.004) does not.
3680 [Or at least, is documented not to: some releases seem to be in conflict
3681 with the documentation.] However, when a pattern was studied with
3682 pcre_study() and all its branches started with \s, PCRE still included VT
3683 as a possible starting character. Of course, this did no harm; it just
3684 caused an unnecessary match attempt.
3686 22. Removed a now-redundant internal flag bit that recorded the fact that case
3687 dependency changed within the pattern. This was once needed for "required
3688 byte" processing, but is no longer used. This recovers a now-scarce options
3689 bit. Also moved the least significant internal flag bit to the most-
3690 significant bit of the word, which was not previously used (hangover from
3691 the days when it was an int rather than a uint) to free up another bit for
3692 the future.
3694 23. Added support for CRLF line endings as well as CR and LF. As well as the
3695 default being selectable at build time, it can now be changed at runtime
3696 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3697 specify that it is scanning data with non-default line endings.
3699 24. Changed the definition of CXXLINK to make it agree with the definition of
3700 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3702 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3703 recursions. This makes a big different to stack usage for some patterns.
3705 26. If a subpattern containing a named recursion or subroutine reference such
3706 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3707 the space required for the compiled pattern went wrong and gave too small a
3708 value. Depending on the environment, this could lead to "Failed: internal
3709 error: code overflow at offset 49" or "glibc detected double free or
3710 corruption" errors.
3712 27. Applied patches from Google (a) to support the new newline modes and (b) to
3713 advance over multibyte UTF-8 characters in GlobalReplace.
3715 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3716 difference for some implementation of PCRE in some Windows version.
3718 29. Added some extra testing facilities to pcretest:
3720 \q<number> in a data line sets the "match limit" value
3721 \Q<number> in a data line sets the "match recursion limt" value
3722 -S <number> sets the stack size, where <number> is in megabytes
3724 The -S option isn't available for Windows.
3727 Version 6.6 06-Feb-06
3728 ---------------------
3730 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3731 in pcreposix.h. I have copied the definition from pcre.h.
3733 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3734 because pcre.h is no longer a built file.
3736 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3737 not normally included in the compiled code.
3740 Version 6.5 01-Feb-06
3741 ---------------------
3743 1. When using the partial match feature with pcre_dfa_exec(), it was not
3744 anchoring the second and subsequent partial matches at the new starting
3745 point. This could lead to incorrect results. For example, with the pattern
3746 /1234/, partially matching against "123" and then "a4" gave a match.
3748 2. Changes to pcregrep:
3750 (a) All non-match returns from pcre_exec() were being treated as failures
3751 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3752 error message is output. Some extra information is given for the
3754 probably the only errors that are likely to be caused by users (by
3755 specifying a regex that has nested indefinite repeats, for instance).
3756 If there are more than 20 of these errors, pcregrep is abandoned.
3758 (b) A binary zero was treated as data while matching, but terminated the
3759 output line if it was written out. This has been fixed: binary zeroes
3760 are now no different to any other data bytes.
3762 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3763 used to set a locale for matching. The --locale=xxxx long option has
3764 been added (no short equivalent) to specify a locale explicitly on the
3765 pcregrep command, overriding the environment variables.
3767 (d) When -B was used with -n, some line numbers in the output were one less
3768 than they should have been.
3770 (e) Added the -o (--only-matching) option.
3772 (f) If -A or -C was used with -c (count only), some lines of context were
3773 accidentally printed for the final match.
3775 (g) Added the -H (--with-filename) option.
3777 (h) The combination of options -rh failed to suppress file names for files
3778 that were found from directory arguments.
3780 (i) Added the -D (--devices) and -d (--directories) options.
3782 (j) Added the -F (--fixed-strings) option.
3784 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3786 (l) Added the --colo(u)r option.
3788 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3789 is not present by default.
3791 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3792 items such as (?R) or (?1), when the recursion could match a number of
3793 alternatives. If it matched one of the alternatives, but subsequently,
3794 outside the recursion, there was a failure, the code tried to back up into
3795 the recursion. However, because of the way PCRE is implemented, this is not
3796 possible, and the result was an incorrect result from the match.
3798 In order to prevent this happening, the specification of recursion has
3799 been changed so that all such subpatterns are automatically treated as
3800 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3802 4. I had overlooked the fact that, in some locales, there are characters for
3803 which isalpha() is true but neither isupper() nor islower() are true. In
3804 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3805 and ordfeminine) are like this. This affected the treatment of \w and \W
3806 when they appeared in character classes, but not when they appeared outside
3807 a character class. The bit map for "word" characters is now created
3808 separately from the results of isalnum() instead of just taking it from the
3809 upper, lower, and digit maps. (Plus the underscore character, of course.)
3811 5. The above bug also affected the handling of POSIX character classes such as
3812 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3813 permanent tables. Instead, the bit maps for such a class were previously
3814 created as the appropriate unions of the upper, lower, and digit bitmaps.
3815 Now they are created by subtraction from the [[:word:]] class, which has
3816 its own bitmap.
3818 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3819 It is created by subtracting the vertical space characters (\x09, \x0a,
3820 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3821 subtraction was done in the overall bitmap for a character class, meaning
3822 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3823 be recognized. This bug has been fixed.
3825 7. Patches from the folks at Google:
3827 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3828 real life, but is still worth protecting against".
3830 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3831 regular expressions".
3833 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3834 have it.
3836 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3837 "configure" and the latter not, in order to fix a problem somebody had
3838 with compiling the Arg class on HP-UX.
3840 (e) Improve the error-handling of the C++ wrapper a little bit.
3842 (f) New tests for checking recursion limiting.
3844 8. The pcre_memmove() function, which is used only if the environment does not
3845 have a standard memmove() function (and is therefore rarely compiled),
3846 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3847 returning a result (though PCRE never actually uses the result).
3849 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3850 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3851 returned instead of calling malloc() with an overflowing number that would
3852 most likely cause subsequent chaos.
3854 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3856 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3857 with this option is matched, the nmatch and pmatch options of regexec() are
3858 ignored.
3860 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3861 provided in case anyone wants to the the POSIX interface with UTF-8
3862 strings.
3864 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3865 C++ linking (needed for some HP-UX environments).
3867 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3868 (unused parameter) and in the pcre_printint() function (omitted "default"
3869 switch label when the default is to do nothing).
3871 15. Added some code to make it possible, when PCRE is compiled as a C++
3872 library, to replace subject pointers for pcre_exec() with a smart pointer
3873 class, thus making it possible to process discontinuous strings.
3875 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3876 much the same function. They were added by different people who were trying
3877 to make PCRE easy to compile on non-Unix systems. It has been suggested
3878 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3879 for compiling on Windows systems. I have therefore replaced it with
3880 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3881 defaults to "extern" for C or "extern C" for C++, which works fine on
3882 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3883 SCOPE with something explicit in config.h. In addition:
3885 (a) pcreposix.h still had just "extern" instead of either of these macros;
3886 I have replaced it with PCRE_DATA_SCOPE.
3888 (b) Functions such as _pcre_xclass(), which are internal to the library,
3889 but external in the C sense, all had PCRE_EXPORT in their definitions.
3890 This is apparently wrong for the Windows case, so I have removed it.
3891 (It makes no difference on Unix-like systems.)
3893 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3894 of recursive calls to match(). This is different to MATCH_LIMIT because
3895 that limits the total number of calls to match(), not all of which increase
3896 the depth of recursion. Limiting the recursion depth limits the amount of
3897 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3898 when PCRE is compiled, and changed at run time. A patch from Google adds
3899 this functionality to the C++ interface.
3901 18. Changes to the handling of Unicode character properties:
3903 (a) Updated the table to Unicode 4.1.0.
3905 (b) Recognize characters that are not in the table as "Cn" (undefined).
3907 (c) I revised the way the table is implemented to a much improved format
3908 which includes recognition of ranges. It now supports the ranges that
3909 are defined in UnicodeData.txt, and it also amalgamates other
3910 characters into ranges. This has reduced the number of entries in the
3911 table from around 16,000 to around 3,000, thus reducing its size
3912 considerably. I realized I did not need to use a tree structure after
3913 all - a binary chop search is just as efficient. Having reduced the
3914 number of entries, I extended their size from 6 bytes to 8 bytes to
3915 allow for more data.
3917 (d) Added support for Unicode script names via properties such as \p{Han}.
3919 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3920 matching that character.
3922 20. When matching a repeated Unicode property with a minimum greater than zero,
3923 (for example \pL{2,}), PCRE could look past the end of the subject if it
3924 reached it while seeking the minimum number of characters. This could
3925 happen only if some of the characters were more than one byte long, because
3926 there is a check for at least the minimum number of bytes.
3928 21. Refactored the implementation of \p and \P so as to be more general, to
3929 allow for more different types of property in future. This has changed the
3930 compiled form incompatibly. Anybody with saved compiled patterns that use
3931 \p or \P will have to recompile them.
3933 22. Added "Any" and "L&" to the supported property types.
3935 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3936 but give a compile time error if the value is greater than 0xff.
3938 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3939 accidentally not being installed or uninstalled.
3941 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3942 made were to insert the current release number. This seemed silly, because
3943 it made things harder for people building PCRE on systems that don't run
3944 "configure". I have turned pcre.h into a distributed file, no longer built
3945 by "configure", with the version identification directly included. There is
3946 no longer a pcre.h.in file.
3948 However, this change necessitated a change to the pcre-config script as
3949 well. It is built from pcre-config.in, and one of the substitutions was the
3950 release number. I have updated configure.ac so that ./configure now finds
3951 the release number by grepping pcre.h.
3953 26. Added the ability to run the tests under valgrind.
3956 Version 6.4 05-Sep-05
3957 ---------------------
3959 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3960 "--" to be printed when multiple files were scanned, even when none of the
3961 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3962 consider it to be a bug, and have restored the previous behaviour.
3964 2. A couple of code tidies to get rid of compiler warnings.
3966 3. The pcretest program used to cheat by referring to symbols in the library
3967 whose names begin with _pcre_. These are internal symbols that are not
3968 really supposed to be visible externally, and in some environments it is
3969 possible to suppress them. The cheating is now confined to including
3970 certain files from the library's source, which is a bit cleaner.
3972 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3973 file's purpose clearer.
3975 5. Reorganized pcre_ucp_findchar().
3978 Version 6.3 15-Aug-05
3979 ---------------------
3981 1. The file libpcre.pc.in did not have general read permission in the tarball.
3983 2. There were some problems when building without C++ support:
3985 (a) If C++ support was not built, "make install" and "make test" still
3986 tried to test it.
3988 (b) There were problems when the value of CXX was explicitly set. Some
3989 changes have been made to try to fix these, and ...
3991 (c) --disable-cpp can now be used to explicitly disable C++ support.
3993 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3994 backslash in a target when C++ was disabled. This confuses some
3995 versions of "make", apparently. Using an intermediate variable solves
3996 this. (Same for CPP_LOBJ.)
3998 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3999 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4000 necessary on certain architectures.
4002 4. Added a setting of -export-symbols-regex to the link command to remove
4003 those symbols that are exported in the C sense, but actually are local
4004 within the library, and not documented. Their names all begin with
4005 "_pcre_". This is not a perfect job, because (a) we have to except some
4006 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4007 available (and never for static libraries). I have made a note to try to
4008 find a way round (a) in the future.
4011 Version 6.2 01-Aug-05
4012 ---------------------
4014 1. There was no test for integer overflow of quantifier values. A construction
4015 such as {1111111111111111} would give undefined results. What is worse, if
4016 a minimum quantifier for a parenthesized subpattern overflowed and became
4017 negative, the calculation of the memory size went wrong. This could have
4018 led to memory overwriting.
4020 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4022 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4023 operating environments where this matters.
4025 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4026 PCRE options from within the C++ wrapper.
4028 5. Named capturing subpatterns were not being correctly counted when a pattern
4029 was compiled. This caused two problems: (a) If there were more than 100
4030 such subpatterns, the calculation of the memory needed for the whole
4031 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4032 back references of the form \12, where the number was greater than 9, were
4033 not recognized as back references, even though there were sufficient
4034 previous subpatterns.
4036 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4037 versions of gcc, e.g. 2.95.4.
4040 Version 6.1 21-Jun-05
4041 ---------------------
4043 1. There was one reference to the variable "posix" in pcretest.c that was not
4044 surrounded by "#if !defined NOPOSIX".
4046 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4047 the cross-check on the old pcre_info() function, for the benefit of the
4048 cut-down version of PCRE that is currently imported into Exim.
4050 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4051 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4052 patterns that start (?i) but I don't think that matters. The use of (?i) is
4053 just an example; this all applies to the other options as well.
4055 4. Since libtool seems to echo the compile commands it is issuing, the output
4056 from "make" can be reduced a bit by putting "@" in front of each libtool
4057 compile command.
4059 5. Patch from the folks at Google for configure.in to be a bit more thorough
4060 in checking for a suitable C++ installation before trying to compile the
4061 C++ stuff. This should fix a reported problem when a compiler was present,
4062 but no suitable headers.
4064 6. The man pages all had just "PCRE" as their title. I have changed them to
4065 be the relevant file name. I have also arranged that these names are
4066 retained in the file doc/pcre.txt, which is a concatenation in text format
4067 of all the man pages except the little individual ones for each function.
4069 7. The NON-UNIX-USE file had not been updated for the different set of source
4070 files that come with release 6. I also added a few comments about the C++
4071 wrapper.
4074 Version 6.0 07-Jun-05
4075 ---------------------
4077 1. Some minor internal re-organization to help with my DFA experiments.
4079 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4080 didn't matter for the library itself when fully configured, but did matter
4081 when compiling without UCP support, or within Exim, where the ucp files are
4082 not imported.
4084 3. Refactoring of the library code to split up the various functions into
4085 different source modules. The addition of the new DFA matching code (see
4086 below) to a single monolithic source would have made it really too
4087 unwieldy, quite apart from causing all the code to be include in a
4088 statically linked application, when only some functions are used. This is
4089 relevant even without the DFA addition now that patterns can be compiled in
4090 one application and matched in another.
4092 The downside of splitting up is that there have to be some external
4093 functions and data tables that are used internally in different modules of
4094 the library but which are not part of the API. These have all had their
4095 names changed to start with "_pcre_" so that they are unlikely to clash
4096 with other external names.
4098 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4099 a different (DFA) algorithm. Although it is slower than the original
4100 function, it does have some advantages for certain types of matching
4101 problem.
4103 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4104 including restarting after a partial match.
4106 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4107 defined when compiling for Windows was sent to me. I have put it into the
4108 code, though I have no means of testing or verifying it.
4110 7. Added the pcre_refcount() auxiliary function.
4112 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4113 match before or at the first newline in the subject string. In pcretest,
4114 the /f option on a pattern can be used to set this.
4116 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4117 would behave wrongly. This has been present in PCRE since release 4.0.
4119 10. A number of changes to the pcregrep command:
4121 (a) Refactored how -x works; insert ^(...)$ instead of setting
4122 PCRE_ANCHORED and checking the length, in preparation for adding
4123 something similar for -w.
4125 (b) Added the -w (match as a word) option.
4127 (c) Refactored the way lines are read and buffered so as to have more
4128 than one at a time available.
4130 (d) Implemented a pcregrep test script.
4132 (e) Added the -M (multiline match) option. This allows patterns to match
4133 over several lines of the subject. The buffering ensures that at least
4134 8K, or the rest of the document (whichever is the shorter) is available
4135 for matching (and similarly the previous 8K for lookbehind assertions).
4137 (f) Changed the --help output so that it now says
4139 -w, --word-regex(p)
4141 instead of two lines, one with "regex" and the other with "regexp"
4142 because that confused at least one person since the short forms are the
4143 same. (This required a bit of code, as the output is generated
4144 automatically from a table. It wasn't just a text change.)
4146 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4147 option but starts with a hyphen. Could be a pattern or a path name
4148 starting with a hyphen, for instance.
4150 (h) "-" can be given as a file name to represent stdin.
4152 (i) When file names are being printed, "(standard input)" is used for
4153 the standard input, for compatibility with GNU grep. Previously
4154 "<stdin>" was used.
4156 (j) The option --label=xxx can be used to supply a name to be used for
4157 stdin when file names are being printed. There is no short form.
4159 (k) Re-factored the options decoding logic because we are going to add
4160 two more options that take data. Such options can now be given in four
4161 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4163 (l) Added the -A, -B, and -C options for requesting that lines of context
4164 around matches be printed.
4166 (m) Added the -L option to print the names of files that do not contain
4167 any matching lines, that is, the complement of -l.
4169 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4170 continue to scan other files.
4172 (o) The -s option was incorrectly implemented. For compatibility with other
4173 greps, it now suppresses the error message for a non-existent or non-
4174 accessible file (but not the return code). There is a new option called
4175 -q that suppresses the output of matching lines, which was what -s was
4176 previously doing.
4178 (p) Added --include and --exclude options to specify files for inclusion
4179 and exclusion when recursing.
4181 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4182 Hopefully, it now does.
4184 12. Missing cast in pcre_study().
4186 13. Added an "uninstall" target to the makefile.
4188 14. Replaced "extern" in the function prototypes in Makefile.in with
4189 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4190 world, but is set differently for Windows.
4192 15. Added a second compiling function called pcre_compile2(). The only
4193 difference is that it has an extra argument, which is a pointer to an
4194 integer error code. When there is a compile-time failure, this is set
4195 non-zero, in addition to the error test pointer being set to point to an
4196 error message. The new argument may be NULL if no error number is required
4197 (but then you may as well call pcre_compile(), which is now just a
4198 wrapper). This facility is provided because some applications need a
4199 numeric error indication, but it has also enabled me to tidy up the way
4200 compile-time errors are handled in the POSIX wrapper.
4202 16. Added VPATH=.libs to the makefile; this should help when building with one
4203 prefix path and installing with another. (Or so I'm told by someone who
4204 knows more about this stuff than I do.)
4206 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4207 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4208 match everything, including newlines. This is not POSIX-compatible, but
4209 somebody wanted the feature. From pcretest it can be activated by using
4210 both the P and the s flags.
4212 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4214 19. libpcre.pc was being incorrectly installed as executable.
4216 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4217 it now also looks for '\r' so that it will work unmodified on Windows.
4219 21. Added Google's contributed C++ wrapper to the distribution.
4221 22. Added some untidy missing memory free() calls in pcretest, to keep
4222 Electric Fence happy when testing.
4226 Version 5.0 13-Sep-04
4227 ---------------------
4229 1. Internal change: literal characters are no longer packed up into items
4230 containing multiple characters in a single byte-string. Each character
4231 is now matched using a separate opcode. However, there may be more than one
4232 byte in the character in UTF-8 mode.
4234 2. The pcre_callout_block structure has two new fields: pattern_position and
4235 next_item_length. These contain the offset in the pattern to the next match
4236 item, and its length, respectively.
4238 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4239 insertion of callouts before each pattern item. Added the /C option to
4240 pcretest to make use of this.
4242 4. On the advice of a Windows user, the lines
4244 #if defined(_WIN32) || defined(WIN32)
4245 _setmode( _fileno( stdout ), 0x8000 );
4246 #endif /* defined(_WIN32) || defined(WIN32) */
4248 have been added to the source of pcretest. This apparently does useful
4249 magic in relation to line terminators.
4251 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4252 for the benefit of those environments where the "b" makes a difference.
4254 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4255 to know about it. I have put a hack into configure.in that adds in code
4256 to set GCC=yes if CC=icc. This seems to end up at a point in the
4257 generated configure script that is early enough to affect the setting of
4258 compiler options, which is what is needed, but I have no means of testing
4259 whether it really works. (The user who reported this had patched the
4260 generated configure script, which of course I cannot do.)
4262 LATER: After change 22 below (new libtool files), the configure script
4263 seems to know about icc (and also ecc). Therefore, I have commented out
4264 this hack in configure.in.
4266 7. Added support for pkg-config (2 patches were sent in).
4268 8. Negated POSIX character classes that used a combination of internal tables
4269 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4270 [[:^ascii]]. Typically, they would match almost any characters. The other
4271 POSIX classes were not broken in this way.
4273 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4274 to find the match, as PCRE was deluded into thinking that the match had to
4275 start at the start point or following a newline. The same bug applied to
4276 patterns with negative forward assertions or any backward assertions
4277 preceding ".*" at the start, unless the pattern required a fixed first
4278 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4280 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4281 starting at the last subject character, bytes beyond the end of the subject
4282 string were read.
4284 11. Renamed the variable "class" as "classbits" to make life easier for C++
4285 users. (Previously there was a macro definition, but it apparently wasn't
4286 enough.)
4288 12. Added the new field "tables" to the extra data so that tables can be passed
4289 in at exec time, or the internal tables can be re-selected. This allows
4290 a compiled regex to be saved and re-used at a later time by a different
4291 program that might have everything at different addresses.
4293 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4294 -R library as well as a -L library.
4296 14. The debugging options of pcretest (-d on the command line or D on a
4297 pattern) showed incorrect output for anything following an extended class
4298 that contained multibyte characters and which was followed by a quantifier.
4300 15. Added optional support for general category Unicode character properties
4301 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4302 support. It adds about 90K to the size of the library. The meanings of the
4303 inbuilt class escapes such as \d and \s have NOT been changed.
4305 16. Updated pcredemo.c to include calls to free() to release the memory for the
4306 compiled pattern.
4308 17. The generated file chartables.c was being created in the source directory
4309 instead of in the building directory. This caused the build to fail if the
4310 source directory was different from the building directory, and was
4311 read-only.
4313 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4314 file. No doubt somebody will tell me if they don't make sense... Also added
4315 Dan Mooney's comments about building on OpenVMS.
4317 19. Added support for partial matching via the PCRE_PARTIAL option for
4318 pcre_exec() and the \P data escape in pcretest.
4320 20. Extended pcretest with 3 new pattern features:
4322 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4323 write the compiled pattern to the file whose name is "rest-of-line".
4324 This is a straight binary dump of the data, with the saved pointer to
4325 the character tables forced to be NULL. The study data, if any, is
4326 written too. After writing, pcretest reads a new pattern.
4328 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4329 compiled pattern from the given file. There must not be any
4330 occurrences of "<" in the file name (pretty unlikely); if there are,
4331 pcretest will instead treat the initial "<" as a pattern delimiter.
4332 After reading in the pattern, pcretest goes on to read data lines as
4333 usual.
4335 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4336 and 16-bit fields in a compiled pattern, to simulate a pattern that
4337 was compiled on a host of opposite endianness.
4339 21. The pcre-exec() function can now cope with patterns that were compiled on
4340 hosts of opposite endianness, with this restriction:
4342 As for any compiled expression that is saved and used later, the tables
4343 pointer field cannot be preserved; the extra_data field in the arguments
4344 to pcre_exec() should be used to pass in a tables address if a value
4345 other than the default internal tables were used at compile time.
4347 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4348 now diagnosed as an error. Previously, most of the time, a negative number
4349 would have been treated as zero, but if in addition "ovector" was passed as
4350 NULL, a crash could occur.
4352 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4353 new versions from the libtool 1.5 distribution (the last one is a copy of
4354 a file called libtool.m4). This seems to have fixed the need to patch
4355 "configure" to support Darwin 1.3 (which I used to do). However, I still
4356 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4357 workstation).
4359 24. Changed the PCRE licence to be the more standard "BSD" licence.
4362 Version 4.5 01-Dec-03
4363 ---------------------
4365 1. There has been some re-arrangement of the code for the match() function so
4366 that it can be compiled in a version that does not call itself recursively.
4367 Instead, it keeps those local variables that need separate instances for
4368 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4369 needs to "recurse". Keeping track of where control must go is done by means
4370 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4371 hide most of the details from the main code, and operates only if
4372 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4373 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4374 operating.
4376 To make it easier for callers to provide specially tailored get/free
4377 functions for this usage, two new functions, pcre_stack_malloc, and
4378 pcre_stack_free, are used. They are always called in strict stacking order,
4379 and the size of block requested is always the same.
4381 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4382 PCRE has been compiled to use the stack or the heap for recursion. The
4383 -C option of pcretest uses this to show which version is compiled.
4385 A new data escape \S, is added to pcretest; it causes the amounts of store
4386 obtained and freed by both kinds of malloc/free at match time to be added
4387 to the output.
4389 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4390 what's available on my current Linux desktop machine.
4392 3. When matching a UTF-8 string, the test for a valid string at the start has
4393 been extended. If start_offset is not zero, PCRE now checks that it points
4394 to a byte that is the start of a UTF-8 character. If not, it returns
4395 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4396 this is necessary because there may be backward assertions in the pattern.
4397 When matching the same subject several times, it may save resources to use
4398 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4400 4. The code for checking the validity of UTF-8 strings has been tightened so
4401 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4402 containing "overlong sequences".
4404 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4405 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4406 should have been "&", but it just so happened that all the cases this let
4407 through by mistake were picked up later in the function.
4409 6. I had used a variable called "isblank" - this is a C99 function, causing
4410 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4412 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4413 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4414 so the version is shown; (c) stop "make test" echoing "./RunTest".
4416 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4418 9. The prototype for memmove() for systems that don't have it was using
4419 size_t, but the inclusion of the header that defines size_t was later. I've
4420 moved the #includes for the C headers earlier to avoid this.
4422 10. Added some adjustments to the code to make it easier to compiler on certain
4423 special systems:
4425 (a) Some "const" qualifiers were missing.
4426 (b) Added the macro EXPORT before all exported functions; by default this
4427 is defined to be empty.
4428 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4429 that it reads its output file name as an argument instead of writing
4430 to the standard output and assuming this can be redirected.
4432 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4433 class containing characters with values greater than 255, PCRE compilation
4434 went into a loop.
4436 12. A recursive reference to a subpattern that was within another subpattern
4437 that had a minimum quantifier of zero caused PCRE to crash. For example,
4438 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4439 recursion. If the recursively-called subpattern itself had a zero repeat,
4440 that was OK.
4442 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4443 buffer into which it was copied (for escape processing) was still set at
4444 1024, so long lines caused crashes.
4446 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4447 "internal error: code overflow...". This applied to any character class
4448 that was followed by a possessive quantifier.
4450 15. Modified the Makefile to add libpcre.la as a prerequisite for
4451 libpcreposix.la because I was told this is needed for a parallel build to
4452 work.
4454 16. If a pattern that contained .* following optional items at the start was
4455 studied, the wrong optimizing data was generated, leading to matching
4456 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4457 matching string must start with a or b or c. The correct conclusion for
4458 this pattern is that a match can start with any character.
4461 Version 4.4 13-Aug-03
4462 ---------------------
4464 1. In UTF-8 mode, a character class containing characters with values between
4465 127 and 255 was not handled correctly if the compiled pattern was studied.
4466 In fixing this, I have also improved the studying algorithm for such
4467 classes (slightly).
4469 2. Three internal functions had redundant arguments passed to them. Removal
4470 might give a very teeny performance improvement.
4472 3. Documentation bug: the value of the capture_top field in a callout is *one
4473 more than* the number of the hightest numbered captured substring.
4475 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4476 in incorrectly linking with a previously installed version. They now link
4477 explicitly with libpcre.la.
4479 5. configure.in no longer needs to recognize Cygwin specially.
4481 6. A problem in pcre.in for Windows platforms is fixed.
4483 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4484 pcretest, it used to include the size of the study block as part of its
4485 output. Unfortunately, the structure contains a field that has a different
4486 size on different hardware architectures. This meant that the tests that
4487 showed this size failed. As the block is currently always of a fixed size,
4488 this information isn't actually particularly useful in pcretest output, so
4489 I have just removed it.
4491 8. Three pre-processor statements accidentally did not start in column 1.
4492 Sadly, there are *still* compilers around that complain, even though
4493 standard C has not required this for well over a decade. Sigh.
4495 9. In pcretest, the code for checking callouts passed small integers in the
4496 callout_data field, which is a void * field. However, some picky compilers
4497 complained about the casts involved for this on 64-bit systems. Now
4498 pcretest passes the address of the small integer instead, which should get
4499 rid of the warnings.
4501 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4502 both compile and run time, and gives an error if an invalid UTF-8 sequence
4503 is found. There is a option for disabling this check in cases where the
4504 string is known to be correct and/or the maximum performance is wanted.
4506 11. In response to a bug report, I changed one line in Makefile.in from
4508 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4509 to
4510 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4512 to look similar to other lines, but I have no way of telling whether this
4513 is the right thing to do, as I do not use Windows. No doubt I'll get told
4514 if it's wrong...
4517 Version 4.3 21-May-03
4518 ---------------------
4520 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4521 Makefile.
4523 2. Some refactoring to improve the quality of the code:
4525 (i) The utf8_table... variables are now declared "const".
4527 (ii) The code for \cx, which used the "case flipping" table to upper case
4528 lower case letters, now just substracts 32. This is ASCII-specific,
4529 but the whole concept of \cx is ASCII-specific, so it seems
4530 reasonable.
4532 (iii) PCRE was using its character types table to recognize decimal and
4533 hexadecimal digits in the pattern. This is silly, because it handles
4534 only 0-9, a-f, and A-F, but the character types table is locale-
4535 specific, which means strange things might happen. A private
4536 table is now used for this - though it costs 256 bytes, a table is
4537 much faster than multiple explicit tests. Of course, the standard
4538 character types table is still used for matching digits in subject
4539 strings against \d.
4541 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4542 ending in _t are). So I've renamed it as ESC_tee.
4544 3. The first argument for regexec() in the POSIX wrapper should have been
4545 defined as "const".
4547 4. Changed pcretest to use malloc() for its buffers so that they can be
4548 Electric Fenced for debugging.
4550 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4551 to read one or more bytes before the start of the subject string. Often this
4552 had no effect on PCRE's behaviour, but in some circumstances it could
4553 provoke a segmentation fault.
4555 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4556 to try to read one or more bytes before the start of the subject string.
4558 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4559 UTF-8 support could misbehave in various ways if the subject string
4560 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4561 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4562 back over UTF-8 characters.)
4565 Version 4.2 14-Apr-03
4566 ---------------------
4568 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4570 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4571 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4572 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4573 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4574 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4576 Note: automatic setting of the BUILD variables is not yet working
4577 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4578 compile-time but not at link-time
4579 [LINK]: use for linking executables only
4580 make different versions for Windows and non-Windows
4581 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4582 libraries
4583 [LINK_FOR_BUILD]: new variable
4584 [OBJEXT]: use throughout
4585 [EXEEXT]: use throughout
4586 <winshared>: new target
4587 <wininstall>: new target
4588 <dftables.o>: use native compiler
4589 <dftables>: use native linker
4590 <install>: handle Windows platform correctly
4591 <clean>: ditto
4592 <check>: ditto
4593 copy DLL to top builddir before testing
4595 As part of these changes, -no-undefined was removed again. This was reported
4596 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4597 in any case.
4599 3. Some tidies to get rid of compiler warnings:
4601 . In the match_data structure, match_limit was an unsigned long int, whereas
4602 match_call_count was an int. I've made them both unsigned long ints.
4604 . In pcretest the fact that a const uschar * doesn't automatically cast to
4605 a void * provoked a warning.
4607 . Turning on some more compiler warnings threw up some "shadow" variables
4608 and a few more missing casts.
4610 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4611 option, a class that contained a single character with a value between 128
4612 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4614 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4615 option, a class that contained several characters, but with at least one
4616 whose value was between 128 and 255 caused PCRE to crash.
4619 Version 4.1 12-Mar-03
4620 ---------------------
4622 1. Compiling with gcc -pedantic found a couple of places where casts were
4623 needed, and a string in dftables.c that was longer than standard compilers are
4624 required to support.
4626 2. Compiling with Sun's compiler found a few more places where the code could
4627 be tidied up in order to avoid warnings.
4629 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4630 first of these names is deprecated in the latest Autoconf in favour of the name
4631 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4632 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4633 analogy I have changed it to CFLAGS_FOR_BUILD.
4635 4. Added -no-undefined to the linking command in the Makefile, because this is
4636 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4637 linking step for the pcreposix library.
4639 5. PCRE was failing to diagnose the case of two named groups with the same
4640 name.
4642 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4643 literal character that is needed in the subject for a match, and scans along to
4644 ensure that it is present before embarking on the full matching process. This
4645 saves time in cases of nested unlimited repeats that are never going to match.
4646 Problem: the scan can take a lot of time if the subject is very long (e.g.
4647 megabytes), thus penalizing straightforward matches. It is now done only if the
4648 amount of subject to be scanned is less than 1000 bytes.
4650 7. A lesser problem with the same optimization is that it was recording the
4651 first character of an anchored pattern as "needed", thus provoking a search
4652 right along the subject, even when the first match of the pattern was going to
4653 fail. The "needed" character is now not set for anchored patterns, unless it
4654 follows something in the pattern that is of non-fixed length. Thus, it still
4655 fulfils its original purpose of finding quick non-matches in cases of nested
4656 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4659 Version 4.0 17-Feb-03
4660 ---------------------
4662 1. If a comment in an extended regex that started immediately after a meta-item
4663 extended to the end of string, PCRE compiled incorrect data. This could lead to
4664 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4666 2. Moved to autoconf 2.53 and libtool 1.4.2.
4668 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4669 the special perltest8 script is no longer needed - all the tests can be run
4670 from a single perltest script.
4672 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4673 by \s. It has now been removed in PCRE. This means it isn't recognized as
4674 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4675 class [:space:] *does* include VT, thereby creating a mess.
4677 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4678 space and tab.
4680 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4681 its new features into the main test script, reducing the number of scripts.
4683 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4684 were backward compatible, and made the (?i) apply to the whole pattern, as if
4685 /i were given. Now it behaves more logically, and applies the option setting
4686 only to what follows. PCRE has been changed to follow suit. However, if it
4687 finds options settings right at the start of the pattern, it extracts them into
4688 the global options, as before. Thus, they show up in the info data.
4690 8. Added support for the \Q...\E escape sequence. Characters in between are
4691 treated as literals. This is slightly different from Perl in that $ and @ are
4692 also handled as literals inside the quotes. In Perl, they will cause variable
4693 interpolation. Note the following examples:
4695 Pattern PCRE matches Perl matches
4697 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4698 \Qabc\$xyz\E abc\$xyz abc\$xyz
4699 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4701 For compatibility with Perl, \Q...\E sequences are recognized inside character
4702 classes as well as outside them.
4704 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4705 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4706 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4707 signed/unsigned warnings.
4709 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4710 option for pcretest, so I've replaced it by a simple function that does just
4711 that job.
4713 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4714 "pcregrep -".
4716 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4717 Java package. This provides some syntactic sugar for simple cases of what my
4718 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4719 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4720 item, you can use this simplified notation. Note that only makes sense with
4721 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4722 greediness, whatever the setting of the PCRE_UNGREEDY option.
4724 13. A change of greediness default within a pattern was not taking effect at
4725 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4726 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4727 was abstracted outside.
4729 14. PCRE now supports the \G assertion. It is true when the current matching
4730 position is at the start point of the match. This differs from \A when the
4731 starting offset is non-zero. Used with the /g option of pcretest (or similar
4732 code), it works in the same way as it does for Perl's /g option. If all
4733 alternatives of a regex begin with \G, the expression is anchored to the start
4734 match position, and the "anchored" flag is set in the compiled expression.
4736 15. Some bugs concerning the handling of certain option changes within patterns
4737 have been fixed. These applied to options other than (?ims). For example,
4738 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4739 been the other way round. Some of this was related to change 7 above.
4741 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4742 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4743 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4744 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4746 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4747 mode. Unlike ".", it always matches newline, whatever the setting of
4748 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4749 assertions. Perl allows it, but it doesn't (in general) work because it can't
4750 calculate the length of the lookbehind. At least, that's the case for Perl
4751 5.8.0 - I've been told they are going to document that it doesn't work in
4752 future.
4754 18. Added an error diagnosis for escapes that PCRE does not support: these are
4755 \L, \l, \N, \P, \p, \U, \u, and \X.
4757 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4758 reading past the end of the pattern in cases such as /[abcd/.
4760 20. PCRE was getting more memory than necessary for patterns with classes that
4761 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4763 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4764 compiling PCRE for use with Virtual Pascal.
4766 22. Small fix to the Makefile to make it work properly if the build is done
4767 outside the source tree.
4769 23. Added a new extension: a condition to go with recursion. If a conditional
4770 subpattern starts with (?(R) the "true" branch is used if recursion has
4771 happened, whereas the "false" branch is used only at the top level.
4773 24. When there was a very long string of literal characters (over 255 bytes
4774 without UTF support, over 250 bytes with UTF support), the computation of how
4775 much memory was required could be incorrect, leading to segfaults or other
4776 strange effects.
4778 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4779 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4780 there was a subsequent back reference to those brackets. This meant that, for
4781 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4782 possible to check for precisely this case. All we can do is abandon the
4783 optimization if .* occurs inside capturing brackets when there are any back
4784 references whatsoever. (See below for a better fix that came later.)
4786 26. The handling of the optimization for finding the first character of a
4787 non-anchored pattern, and for finding a character that is required later in the
4788 match were failing in some cases. This didn't break the matching; it just
4789 failed to optimize when it could. The way this is done has been re-implemented.
4791 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4793 28. Added a new feature that provides some of the functionality that Perl
4794 provides with (?{...}). The facility is termed a "callout". The way it is done
4795 in PCRE is for the caller to provide an optional function, by setting
4796 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4797 global variable. By default it is unset, which disables all calling out. To get
4798 the function called, the regex must include (?C) at appropriate points. This
4799 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4800 This provides a means of identifying different callout points. When PCRE
4801 reaches such a point in the regex, if pcre_callout has been set, the external
4802 function is called. It is provided with data in a structure called
4803 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4804 matching continues; if it returns a non-zero value, the match at the current
4805 point fails. However, backtracking will occur if possible. [This was changed
4806 later and other features added - see item 49 below.]
4808 29. pcretest is upgraded to test the callout functionality. It provides a
4809 callout function that displays information. By default, it shows the start of
4810 the match and the current position in the text. There are some new data escapes
4811 to vary what happens:
4813 \C+ in addition, show current contents of captured substrings
4814 \C- do not supply a callout function
4815 \C!n return 1 when callout number n is reached
4816 \C!n!m return 1 when callout number n is reached for the mth time
4818 30. If pcregrep was called with the -l option and just a single file name, it
4819 output "<stdin>" if a match was found, instead of the file name.
4821 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4822 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4823 pcre_exec(). This saves a malloc/free per call. The default value of
4824 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4825 when configuring.
4827 32. The default maximum size of a compiled pattern is 64K. There have been a
4828 few cases of people hitting this limit. The code now uses macros to handle the
4829 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4830 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4831 configuring. Tests 2 and 5 work only with 2-byte links because they output
4832 debugging information about compiled patterns.
4834 33. Internal code re-arrangements:
4836 (a) Moved the debugging function for printing out a compiled regex into
4837 its own source file (printint.c) and used #include to pull it into
4838 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4839 separate copies.
4841 (b) Defined the list of op-code names for debugging as a macro in
4842 internal.h so that it is next to the definition of the opcodes.
4844 (c) Defined a table of op-code lengths for simpler skipping along compiled
4845 code. This is again a macro in internal.h so that it is next to the
4846 definition of the opcodes.
4848 34. Added support for recursive calls to individual subpatterns, along the
4849 lines of Robin Houston's patch (but implemented somewhat differently).
4851 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4852 allow it to read and process whole directories in Win32. This code was
4853 contributed by Lionel Fourquaux; it has not been tested by me.
4855 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4856 used to name a group. Names consist of alphanumerics and underscores, and must
4857 be unique. Back references use the syntax (?P=name) and recursive calls use
4858 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4859 numbers. The function pcre_fullinfo() can be used after compilation to extract
4860 a name/number map. There are three relevant calls:
4862 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4863 PCRE_INFO_NAMECOUNT yields the number of entries
4864 PCRE_INFO_NAMETABLE yields a pointer to the map.
4866 The map is a vector of fixed-size entries. The size of each entry depends on
4867 the length of the longest name used. The first two bytes of each entry are the
4868 group number, most significant byte first. There follows the corresponding
4869 name, zero terminated. The names are in alphabetical order.
4871 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4872 case instead of 255. Making it the same both with and without UTF-8 support
4873 means that the same test output works with both.
4875 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4876 calling malloc() with a zero argument.
4878 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4879 optimization. I've improved things by keeping a bitmap of backreferences with
4880 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4881 fact referenced, the optimization can be applied. It is unlikely that a
4882 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4883 the match to follow \n) will appear inside brackets with a number greater than
4884 31, but if it does, any back reference > 31 suppresses the optimization.
4886 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4887 of disabling numbered capturing parentheses. Any opening parenthesis that is
4888 not followed by ? behaves as if it were followed by ?: but named parentheses
4889 can still be used for capturing (and they will acquire numbers in the usual
4890 way).
4892 41. Redesigned the return codes from the match() function into yes/no/error so
4893 that errors can be passed back from deep inside the nested calls. A malloc
4894 failure while inside a recursive subpattern call now causes the
4895 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4897 42. It is now possible to set a limit on the number of times the match()
4898 function is called in a call to pcre_exec(). This facility makes it possible to
4899 limit the amount of recursion and backtracking, though not in a directly
4900 obvious way, because the match() function is used in a number of different
4901 circumstances. The count starts from zero for each position in the subject
4902 string (for non-anchored patterns). The default limit is, for compatibility, a
4903 large number, namely 10 000 000. You can change this in two ways:
4905 (a) When configuring PCRE before making, you can use --with-match-limit=n
4906 to set a default value for the compiled library.
4908 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4909 a different value is set. See 45 below.
4911 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4913 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4914 of things that can be changed at compile time. The first argument specifies
4915 what is wanted and the second points to where the information is to be placed.
4916 The current list of available information is:
4920 The output is an integer that is set to one if UTF-8 support is available;
4921 otherwise it is set to zero.
4925 The output is an integer that it set to the value of the code that is used for
4926 newline. It is either LF (10) or CR (13).
4930 The output is an integer that contains the number of bytes used for internal
4931 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4935 The output is an integer that contains the threshold above which the POSIX
4936 interface uses malloc() for output vectors. See item 31 above.
4940 The output is an unsigned integer that contains the default limit of the number
4941 of match() calls in a pcre_exec() execution. See 42 above.
4943 44. pcretest has been upgraded by the addition of the -C option. This causes it
4944 to extract all the available output from the new pcre_config() function, and to
4945 output it. The program then exits immediately.
4947 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4948 order to support additional features. One way would have been to define
4949 pcre_exec2() (for example) with extra arguments, but this would not have been
4950 extensible, and would also have required all calls to the original function to
4951 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4952 is used for passing in "extra" data from pcre_study().
4954 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4955 contains the following fields:
4957 flags a bitmap indicating which of the following fields are set
4958 study_data opaque data from pcre_study()
4959 match_limit a way of specifying a limit on match() calls for a specific
4960 call to pcre_exec()
4961 callout_data data for callouts (see 49 below)
4963 The flag bits are also defined in pcre.h, and are
4969 The pcre_study() function now returns one of these new pcre_extra blocks, with
4970 the actual study data pointed to by the study_data field, and the
4971 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4972 before. That is, this change is entirely upwards-compatible and requires no
4973 change to existing code.
4975 If you want to pass in additional data to pcre_exec(), you can either place it
4976 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4977 block.
4979 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4980 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4981 times with different match limits, until it finds the minimum value needed for
4982 pcre_exec() to complete. The value is then output. This can be instructive; for
4983 most simple matches the number is quite small, but for pathological cases it
4984 gets very large very quickly.
4986 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4987 returns the size of the data block pointed to by the study_data field in a
4988 pcre_extra block, that is, the value that was passed as the argument to
4989 pcre_malloc() when PCRE was getting memory in which to place the information
4990 created by pcre_study(). The fourth argument should point to a size_t variable.
4991 pcretest has been extended so that this information is shown after a successful
4992 pcre_study() call when information about the compiled regex is being displayed.
4994 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4995 because what follows is always an absolute path. (Later: it turns out that this
4996 is more than cosmetic for MinGW, because it doesn't like empty path
4997 components.)
4999 49. Some changes have been made to the callout feature (see 28 above):
5001 (i) A callout function now has three choices for what it returns:
5003 0 => success, carry on matching
5004 > 0 => failure at this point, but backtrack if possible
5005 < 0 => serious error, return this value from pcre_exec()
5007 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5008 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5009 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5010 use by callout functions. It will never be used by PCRE itself.
5012 (ii) The pcre_extra structure (see 45 above) has a void * field called
5013 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5014 pcre_callout_block structure has a field of the same name. The contents of
5015 the field passed in the pcre_extra structure are passed to the callout
5016 function in the corresponding field in the callout block. This makes it
5017 easier to use the same callout-containing regex from multiple threads. For
5018 testing, the pcretest program has a new data escape
5020 \C*n pass the number n (may be negative) as callout_data
5022 If the callout function in pcretest receives a non-zero value as
5023 callout_data, it returns that value.
5025 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5026 there were some redundant $(CFLAGS) in commands that are now specified as
5027 $(LINK), which already includes $(CFLAGS).
5029 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5030 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5031 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5032 one-byte characters throughout. Note that case-insensitive matching applies
5033 only to characters whose values are less than 256. PCRE doesn't support the
5034 notion of cases for higher-valued characters.
5036 (i) A character class whose characters are all within 0-255 is handled as
5037 a bit map, and the map is inverted for negative classes. Previously, a
5038 character > 255 always failed to match such a class; however it should
5039 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5041 (ii) A negated character class with a single character < 255 is coded as
5042 "not this character" (OP_NOT). This wasn't working properly when the test
5043 character was multibyte, either singly or repeated.
5045 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5046 mode, for example: \x{100}{2,3}.
5048 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5049 singly or repeated) now correctly test multibyte characters. However,
5050 PCRE doesn't recognize any characters with values greater than 255 as
5051 digits, spaces, or word characters. Such characters always match \D, \S,
5052 and \W, and never match \d, \s, or \w.
5054 (v) Classes may now contain characters and character ranges with values
5055 greater than 255. For example: [ab\x{100}-\x{400}].
5057 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5058 PCRE in UTF-8 mode.
5060 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5061 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5062 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5063 value.)
5065 53. The single man page has become too large. I have therefore split it up into
5066 a number of separate man pages. These also give rise to individual HTML pages;
5067 these are now put in a separate directory, and there is an index.html page that
5068 lists them all. Some hyperlinking between the pages has been installed.
5070 54. Added convenience functions for handling named capturing parentheses.
5072 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5073 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5074 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5075 are faulted.
5077 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5078 calling configure. These values are used when compiling the dftables.c program
5079 which is run to generate the source of the default character tables. They
5080 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5081 you will need to set these values.
5083 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5086 Version 3.9 02-Jan-02
5087 ---------------------
5089 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5091 2. If --disable-static was given, the building process failed when trying to
5092 build pcretest and pcregrep. (For some reason it was using libtool to compile
5093 them, which is not right, as they aren't part of the library.)
5096 Version 3.8 18-Dec-01
5097 ---------------------
5099 1. The experimental UTF-8 code was completely screwed up. It was packing the
5100 bytes in the wrong order. How dumb can you get?
5103 Version 3.7 29-Oct-01
5104 ---------------------
5106 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5107 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5108 this didn't happen under Solaris 8, where I normally test things.
5110 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5111 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5112 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5113 this fix an hour or so after the initial 3.7 release.)
5116 Version 3.6 23-Oct-01
5117 ---------------------
5119 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5120 offsets passed as NULL with zero offset count.
5122 2. The config.guess and config.sub files had not been updated when I moved to
5123 the latest autoconf.
5126 Version 3.5 15-Aug-01
5127 ---------------------
5129 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5130 had been forgotten.
5132 2. By using declared but undefined structures, we can avoid using "void"
5133 definitions in pcre.h while keeping the internal definitions of the structures
5134 private.
5136 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5137 user point of view, this means that both static and shared libraries are built
5138 by default, but this can be individually controlled. More of the work of
5139 handling this static/shared cases is now inside libtool instead of PCRE's make
5140 file.
5142 4. The pcretest utility is now installed along with pcregrep because it is
5143 useful for users (to test regexs) and by doing this, it automatically gets
5144 relinked by libtool. The documentation has been turned into a man page, so
5145 there are now .1, .txt, and .html versions in /doc.
5147 5. Upgrades to pcregrep:
5148 (i) Added long-form option names like gnu grep.
5149 (ii) Added --help to list all options with an explanatory phrase.
5150 (iii) Added -r, --recursive to recurse into sub-directories.
5151 (iv) Added -f, --file to read patterns from a file.
5153 6. pcre_exec() was referring to its "code" argument before testing that
5154 argument for NULL (and giving an error if it was NULL).
5156 7. Upgraded Makefile.in to allow for compiling in a different directory from
5157 the source directory.
5159 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5160 options bits, the pointer it was passed was to an int instead of to an unsigned
5161 long int. This mattered only on 64-bit systems.
5163 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5164 generated) instead of pcre.in, which it its source. Also made the same change
5165 in several of the .c files.
5167 10. A new release of gcc defines printf() as a macro, which broke pcretest
5168 because it had an ifdef in the middle of a string argument for printf(). Fixed
5169 by using separate calls to printf().
5171 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5172 script, to force use of CR or LF instead of \n in the source. On non-Unix
5173 systems, the value can be set in config.h.
5175 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5176 absolute limit. Changed the text of the error message to make this clear, and
5177 likewise updated the man page.
5179 13. The limit of 99 on the number of capturing subpatterns has been removed.
5180 The new limit is 65535, which I hope will not be a "real" limit.
5183 Version 3.4 22-Aug-00
5184 ---------------------
5186 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5188 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5191 Version 3.3 01-Aug-00
5192 ---------------------
5194 1. If an octal character was given, but the value was greater than \377, it
5195 was not getting masked to the least significant bits, as documented. This could
5196 lead to crashes in some systems.
5198 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5199 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5201 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5202 These just pass their arguments on to (pcre_free)(), but they are provided
5203 because some uses of PCRE bind it to non-C systems that can call its functions,
5204 but cannot call free() or pcre_free() directly.
5206 4. Add "make test" as a synonym for "make check". Corrected some comments in
5207 the Makefile.
5209 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5210 Makefile.
5212 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5213 command called pgrep for grepping around the active processes.
5215 7. Added the beginnings of support for UTF-8 character strings.
5217 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5218 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5219 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5220 out for the ar command.)
5223 Version 3.2 12-May-00
5224 ---------------------
5226 This is purely a bug fixing release.
5228 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5229 of ZA. This was just one example of several cases that could provoke this bug,
5230 which was introduced by change 9 of version 2.00. The code for breaking
5231 infinite loops after an iteration that matches an empty string was't working
5232 correctly.
5234 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5235 when matched against abbab (for example). After matching an empty string, it
5236 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5237 caused it to match further down the string than it should.
5239 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5240 was there because it doesn't seem to be needed, and it causes trouble on some
5241 systems, as it is not a Standard C header. It has been removed.
5243 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5244 were reported on the Macintosh. The changes were from
5246 while ((c = *(++ptr)) != 0 && c != '\n');
5247 to
5248 while ((c = *(++ptr)) != 0 && c != '\n') ;
5250 Totally extraordinary, but if that's what it takes...
5252 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5253 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5254 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5255 assumes the way PCRE uses memmove() (always moving upwards).
5257 6. PCRE is being used in one environment where strchr() is not available. There
5258 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5259 faster code anyway.
5262 Version 3.1 09-Feb-00
5263 ---------------------
5265 The only change in this release is the fixing of some bugs in Makefile.in for
5266 the "install" target:
5268 (1) It was failing to install pcreposix.h.
5270 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5273 Version 3.0 01-Feb-00
5274 ---------------------
5276 1. Add support for the /+ modifier to perltest (to output $` like it does in
5277 pcretest).
5279 2. Add support for the /g modifier to perltest.
5281 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5282 matches null strings.
5284 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5285 pattern. Perl treats empty patterns specially - it reuses the most recent
5286 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5287 effect.
5289 5. The POSIX interface was broken in that it was just handing over the POSIX
5290 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5291 required a bigger vector, with some working space on the end. This means that
5292 the POSIX wrapper now has to get and free some memory, and copy the results.
5294 6. Added some simple autoconf support, placing the test data and the
5295 documentation in separate directories, re-organizing some of the
5296 information files, and making it build pcre-config (a GNU standard). Also added
5297 libtool support for building PCRE as a shared library, which is now the
5298 default.
5300 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5301 09 are not valid octal constants. Single digits will be used for minor values
5302 less than 10.
5304 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5305 existing programs that set these in the POSIX interface can use PCRE without
5306 modification.
5308 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5309 return all that pcre_info() returns, plus additional data. The pcre_info()
5310 function is retained for compatibility, but is considered to be obsolete.
5312 10. Added experimental recursion feature (?R) to handle one common case that
5313 Perl 5.6 will be able to do with (?p{...}).
5315 11. Added support for POSIX character classes like [:alpha:], which Perl is
5316 adopting.
5319 Version 2.08 31-Aug-99
5320 ----------------------
5322 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5323 trying to match at the startoffset position, but instead was moving forward to
5324 the next newline as if a previous match had failed.
5326 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5327 and could get into a loop if a null string was matched other than at the start
5328 of the subject.
5330 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5331 be distinguished at compile time, and for completeness also added PCRE_DATE.
5333 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5334 in GnuWin32 environments.
5337 Version 2.07 29-Jul-99
5338 ----------------------
5340 1. The documentation is now supplied in plain text form and HTML as well as in
5341 the form of man page sources.
5343 2. C++ compilers don't like assigning (void *) values to other pointer types.
5344 In particular this affects malloc(). Although there is no problem in Standard
5345 C, I've put in casts to keep C++ compilers happy.
5347 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5348 should be (const char *).
5350 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5351 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5352 However, I haven't made this a standard facility. The documentation doesn't
5353 mention it, and the Makefile doesn't support it.
5355 5. The Makefile now contains an "install" target, with editable destinations at
5356 the top of the file. The pcretest program is not installed.
5358 6. pgrep -V now gives the PCRE version number and date.
5360 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5361 causing the entire string to be ignored, instead of just the last character.
5363 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5364 non-matching string, it can take a very, very long time, even for strings of
5365 quite modest length, because of the nested recursion. PCRE now does better in
5366 some of these cases. It does this by remembering the last required literal
5367 character in the pattern, and pre-searching the subject to ensure it is present
5368 before running the real match. In other words, it applies a heuristic to detect
5369 some types of certain failure quickly, and in the above example, if presented
5370 with a string that has no trailing " it gives "no match" very quickly.
5372 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5373 other alternatives are tried instead.
5376 Version 2.06 09-Jun-99
5377 ----------------------
5379 1. Change pcretest's output for amount of store used to show just the code
5380 space, because the remainder (the data block) varies in size between 32-bit and
5381 64-bit systems.
5383 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5384 start matching at. This allows lookbehinds to work when searching for multiple
5385 occurrences in a string.
5387 3. Added additional options to pcretest for testing multiple occurrences:
5389 /+ outputs the rest of the string that follows a match
5390 /g loops for multiple occurrences, using the new startoffset argument
5391 /G loops for multiple occurrences by passing an incremented pointer
5393 4. PCRE wasn't doing the "first character" optimization for patterns starting
5394 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5395 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5396 the letter 'x'. On long subject strings, this gives a significant speed-up.
5399 Version 2.05 21-Apr-99
5400 ----------------------
5402 1. Changed the type of magic_number from int to long int so that it works
5403 properly on 16-bit systems.
5405 2. Fixed a bug which caused patterns starting with .* not to work correctly
5406 when the subject string contained newline characters. PCRE was assuming
5407 anchoring for such patterns in all cases, which is not correct because .* will
5408 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5409 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5410 must be retried after every newline in the subject.
5413 Version 2.04 18-Feb-99
5414 ----------------------
5416 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5417 computation of the store needed to hold the pattern was incorrect (too large).
5418 If such patterns were nested a few deep, this could multiply and become a real
5419 problem.
5421 2. Added /M option to pcretest to show the memory requirement of a specific
5422 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5424 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5425 compiled in such a way that the backtracking after subsequent failure was
5426 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5427 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5430 Version 2.03 02-Feb-99
5431 ----------------------
5433 1. Fixed typo and small mistake in man page.
5435 2. Added 4th condition (GPL supersedes if conflict) and created separate
5436 LICENCE file containing the conditions.
5438 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5439 Perl, that is the internal \ allows the delimiter to be included in the
5440 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5441 the final delimiter, add \ to the end of the pattern (to test the error).
5443 4. Added the convenience functions for extracting substrings after a successful
5444 match. Updated pcretest to make it able to test these functions.
5447 Version 2.02 14-Jan-99
5448 ----------------------
5450 1. Initialized the working variables associated with each extraction so that
5451 their saving and restoring doesn't refer to uninitialized store.
5453 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5454 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5455 fix the problem.
5457 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5458 calls, and wasn't printing the correct value for compiling calls. Increased the
5459 default value of LOOPREPEAT, and the number of significant figures in the
5460 times.
5462 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5464 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5465 a building problem on Windows NT with a FAT file system.
5468 Version 2.01 21-Oct-98
5469 ----------------------
5471 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5472 to character tables built by pcre_maketables() in the current locale. If NULL
5473 is passed, the default tables are used.
5476 Version 2.00 24-Sep-98
5477 ----------------------
5479 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5480 it any more.
5482 2. Allow quantification of (?>) groups, and make it work correctly.
5484 3. The first character computation wasn't working for (?>) groups.
5486 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5487 end of the subject) and add 5.005's \z, which really does match only at the
5488 very end of the subject.
5490 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5492 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5493 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5494 localized options. All options to pcre_study() were also removed.
5496 7. Add other new features from 5.005:
5498 $(?<= positive lookbehind
5499 $(?<! negative lookbehind
5500 (?imsx-imsx) added the unsetting capability
5501 such a setting is global if at outer level; local otherwise
5502 (?imsx-imsx:) non-capturing groups with option setting
5503 (?(cond)re|re) conditional pattern matching
5505 A backreference to itself in a repeated group matches the previous
5506 captured string.
5508 8. General tidying up of studying (both automatic and via "study")
5509 consequential on the addition of new assertions.
5511 9. As in 5.005, unlimited repeated groups that could match an empty substring
5512 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5513 runtime if any iteration does actually match an empty substring.
5515 10. Include the RunTest script in the distribution.
5517 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5518 discrepancies, some of which were old and were also with respect to 5.004. They
5519 have now been fixed.
5522 Version 1.09 28-Apr-98
5523 ----------------------
5525 1. A negated single character class followed by a quantifier with a minimum
5526 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5527 program crashes, or just wrong answers. This did not apply to negated classes
5528 containing more than one character, or to minima other than one.
5531 Version 1.08 27-Mar-98
5532 ----------------------
5534 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5536 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5537 latter must appear before anything that relies on it in the pattern.
5540 Version 1.07 16-Feb-98
5541 ----------------------
5543 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5544 repeat of a potentially empty string).
5547 Version 1.06 23-Jan-98
5548 ----------------------
5550 1. Added Markus Oberhumer's little patches for C++.
5552 2. Literal strings longer than 255 characters were broken.
5555 Version 1.05 23-Dec-97
5556 ----------------------
5558 1. Negated character classes containing more than one character were failing if
5559 PCRE_CASELESS was set at run time.
5562 Version 1.04 19-Dec-97
5563 ----------------------
5565 1. Corrected the man page, where some "const" qualifiers had been omitted.
5567 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5568 input syntax.
5570 3. Fixed memory leak which occurred when a regex with back references was
5571 matched with an offsets vector that wasn't big enough. The temporary memory
5572 that is used in this case wasn't being freed if the match failed.
5574 4. Tidied pcretest to ensure it frees memory that it gets.
5576 5. Temporary memory was being obtained in the case where the passed offsets
5577 vector was exactly big enough.
5579 6. Corrected definition of offsetof() from change 5 below.
5581 7. I had screwed up change 6 below and broken the rules for the use of
5582 setjmp(). Now fixed.
5585 Version 1.03 18-Dec-97
5586 ----------------------
5588 1. A erroneous regex with a missing opening parenthesis was correctly
5589 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5590 on some systems.
5592 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5593 it was reported that one broken compiler failed on the former because "code" is
5594 also an independent variable.
5596 3. The erroneous regex a[]b caused an array overrun reference.
5598 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5599 fail on data ending with that character. (It was going on too far, and checking
5600 the next character, typically a binary zero.) This was specific to the
5601 optimized code for single-character negative classes.
5603 5. Added a contributed patch from the TIN world which does the following:
5605 + Add an undef for memmove, in case the the system defines a macro for it.
5607 + Add a definition of offsetof(), in case there isn't one. (I don't know
5608 the reason behind this - offsetof() is part of the ANSI standard - but
5609 it does no harm).
5611 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5612 most of the places where whitespace preceded '#'. I have given up and
5613 allowed the remaining 2 cases to be at the margin.
5615 + Rename some variables in pcre to eliminate shadowing. This seems very
5616 pedantic, but does no harm, of course.
5618 6. Moved the call to setjmp() into its own function, to get rid of warnings
5619 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5621 7. Constructs such as \d{8,} were compiling into the equivalent of
5622 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5623 outcome, but in this particular case used more store than had been allocated,
5624 which caused the bug to be discovered because it threw up an internal error.
5626 8. The debugging code in both pcre and pcretest for outputting the compiled
5627 form of a regex was going wrong in the case of back references followed by
5628 curly-bracketed repeats.
5631 Version 1.02 12-Dec-97
5632 ----------------------
5634 1. Typos in pcre.3 and comments in the source fixed.
5636 2. Applied a contributed patch to get rid of places where it used to remove
5637 'const' from variables, and fixed some signed/unsigned and uninitialized
5638 variable warnings.
5640 3. Added the "runtest" target to Makefile.
5642 4. Set default compiler flag to -O2 rather than just -O.
5645 Version 1.01 19-Nov-97
5646 ----------------------
5648 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5649 like /([ab]*)*/, that is, for classes with more than one character in them.
5651 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5652 as /((?>a*))*/ (a PCRE_EXTRA facility).
5655 Version 1.00 18-Nov-97
5656 ----------------------
5658 1. Added compile-time macros to support systems such as SunOS4 which don't have
5659 memmove() or strerror() but have other things that can be used instead.
5661 2. Arranged that "make clean" removes the executables.
5664 Version 0.99 27-Oct-97
5665 ----------------------
5667 1. Fixed bug in code for optimizing classes with only one character. It was
5668 initializing a 32-byte map regardless, which could cause it to run off the end
5669 of the memory it had got.
5671 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5674 Version 0.98 22-Oct-97
5675 ----------------------
5677 1. Fixed bug in code for handling temporary memory usage when there are more
5678 back references than supplied space in the ovector. This could cause segfaults.
5681 Version 0.97 21-Oct-97
5682 ----------------------
5684 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5686 2. Optimized negated single characters not to use a bit map.
5688 3. Brought error texts together as macro definitions; clarified some of them;
5689 fixed one that was wrong - it said "range out of order" when it meant "invalid
5690 escape sequence".
5692 4. Changed some char * arguments to const char *.
5694 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5696 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5697 pcretest.
5700 Version 0.96 16-Oct-97
5701 ----------------------
5703 1. Added a simple "pgrep" utility to the distribution.
5705 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5706 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5707 where "ddd" means "one or more decimal digits".
5709 3. Fixed serious bug. If a pattern had a back reference, but the call to
5710 pcre_exec() didn't supply a large enough ovector to record the related
5711 identifying subpattern, the match always failed. PCRE now remembers the number
5712 of the largest back reference, and gets some temporary memory in which to save
5713 the offsets during matching if necessary, in order to ensure that
5714 backreferences always work.
5716 4. Increased the compatibility with Perl in a number of ways:
5718 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5719 to request this handling. The option can be set at compile or exec time.
5721 (b) $ matches before a terminating newline by default; an option
5722 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5723 mode). The option can be set at compile or exec time.
5725 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5726 the same as Perl's. If the decimal number it represents is less than 10
5727 or there aren't that many previous left capturing parentheses, an octal
5728 escape is read. Inside a character class, it's always an octal escape,
5729 even if it is a single digit.
5731 (d) An escaped but undefined alphabetic character is taken as a literal,
5732 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5733 escapes.
5735 (e) {0} is now permitted. (The previous item is removed from the compiled
5736 pattern).
5738 5. Changed all the names of code files so that the basic parts are no longer
5739 than 10 characters, and abolished the teeny "globals.c" file.
5741 6. Changed the handling of character classes; they are now done with a 32-byte
5742 bit map always.
5744 7. Added the -d and /D options to pcretest to make it possible to look at the
5745 internals of compilation without having to recompile pcre.
5748 Version 0.95 23-Sep-97
5749 ----------------------
5751 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5752 \x20 at the start of a run of normal characters. These were being treated as
5753 real characters, instead of the source characters being re-checked.
5756 Version 0.94 18-Sep-97
5757 ----------------------
5759 1. The functions are now thread-safe, with the caveat that the global variables
5760 containing pointers to malloc() and free() or alternative functions are the
5761 same for all threads.
5763 2. Get pcre_study() to generate a bitmap of initial characters for non-
5764 anchored patterns when this is possible, and use it if passed to pcre_exec().
5767 Version 0.93 15-Sep-97
5768 ----------------------
5770 1. /(b)|(:+)/ was computing an incorrect first character.
5772 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5773 but not actually doing anything yet.
5775 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5776 as Perl does (e.g. [-az] or [az-]).
5778 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5779 all possible positions.
5781 5. Split up into different modules to avoid including unneeded functions in a
5782 compiled binary. However, compile and exec are still in one module. The "study"
5783 function is split off.
5785 6. The character tables are now in a separate module whose source is generated
5786 by an auxiliary program - but can then be edited by hand if required. There are
5787 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5788 toupper() in the code.
5790 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5791 make them global. Abolish the function for setting them, as the caller can now
5792 set them directly.
5795 Version 0.92 11-Sep-97
5796 ----------------------
5798 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5799 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5801 2. Caseless matching was not working in character classes if the characters in
5802 the pattern were in upper case.
5804 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5806 4. Make PCRE_ANCHORED public and accept as a compile option.
5808 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5809 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5810 pass them.
5812 6. Give an error if bad option bits passed at compile or run time.
5814 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5815 pcretest to cause it to pass that flag.
5817 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5818 options, and the first character, if set.
5820 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5823 Version 0.91 10-Sep-97
5824 ----------------------
5826 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5827 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5829 2. PCRE was looping on encountering an indefinitely repeated back reference to
5830 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5831 Perl does - treats the match as successful.
5833 ****


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

  ViewVC Help
Powered by ViewVC 1.1.5