4 Version 8.35 04-April-2014
5 --------------------------
7 1. A new flag is set, when property checks are present in an XCLASS.
8 When this flag is not set, PCRE can perform certain optimizations
9 such as studying these XCLASS-es.
11 2. The auto-possessification of character sets were improved: a normal
12 and an extended character set can be compared now. Furthermore
13 the JIT compiler optimizes more character set checks.
15 3. Got rid of some compiler warnings for potentially uninitialized variables
16 that show up only when compiled with -O2.
18 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
19 of a match later then the end of the match. The pcretest program was not
20 handling the case sensibly - it was outputting from the start to the next
21 binary zero. It now reports this situation in a message, and outputs the
22 text from the end to the start.
24 5. Fast forward search is improved in JIT. Instead of the first three
25 characters, any three characters with fixed position can be searched.
26 Search order: first, last, middle.
28 6. Improve character range checks in JIT. Characters are read by an inprecise
29 function now, which returns with an unknown value if the character code is
30 above a certain treshold (e.g: 256). The only limitation is that the value
31 must be bigger than the treshold as well. This function is useful, when
32 the characters above the treshold are handled in the same way.
34 7. The macros whose names start with RAWUCHAR are placeholders for a future
35 mode in which only the bottom 21 bits of 32-bit data items are used. To
36 make this more memorable for those maintaining the code, the names have
37 been changed to start with UCHAR21, and an extensive comment has been added
40 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
41 to the export list in Makefile.am (they were accidentally omitted from the
44 9. The informational output from pcretest used the phrase "starting byte set"
45 which is inappropriate for the 16-bit and 32-bit libraries. As the output
46 for "first char" and "need char" really means "non-UTF-char", I've changed
47 "byte" to "char", and slightly reworded the output. The documentation about
48 these values has also been (I hope) clarified.
50 10. Another JIT related optimization: use table jumps for selecting the correct
51 backtracking path, when more than four alternatives are present inside a
54 11. Empty match is not possible, when the minimum length is greater than zero,
55 and there is no \K in the pattern. JIT should avoid empty match checks in
58 12. In a caseless character class with UCP support, when a character with more
59 than one alternative case was not the first character of a range, not all
60 the alternative cases were added to the class. For example, s and \x{17f}
61 are both alternative cases for S: the class [RST] was handled correctly,
64 13. The configure.ac file always checked for pthread support when JIT was
65 enabled. This is not used in Windows, so I have put this test inside a
66 check for the presence of windows.h (which was already tested for).
68 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
69 The algorithm provides a way to skip certain starting offsets, and usually
70 faster than linear prefix searches.
72 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
73 as for 'fr_FR' and 'french'. For some reason, however, it then used the
74 Windows-specific input and output files, which have 'french' screwed in.
75 So this could never have worked. One of the problems with locales is that
76 they aren't always the same. I have now updated RunTest so that it checks
77 the output of the locale test (test 3) against three different output
78 files, and it allows the test to pass if any one of them matches. With luck
79 this should make the test pass on some versions of Solaris where it was
80 failing. Because of the uncertainty, the script did not used to stop if
81 test 3 failed; it now does. If further versions of a French locale ever
82 come to light, they can now easily be added.
84 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
85 there was a message during ./configure, but it did not stop. This now
86 provokes an error. The invalid example in README has been corrected.
87 If a value less than the minimum is given, the minimum value has always
88 been used, but now a warning is given.
90 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
91 was a bug in the test system, which is now fixed. Also, the list of various
92 configurations that are tested for each release did not have one with both
93 16/32 bits and --enable-bar-anycrlf. It now does.
95 18. pcretest was missing "-C bsr" for displaying the \R default setting.
97 19. Little endian PowerPC systems are supported now by the JIT compiler.
99 20. The fast forward newline mechanism could enter to an infinite loop on
100 certain invalid UTF-8 input. Although we don't support these cases
101 this issue can be fixed by a performance optimization.
103 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
104 not take account if existing stack usage. There is now a new global
105 variable called pcre_stack_guard that can be set to point to an external
106 function to check stack availability. It is called at the start of
107 processing every parenthesized group.
109 22. A typo in the code meant that in ungreedy mode the max/min qualifier
110 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
113 23. When UTF was disabled, the JIT program reported some incorrect compile
114 errors. These messages are silenced now.
116 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
119 25. Change all the temporary files used in RunGrepTest to be different to those
120 used by RunTest so that the tests can be run simultaneously, for example by
124 Version 8.34 15-December-2013
125 -----------------------------
127 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
128 executable memory. Patch inspired by Carsten Klein.
130 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
131 this macro is never tested and has no effect, because the work to support
132 coverage involves only compiling and linking options and special targets in
133 the Makefile. The comment in config.h implied that defining the macro would
134 enable coverage support, which is totally false. There was also support for
135 setting this macro in the CMake files (my fault, I just copied it from
136 configure). SUPPORT_GCOV has now been removed.
138 3. Make a small performance improvement in strlen16() and strlen32() in
141 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
142 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
143 these cases). There was also one in pcretest.c.
145 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
147 6. In UTF mode, the code for checking whether a group could match an empty
148 string (which is used for indefinitely repeated groups to allow for
149 breaking an infinite loop) was broken when the group contained a repeated
150 negated single-character class with a character that occupied more than one
151 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
152 UTF-8 mode). The effect was undefined: the group might or might not be
153 deemed as matching an empty string, or the program might have crashed.
155 7. The code for checking whether a group could match an empty string was not
156 recognizing that \h, \H, \v, \V, and \R must match a character.
158 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
159 an empty string. If it can, pcretest shows this in its information output.
161 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
162 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
163 matched by pcre_exec() without using JIT:
165 (a) If the rest of the pattern did not match after a maximal run of
166 grapheme clusters, the code for backing up to try with fewer of them
167 did not always back up over a full grapheme when characters that do not
168 have the modifier quality were involved, e.g. Hangul syllables.
170 (b) If the match point in a subject started with modifier character, and
171 there was no match, the code could incorrectly back up beyond the match
172 point, and potentially beyond the first character in the subject,
173 leading to a segfault or an incorrect match result.
175 10. A conditional group with an assertion condition could lead to PCRE
176 recording an incorrect first data item for a match if no other first data
177 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
178 first data item, and therefore matched "ca" after "c" instead of at the
181 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
182 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
184 12. The source of pcregrep now includes z/OS-specific code so that it can be
185 compiled for z/OS as part of the special z/OS distribution.
187 13. Added the -T and -TM options to pcretest.
189 14. The code in pcre_compile.c for creating the table of named capturing groups
190 has been refactored. Instead of creating the table dynamically during the
191 actual compiling pass, the information is remembered during the pre-compile
192 pass (on the stack unless there are more than 20 named groups, in which
193 case malloc() is used) and the whole table is created before the actual
194 compile happens. This has simplified the code (it is now nearly 150 lines
195 shorter) and prepared the way for better handling of references to groups
196 with duplicate names.
198 15. A back reference to a named subpattern when there is more than one of the
199 same name now checks them in the order in which they appear in the pattern.
200 The first one that is set is used for the reference. Previously only the
201 first one was inspected. This change makes PCRE more compatible with Perl.
203 16. Unicode character properties were updated from Unicode 6.3.0.
205 17. The compile-time code for auto-possessification has been refactored, based
206 on a patch by Zoltan Herczeg. It now happens after instead of during
207 compilation. The code is cleaner, and more cases are handled. The option
208 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
209 options in pcretest are provided to set it. It can also be set by
210 (*NO_AUTO_POSSESS) at the start of a pattern.
212 18. The character VT has been added to the default ("C" locale) set of
213 characters that match \s and are generally treated as white space,
214 following this same change in Perl 5.18. There is now no difference between
215 "Perl space" and "POSIX space". Whether VT is treated as white space in
216 other locales depends on the locale.
218 19. The code for checking named groups as conditions, either for being set or
219 for being recursed, has been refactored (this is related to 14 and 15
220 above). Processing unduplicated named groups should now be as fast at
221 numerical groups, and processing duplicated groups should be faster than
224 20. Two patches to the CMake build system, by Alexander Barkov:
226 (1) Replace the "source" command by "." in CMakeLists.txt because
227 "source" is a bash-ism.
229 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
230 without these the CMake build does not work on Solaris.
232 21. Perl has changed its handling of \8 and \9. If there is no previously
233 encountered capturing group of those numbers, they are treated as the
234 literal characters 8 and 9 instead of a binary zero followed by the
235 literals. PCRE now does the same.
237 22. Following Perl, added \o{} to specify codepoints in octal, making it
238 possible to specify values greater than 0777 and also making them
241 23. Perl now gives an error for missing closing braces after \x{... instead of
242 treating the string as literal. PCRE now does the same.
244 24. RunTest used to grumble if an inappropriate test was selected explicitly,
245 but just skip it when running all tests. This make it awkward to run ranges
246 of tests when one of them was inappropriate. Now it just skips any
247 inappropriate tests, as it always did when running all tests.
249 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
250 character types such as \d or \w, too many callouts were inserted, and the
251 data that they returned was rubbish.
253 26. In UCP mode, \s was not matching two of the characters that Perl matches,
254 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
255 were matched by \h. The code has now been refactored so that the lists of
256 the horizontal and vertical whitespace characters used for \h and \v (which
257 are defined only in one place) are now also used for \s.
259 27. Add JIT support for the 64 bit TileGX architecture.
260 Patch by Jiong Wang (Tilera Corporation).
262 28. Possessive quantifiers for classes (both explicit and automatically
263 generated) now use special opcodes instead of wrapping in ONCE brackets.
265 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
266 (because it's meaningless), this was not happening when PCRE_CASELESS was
267 set. Not wrong, but inefficient.
269 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
270 properties for \w, \d, etc) is present in a test regex. Otherwise if the
271 test contains no characters greater than 255, Perl doesn't realise it
272 should be using Unicode semantics.
274 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
275 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
276 does in Unicode mode.
278 32. Added the "forbid" facility to pcretest so that putting tests into the
279 wrong test files can sometimes be quickly detected.
281 33. There is now a limit (default 250) on the depth of nesting of parentheses.
282 This limit is imposed to control the amount of system stack used at compile
283 time. It can be changed at build time by --with-parens-nest-limit=xxx or
284 the equivalent in CMake.
286 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
287 errors. Perl warns for these when in warning mode, but PCRE has no facility
290 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
291 However, this was not working for (?!) because it is optimized to (*FAIL),
292 for which PCRE does not allow quantifiers. The optimization is now disabled
293 when a quantifier follows (?!). I can't see any use for this, but it makes
296 36. Perl no longer allows group names to start with digits, so I have made this
297 change also in PCRE. It simplifies the code a bit.
299 37. In extended mode, Perl ignores spaces before a + that indicates a
300 possessive quantifier. PCRE allowed a space before the quantifier, but not
301 before the possessive +. It now does.
303 38. The use of \K (reset reported match start) within a repeated possessive
304 group such as (a\Kb)*+ was not working.
306 40. Document that the same character tables must be used at compile time and
307 run time, and that the facility to pass tables to pcre_exec() and
308 pcre_dfa_exec() is for use only with saved/restored patterns.
310 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
311 features for Builds with MSVC:
313 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
314 builds). The .rc files can be used to set FileDescription and many other
317 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
318 This allows higher-level build scripts which want .pdb files to avoid
319 hard-coding the exact files needed."
321 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
322 mean "start of word" and "end of word", respectively, as a transition aid.
324 43. A minimizing repeat of a class containing codepoints greater than 255 in
325 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
326 compiled to use the heap for recursion.
328 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
332 Version 8.33 28-May-2013
333 ------------------------
335 1. Added 'U' to some constants that are compared to unsigned integers, to
336 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
337 variables that are added to signed variables, to ensure the result is
338 signed and can be negated.
340 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
343 3. Revise the creation of config.h.generic so that all boolean macros are
344 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
345 overriding via -D on the command line possible.
347 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
348 to unsigned int is reported to make a quite noticeable speed difference in
349 a specific Windows environment. Testing on Linux did also appear to show
350 some benefit (and it is clearly not harmful). Also fixed the definition of
351 Xop which should be unsigned.
353 5. Related to (4), changing the definition of the intermediate variable cc
354 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
357 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
360 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
363 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
365 9. Optimizing fast_forward_start_bits in JIT.
367 10. Adding support for callouts in JIT, and fixing some issues revealed
368 during this work. Namely:
370 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
372 (b) Minimum length was not checked before the matching is started.
374 11. The value of capture_last that is passed to callouts was incorrect in some
375 cases when there was a capture on one path that was subsequently abandoned
376 after a backtrack. Also, the capture_last value is now reset after a
377 recursion, since all captures are also reset in this case.
379 12. The interpreter no longer returns the "too many substrings" error in the
380 case when an overflowing capture is in a branch that is subsequently
381 abandoned after a backtrack.
383 13. In the pathological case when an offset vector of size 2 is used, pcretest
384 now prints out the matched string after a yield of 0 or 1.
386 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
387 Only supported by the JIT compiler at the moment.
389 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
391 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
392 offset of the starting point of the matching process, provided the offsets
393 vector is large enough.
395 17. The \A escape now records a lookbehind value of 1, though its execution
396 does not actually inspect the previous character. This is to ensure that,
397 in partial multi-segment matching, at least one character from the old
398 segment is retained when a new segment is processed. Otherwise, if there
399 are no lookbehinds in the pattern, \A might match incorrectly at the start
402 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
404 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
405 gives some modest performance improvement in 8-bit mode.
407 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
408 be expressed in certain programming languages using Universal Character
411 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
412 which points out that "non characters" are not "characters that may not
413 appear in Unicode strings" but rather "characters that are reserved for
414 internal use and have only local meaning".
416 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
417 there was a conditional group that depended on an assertion, if the
418 assertion was false, the callout that immediately followed the alternation
419 in the condition was skipped when pcre_exec() was used for matching.
421 23. Allow an explicit callout to be inserted before an assertion that is the
422 condition for a conditional group, for compatibility with automatic
423 callouts, which always insert a callout at this point.
425 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
426 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
428 25. (*PRUNE) is now supported by the JIT compiler.
430 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
432 27. Fix the case where there are two or more SKIPs with arguments that may be
435 28. (*SKIP) is now supported by the JIT compiler.
437 29. (*THEN) is now supported by the JIT compiler.
439 30. Update RunTest with additional test selector options.
441 31. The way PCRE handles backtracking verbs has been changed in two ways.
443 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
444 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
445 backtracking. In some cases this makes it more Perl-compatible, but Perl's
446 rather obscure rules do not always do the same thing.
448 (2) Previously, backtracking verbs were confined within assertions. This is
449 no longer the case for positive assertions, except for (*ACCEPT). Again,
450 this sometimes improves Perl compatibility, and sometimes does not.
452 32. A number of tests that were in test 2 because Perl did things differently
453 have been moved to test 1, because either Perl or PCRE has changed, and
454 these tests are now compatible.
456 32. Backtracking control verbs are now handled in the same way in JIT and
459 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
460 contained a forward subroutine reference caused a compile error.
462 34. Auto-detect and optimize limited repetitions in JIT.
464 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
467 36. In the interpreter, maximizing pattern repetitions for characters and
468 character types now use tail recursion, which reduces stack usage.
470 37. The value of the max lookbehind was not correctly preserved if a compiled
471 and saved regex was reloaded on a host of different endianness.
473 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
474 of the compiled pattern block, expand the flags field from 16 to 32 bits
475 because it was almost full.
477 39. Try madvise first before posix_madvise.
479 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
480 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
481 original change locked out all matches of empty strings. This has been
482 changed so that one match of an empty string per line is recognized.
483 Subsequent searches on the same line (for colouring or for --only-matching,
484 for example) do not recognize empty strings.
486 41. Applied a user patch to fix a number of spelling mistakes in comments.
488 42. Data lines longer than 65536 caused pcretest to crash.
490 43. Clarified the data type for length and startoffset arguments for pcre_exec
491 and pcre_dfa_exec in the function-specific man pages, where they were
492 explicitly stated to be in bytes, never having been updated. I also added
493 some clarification to the pcreapi man page.
495 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
496 a segmentation fault.
499 Version 8.32 30-November-2012
500 -----------------------------
502 1. Improved JIT compiler optimizations for first character search and single
505 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
506 Patch by Daniel Richard G.
508 3. Single character iterator optimizations in the JIT compiler.
510 4. Improved JIT compiler optimizations for character ranges.
512 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
513 Reported by Giuseppe D'Angelo.
515 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
516 of a line, was being set incorrectly in cases where .* appeared inside
517 atomic brackets at the start of a pattern, or where there was a subsequent
520 7. Improved instruction cache flush for POWER/PowerPC.
521 Patch by Daniel Richard G.
523 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
526 (a) There is now no limit to the number of patterns to be matched.
528 (b) An error is given if a pattern is too long.
530 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
533 (d) --exclude-from and --include-from (multiple use) have been added.
535 (e) Exclusions and inclusions now apply to all files and directories, not
536 just to those obtained from scanning a directory recursively.
538 (f) Multiple uses of -f and --file-list are now supported.
540 (g) In a Windows environment, the default for -d has been changed from
541 "read" (the GNU grep default) to "skip", because otherwise the presence
542 of a directory in the file list provokes an error.
544 (h) The documentation has been revised and clarified in places.
546 9. Improve the matching speed of capturing brackets.
548 10. Changed the meaning of \X so that it now matches a Unicode extended
551 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
552 out POSIX threads when JIT support is configured.
554 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
556 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
557 structure could go wrong in environments where size_t is not the same size
560 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
563 15. The EBCDIC support had decayed; later updates to the code had included
564 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
565 general tidy up of EBCDIC-related issues, and the documentation was also
566 not quite right. There is now a test that can be run on ASCII systems to
567 check some of the EBCDIC-related things (but is it not a full test).
569 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
570 in a small tidy to the code.
572 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
574 18. If the --only-matching (-o) option in pcregrep is specified multiple
575 times, each one causes appropriate output. For example, -o1 -o2 outputs the
576 substrings matched by the 1st and 2nd capturing parentheses. A separating
577 string can be specified by --om-separator (default empty).
579 19. Improving the first n character searches.
581 20. Turn case lists for horizontal and vertical white space into macros so that
582 they are defined only once.
584 21. This set of changes together give more compatible Unicode case-folding
585 behaviour for characters that have more than one other case when UCP
586 support is available.
588 (a) The Unicode property table now has offsets into a new table of sets of
589 three or more characters that are case-equivalent. The MultiStage2.py
590 script that generates these tables (the pcre_ucd.c file) now scans
591 CaseFolding.txt instead of UnicodeData.txt for character case
594 (b) The code for adding characters or ranges of characters to a character
595 class has been abstracted into a generalized function that also handles
596 case-independence. In UTF-mode with UCP support, this uses the new data
597 to handle characters with more than one other case.
599 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
600 whose other case is greater than 256 are now correctly matched
601 caselessly. Previously, the high codepoint matched the low one, but not
604 (d) The processing of \h, \H, \v, and \ in character classes now makes use
605 of the new class addition function, using character lists defined as
606 macros alongside the case definitions of 20 above.
608 (e) Caseless back references now work with characters that have more than
611 (f) General caseless matching of characters with more than one other case
614 22. Unicode character properties were updated from Unicode 6.2.0
616 23. Improved CMake support under Windows. Patch by Daniel Richard G.
618 24. Add support for 32-bit character strings, and UTF-32
620 25. Major JIT compiler update (code refactoring and bugfixing).
621 Experimental Sparc 32 support is added.
623 26. Applied a modified version of Daniel Richard G's patch to create
624 pcre.h.generic and config.h.generic by "make" instead of in the
625 PrepareRelease script.
627 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
628 pcre_compile.c when checking for a zero character.
630 28. Introducing a native interface for JIT. Through this interface, the compiled
631 machine code can be directly executed. The purpose of this interface is to
632 provide fast pattern matching, so several sanity checks are not performed.
633 However, feature tests are still performed. The new interface provides
634 1.4x speedup compared to the old one.
636 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
637 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
638 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
640 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
641 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
642 the "old" RFC 2279). Instead, it ended up passing a negative length to
645 31. Add support for GCC's visibility feature to hide internal functions.
647 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
648 "unknown -C option" after outputting 0 or 1.
650 33. There is now support for generating a code coverage report for the test
651 suite in environments where gcc is the compiler and lcov is installed. This
652 is mainly for the benefit of the developers.
654 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
655 unaddressable using valgrind annotations, allowing valgrind to detect
656 invalid memory accesses. This is mainly for the benefit of the developers.
658 25. (*UTF) can now be used to start a pattern in any of the three libraries.
660 26. Give configure error if --enable-cpp but no C++ compiler found.
663 Version 8.31 06-July-2012
664 -------------------------
666 1. Fixing a wrong JIT test case and some compiler warnings.
668 2. Removed a bashism from the RunTest script.
670 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
671 to unsigned type, result still unsigned" that was given by an MS compiler
672 on encountering the code "-sizeof(xxx)".
674 4. Partial matching support is added to the JIT compiler.
676 5. Fixed several bugs concerned with partial matching of items that consist
677 of more than one character:
679 (a) /^(..)\1/ did not partially match "aba" because checking references was
680 done on an "all or nothing" basis. This also applied to repeated
683 (b) \R did not give a hard partial match if \r was found at the end of the
686 (c) \X did not give a hard partial match after matching one or more
687 characters at the end of the subject.
689 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
690 a partial match for the string "\r".
692 (e) When newline was set to CRLF, the metacharacter "." did not recognize
693 a partial match for a CR character at the end of the subject string.
695 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
696 running pcretest, the text "(JIT)" added to the output whenever JIT is
697 actually used to run the match.
699 7. Individual JIT compile options can be set in pcretest by following -s+[+]
700 or /S+[+] with a digit between 1 and 7.
702 8. OP_NOT now supports any UTF character not just single-byte ones.
704 9. (*MARK) control verb is now supported by the JIT compiler.
706 10. The command "./RunTest list" lists the available tests without actually
707 running any of them. (Because I keep forgetting what they all are.)
709 11. Add PCRE_INFO_MAXLOOKBEHIND.
711 12. Applied a (slightly modified) user-supplied patch that improves performance
712 when the heap is used for recursion (compiled with --disable-stack-for-
713 recursion). Instead of malloc and free for each heap frame each time a
714 logical recursion happens, frames are retained on a chain and re-used where
715 possible. This sometimes gives as much as 30% improvement.
717 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
720 14. As documented, (*COMMIT) is now confined to within a positive assertion.
722 15. It is now possible to link pcretest with libedit as an alternative to
725 16. (*COMMIT) control verb is now supported by the JIT compiler.
727 17. The Unicode data tables have been updated to Unicode 6.1.0.
729 18. Added --file-list option to pcregrep.
731 19. Added binary file support to pcregrep, including the -a, --binary-files,
732 -I, and --text options.
734 20. The madvise function is renamed for posix_madvise for QNX compatibility
735 reasons. Fixed by Giuseppe D'Angelo.
737 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
739 22. Changed the option for creating symbolic links for 16-bit man pages from
740 -s to -sf so that re-installing does not cause issues.
742 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
744 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
745 matches in certain environments (the workspace was not being correctly
746 retained). Also added to pcre_dfa_exec() a simple plausibility check on
747 some of the workspace data at the beginning of a restart.
749 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
750 was not doing so when it should - probably a typo introduced by SVN 528
753 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
754 \w+ when the character tables indicated that \x{c4} was a word character.
755 There were several related cases, all because the tests for doing a table
756 lookup were testing for characters less than 127 instead of 255.
758 27. If a pattern contains capturing parentheses that are not used in a match,
759 their slots in the ovector are set to -1. For those that are higher than
760 any matched groups, this happens at the end of processing. In the case when
761 there were back references that the ovector was too small to contain
762 (causing temporary malloc'd memory to be used during matching), and the
763 highest capturing number was not used, memory off the end of the ovector
764 was incorrectly being set to -1. (It was using the size of the temporary
765 memory instead of the true size.)
767 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
768 ovector size, it uses memory at the end of the block that it has got.
770 29. Check for an overlong MARK name and give an error at compile time. The
771 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
773 30. JIT compiler update.
775 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
778 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
780 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
782 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
783 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
786 35. Improve JIT code generation for greedy plus quantifier.
788 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
789 1 to "aa" instead of to an empty string. The bug affected repeated groups
790 that could potentially match an empty string.
792 37. Optimizing single character iterators in JIT.
794 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
795 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
796 codepoints that are too big for the mode are faulted, and in a UTF mode,
797 disallowed codepoints are also faulted.
799 39. If PCRE was compiled with UTF support, in three places in the DFA
800 matcher there was code that should only have been obeyed in UTF mode, but
801 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
802 processing when bytes with values greater than 127 were present. In 16-bit
803 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
804 both cases the values are those that cannot be the first data item in a UTF
805 character. The three items that might have provoked this were recursions,
806 possessively repeated groups, and atomic groups.
808 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
809 and pcregrep, because some OS require shared objects to be explicitly
810 passed to ld, causing the link step to fail if they are not.
812 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
813 mode, patterns that started with \h* or \R* might be incorrectly matched.
816 Version 8.30 04-February-2012
817 -----------------------------
819 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
820 name is defined in ctype.h.
822 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
823 only in quite long subpatterns.
825 3. Removed the function pcre_info(), which has been obsolete and deprecated
826 since it was replaced by pcre_fullinfo() in February 2000.
828 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
829 match a (*MARK), and the match failed at the start of the subject, a
830 reference to memory before the start of the subject could occur. This bug
831 was introduced by fix 17 of release 8.21.
833 5. A reference to an unset group with zero minimum repetition was giving
834 totally wrong answers (in non-JavaScript-compatibility mode). For example,
835 /(another)?(\1?)test/ matched against "hello world test". This bug was
836 introduced in release 8.13.
838 6. Add support for 16-bit character strings (a large amount of work involving
839 many changes and refactorings).
841 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
842 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
845 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
846 rounding is not applied in this particular case).
848 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
849 if they appear, or are escaped, in patterns.
851 10. Get rid of a number of -Wunused-but-set-variable warnings.
853 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
854 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
855 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
856 also returns the mark "x". This bug applied to capturing parentheses,
857 non-capturing parentheses, and atomic parentheses. It also applied to some
860 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
861 information out of configure.ac instead of relying on pcre.h.generic, which
862 is not stored in the repository.
864 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
867 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
869 15. Applied Graycode's patch to put the top-level frame on the stack rather
870 than the heap when not using the stack for recursion. This gives a
871 performance improvement in many cases when recursion is not deep.
873 16. Experimental code added to "pcretest -C" to output the stack frame size.
876 Version 8.21 12-Dec-2011
877 ------------------------
879 1. Updating the JIT compiler.
881 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
884 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
885 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
886 calling _pcre_jit_exec. Some extra comments are added.
888 4. (*MARK) settings inside atomic groups that do not contain any capturing
889 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
890 was introduced by change 18 for 8.20.
892 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
895 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
896 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
897 This bug was probably introduced by change 9 of 8.13.
899 7. While fixing 6 above, I noticed that a number of other items were being
900 incorrectly rejected as "not fixed length". This arose partly because newer
901 opcodes had not been added to the fixed-length checking code. I have (a)
902 corrected the bug and added tests for these items, and (b) arranged for an
903 error to occur if an unknown opcode is encountered while checking for fixed
904 length instead of just assuming "not fixed length". The items that were
905 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
906 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
907 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
909 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
910 being incorrectly compiled and would have given unpredicatble results.
912 9. A possessively repeated subpattern with minimum repeat count greater than
913 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
914 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
915 the first (A) could occur when it should not.
917 10. Add a cast and remove a redundant test from the code.
919 11. JIT should use pcre_malloc/pcre_free for allocation.
921 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
922 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
923 is anything other than /usr, -L is still shown).
925 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
927 14. Perl does not support \N without a following name in a [] class; PCRE now
930 15. If a forward reference was repeated with an upper limit of around 2000,
931 it caused the error "internal error: overran compiling workspace". The
932 maximum number of forward references (including repeats) was limited by the
933 internal workspace, and dependent on the LINK_SIZE. The code has been
934 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
935 the default depends on LINK_SIZE. There is a new upper limit (for safety)
936 of around 200,000 forward references. While doing this, I also speeded up
937 the filling in of repeated forward references.
939 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
940 incorrectly expecting the subject to contain another "a" after the start.
942 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
943 in the match, the SKIP should be ignored. This was not happening; instead
944 the SKIP was being treated as NOMATCH. For patterns such as
945 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
948 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
949 now much more compatible with Perl, in particular in cases where the result
950 is a non-match for a non-anchored pattern. For example, if
951 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
952 "m", where previously it did not return a name. A side effect of this
953 change is that for partial matches, the last encountered mark name is
954 returned, as for non matches. A number of tests that were previously not
955 Perl-compatible have been moved into the Perl-compatible test files. The
956 refactoring has had the pleasing side effect of removing one argument from
957 the match() function, thus reducing its stack requirements.
959 19. If the /S+ option was used in pcretest to study a pattern using JIT,
960 subsequent uses of /S (without +) incorrectly behaved like /S+.
962 21. Retrieve executable code size support for the JIT compiler and fixing
965 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
966 not work when the shorter character appeared right at the end of the
969 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
972 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
973 output it when the /M option is used in pcretest.
975 25. The CheckMan script was not being included in the distribution. Also, added
976 an explicit "perl" to run Perl scripts from the PrepareRelease script
977 because this is reportedly needed in Windows.
979 26. If study data was being save in a file and studying had not found a set of
980 "starts with" bytes for the pattern, the data written to the file (though
981 never used) was taken from uninitialized memory and so caused valgrind to
984 27. Updated RunTest.bat as provided by Sheri Pierce.
986 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
988 29. Computation of memory usage for the table of capturing group names was
989 giving an unnecessarily large value.
992 Version 8.20 21-Oct-2011
993 ------------------------
995 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
996 a POSIX class. After further experiments with Perl, which convinced me that
997 Perl has bugs and confusions, a closing square bracket is no longer allowed
998 in a POSIX name. This bug also affected patterns with classes that started
1001 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1002 captured substring, but while checking the failing first alternative,
1003 substring 1 is temporarily captured. If the output vector supplied to
1004 pcre_exec() was not big enough for this capture, the yield of the function
1005 was still zero ("insufficient space for captured substrings"). This cannot
1006 be totally fixed without adding another stack variable, which seems a lot
1007 of expense for a edge case. However, I have improved the situation in cases
1008 such as /(a)(b)x|abc/ matched against "abc", where the return code
1009 indicates that fewer than the maximum number of slots in the ovector have
1012 3. Related to (2) above: when there are more back references in a pattern than
1013 slots in the output vector, pcre_exec() uses temporary memory during
1014 matching, and copies in the captures as far as possible afterwards. It was
1015 using the entire output vector, but this conflicts with the specification
1016 that only 2/3 is used for passing back captured substrings. Now it uses
1017 only the first 2/3, for compatibility. This is, of course, another edge
1020 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1021 main code base, and can be used by building with --enable-jit. When this is
1022 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1023 runtime --no-jit option is given.
1025 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1026 ovector, the return from the function was zero, implying that there were
1027 other matches that did not fit. The correct "exactly full" value is now
1030 6. If a subpattern that was called recursively or as a subroutine contained
1031 (*PRUNE) or any other control that caused it to give a non-standard return,
1032 invalid errors such as "Error -26 (nested recursion at the same subject
1033 position)" or even infinite loops could occur.
1035 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1036 computing the minimum length on reaching *ACCEPT, and so ended up with the
1037 wrong value of 1 rather than 0. Further investigation indicates that
1038 computing a minimum subject length in the presence of *ACCEPT is difficult
1039 (think back references, subroutine calls), and so I have changed the code
1040 so that no minimum is registered for a pattern that contains *ACCEPT.
1042 8. If (*THEN) was present in the first (true) branch of a conditional group,
1043 it was not handled as intended. [But see 16 below.]
1045 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1048 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1049 the first byte in a match must be "a".
1051 11. Change 17 for 8.13 increased the recursion depth for patterns like
1052 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1053 pattern contains any instances of (*THEN). If it does not, the old
1054 optimizations are restored. It would be nice to do this on a per-group
1055 basis, but at the moment that is not feasible.
1057 12. In some environments, the output of pcretest -C is CRLF terminated. This
1058 broke RunTest's code that checks for the link size. A single white space
1059 character after the value is now allowed for.
1061 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1062 For "fr", it uses the Windows-specific input and output files.
1064 14. If (*THEN) appeared in a group that was called recursively or as a
1065 subroutine, it did not work as intended. [But see next item.]
1067 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1068 pattern fragments (but not containing any | characters). If A and B are
1069 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1070 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1071 D. In other words, Perl considers parentheses that do not contain any |
1072 characters to be part of a surrounding alternative, whereas PCRE was
1073 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1074 differently. PCRE now behaves in the same way as Perl, except in the case
1075 of subroutine/recursion calls such as (?1) which have in any case always
1076 been different (but PCRE had them first :-).
1078 16. Related to 15 above: Perl does not treat the | in a conditional group as
1079 creating alternatives. Such a group is treated in the same way as an
1080 ordinary group without any | characters when processing (*THEN). PCRE has
1081 been changed to match Perl's behaviour.
1083 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1084 RunGrepTest script failed.
1086 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1087 inevitable for groups that contain captures, but it can lead to a lot of
1088 stack use in large patterns. The old behaviour has been restored for atomic
1089 groups that do not contain any capturing parentheses.
1091 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1092 suppress the check for a minimum subject length at run time. (If it was
1093 given to pcre_exec() or pcre_dfa_exec() it did work.)
1095 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1096 fail to work when decoding hex characters in data strings in EBCDIC
1099 21. It appears that in at least one Mac OS environment, the isxdigit() function
1100 is implemented as a macro that evaluates to its argument more than once,
1101 contravening the C 90 Standard (I haven't checked a later standard). There
1102 was an instance in pcretest which caused it to go wrong when processing
1103 \x{...} escapes in subject strings. The has been rewritten to avoid using
1104 things like p++ in the argument of isxdigit().
1107 Version 8.13 16-Aug-2011
1108 ------------------------
1110 1. The Unicode data tables have been updated to Unicode 6.0.0.
1112 2. Two minor typos in pcre_internal.h have been fixed.
1114 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1115 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1116 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1118 4. There were a number of related bugs in the code for matching backrefences
1119 caselessly in UTF-8 mode when codes for the characters concerned were
1120 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1121 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1122 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1123 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1124 2-byte code at the end of the subject (it thought there wasn't enough data
1127 5. Comprehensive information about what went wrong is now returned by
1128 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1129 as the output vector has at least 2 elements. The offset of the start of
1130 the failing character and a reason code are placed in the vector.
1132 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1133 now returned is for the first byte of the failing character, instead of the
1134 last byte inspected. This is an incompatible change, but I hope it is small
1135 enough not to be a problem. It makes the returned offset consistent with
1136 pcre_exec() and pcre_dfa_exec().
1138 7. pcretest now gives a text phrase as well as the error number when
1139 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1140 failure, the offset and reason code are output.
1142 8. When \R was used with a maximizing quantifier it failed to skip backwards
1143 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1144 back over a single character (\n). This seems wrong (because it treated the
1145 two characters as a single entity when going forwards), conflicts with the
1146 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1147 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1150 9. Some internal refactoring has changed the processing so that the handling
1151 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1152 time (the PCRE_DOTALL option was changed this way some time ago: version
1153 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1154 which was always a bit of a fudge. It also means that there is one less
1155 argument for the match() function, which reduces its stack requirements
1156 slightly. This change also fixes an incompatibility with Perl: the pattern
1157 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1159 10. More internal refactoring has drastically reduced the number of recursive
1160 calls to match() for possessively repeated groups such as (abc)++ when
1163 11. While implementing 10, a number of bugs in the handling of groups were
1164 discovered and fixed:
1166 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1167 (a|)*(?1) gave a compile-time internal error.
1168 ((a|)+)+ did not notice that the outer group could match an empty string.
1169 (^a|^)+ was not marked as anchored.
1170 (.*a|.*)+ was not marked as matching at start or after a newline.
1172 12. Yet more internal refactoring has removed another argument from the match()
1173 function. Special calls to this function are now indicated by setting a
1174 value in a variable in the "match data" data block.
1176 13. Be more explicit in pcre_study() instead of relying on "default" for
1177 opcodes that mean there is no starting character; this means that when new
1178 ones are added and accidentally left out of pcre_study(), testing should
1181 14. The -s option of pcretest has been documented for ages as being an old
1182 synonym of -m (show memory usage). I have changed it to mean "force study
1183 for every regex", that is, assume /S for every regex. This is similar to -i
1184 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1185 using it. It makes it easier to run collections of tests with and without
1186 study enabled, and thereby test pcre_study() more easily. All the standard
1187 tests are now run with and without -s (but some patterns can be marked as
1188 "never study" - see 20 below).
1190 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1191 restoration of the capturing data to the outer values was not happening
1194 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1195 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1196 pattern had matched an empty string, and so incorrectly returned a no
1199 17. There was optimizing code for the last branch of non-capturing parentheses,
1200 and also for the obeyed branch of a conditional subexpression, which used
1201 tail recursion to cut down on stack usage. Unfortunately, now that there is
1202 the possibility of (*THEN) occurring in these branches, tail recursion is
1203 no longer possible because the return has to be checked for (*THEN). These
1204 two optimizations have therefore been removed. [But see 8.20/11 above.]
1206 18. If a pattern containing \R was studied, it was assumed that \R always
1207 matched two bytes, thus causing the minimum subject length to be
1208 incorrectly computed because \R can also match just one byte.
1210 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1211 was incorrectly computed.
1213 20. If /S is present twice on a test pattern in pcretest input, it now
1214 *disables* studying, thereby overriding the use of -s on the command line
1215 (see 14 above). This is necessary for one or two tests to keep the output
1216 identical in both cases.
1218 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1219 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1221 22. When an atomic group that contained a capturing parenthesis was
1222 successfully matched, but the branch in which it appeared failed, the
1223 capturing was not being forgotten if a higher numbered group was later
1224 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1225 group 1 to "a", when in fact it should be unset. This applied to multi-
1226 branched capturing and non-capturing groups, repeated or not, and also to
1227 positive assertions (capturing in negative assertions does not happen
1228 in PCRE) and also to nested atomic groups.
1230 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1231 subject after a captured substring, to make it easier to tell which of a
1232 number of identical substrings has been captured.
1234 24. The way atomic groups are processed by pcre_exec() has been changed so that
1235 if they are repeated, backtracking one repetition now resets captured
1236 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1237 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1238 "aaa". Previously, it would have been "a". As part of this code
1239 refactoring, the way recursive calls are handled has also been changed.
1241 25. If an assertion condition captured any substrings, they were not passed
1242 back unless some other capturing happened later. For example, if
1243 (?(?=(a))a) was matched against "a", no capturing was returned.
1245 26. When studying a pattern that contained subroutine calls or assertions,
1246 the code for finding the minimum length of a possible match was handling
1247 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1248 group 1 called group 2 while simultaneously a separate group 2 called group
1249 1). A stack overflow occurred in this case. I have fixed this by limiting
1250 the recursion depth to 10.
1252 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1253 Fortmann. This supports explicit test numbers on the command line, and has
1254 argument validation and error reporting.
1256 28. An instance of \X with an unlimited repeat could fail if at any point the
1257 first character it looked at was a mark character.
1259 29. Some minor code refactoring concerning Unicode properties and scripts
1260 should reduce the stack requirement of match() slightly.
1262 30. Added the '=' option to pcretest to check the setting of unused capturing
1263 slots at the end of the pattern, which are documented as being -1, but are
1264 not included in the return count.
1266 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1267 compiled something random. Now it gives a compile-time error (as does
1270 32. A *MARK encountered during the processing of a positive assertion is now
1271 recorded and passed back (compatible with Perl).
1273 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1274 had alternative anchored branches, the search for a second match in a line
1275 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1276 matched the line "0102" twice. The same bug affected patterns that started
1277 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1280 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1281 does, and because of capturing effects, quantifying parenthesized
1282 assertions may at times be useful. Quantifiers are now allowed for
1283 parenthesized assertions.
1285 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1287 36. \g was being checked for fancy things in a character class, when it should
1288 just be a literal "g".
1290 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1291 appearance of a nested POSIX class supersedes an apparent external class.
1292 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1293 unescaped square brackets may also appear as part of class names. For
1294 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1295 more like Perl. (But see 8.20/1 above.)
1297 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1298 was because it thought it was \N{name}, which is not supported).
1300 39. Add minix to OS list not supporting the -S option in pcretest.
1302 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1303 cannot analyze patterns in sufficient detail to catch mutual recursions
1304 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1305 subgroup is called recursively as a subpattern for a second time at the
1306 same position in the subject string. In previous releases this might have
1307 been caught by the recursion limit, or it might have run out of stack.
1309 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1310 happen only once. PCRE was, however incorrectly giving a compile time error
1311 "recursive call could loop indefinitely" because it cannot analyze the
1312 pattern in sufficient detail. The compile time test no longer happens when
1313 PCRE is compiling a conditional subpattern, but actual runaway loops are
1314 now caught at runtime (see 40 above).
1316 42. It seems that Perl allows any characters other than a closing parenthesis
1317 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1318 has been changed to be the same.
1320 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1321 as not to get warnings when autogen.sh is called. Also changed
1322 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1324 44. To help people who use pcregrep to scan files containing exceedingly long
1325 lines, the following changes have been made:
1327 (a) The default value of the buffer size parameter has been increased from
1328 8K to 20K. (The actual buffer used is three times this size.)
1330 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1333 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1334 to be set at run time.
1336 (d) Numerical values in pcregrep options can be followed by K or M, for
1337 example --buffer-size=50K.
1339 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1340 given and the return code is set to 2.
1342 45. Add a pointer to the latest mark to the callout data block.
1344 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1345 partial match of an empty string instead of no match. This was specific to
1348 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1349 complete match instead of a partial match. This bug was dependent on both
1350 the PCRE_UTF8 and PCRE_DOTALL options being set.
1352 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1353 starting byte set, because \b was not being ignored.
1356 Version 8.12 15-Jan-2011
1357 ------------------------
1359 1. Fixed some typos in the markup of the man pages, and wrote a script that
1360 checks for such things as part of the documentation building process.
1362 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1363 --match-limit and --recursion-limit options (added for 8.11). In
1364 particular, this made one of the standard tests fail. (The integer value
1365 went into the wrong half of a long int.)
1367 3. If the --colour option was given to pcregrep with -v (invert match), it
1368 did strange things, either producing crazy output, or crashing. It should,
1369 of course, ignore a request for colour when reporting lines that do not
1372 4. Another pcregrep bug caused similar problems if --colour was specified with
1373 -M (multiline) and the pattern match finished with a line ending.
1375 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1376 matched in multiline mode, the following line was shown as part of the
1377 match. This seems wrong, so I have changed it.
1379 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1380 the check for further matches in the same line (so they could be coloured)
1381 to overrun the end of the current line. If another match was found, it was
1382 incorrectly shown (and then shown again when found in the next line).
1384 7. If pcregrep was compiled under Windows, there was a reference to the
1385 function pcregrep_exit() before it was defined. I am assuming this was
1386 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1387 reported by a user. I've moved the definition above the reference.
1390 Version 8.11 10-Dec-2010
1391 ------------------------
1393 1. (*THEN) was not working properly if there were untried alternatives prior
1394 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1395 backtracked to try for "b" instead of moving to the next alternative branch
1396 at the same level (in this case, to look for "c"). The Perl documentation
1397 is clear that when (*THEN) is backtracked onto, it goes to the "next
1398 alternative in the innermost enclosing group".
1400 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1401 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1402 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1403 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1406 3. If \s appeared in a character class, it removed the VT character from
1407 the class, even if it had been included by some previous item, for example
1408 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1409 of \s, but is part of the POSIX "space" class.)
1411 4. A partial match never returns an empty string (because you can always
1412 match an empty string at the end of the subject); however the checking for
1413 an empty string was starting at the "start of match" point. This has been
1414 changed to the "earliest inspected character" point, because the returned
1415 data for a partial match starts at this character. This means that, for
1416 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1417 (previously it gave "no match").
1419 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1420 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1421 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1422 has an implication that the given string is incomplete (because a partial
1423 match is preferred over a full match). For this reason, these items now
1424 give a partial match in this situation. [Aside: previously, the one case
1425 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1426 match rather than a full match, which was wrong by the old rules, but is
1429 6. There was a bug in the handling of #-introduced comments, recognized when
1430 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1431 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1432 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1433 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1434 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1435 places in pcre_compile().
1437 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1438 comments when looking ahead for named forward references to subpatterns,
1439 the only newline sequence it recognized was NL. It now handles newlines
1440 according to the set newline convention.
1442 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1443 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1444 cater for a lack of strerror(). These oversights have been fixed.
1446 9. Added --match-limit and --recursion-limit to pcregrep.
1448 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1450 11. When the -o option was used, pcregrep was setting a return code of 1, even
1451 when matches were found, and --line-buffered was not being honoured.
1453 12. Added an optional parentheses number to the -o and --only-matching options
1456 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1457 can match an empty string. The code to do it in pcretest and pcredemo
1460 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1461 only one byte after an empty string match just before CRLF (this case
1462 just got forgotten; "any" and "anycrlf" were OK).
1464 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1465 mode when an empty string match preceded an ASCII character followed by
1466 a non-ASCII character. (The code for advancing by one character rather
1467 than one byte was nonsense.)
1469 (c) The pcredemo.c sample program did not have any code at all to handle
1470 the cases when CRLF is a valid newline sequence.
1472 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1473 as a starting offset was within the subject string. There is now a new
1474 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1475 negative or greater than the length of the string. In order to test this,
1476 pcretest is extended to allow the setting of negative starting offsets.
1478 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1479 starting offset points to the beginning of a UTF-8 character was
1480 unnecessarily clumsy. I tidied it up.
1482 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1483 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1485 17. Nobody had reported that the --include_dir option, which was added in
1486 release 7.7 should have been called --include-dir (hyphen, not underscore)
1487 for compatibility with GNU grep. I have changed it to --include-dir, but
1488 left --include_dir as an undocumented synonym, and the same for
1489 --exclude-dir, though that is not available in GNU grep, at least as of
1492 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1493 characters from a string of bytes) have been redefined so as not to use
1494 loops, in order to improve performance in some environments. At the same
1495 time, I abstracted some of the common code into auxiliary macros to save
1496 repetition (this should not affect the compiled code).
1498 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1499 compile-time error is now given if \c is not followed by an ASCII
1500 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1501 different, and any byte value is allowed.)
1503 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1504 START_OPTIMIZE option, which is now allowed at compile time - but just
1505 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1506 to pcregrep and other applications that have no direct access to PCRE
1507 options. The new /Y option in pcretest sets this option when calling
1510 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1511 back references. Groups containing recursive back references were forced to
1512 be atomic by that change, but in the case of named groups, the amount of
1513 memory required was incorrectly computed, leading to "Failed: internal
1514 error: code overflow". This has been fixed.
1516 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1517 pcretest.c, to avoid build problems in some Borland environments.
1520 Version 8.10 25-Jun-2010
1521 ------------------------
1523 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1526 2. (*ACCEPT) was not working when inside an atomic group.
1528 3. Inside a character class, \B is treated as a literal by default, but
1529 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1530 causes the error). The code is unchanged, but I tidied the documentation.
1532 4. Inside a character class, PCRE always treated \R and \X as literals,
1533 whereas Perl faults them if its -w option is set. I have changed PCRE so
1534 that it faults them when PCRE_EXTRA is set.
1536 5. Added support for \N, which always matches any character other than
1537 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1539 6. When compiling pcregrep with newer versions of gcc which may have
1540 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1541 declared with attribute warn_unused_result" were given. Just casting the
1542 result to (void) does not stop the warnings; a more elaborate fudge is
1543 needed. I've used a macro to implement this.
1545 7. Minor change to pcretest.c to avoid a compiler warning.
1547 8. Added four artifical Unicode properties to help with an option to make
1548 \s etc use properties (see next item). The new properties are: Xan
1549 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1551 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1552 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1553 this option. Modified pcretest to add /W to test this facility. Added
1554 REG_UCP to make it available via the POSIX interface.
1556 10. Added --line-buffered to pcregrep.
1558 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1559 studied, and the match started with a letter with a code point greater than
1560 127 whose first byte was different to the first byte of the other case of
1561 the letter, the other case of this starting letter was not recognized
1564 12. If a pattern that was studied started with a repeated Unicode property
1565 test, for example, \p{Nd}+, there was the theoretical possibility of
1566 setting up an incorrect bitmap of starting bytes, but fortunately it could
1567 not have actually happened in practice until change 8 above was made (it
1568 added property types that matched character-matching opcodes).
1570 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1571 possible starting bytes for non-anchored patterns.
1573 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1574 \R, and also a number of cases that involve Unicode properties, both
1575 explicit and implicit when PCRE_UCP is set.
1577 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1578 input, it could crash or give wrong results if characters with values
1579 greater than 0xc0 were present in the subject string. (Detail: it assumed
1580 UTF-8 input when processing these items.)
1582 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1583 size_t is 64-bit (#991).
1585 17. Added a check for running out of memory when PCRE is compiled with
1586 --disable-stack-for-recursion (#990).
1588 18. If the last data line in a file for pcretest does not have a newline on
1589 the end, a newline was missing in the output.
1591 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1592 less than 128) in its various bitmaps. However, there is a facility for
1593 generating tables according to the current locale when PCRE is compiled. It
1594 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1595 characters, are recognized by isspace() and therefore were getting set in
1596 these tables, and indeed these tables seem to approximate to ISO 8859. This
1597 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1598 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1599 which of course cannot start UTF-8 characters. I have changed the code so
1600 that only real ASCII characters (less than 128) and the correct starting
1601 bytes for UTF-8 encodings are set for characters greater than 127 when in
1602 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1605 20. Added the /T option to pcretest so as to be able to run tests with non-
1606 standard character tables, thus making it possible to include the tests
1607 used for 19 above in the standard set of tests.
1609 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1610 reference to a subpattern the other side of a comment that contains an
1611 opening parenthesis caused either an internal compiling error, or a
1612 reference to the wrong subpattern.
1615 Version 8.02 19-Mar-2010
1616 ------------------------
1618 1. The Unicode data tables have been updated to Unicode 5.2.0.
1620 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1623 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1624 original author of that file, following a query about its status.
1626 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1627 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1629 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1630 quantifier applied to a forward-referencing subroutine call, could compile
1631 incorrect code or give the error "internal error: previously-checked
1632 referenced subpattern not found".
1634 6. Both MS Visual Studio and Symbian OS have problems with initializing
1635 variables to point to external functions. For these systems, therefore,
1636 pcre_malloc etc. are now initialized to local functions that call the
1637 relevant global functions.
1639 7. There were two entries missing in the vectors called coptable and poptable
1640 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1641 I've fixed the data, and added a kludgy way of testing at compile time that
1642 the lengths are correct (equal to the number of opcodes).
1644 8. Following on from 7, I added a similar kludge to check the length of the
1645 eint vector in pcreposix.c.
1647 9. Error texts for pcre_compile() are held as one long string to avoid too
1648 much relocation at load time. To find a text, the string is searched,
1649 counting zeros. There was no check for running off the end of the string,
1650 which could happen if a new error number was added without updating the
1653 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1655 11. \K was not working if it appeared in an atomic group or in a group that
1656 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1657 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1658 the assertion is positive, but not if it is negative.
1660 12. Change 11 fortuitously reduced the size of the stack frame used in the
1661 "match()" function of pcre_exec.c by one pointer. Forthcoming
1662 implementation of support for (*MARK) will need an extra pointer on the
1663 stack; I have reserved it now, so that the stack frame size does not
1666 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1667 item in branch that calls a recursion is a subroutine call - as in the
1668 second branch in the above example - was incorrectly given the compile-
1669 time error "recursive call could loop indefinitely" because pcre_compile()
1670 was not correctly checking the subroutine for matching a non-empty string.
1672 14. The checks for overrunning compiling workspace could trigger after an
1673 overrun had occurred. This is a "should never occur" error, but it can be
1674 triggered by pathological patterns such as hundreds of nested parentheses.
1675 The checks now trigger 100 bytes before the end of the workspace.
1677 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1680 Version 8.01 19-Jan-2010
1681 ------------------------
1683 1. If a pattern contained a conditional subpattern with only one branch (in
1684 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1685 computed the wrong minimum data length (which is of course zero for such
1686 subpatterns). This could cause incorrect "no match" results.
1688 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1689 the pattern is reset in the first branch, pcre_compile() failed with
1690 "internal error: code overflow at offset...". This happened only when
1691 the reset was to the original external option setting. (An optimization
1692 abstracts leading options settings into an external setting, which was the
1695 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1696 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1697 assertion pattern did not match (meaning that the assertion was true), it
1698 was incorrectly treated as false if the SKIP had been reached during the
1699 matching. This also applied to assertions used as conditions.
1701 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1702 assertion subpattern, including such a pattern used as a condition,
1703 unpredictable results occurred, instead of the error return
1704 PCRE_ERROR_DFA_UITEM.
1706 5. The C++ GlobalReplace function was not working like Perl for the special
1707 situation when an empty string is matched. It now does the fancy magic
1708 stuff that is necessary.
1710 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1711 removed. (These were left over from very, very early versions of PCRE.)
1713 7. Some cosmetic changes to the code to make life easier when compiling it
1714 as part of something else:
1716 (a) Change DEBUG to PCRE_DEBUG.
1718 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1719 called "current" as "current_branch", to prevent a collision with the
1720 Linux macro when compiled as a kernel module.
1722 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1723 prevent a collision with the Linux macro when compiled as a kernel
1726 8. In pcre_compile() there are some checks for integer overflows that used to
1727 cast potentially large values to (double). This has been changed to that
1728 when building, a check for int64_t is made, and if it is found, it is used
1729 instead, thus avoiding the use of floating point arithmetic. (There is no
1730 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1733 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1734 2005 (difference between two addresses compared to an unsigned value).
1736 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1737 custom one, because of the following reported problem in Windows:
1739 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1741 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1742 therefore missing the function definition.
1743 - The compiler thus generates a "C" signature for the test function.
1744 - The linker fails to find the "C" function.
1745 - PCRE fails to configure if asked to do so against libbz2.
1747 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1748 messages were output:
1750 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1751 rerunning libtoolize, to keep the correct libtool macros in-tree.
1752 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1754 I have done both of these things.
1756 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1757 most of the time, it *can* run out if it is given a pattern that contains a
1758 runaway infinite recursion. I updated the discussion in the pcrestack man
1761 13. Now that we have gone to the x.xx style of version numbers, the minor
1762 version may start with zero. Using 08 or 09 is a bad idea because users
1763 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1764 interpreted as invalid octal numbers. I've updated the previous comment in
1765 configure.ac, and also added a check that gives an error if 08 or 09 are
1768 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1769 causing partial matching to fail when the end of the subject matched \W
1770 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1772 15. There were some discrepancies between the declarations in pcre_internal.h
1773 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1774 their definitions. The declarations used "const uschar *" and the
1775 definitions used USPTR. Even though USPTR is normally defined as "const
1776 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1777 reported that: "This difference in casting confuses some C++ compilers, for
1778 example, SunCC recognizes above declarations as different functions and
1779 generates broken code for hbpcre." I have changed the declarations to use
1782 16. GNU libtool is named differently on some systems. The autogen.sh script now
1783 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1786 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1787 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1788 comment: "Figure out how to create a longlong from a string: strtoll and
1789 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1790 instance, but it only takes 2 args instead of 3!"
1792 18. A subtle bug concerned with back references has been fixed by a change of
1793 specification, with a corresponding code fix. A pattern such as
1794 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1795 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1796 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1797 same bug. Such groups have to be quantified to be useful, or contained
1798 inside another quantified group. (If there's no repetition, the reference
1799 can never match.) The problem arises because, having left the group and
1800 moved on to the rest of the pattern, a later failure that backtracks into
1801 the group uses the captured value from the final iteration of the group
1802 rather than the correct earlier one. I have fixed this in PCRE by forcing
1803 any group that contains a reference to itself to be an atomic group; that
1804 is, there cannot be any backtracking into it once it has completed. This is
1805 similar to recursive and subroutine calls.
1808 Version 8.00 19-Oct-09
1809 ----------------------
1811 1. The table for translating pcre_compile() error codes into POSIX error codes
1812 was out-of-date, and there was no check on the pcre_compile() error code
1813 being within the table. This could lead to an OK return being given in
1816 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1817 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1818 in a Windows environment.
1820 3. The pcregrep --count option prints the count for each file even when it is
1821 zero, as does GNU grep. However, pcregrep was also printing all files when
1822 --files-with-matches was added. Now, when both options are given, it prints
1823 counts only for those files that have at least one match. (GNU grep just
1824 prints the file name in this circumstance, but including the count seems
1825 more useful - otherwise, why use --count?) Also ensured that the
1826 combination -clh just lists non-zero counts, with no names.
1828 4. The long form of the pcregrep -F option was incorrectly implemented as
1829 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1830 but it seems right to fix it, and I didn't think it was worth preserving
1833 5. The command line items --regex=pattern and --regexp=pattern were not
1834 recognized by pcregrep, which required --regex pattern or --regexp pattern
1835 (with a space rather than an '='). The man page documented the '=' forms,
1836 which are compatible with GNU grep; these now work.
1838 6. No libpcreposix.pc file was created for pkg-config; there was just
1839 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1841 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1842 when UCP support is not needed, by modifying the Python script that
1843 generates it from Unicode data files. This should not matter if the module
1844 is correctly used as a library, but I received one complaint about 50K of
1845 unwanted data. My guess is that the person linked everything into his
1846 program rather than using a library. Anyway, it does no harm.
1848 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1849 was a minimum greater than 1 for a wide character in a possessive
1850 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1851 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1852 character. Chaos in the form of incorrect output or a compiling loop could
1855 9. The restrictions on what a pattern can contain when partial matching is
1856 requested for pcre_exec() have been removed. All patterns can now be
1857 partially matched by this function. In addition, if there are at least two
1858 slots in the offset vector, the offset of the earliest inspected character
1859 for the match and the offset of the end of the subject are set in them when
1860 PCRE_ERROR_PARTIAL is returned.
1862 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1863 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1864 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1865 and may be more useful for multi-segment matching.
1867 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1868 used to be given if ever the end of the subject was reached; now it is
1869 given only if matching could not proceed because another character was
1870 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1871 string "Z", which now yields "no match" instead of "partial match". In the
1872 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1873 final character ended with (*FAIL).
1875 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1876 if the pattern had a "must contain" character that was already found in the
1877 earlier partial match, unless partial matching was again requested. For
1878 example, with the pattern /dog.(body)?/, the "must contain" character is
1879 "g". If the first part-match was for the string "dog", restarting with
1880 "sbody" failed. This bug has been fixed.
1882 13. The string returned by pcre_dfa_exec() after a partial match has been
1883 changed so that it starts at the first inspected character rather than the
1884 first character of the match. This makes a difference only if the pattern
1885 starts with a lookbehind assertion or \b or \B (\K is not supported by
1886 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1887 matching functions compatible, and I think it's the right thing to do.
1889 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1890 so that the demonstration program is easily available in environments where
1891 PCRE has not been installed from source.
1893 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1894 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1897 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1898 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1899 is not the first non-POSIX option to be added. Clearly some people find
1900 these options useful.
1902 17. If a caller to the POSIX matching function regexec() passes a non-zero
1903 value for nmatch with a NULL value for pmatch, the value of
1904 nmatch is forced to zero.
1906 18. RunGrepTest did not have a test for the availability of the -u option of
1907 the diff command, as RunTest does. It now checks in the same way as
1908 RunTest, and also checks for the -b option.
1910 19. If an odd number of negated classes containing just a single character
1911 interposed, within parentheses, between a forward reference to a named
1912 subpattern and the definition of the subpattern, compilation crashed with
1913 an internal error, complaining that it could not find the referenced
1914 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1915 [The bug was that it was starting one character too far in when skipping
1916 over the character class, thus treating the ] as data rather than
1917 terminating the class. This meant it could skip too much.]
1919 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1920 /g option in pcretest when the pattern contains \K, which makes it possible
1921 to have an empty string match not at the start, even when the pattern is
1922 anchored. Updated pcretest and pcredemo to use this option.
1924 21. If the maximum number of capturing subpatterns in a recursion was greater
1925 than the maximum at the outer level, the higher number was returned, but
1926 with unset values at the outer level. The correct (outer level) value is
1929 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1930 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1931 make it do so. The string so far is captured, making this feature
1932 compatible with Perl.
1934 23. The tests have been re-organized, adding tests 11 and 12, to make it
1935 possible to check the Perl 5.10 features against Perl 5.10.
1937 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1938 pattern matches a fixed length string. PCRE did not allow this; now it
1939 does. Neither allows recursion.
1941 25. I finally figured out how to implement a request to provide the minimum
1942 length of subject string that was needed in order to match a given pattern.
1943 (It was back references and recursion that I had previously got hung up
1944 on.) This code has now been added to pcre_study(); it finds a lower bound
1945 to the length of subject needed. It is not necessarily the greatest lower
1946 bound, but using it to avoid searching strings that are too short does give
1947 some useful speed-ups. The value is available to calling programs via
1950 26. While implementing 25, I discovered to my embarrassment that pcretest had
1951 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1952 study optimizations had never been tested with that matching function.
1953 Oops. What is worse, even when it was passed study data, there was a bug in
1954 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1955 were also very few tests of studied patterns with pcre_dfa_exec().
1957 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1958 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1959 on the other side of the coin, they are no longer allowed to have different
1960 names, because these cannot be distinguished in PCRE, and this has caused
1961 confusion. (This is a difference from Perl.)
1963 28. When duplicate subpattern names are present (necessarily with different
1964 numbers, as required by 27 above), and a test is made by name in a
1965 conditional pattern, either for a subpattern having been matched, or for
1966 recursion in such a pattern, all the associated numbered subpatterns are
1967 tested, and the overall condition is true if the condition is true for any
1968 one of them. This is the way Perl works, and is also more like the way
1969 testing by number works.
1972 Version 7.9 11-Apr-09
1973 ---------------------
1975 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1976 (pcretest), all targets were linked against these libraries. This included
1977 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1978 libraries. This caused unwanted dependencies to be created. This problem
1979 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1980 pcretest is linked with readline.
1982 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1983 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1984 moved outside it again, because FALSE and TRUE are already defined in AIX,
1987 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1988 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1990 4. The pcregrep documentation said spaces were inserted as well as colons (or
1991 hyphens) following file names and line numbers when outputting matching
1992 lines. This is not true; no spaces are inserted. I have also clarified the
1993 wording for the --colour (or --color) option.
1995 5. In pcregrep, when --colour was used with -o, the list of matching strings
1996 was not coloured; this is different to GNU grep, so I have changed it to be
1999 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2000 each matching line was coloured. Now it goes on to look for further matches
2001 of any of the test patterns, which is the same behaviour as GNU grep.
2003 7. A pattern that could match an empty string could cause pcregrep to loop; it
2004 doesn't make sense to accept an empty string match in pcregrep, so I have
2005 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2006 seems to be how GNU grep behaves. [But see later change 40 for release
2009 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2010 start or after a newline", because the conditional assertion was not being
2011 correctly handled. The rule now is that both the assertion and what follows
2012 in the first alternative must satisfy the test.
2014 9. If auto-callout was enabled in a pattern with a conditional group whose
2015 condition was an assertion, PCRE could crash during matching, both with
2016 pcre_exec() and pcre_dfa_exec().
2018 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2021 11. Unicode property support in character classes was not working for
2022 characters (bytes) greater than 127 when not in UTF-8 mode.
2024 12. Added the -M command line option to pcretest.
2026 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2028 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2030 16. Added comments and documentation about mis-use of no_arg in the C++
2033 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2034 from Martin Jerabek that uses macro names for all relevant character and
2037 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2038 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2039 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2040 these, but not everybody uses configure.
2042 19. A conditional group that had only one branch was not being correctly
2043 recognized as an item that could match an empty string. This meant that an
2044 enclosing group might also not be so recognized, causing infinite looping
2045 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2046 with the subject "ab", where knowledge that the repeated group can match
2047 nothing is needed in order to break the loop.
2049 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2050 exec(), but without supplying a callout function, matching went wrong.
2052 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2053 leak if the size of the offset vector was greater than 30. When the vector
2054 is smaller, the saved offsets during recursion go onto a local stack
2055 vector, but for larger vectors malloc() is used. It was failing to free
2056 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2059 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2060 heapframe that is used only when UTF-8 support is enabled. This caused no
2061 problem, but was untidy.
2063 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2064 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2065 included within another project.
2067 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2068 slightly modified by me:
2070 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2071 not building pcregrep.
2073 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2074 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2076 25. Forward references, both numeric and by name, in patterns that made use of
2077 duplicate group numbers, could behave incorrectly or give incorrect errors,
2078 because when scanning forward to find the reference group, PCRE was not
2079 taking into account the duplicate group numbers. A pattern such as
2080 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2082 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2083 the feature of a custom pointer more persuasive (as requested by a user).
2085 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2086 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2089 28. Added support for (*UTF8) at the start of a pattern.
2091 29. Arrange for flags added by the "release type" setting in CMake to be shown
2092 in the configuration summary.
2095 Version 7.8 05-Sep-08
2096 ---------------------
2098 1. Replaced UCP searching code with optimized version as implemented for Ad
2099 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2100 stage table and inline lookup instead of a function, giving speed ups of 2
2101 to 5 times on some simple patterns that I tested. Permission was given to
2102 distribute the MultiStage2.py script that generates the tables (it's not in
2103 the tarball, but is in the Subversion repository).
2105 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2108 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2109 a group with a zero qualifier. The result of the study could be incorrect,
2110 or the function might crash, depending on the pattern.
2112 4. Caseless matching was not working for non-ASCII characters in back
2113 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2114 It now works when Unicode Property Support is available.
2116 5. In pcretest, an escape such as \x{de} in the data was always generating
2117 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2118 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2121 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2123 7. Added two (int) casts to pcregrep when printing the difference of two
2124 pointers, in case they are 64-bit values.
2126 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2129 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2130 and a #define of that name to empty if it is not externally set. This is to
2131 allow users of MSVC to set it if necessary.
2133 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2134 the convenience functions in the pcre_get.c source file.
2136 11. An option change at the start of a pattern that had top-level alternatives
2137 could cause overwriting and/or a crash. This command provoked a crash in
2140 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2142 This potential security problem was recorded as CVE-2008-2371.
2144 12. For a pattern where the match had to start at the beginning or immediately
2145 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2146 pcre_dfa_exec() could read past the end of the passed subject if there was
2147 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2148 pcretest so that it places the subject at the end of its malloc-ed buffer.
2150 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2151 exec() might read past the end of the data buffer in UTF-8 mode.
2153 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2154 the data contained the byte 0x85 as part of a UTF-8 character within its
2155 first line. This applied both to normal and DFA matching.
2157 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2158 /^[^d]*?$/8 failed to match "abc".
2160 16. Added a missing copyright notice to pcrecpp_internal.h.
2162 17. Make it more clear in the documentation that values returned from
2163 pcre_exec() in ovector are byte offsets, not character counts.
2165 18. Tidied a few places to stop certain compilers from issuing warnings.
2167 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2168 supplied by Stefan Weber. I made a further small update for 7.8 because
2169 there is a change of source arrangements: the pcre_searchfuncs.c module is
2170 replaced by pcre_ucd.c.
2173 Version 7.7 07-May-08
2174 ---------------------
2176 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2177 a string to a long long, pretend we don't even have a long long." This is
2178 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2180 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2181 pre-7.6 versions, which defined a global no_arg variable instead of putting
2182 it in the RE class. (See also #8 below.)
2184 3. Remove a line of dead code, identified by coverity and reported by Nuno
2187 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2189 (1) The include/exclude patterns were being applied to the whole pathnames
2190 of files, instead of just to the final components.
2192 (2) If there was more than one level of directory, the subdirectories were
2193 skipped unless they satisfied the include/exclude conditions. This is
2194 inconsistent with GNU grep (and could even be seen as contrary to the
2195 pcregrep specification - which I improved to make it absolutely clear).
2196 The action now is always to scan all levels of directory, and just
2197 apply the include/exclude patterns to regular files.
2199 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2200 --exclude_dir in the tests to avoid scanning .svn directories.
2202 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2203 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2204 doesn't support NULs in patterns.
2206 7. Added some missing "const"s to declarations of static tables in
2207 pcre_compile.c and pcre_dfa_exec.c.
2209 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2210 caused by fix #2 above. (Subsequently also a second patch to fix the
2211 first patch. And a third patch - this was a messy problem.)
2213 9. Applied Craig's patch to remove the use of push_back().
2215 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2216 matching function regexec().
2218 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2219 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2220 references. PCRE supports relative numbers with this syntax (I don't think
2223 12. Previously, a group with a zero repeat such as (...){0} was completely
2224 omitted from the compiled regex. However, this means that if the group
2225 was called as a subroutine from elsewhere in the pattern, things went wrong
2226 (an internal error was given). Such groups are now left in the compiled
2227 pattern, with a new opcode that causes them to be skipped at execution
2230 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2231 to the way PCRE behaves:
2233 (a) A lone ] character is dis-allowed (Perl treats it as data).
2235 (b) A back reference to an unmatched subpattern matches an empty string
2236 (Perl fails the current match path).
2238 (c) A data ] in a character class must be notated as \] because if the
2239 first data character in a class is ], it defines an empty class. (In
2240 Perl it is not possible to have an empty class.) The empty class []
2241 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2242 The negative empty class [^] matches any one character, independently
2243 of the DOTALL setting.
2245 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2246 non-existent subpattern following a character class starting with ']' and
2247 containing () gave an internal compiling error instead of "reference to
2248 non-existent subpattern". Fortunately, when the pattern did exist, the
2249 compiled code was correct. (When scanning forwards to check for the
2250 existence of the subpattern, it was treating the data ']' as terminating
2251 the class, so got the count wrong. When actually compiling, the reference
2252 was subsequently set up correctly.)
2254 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2255 it was being rejected as not supported by pcre_dfa_exec(), even though
2256 other assertions are supported. I have made pcre_dfa_exec() support
2259 16. The implementation of 13c above involved the invention of a new opcode,
2260 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2261 cannot be changed at match time, I realized I could make a small
2262 improvement to matching performance by compiling OP_ALLANY instead of
2263 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2266 17. Compiling pcretest on Windows with readline support failed without the
2267 following two fixes: (1) Make the unistd.h include conditional on
2268 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2270 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2271 ncurses library to be included for pcretest when ReadLine support is
2272 requested, but also to allow for it to be overridden. This patch came from
2275 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2276 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2277 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2281 Version 7.6 28-Jan-08
2282 ---------------------
2284 1. A character class containing a very large number of characters with
2285 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2288 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2289 HAVE_LONG_LONG is not defined.
2291 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2292 bring them up to date and include new features. This patch includes:
2294 - Fixed PH's badly added libz and libbz2 support.
2295 - Fixed a problem with static linking.
2296 - Added pcredemo. [But later removed - see 7 below.]
2297 - Fixed dftables problem and added an option.
2298 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2300 - Added readline support for pcretest.
2301 - Added an listing of the option settings after cmake has run.
2303 4. A user submitted a patch to Makefile that makes it easy to create
2304 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2305 Makefile.am that cause it to include this special target, without
2306 affecting anything else. Note that the same mingw target plus all
2307 the other distribution libraries and programs are now supported
2308 when configuring with CMake (see 6 below) instead of with
2311 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2312 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2313 exported in the Windows port". It has not yet been confirmed that the patch
2314 solves the problem, but it does no harm.
2316 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2317 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2318 with CMake, and also correct the comment about stack recursion.
2320 7. Remove the automatic building of pcredemo from the ./configure system and
2321 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2322 of a program that users should build themselves after PCRE is installed, so
2323 building it automatically is not really right. What is more, it gave
2324 trouble in some build environments.
2326 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2329 Version 7.5 10-Jan-08
2330 ---------------------
2332 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2333 values in parens when parsing an RE using the C++ wrapper."
2335 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2336 Characters greater than 255 were excluded from the class instead of being
2339 3. The same bug as (2) above applied to negated POSIX classes such as
2342 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2343 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2346 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2347 first named subpattern) and a construct such as (?&a) would reference the
2348 first named subpattern whose name started with "a" (in other words, the
2349 length check was missing). Both these problems are fixed. "Subpattern name
2350 expected" is now given for (?&) (a zero-length name), and this patch also
2351 makes it give the same error for \k'' (previously it complained that that
2352 was a reference to a non-existent subpattern).
2354 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2355 this is right because (?- can be followed by option settings as well as by
2356 digits. I have, however, made the messages clearer.
2358 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2359 than the number used in the conditional) now cause a compile-time error.
2360 This is actually not compatible with Perl, which accepts such patterns, but
2361 treats the conditional as always being FALSE (as PCRE used to), but it
2362 seems to me that giving a diagnostic is better.
2364 8. Change "alphameric" to the more common word "alphanumeric" in comments
2367 9. Fix two occurrences of "backslash" in comments that should have been
2370 10. Remove two redundant lines of code that can never be obeyed (their function
2371 was moved elsewhere).
2373 11. The program that makes PCRE's Unicode character property table had a bug
2374 which caused it to generate incorrect table entries for sequences of
2375 characters that have the same character type, but are in different scripts.
2376 It amalgamated them into a single range, with the script of the first of
2377 them. In other words, some characters were in the wrong script. There were
2378 thirteen such cases, affecting characters in the following ranges:
2394 12. The -o option (show only the matching part of a line) for pcregrep was not
2395 compatible with GNU grep in that, if there was more than one match in a
2396 line, it showed only the first of them. It now behaves in the same way as
2399 13. If the -o and -v options were combined for pcregrep, it printed a blank
2400 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2401 does the same. The return code can be used to tell if there were any
2404 14. Added --file-offsets and --line-offsets to pcregrep.
2406 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2407 infinitely looping recursion. The bug was that positive lookaheads were not
2408 being skipped when checking for a possible empty match (negative lookaheads
2409 and both kinds of lookbehind were skipped).
2411 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2412 inclusion of <windows.h> to before rather than after the definition of
2413 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2415 17. Specifying a possessive quantifier with a specific limit for a Unicode
2416 character property caused pcre_compile() to compile bad code, which led at
2417 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2418 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2419 caused the error; without that there was no problem.
2421 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2423 19. Added --enable-pcretest-libreadline.
2425 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2426 RE::GlobalReplace(). As a result, the number of replacements returned was
2427 double what it should be. I removed one of the increments, but Craig sent a
2428 later patch that removed the other one (the right fix) and added unit tests
2429 that check the return values (which was not done before).
2431 21. Several CMake things:
2433 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2434 the names libpcre and libpcreposix, not just pcre and pcreposix.
2436 (2) The above change means that pcretest and pcregrep are now correctly
2437 linked with the newly-built libraries, not previously installed ones.
2439 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2441 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2442 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2443 UTF-8 newline character). The key issue is that the pattern starts .*;
2444 this means that the match must be either at the beginning, or after a
2445 newline. The bug was in the code for advancing after a failed match and
2446 checking that the new position followed a newline. It was not taking
2447 account of UTF-8 characters correctly.
2449 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2450 character classes. PCRE was not treating the sequence [:...:] as a
2451 character class unless the ... were all letters. Perl, however, seems to
2452 allow any characters between [: and :], though of course it rejects as
2453 unknown any "names" that contain non-letters, because all the known class
2454 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2455 for example, whereas PCRE did not - it did not recognize a POSIX character
2456 class. This seemed a bit dangerous, so the code has been changed to be
2457 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2458 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2459 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2460 Perl does, and where it didn't before.
2462 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2463 Windows environments %n is disabled by default.
2466 Version 7.4 21-Sep-07
2467 ---------------------
2469 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2470 means that a class such as [\s] counted as "explicit reference to CR or
2471 LF". That isn't really right - the whole point of the change was to try to
2472 help when there was an actual mention of one of the two characters. So now
2473 the change happens only if \r or \n (or a literal CR or LF) character is
2476 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2477 of both had grown to the point where there were only 3 bits left.
2478 Fortunately, there was spare space in the data structure, and so I have
2479 moved the internal flags into a new 16-bit field to free up more option
2482 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2483 but did not set the internal JCHANGED flag - either of these is enough to
2484 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2485 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2486 start sets both bits.
2488 4. Added options (at build time, compile time, exec time) to change \R from
2489 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2491 5. doc/pcresyntax.html was missing from the distribution.
2493 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2494 compatibility, even though it is no longer used.
2496 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2497 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2498 windows.h file is present (where different names are used). [This was
2499 reversed later after testing - see 16 below.]
2501 8. Changed all #include <config.h> to #include "config.h". There were also
2502 some further <pcre.h> cases that I changed to "pcre.h".
2504 9. When pcregrep was used with the --colour option, it missed the line ending
2505 sequence off the lines that it output.
2507 10. It was pointed out to me that arrays of string pointers cause lots of
2508 relocations when a shared library is dynamically loaded. A technique of
2509 using a single long string with a table of offsets can drastically reduce
2510 these. I have refactored PCRE in four places to do this. The result is
2514 After changing UCP table: 187
2515 After changing error message table: 43
2516 After changing table of "verbs" 36
2517 After changing table of Posix names 22
2519 Thanks to the folks working on Gregex for glib for this insight.
2521 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2522 unicode-properties was also set.
2524 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2526 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2527 checked only for CRLF.
2529 14. Added casts to pcretest.c to avoid compiler warnings.
2531 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2533 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2534 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2535 entirely. This removes changes made in 7 above.
2537 17. The CMake files have been updated, and there is now more information about
2538 building with CMake in the NON-UNIX-USE document.
2541 Version 7.3 28-Aug-07
2542 ---------------------
2544 1. In the rejigging of the build system that eventually resulted in 7.1, the
2545 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2546 brackets there is not right, since it causes compilers to look for an
2547 installed pcre.h, not the version that is in the source that is being
2548 compiled (which of course may be different). I have changed it back to:
2552 I have a vague recollection that the change was concerned with compiling in
2553 different directories, but in the new build system, that is taken care of
2554 by the VPATH setting the Makefile.
2556 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2557 when the subject happened to end in the byte 0x85 (e.g. if the last
2558 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2559 characters but of course it shouldn't be taken as a newline when it is part
2560 of another character. The bug was that, for an unlimited repeat of . in
2561 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2562 characters when looking for a newline.
2564 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2566 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2569 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2570 long printing in the pcrecpp unittest when running under MinGW.
2572 6. ESC_K was left out of the EBCDIC table.
2574 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2575 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2576 limit also applies to "virtual nesting" when a pattern is recursive, and in
2577 this case 1000 isn't so big. I have been able to remove this limit at the
2578 expense of backing off one optimization in certain circumstances. Normally,
2579 when pcre_exec() would call its internal match() function recursively and
2580 immediately return the result unconditionally, it uses a "tail recursion"
2581 feature to save stack. However, when a subpattern that can match an empty
2582 string has an unlimited repetition quantifier, it no longer makes this
2583 optimization. That gives it a stack frame in which to save the data for
2584 checking that an empty string has been matched. Previously this was taken
2585 from the 1000-entry workspace that had been reserved. So now there is no
2586 explicit limit, but more stack is used.
2588 8. Applied Daniel's patches to solve problems with the import/export magic
2589 syntax that is required for Windows, and which was going wrong for the
2590 pcreposix and pcrecpp parts of the library. These were overlooked when this
2591 problem was solved for the main library.
2593 9. There were some crude static tests to avoid integer overflow when computing
2594 the size of patterns that contain repeated groups with explicit upper
2595 limits. As the maximum quantifier is 65535, the maximum group length was
2596 set at 30,000 so that the product of these two numbers did not overflow a
2597 32-bit integer. However, it turns out that people want to use groups that
2598 are longer than 30,000 bytes (though not repeat them that many times).
2599 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2600 made it possible to implement the integer overflow checks in a much more
2601 dynamic way, which I have now done. The artificial limitation on group
2602 length has been removed - we now have only the limit on the total length of
2603 the compiled pattern, which depends on the LINK_SIZE setting.
2605 10. Fixed a bug in the documentation for get/copy named substring when
2606 duplicate names are permitted. If none of the named substrings are set, the
2607 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2610 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2611 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2612 because the ] is interpreted as the first data character and the
2613 terminating ] is not found. PCRE has been made compatible with Perl in this
2614 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2615 cause memory overwriting.
2617 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2618 string has been matched (to stop an infinite loop). It was not recognizing
2619 a conditional subpattern that could match an empty string if that
2620 subpattern was within another subpattern. For example, it looped when
2621 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2622 condition was not nested. This bug has been fixed.
2624 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2625 past the start of the subject in the presence of bytes with the top bit
2626 set, for example "\x8aBCD".
2628 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2629 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2631 14. Optimized (?!) to (*FAIL).
2633 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2634 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2635 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2636 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2637 does: it's just the validity check that is more restrictive.
2639 16. Inserted checks for integer overflows during escape sequence (backslash)
2640 processing, and also fixed erroneous offset values for syntax errors during
2641 backslash processing.
2643 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2644 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2646 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2649 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2650 something other than just ASCII characters) inside a group that had an
2651 unlimited repeat caused a loop at compile time (while checking to see
2652 whether the group could match an empty string).
2654 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2655 [\P{Any}] did so. (Error in the code for printing property names.)
2657 21. An orphan \E inside a character class could cause a crash.
2659 22. A repeated capturing bracket such as (A)? could cause a wild memory
2660 reference during compilation.
2662 23. There are several functions in pcre_compile() that scan along a compiled
2663 expression for various reasons (e.g. to see if it's fixed length for look
2664 behind). There were bugs in these functions when a repeated \p or \P was
2665 present in the pattern. These operators have additional parameters compared
2666 with \d, etc, and these were not being taken into account when moving along
2667 the compiled data. Specifically:
2669 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2672 (b) An item such as \pL+ within a repeated group could cause crashes or
2675 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2676 "reference to non-existent subpattern" error.
2678 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2680 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2681 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2683 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2685 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2686 character were causing crashes (broken optimization).
2688 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2689 \p or \P) caused a compile-time loop.
2691 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2692 break. For example, the unstudied pattern [\r\n]A does not match the string
2693 "\r\nA" because change 7.0/46 below moves the current point on by two
2694 characters after failing to match at the start. However, the pattern \nA
2695 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2696 the same is true. There doesn't seem any very clean way out of this, but
2697 what I have chosen to do makes the common cases work: PCRE now takes note
2698 of whether there can be an explicit match for \r or \n anywhere in the
2699 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2700 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2701 pattern has explicit CR or LF references.
2703 29. Added (*CR) etc for changing newline setting at start of pattern.
2706 Version 7.2 19-Jun-07
2707 ---------------------
2709 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2710 which is apparently normally available under Windows.
2712 2. Re-jig the pcregrep tests with different newline settings in an attempt
2713 to make them independent of the local environment's newline setting.
2715 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2717 4. Some of the "internals" tests were previously cut out when the link size
2718 was not 2, because the output contained actual offsets. The recent new
2719 "Z" feature of pcretest means that these can be cut out, making the tests
2720 usable with all link sizes.
2722 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2723 stack recursion. This gives a massive performance boost under BSD, but just
2724 a small improvement under Linux. However, it saves one field in the frame
2727 6. Added more features from the forthcoming Perl 5.10:
2729 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2730 recursion call. It refers to the nth most recently opened parentheses.
2732 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2733 to be opened parentheses.
2735 (c) Conditions that refer to capturing parentheses can be specified
2736 relatively, for example, (?(-2)... or (?(+3)...
2738 (d) \K resets the start of the current match so that everything before
2741 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2743 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2746 (g) (?| introduces a group in which the numbering of parentheses in each
2747 alternative starts with the same number.
2749 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2751 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2754 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2755 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2756 for detecting groups that can match an empty string.
2758 9. A pattern with a very large number of alternatives (more than several
2759 hundred) was running out of internal workspace during the pre-compile
2760 phase, where pcre_compile() figures out how much memory will be needed. A
2761 bit of new cunning has reduced the workspace needed for groups with
2762 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2763 workspace instead of running out of the 4096 that are available.
2765 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2767 11. Applied patch from Google to remove an optimization that didn't quite work.
2768 The report of the bug said:
2770 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2771 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2772 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2774 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2775 it matched the wrong number of bytes.
2778 Version 7.1 24-Apr-07
2779 ---------------------
2781 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2782 that is more "standard", making use of automake and other Autotools. There
2783 is some re-arrangement of the files and adjustment of comments consequent
2786 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2787 for recursive directory scanning broke on some systems because the files
2788 are not scanned in any specific order and on different systems the order
2789 was different. A call to "sort" has been inserted into RunGrepTest for the
2790 approprate test as a short-term fix. In the longer term there may be an
2793 3. I had an email from Eric Raymond about problems translating some of PCRE's
2794 man pages to HTML (despite the fact that I distribute HTML pages, some
2795 people do their own conversions for various reasons). The problems
2796 concerned the use of low-level troff macros .br and .in. I have therefore
2797 removed all such uses from the man pages (some were redundant, some could
2798 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2799 HTML has been updated to handle .nf/.fi and to complain if it encounters
2802 4. Updated comments in configure.ac that get placed in config.h.in and also
2803 arranged for config.h to be included in the distribution, with the name
2804 config.h.generic, for the benefit of those who have to compile without
2805 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2807 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2808 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2809 makevp.bat for the current PCRE, using the additional files
2810 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2812 6. A Windows user reported a minor discrepancy with test 2, which turned out
2813 to be caused by a trailing space on an input line that had got lost in his
2814 copy. The trailing space was an accident, so I've just removed it.
2816 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2819 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2820 as "const" (a) because they are and (b) because it helps the PHP
2821 maintainers who have recently made a script to detect big data structures
2822 in the php code that should be moved to the .rodata section. I remembered
2823 to update Builducptable as well, so it won't revert if ucptable.h is ever
2826 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2827 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2828 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2829 support is not required. This saves 1.5-2K of code, which is important in
2832 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2833 so as not to refer to the tables, even though these functions will never be
2834 called when UTF-8 support is disabled. Otherwise there are problems with a
2837 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2839 (a) It was defining its arguments as char * instead of void *.
2841 (b) It was assuming that all moves were upwards in memory; this was true
2842 a long time ago when I wrote it, but is no longer the case.
2844 The emulated memove() is provided for those environments that have neither
2845 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2846 is clearly not the case, as these two bugs were recently reported.
2848 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2849 and Detrail to create the HTML documentation, the .txt form of the man
2850 pages, and it removes trailing spaces from listed files. It also creates
2851 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2852 case, it wraps all the #defines with #ifndefs. This script should be run
2855 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2856 with Unicode property support.
2858 (a) For a maximizing quantifier, if the two different cases of the
2859 character were of different lengths in their UTF-8 codings (there are
2860 some cases like this - I found 11), and the matching function had to
2861 back up over a mixture of the two cases, it incorrectly assumed they
2862 were both the same length.
2864 (b) When PCRE was configured to use the heap rather than the stack for
2865 recursion during matching, it was not correctly preserving the data for
2866 the other case of a UTF-8 character when checking ahead for a match
2867 while processing a minimizing repeat. If the check also involved
2868 matching a wide character, but failed, corruption could cause an
2869 erroneous result when trying to check for a repeat of the original
2872 13. Some tidying changes to the testing mechanism:
2874 (a) The RunTest script now detects the internal link size and whether there
2875 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2876 values substituted by "configure". (The RunGrepTest script already did
2877 this for UTF-8.) The configure.ac script no longer substitutes the
2880 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2881 with length and offset values. This means that the output is different
2882 for different internal link sizes. Test 2 is skipped for link sizes
2883 other than 2 because of this, bypassing the problem. Unfortunately,
2884 there was also a test in test 3 (the locale tests) that used /B and
2885 failed for link sizes other than 2. Rather than cut the whole test out,
2886 I have added a new /Z option to pcretest that replaces the length and
2887 offset values with spaces. This is now used to make test 3 independent
2888 of link size. (Test 2 will be tidied up later.)
2890 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2891 segmentation fault instead of returning the appropriate error message.
2893 15. In multiline mode when the newline sequence was set to "any", the pattern
2894 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2895 This doesn't seem right; it now treats the CRLF combination as the line
2896 ending, and so does not match in that case. It's only a pattern such as ^$
2897 that would hit this one: something like ^ABC$ would have failed after \r
2898 and then tried again after \r\n.
2900 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2901 in an attempt to make files that differ only in their line terminators
2902 compare equal. This works on Linux.
2904 17. Under certain error circumstances pcregrep might try to free random memory
2905 as it exited. This is now fixed, thanks to valgrind.
2907 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2908 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2909 was because its rules for how to advance for /g after matching an empty
2910 string at the end of a line did not allow for this case. They now check for
2913 20. pcretest is supposed to handle patterns and data of any length, by
2914 extending its buffers when necessary. It was getting this wrong when the
2915 buffer for a data line had to be extended.
2917 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2918 CRLF as a newline sequence.
2920 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2921 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2922 I have nevertheless tidied it up.
2924 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2926 24. Added a man page for pcre-config.
2929 Version 7.0 19-Dec-06
2930 ---------------------
2932 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2933 moving to gcc 4.1.1.
2935 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2936 sys/time.h, which is documented as needed for this function. It doesn't
2937 seem to matter on Linux, but it showed up on some releases of OS X.
2939 3. It seems that there are systems where bytes whose values are greater than
2940 127 match isprint() in the "C" locale. The "C" locale should be the
2941 default when a C program starts up. In most systems, only ASCII printing
2942 characters match isprint(). This difference caused the output from pcretest
2943 to vary, making some of the tests fail. I have changed pcretest so that:
2945 (a) When it is outputting text in the compiled version of a pattern, bytes
2946 other than 32-126 are always shown as hex escapes.
2948 (b) When it is outputting text that is a matched part of a subject string,
2949 it does the same, unless a different locale has been set for the match
2950 (using the /L modifier). In this case, it uses isprint() to decide.
2952 4. Fixed a major bug that caused incorrect computation of the amount of memory
2953 required for a compiled pattern when options that changed within the
2954 pattern affected the logic of the preliminary scan that determines the
2955 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2956 that the computed length was too small. The symptoms of this bug were
2957 either the PCRE error "internal error: code overflow" from pcre_compile(),
2958 or a glibc crash with a message such as "pcretest: free(): invalid next
2959 size (fast)". Examples of patterns that provoked this bug (shown in
2960 pcretest format) are:
2963 /(?x)(?-x: \s*#\s*)/
2967 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2968 is now done differently.
2970 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2971 wrapper classes; (b) implement a new function in the C++ scanner that is
2972 more efficient than the old way of doing things because it avoids levels of
2973 recursion in the regex matching; (c) add a paragraph to the documentation
2974 for the FullMatch() function.
2976 6. The escape sequence \n was being treated as whatever was defined as
2977 "newline". Not only was this contrary to the documentation, which states
2978 that \n is character 10 (hex 0A), but it also went horribly wrong when
2979 "newline" was defined as CRLF. This has been fixed.
2981 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2982 was being set to -1 for the "end of line" case (supposedly a value that no
2983 character can have). Though this value is never used (the check for end of
2984 line is "zero bytes in current character"), it caused compiler complaints.
2985 I've changed it to 0xffffffff.
2987 8. In pcre_version.c, the version string was being built by a sequence of
2988 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2989 string (as it is for production releases) called a macro with an empty
2990 argument. The C standard says the result of this is undefined. The gcc
2991 compiler treats it as an empty string (which was what was wanted) but it is
2992 reported that Visual C gives an error. The source has been hacked around to
2995 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2996 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2997 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3000 10. Originally, pcretest opened its input and output without "b"; then I was
3001 told that "b" was needed in some environments, so it was added for release
3002 5.0 to both the input and output. (It makes no difference on Unix-like
3003 systems.) Later I was told that it is wrong for the input on Windows. I've
3004 now abstracted the modes into two macros, to make it easier to fiddle with
3005 them, and removed "b" from the input mode under Windows.
3007 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3009 12. Added -help and --help to pcretest as an official way of being reminded
3012 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3013 and pcrecpp.cc because they annoy compilers at high warning levels.
3015 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3017 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3018 scripts are not C programs :-) and which was not noticed because it works
3021 16. pcretest is supposed to handle any length of pattern and data line (as one
3022 line or as a continued sequence of lines) by extending its input buffer if
3023 necessary. This feature was broken for very long pattern lines, leading to
3024 a string of junk being passed to pcre_compile() if the pattern was longer
3027 17. I have done a major re-factoring of the way pcre_compile() computes the
3028 amount of memory needed for a compiled pattern. Previously, there was code
3029 that made a preliminary scan of the pattern in order to do this. That was
3030 OK when PCRE was new, but as the facilities have expanded, it has become
3031 harder and harder to keep it in step with the real compile phase, and there
3032 have been a number of bugs (see for example, 4 above). I have now found a
3033 cunning way of running the real compile function in a "fake" mode that
3034 enables it to compute how much memory it would need, while actually only
3035 ever using a few hundred bytes of working memory and without too many
3036 tests of the mode. This should make future maintenance and development
3037 easier. A side effect of this work is that the limit of 200 on the nesting
3038 depth of parentheses has been removed (though this was never a serious
3039 limitation, I suspect). However, there is a downside: pcre_compile() now
3040 runs more slowly than before (30% or more, depending on the pattern). I
3041 hope this isn't a big issue. There is no effect on runtime performance.
3043 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3044 newline (only possible for the last line of a file) and it was a
3045 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3047 19. Added additional timing features to pcretest. (1) The -tm option now times
3048 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3049 separate command line item, by a number that specifies the number of
3050 repeats to use when timing. The default is 50000; this gives better
3051 precision, but takes uncomfortably long for very large patterns.
3053 20. Extended pcre_study() to be more clever in cases where a branch of a
3054 subpattern has no definite first character. For example, (a*|b*)[cd] would
3055 previously give no result from pcre_study(). Now it recognizes that the
3056 first character must be a, b, c, or d.
3058 21. There was an incorrect error "recursive call could loop indefinitely" if
3059 a subpattern (or the entire pattern) that was being tested for matching an
3060 empty string contained only one non-empty item after a nested subpattern.
3061 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3062 incorrectly, because the \d was being skipped in the check.
3064 22. The pcretest program now has a new pattern option /B and a command line
3065 option -b, which is equivalent to adding /B to every pattern. This causes
3066 it to show the compiled bytecode, without the additional information that
3067 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3068 is the same as /B/I).
3070 23. A new optimization is now able automatically to treat some sequences such
3071 as a*b as a*+b. More specifically, if something simple (such as a character
3072 or a simple class like \d) has an unlimited quantifier, and is followed by
3073 something that cannot possibly match the quantified thing, the quantifier
3074 is automatically "possessified".
3076 24. A recursive reference to a subpattern whose number was greater than 39
3077 went wrong under certain circumstances in UTF-8 mode. This bug could also
3078 have affected the operation of pcre_study().
3080 25. Realized that a little bit of performance could be had by replacing
3081 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3083 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3085 27. Possessive quantifiers such as a++ were previously implemented by turning
3086 them into atomic groups such as ($>a+). Now they have their own opcodes,
3087 which improves performance. This includes the automatically created ones
3090 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3091 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3092 the first matched character to be a colon. This applied both to named and
3095 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3097 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3098 building PCRE on Windows easier, so I have included it in the distribution.
3100 31. There is now a check in pcretest against a ridiculously large number being
3101 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3102 loop, the loop is abandoned.
3104 32. Forward references to subpatterns in conditions such as (?(2)...) where
3105 subpattern 2 is defined later cause pcre_compile() to search forwards in
3106 the pattern for the relevant set of parentheses. This search went wrong
3107 when there were unescaped parentheses in a character class, parentheses
3108 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3110 33. "Subroutine" calls and backreferences were previously restricted to
3111 referencing subpatterns earlier in the regex. This restriction has now
3114 34. Added a number of extra features that are going to be in Perl 5.10. On the
3115 whole, these are just syntactic alternatives for features that PCRE had
3116 previously implemented using the Python syntax or my own invention. The
3117 other formats are all retained for compatibility.
3119 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3120 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3121 also .NET compatible.
3123 (b) A recursion or subroutine call to a named group can now be defined as
3124 (?&name) as well as (?P>name).
3126 (c) A backreference to a named group can now be defined as \k<name> or
3127 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3128 5.10, are also .NET compatible.
3130 (d) A conditional reference to a named group can now use the syntax
3131 (?(<name>) or (?('name') as well as (?(name).
3133 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3134 groups (named and numbered) that are never evaluated inline, but can be
3135 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3136 is always false. There may be only one alternative in such a group.
3138 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3139 as the simple (?(R). The condition is true only if the most recent
3140 recursion is that of the given number or name. It does not search out
3141 through the entire recursion stack.
3143 (g) The escape \gN or \g{N} has been added, where N is a positive or
3144 negative number, specifying an absolute or relative reference.
3146 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3147 some "unreachable code" warnings.
3149 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3150 things, this adds five new scripts.
3152 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3153 There were also incompatibilities regarding the handling of \Q..\E inside
3154 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3155 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3157 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3158 matches an empty string, and forcibly breaks the loop. There were bugs in
3159 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3160 against aaaa the result was just "a" rather than "aaaa", for example. Two
3161 separate and independent bugs (that affected different cases) have been
3164 39. Refactored the code to abolish the use of different opcodes for small
3165 capturing bracket numbers. This is a tidy that I avoided doing when I
3166 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3167 The new approach is not only tidier, it makes it possible to reduce the
3168 memory needed to fix the previous bug (38).
3170 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3171 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3172 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3175 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3178 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3179 copy construction and assignment in the C++ wrapper.
3181 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3182 couple of bugs that could have given wrong results in the "--newline=crlf"
3185 44. Added a number of casts and did some reorganization of signed/unsigned int
3186 variables following suggestions from Dair Grant. Also renamed the variable
3187 "this" as "item" because it is a C++ keyword.
3189 45. Arranged for dftables to add
3191 #include "pcre_internal.h"
3193 to pcre_chartables.c because without it, gcc 4.x may remove the array
3194 definition from the final binary if PCRE is built into a static library and
3195 dead code stripping is activated.
3197 46. For an unanchored pattern, if a match attempt fails at the start of a
3198 newline sequence, and the newline setting is CRLF or ANY, and the next two
3199 characters are CRLF, advance by two characters instead of one.
3202 Version 6.7 04-Jul-06
3203 ---------------------
3205 1. In order to handle tests when input lines are enormously long, pcretest has
3206 been re-factored so that it automatically extends its buffers when
3207 necessary. The code is crude, but this _is_ just a test program. The
3208 default size has been increased from 32K to 50K.
3210 2. The code in pcre_study() was using the value of the re argument before
3211 testing it for NULL. (Of course, in any sensible call of the function, it
3214 3. The memmove() emulation function in pcre_internal.h, which is used on
3215 systems that lack both memmove() and bcopy() - that is, hardly ever -
3216 was missing a "static" storage class specifier.
3218 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3219 containing an extended class (one that cannot be represented by a bitmap
3220 because it contains high-valued characters or Unicode property items, e.g.
3221 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3222 pattern, but PCRE should not loop if you do not (it no longer does).
3223 [Detail: two cases were found: (a) a repeated subpattern containing an
3224 extended class; (b) a recursive reference to a subpattern that followed a
3225 previous extended class. It wasn't skipping over the extended class
3226 correctly when UTF-8 mode was not set.]
3228 5. A negated single-character class was not being recognized as fixed-length
3229 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3230 compile error "lookbehind assertion is not fixed length".
3232 6. The RunPerlTest auxiliary script was showing an unexpected difference
3233 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3234 write a Perl script that can interpret lines of an input file either as
3235 byte characters or as UTF-8, which is what "perltest" was being required to
3236 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3237 can't do is switch easily at run time between having the "use utf8;" pragma
3238 or not. In the end, I fudged it by using the RunPerlTest script to insert
3239 "use utf8;" explicitly for the UTF-8 tests.
3241 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3242 the end of the subject string, contrary to the documentation and to what
3243 Perl does. This was true of both matching functions. Now it matches only at
3244 the start of the subject and immediately after *internal* newlines.
3246 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3247 a pointer to an int instead of a pointer to an unsigned long int. This
3248 caused problems on 64-bit systems.
3250 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3251 instance of the 'standard' template library not being so standard".
3253 10. There was no check on the number of named subpatterns nor the maximum
3254 length of a subpattern name. The product of these values is used to compute
3255 the size of the memory block for a compiled pattern. By supplying a very
3256 long subpattern name and a large number of named subpatterns, the size
3257 computation could be caused to overflow. This is now prevented by limiting
3258 the length of names to 32 characters, and the number of named subpatterns
3261 11. Subpatterns that are repeated with specific counts have to be replicated in
3262 the compiled pattern. The size of memory for this was computed from the
3263 length of the subpattern and the repeat count. The latter is limited to
3264 65535, but there was no limit on the former, meaning that integer overflow
3265 could in principle occur. The compiled length of a repeated subpattern is
3266 now limited to 30,000 bytes in order to prevent this.
3268 12. Added the optional facility to have named substrings with the same name.
3270 13. Added the ability to use a named substring as a condition, using the
3271 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3272 are numbers (not recommended). Forward references are permitted.
3274 14. Added forward references in named backreferences (if you see what I mean).
3276 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3277 pattern could run off the end of the subject. For example, the pattern
3278 "(?s)(.{1,5})"8 did this with the subject "ab".
3280 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3281 PCRE_CASELESS was set when matching characters that were quantified with ?
3284 17. A character class other than a single negated character that had a minimum
3285 but no maximum quantifier - for example [ab]{6,} - was not handled
3286 correctly by pce_dfa_exec(). It would match only one character.
3288 18. A valid (though odd) pattern that looked like a POSIX character
3289 class but used an invalid character after [ (for example [[,abc,]]) caused
3290 pcre_compile() to give the error "Failed: internal error: code overflow" or
3291 in some cases to crash with a glibc free() error. This could even happen if
3292 the pattern terminated after [[ but there just happened to be a sequence of
3293 letters, a binary zero, and a closing ] in the memory that followed.
3295 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3296 over the years. Originally (before any Unicode support), just the bottom 8
3297 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3298 output from "man perlunicode" includes this:
3300 The regular expression compiler produces polymorphic opcodes. That
3301 is, the pattern adapts to the data and automatically switches to
3302 the Unicode character scheme when presented with Unicode data--or
3303 instead uses a traditional byte scheme when presented with byte
3306 Sadly, a wide octal escape does not cause a switch, and in a string with
3307 no other multibyte characters, these octal escapes are treated as before.
3308 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3309 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3312 I have not perpetrated such confusion in PCRE. Up till now, it took just
3313 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3314 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3315 translate to the appropriate multibyte character.
3317 29. Applied some refactoring to reduce the number of warnings from Microsoft
3318 and Borland compilers. This has included removing the fudge introduced
3319 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3320 a warning about an unused variable.
3322 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3323 characters since release 4.0, because Perl (from release 5.004) does not.
3324 [Or at least, is documented not to: some releases seem to be in conflict
3325 with the documentation.] However, when a pattern was studied with
3326 pcre_study() and all its branches started with \s, PCRE still included VT
3327 as a possible starting character. Of course, this did no harm; it just
3328 caused an unnecessary match attempt.
3330 22. Removed a now-redundant internal flag bit that recorded the fact that case
3331 dependency changed within the pattern. This was once needed for "required
3332 byte" processing, but is no longer used. This recovers a now-scarce options
3333 bit. Also moved the least significant internal flag bit to the most-
3334 significant bit of the word, which was not previously used (hangover from
3335 the days when it was an int rather than a uint) to free up another bit for
3338 23. Added support for CRLF line endings as well as CR and LF. As well as the
3339 default being selectable at build time, it can now be changed at runtime
3340 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3341 specify that it is scanning data with non-default line endings.
3343 24. Changed the definition of CXXLINK to make it agree with the definition of
3344 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3346 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3347 recursions. This makes a big different to stack usage for some patterns.
3349 26. If a subpattern containing a named recursion or subroutine reference such
3350 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3351 the space required for the compiled pattern went wrong and gave too small a
3352 value. Depending on the environment, this could lead to "Failed: internal
3353 error: code overflow at offset 49" or "glibc detected double free or
3356 27. Applied patches from Google (a) to support the new newline modes and (b) to
3357 advance over multibyte UTF-8 characters in GlobalReplace.
3359 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3360 difference for some implementation of PCRE in some Windows version.
3362 29. Added some extra testing facilities to pcretest:
3364 \q<number> in a data line sets the "match limit" value
3365 \Q<number> in a data line sets the "match recursion limt" value
3366 -S <number> sets the stack size, where <number> is in megabytes
3368 The -S option isn't available for Windows.
3371 Version 6.6 06-Feb-06
3372 ---------------------
3374 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3375 in pcreposix.h. I have copied the definition from pcre.h.
3377 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3378 because pcre.h is no longer a built file.
3380 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3381 not normally included in the compiled code.
3384 Version 6.5 01-Feb-06
3385 ---------------------
3387 1. When using the partial match feature with pcre_dfa_exec(), it was not
3388 anchoring the second and subsequent partial matches at the new starting
3389 point. This could lead to incorrect results. For example, with the pattern
3390 /1234/, partially matching against "123" and then "a4" gave a match.
3392 2. Changes to pcregrep:
3394 (a) All non-match returns from pcre_exec() were being treated as failures
3395 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3396 error message is output. Some extra information is given for the
3397 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3398 probably the only errors that are likely to be caused by users (by
3399 specifying a regex that has nested indefinite repeats, for instance).
3400 If there are more than 20 of these errors, pcregrep is abandoned.
3402 (b) A binary zero was treated as data while matching, but terminated the
3403 output line if it was written out. This has been fixed: binary zeroes
3404 are now no different to any other data bytes.
3406 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3407 used to set a locale for matching. The --locale=xxxx long option has
3408 been added (no short equivalent) to specify a locale explicitly on the
3409 pcregrep command, overriding the environment variables.
3411 (d) When -B was used with -n, some line numbers in the output were one less
3412 than they should have been.
3414 (e) Added the -o (--only-matching) option.
3416 (f) If -A or -C was used with -c (count only), some lines of context were
3417 accidentally printed for the final match.
3419 (g) Added the -H (--with-filename) option.
3421 (h) The combination of options -rh failed to suppress file names for files
3422 that were found from directory arguments.
3424 (i) Added the -D (--devices) and -d (--directories) options.
3426 (j) Added the -F (--fixed-strings) option.
3428 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3430 (l) Added the --colo(u)r option.
3432 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3433 is not present by default.
3435 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3436 items such as (?R) or (?1), when the recursion could match a number of
3437 alternatives. If it matched one of the alternatives, but subsequently,
3438 outside the recursion, there was a failure, the code tried to back up into
3439 the recursion. However, because of the way PCRE is implemented, this is not
3440 possible, and the result was an incorrect result from the match.
3442 In order to prevent this happening, the specification of recursion has
3443 been changed so that all such subpatterns are automatically treated as
3444 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3446 4. I had overlooked the fact that, in some locales, there are characters for
3447 which isalpha() is true but neither isupper() nor islower() are true. In
3448 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3449 and ordfeminine) are like this. This affected the treatment of \w and \W
3450 when they appeared in character classes, but not when they appeared outside
3451 a character class. The bit map for "word" characters is now created
3452 separately from the results of isalnum() instead of just taking it from the
3453 upper, lower, and digit maps. (Plus the underscore character, of course.)
3455 5. The above bug also affected the handling of POSIX character classes such as
3456 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3457 permanent tables. Instead, the bit maps for such a class were previously
3458 created as the appropriate unions of the upper, lower, and digit bitmaps.
3459 Now they are created by subtraction from the [[:word:]] class, which has
3462 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3463 It is created by subtracting the vertical space characters (\x09, \x0a,
3464 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3465 subtraction was done in the overall bitmap for a character class, meaning
3466 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3467 be recognized. This bug has been fixed.
3469 7. Patches from the folks at Google:
3471 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3472 real life, but is still worth protecting against".
3474 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3475 regular expressions".
3477 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3480 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3481 "configure" and the latter not, in order to fix a problem somebody had
3482 with compiling the Arg class on HP-UX.
3484 (e) Improve the error-handling of the C++ wrapper a little bit.
3486 (f) New tests for checking recursion limiting.
3488 8. The pcre_memmove() function, which is used only if the environment does not
3489 have a standard memmove() function (and is therefore rarely compiled),
3490 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3491 returning a result (though PCRE never actually uses the result).
3493 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3494 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3495 returned instead of calling malloc() with an overflowing number that would
3496 most likely cause subsequent chaos.
3498 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3500 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3501 with this option is matched, the nmatch and pmatch options of regexec() are
3504 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3505 provided in case anyone wants to the the POSIX interface with UTF-8
3508 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3509 C++ linking (needed for some HP-UX environments).
3511 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3512 (unused parameter) and in the pcre_printint() function (omitted "default"
3513 switch label when the default is to do nothing).
3515 15. Added some code to make it possible, when PCRE is compiled as a C++
3516 library, to replace subject pointers for pcre_exec() with a smart pointer
3517 class, thus making it possible to process discontinuous strings.
3519 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3520 much the same function. They were added by different people who were trying
3521 to make PCRE easy to compile on non-Unix systems. It has been suggested
3522 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3523 for compiling on Windows systems. I have therefore replaced it with
3524 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3525 defaults to "extern" for C or "extern C" for C++, which works fine on
3526 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3527 SCOPE with something explicit in config.h. In addition:
3529 (a) pcreposix.h still had just "extern" instead of either of these macros;
3530 I have replaced it with PCRE_DATA_SCOPE.
3532 (b) Functions such as _pcre_xclass(), which are internal to the library,
3533 but external in the C sense, all had PCRE_EXPORT in their definitions.
3534 This is apparently wrong for the Windows case, so I have removed it.
3535 (It makes no difference on Unix-like systems.)
3537 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3538 of recursive calls to match(). This is different to MATCH_LIMIT because
3539 that limits the total number of calls to match(), not all of which increase
3540 the depth of recursion. Limiting the recursion depth limits the amount of
3541 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3542 when PCRE is compiled, and changed at run time. A patch from Google adds
3543 this functionality to the C++ interface.
3545 18. Changes to the handling of Unicode character properties:
3547 (a) Updated the table to Unicode 4.1.0.
3549 (b) Recognize characters that are not in the table as "Cn" (undefined).
3551 (c) I revised the way the table is implemented to a much improved format
3552 which includes recognition of ranges. It now supports the ranges that
3553 are defined in UnicodeData.txt, and it also amalgamates other
3554 characters into ranges. This has reduced the number of entries in the
3555 table from around 16,000 to around 3,000, thus reducing its size
3556 considerably. I realized I did not need to use a tree structure after
3557 all - a binary chop search is just as efficient. Having reduced the
3558 number of entries, I extended their size from 6 bytes to 8 bytes to
3559 allow for more data.
3561 (d) Added support for Unicode script names via properties such as \p{Han}.
3563 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3564 matching that character.
3566 20. When matching a repeated Unicode property with a minimum greater than zero,
3567 (for example \pL{2,}), PCRE could look past the end of the subject if it
3568 reached it while seeking the minimum number of characters. This could
3569 happen only if some of the characters were more than one byte long, because
3570 there is a check for at least the minimum number of bytes.
3572 21. Refactored the implementation of \p and \P so as to be more general, to
3573 allow for more different types of property in future. This has changed the
3574 compiled form incompatibly. Anybody with saved compiled patterns that use
3575 \p or \P will have to recompile them.
3577 22. Added "Any" and "L&" to the supported property types.
3579 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3580 but give a compile time error if the value is greater than 0xff.
3582 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3583 accidentally not being installed or uninstalled.
3585 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3586 made were to insert the current release number. This seemed silly, because
3587 it made things harder for people building PCRE on systems that don't run
3588 "configure". I have turned pcre.h into a distributed file, no longer built
3589 by "configure", with the version identification directly included. There is
3590 no longer a pcre.h.in file.
3592 However, this change necessitated a change to the pcre-config script as
3593 well. It is built from pcre-config.in, and one of the substitutions was the
3594 release number. I have updated configure.ac so that ./configure now finds
3595 the release number by grepping pcre.h.
3597 26. Added the ability to run the tests under valgrind.
3600 Version 6.4 05-Sep-05
3601 ---------------------
3603 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3604 "--" to be printed when multiple files were scanned, even when none of the
3605 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3606 consider it to be a bug, and have restored the previous behaviour.
3608 2. A couple of code tidies to get rid of compiler warnings.
3610 3. The pcretest program used to cheat by referring to symbols in the library
3611 whose names begin with _pcre_. These are internal symbols that are not
3612 really supposed to be visible externally, and in some environments it is
3613 possible to suppress them. The cheating is now confined to including
3614 certain files from the library's source, which is a bit cleaner.
3616 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3617 file's purpose clearer.
3619 5. Reorganized pcre_ucp_findchar().
3622 Version 6.3 15-Aug-05
3623 ---------------------
3625 1. The file libpcre.pc.in did not have general read permission in the tarball.
3627 2. There were some problems when building without C++ support:
3629 (a) If C++ support was not built, "make install" and "make test" still
3632 (b) There were problems when the value of CXX was explicitly set. Some
3633 changes have been made to try to fix these, and ...
3635 (c) --disable-cpp can now be used to explicitly disable C++ support.
3637 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3638 backslash in a target when C++ was disabled. This confuses some
3639 versions of "make", apparently. Using an intermediate variable solves
3640 this. (Same for CPP_LOBJ.)
3642 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3643 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3644 necessary on certain architectures.
3646 4. Added a setting of -export-symbols-regex to the link command to remove
3647 those symbols that are exported in the C sense, but actually are local
3648 within the library, and not documented. Their names all begin with
3649 "_pcre_". This is not a perfect job, because (a) we have to except some
3650 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3651 available (and never for static libraries). I have made a note to try to
3652 find a way round (a) in the future.
3655 Version 6.2 01-Aug-05
3656 ---------------------
3658 1. There was no test for integer overflow of quantifier values. A construction
3659 such as {1111111111111111} would give undefined results. What is worse, if
3660 a minimum quantifier for a parenthesized subpattern overflowed and became
3661 negative, the calculation of the memory size went wrong. This could have
3662 led to memory overwriting.
3664 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3666 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3667 operating environments where this matters.
3669 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3670 PCRE options from within the C++ wrapper.
3672 5. Named capturing subpatterns were not being correctly counted when a pattern
3673 was compiled. This caused two problems: (a) If there were more than 100
3674 such subpatterns, the calculation of the memory needed for the whole
3675 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3676 back references of the form \12, where the number was greater than 9, were
3677 not recognized as back references, even though there were sufficient
3678 previous subpatterns.
3680 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3681 versions of gcc, e.g. 2.95.4.
3684 Version 6.1 21-Jun-05
3685 ---------------------
3687 1. There was one reference to the variable "posix" in pcretest.c that was not
3688 surrounded by "#if !defined NOPOSIX".
3690 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3691 the cross-check on the old pcre_info() function, for the benefit of the
3692 cut-down version of PCRE that is currently imported into Exim.
3694 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3695 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3696 patterns that start (?i) but I don't think that matters. The use of (?i) is
3697 just an example; this all applies to the other options as well.
3699 4. Since libtool seems to echo the compile commands it is issuing, the output
3700 from "make" can be reduced a bit by putting "@" in front of each libtool
3703 5. Patch from the folks at Google for configure.in to be a bit more thorough
3704 in checking for a suitable C++ installation before trying to compile the
3705 C++ stuff. This should fix a reported problem when a compiler was present,
3706 but no suitable headers.
3708 6. The man pages all had just "PCRE" as their title. I have changed them to
3709 be the relevant file name. I have also arranged that these names are
3710 retained in the file doc/pcre.txt, which is a concatenation in text format
3711 of all the man pages except the little individual ones for each function.
3713 7. The NON-UNIX-USE file had not been updated for the different set of source
3714 files that come with release 6. I also added a few comments about the C++
3718 Version 6.0 07-Jun-05
3719 ---------------------
3721 1. Some minor internal re-organization to help with my DFA experiments.
3723 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3724 didn't matter for the library itself when fully configured, but did matter
3725 when compiling without UCP support, or within Exim, where the ucp files are
3728 3. Refactoring of the library code to split up the various functions into
3729 different source modules. The addition of the new DFA matching code (see
3730 below) to a single monolithic source would have made it really too
3731 unwieldy, quite apart from causing all the code to be include in a
3732 statically linked application, when only some functions are used. This is
3733 relevant even without the DFA addition now that patterns can be compiled in
3734 one application and matched in another.
3736 The downside of splitting up is that there have to be some external
3737 functions and data tables that are used internally in different modules of
3738 the library but which are not part of the API. These have all had their
3739 names changed to start with "_pcre_" so that they are unlikely to clash
3740 with other external names.
3742 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3743 a different (DFA) algorithm. Although it is slower than the original
3744 function, it does have some advantages for certain types of matching
3747 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3748 including restarting after a partial match.
3750 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3751 defined when compiling for Windows was sent to me. I have put it into the
3752 code, though I have no means of testing or verifying it.
3754 7. Added the pcre_refcount() auxiliary function.
3756 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3757 match before or at the first newline in the subject string. In pcretest,
3758 the /f option on a pattern can be used to set this.
3760 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3761 would behave wrongly. This has been present in PCRE since release 4.0.
3763 10. A number of changes to the pcregrep command:
3765 (a) Refactored how -x works; insert ^(...)$ instead of setting
3766 PCRE_ANCHORED and checking the length, in preparation for adding
3767 something similar for -w.
3769 (b) Added the -w (match as a word) option.
3771 (c) Refactored the way lines are read and buffered so as to have more
3772 than one at a time available.
3774 (d) Implemented a pcregrep test script.
3776 (e) Added the -M (multiline match) option. This allows patterns to match
3777 over several lines of the subject. The buffering ensures that at least
3778 8K, or the rest of the document (whichever is the shorter) is available
3779 for matching (and similarly the previous 8K for lookbehind assertions).
3781 (f) Changed the --help output so that it now says
3785 instead of two lines, one with "regex" and the other with "regexp"
3786 because that confused at least one person since the short forms are the
3787 same. (This required a bit of code, as the output is generated
3788 automatically from a table. It wasn't just a text change.)
3790 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3791 option but starts with a hyphen. Could be a pattern or a path name
3792 starting with a hyphen, for instance.
3794 (h) "-" can be given as a file name to represent stdin.
3796 (i) When file names are being printed, "(standard input)" is used for
3797 the standard input, for compatibility with GNU grep. Previously
3800 (j) The option --label=xxx can be used to supply a name to be used for
3801 stdin when file names are being printed. There is no short form.
3803 (k) Re-factored the options decoding logic because we are going to add
3804 two more options that take data. Such options can now be given in four
3805 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3807 (l) Added the -A, -B, and -C options for requesting that lines of context
3808 around matches be printed.
3810 (m) Added the -L option to print the names of files that do not contain
3811 any matching lines, that is, the complement of -l.
3813 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3814 continue to scan other files.
3816 (o) The -s option was incorrectly implemented. For compatibility with other
3817 greps, it now suppresses the error message for a non-existent or non-
3818 accessible file (but not the return code). There is a new option called
3819 -q that suppresses the output of matching lines, which was what -s was
3822 (p) Added --include and --exclude options to specify files for inclusion
3823 and exclusion when recursing.
3825 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3826 Hopefully, it now does.
3828 12. Missing cast in pcre_study().
3830 13. Added an "uninstall" target to the makefile.
3832 14. Replaced "extern" in the function prototypes in Makefile.in with
3833 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3834 world, but is set differently for Windows.
3836 15. Added a second compiling function called pcre_compile2(). The only
3837 difference is that it has an extra argument, which is a pointer to an
3838 integer error code. When there is a compile-time failure, this is set
3839 non-zero, in addition to the error test pointer being set to point to an
3840 error message. The new argument may be NULL if no error number is required
3841 (but then you may as well call pcre_compile(), which is now just a
3842 wrapper). This facility is provided because some applications need a
3843 numeric error indication, but it has also enabled me to tidy up the way
3844 compile-time errors are handled in the POSIX wrapper.
3846 16. Added VPATH=.libs to the makefile; this should help when building with one
3847 prefix path and installing with another. (Or so I'm told by someone who
3848 knows more about this stuff than I do.)
3850 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3851 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3852 match everything, including newlines. This is not POSIX-compatible, but
3853 somebody wanted the feature. From pcretest it can be activated by using
3854 both the P and the s flags.
3856 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3858 19. libpcre.pc was being incorrectly installed as executable.
3860 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3861 it now also looks for '\r' so that it will work unmodified on Windows.
3863 21. Added Google's contributed C++ wrapper to the distribution.
3865 22. Added some untidy missing memory free() calls in pcretest, to keep
3866 Electric Fence happy when testing.
3870 Version 5.0 13-Sep-04
3871 ---------------------
3873 1. Internal change: literal characters are no longer packed up into items
3874 containing multiple characters in a single byte-string. Each character
3875 is now matched using a separate opcode. However, there may be more than one
3876 byte in the character in UTF-8 mode.
3878 2. The pcre_callout_block structure has two new fields: pattern_position and
3879 next_item_length. These contain the offset in the pattern to the next match
3880 item, and its length, respectively.
3882 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3883 insertion of callouts before each pattern item. Added the /C option to
3884 pcretest to make use of this.
3886 4. On the advice of a Windows user, the lines
3888 #if defined(_WIN32) || defined(WIN32)
3889 _setmode( _fileno( stdout ), 0x8000 );
3890 #endif /* defined(_WIN32) || defined(WIN32) */
3892 have been added to the source of pcretest. This apparently does useful
3893 magic in relation to line terminators.
3895 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3896 for the benefit of those environments where the "b" makes a difference.
3898 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3899 to know about it. I have put a hack into configure.in that adds in code
3900 to set GCC=yes if CC=icc. This seems to end up at a point in the
3901 generated configure script that is early enough to affect the setting of
3902 compiler options, which is what is needed, but I have no means of testing
3903 whether it really works. (The user who reported this had patched the
3904 generated configure script, which of course I cannot do.)
3906 LATER: After change 22 below (new libtool files), the configure script
3907 seems to know about icc (and also ecc). Therefore, I have commented out
3908 this hack in configure.in.
3910 7. Added support for pkg-config (2 patches were sent in).
3912 8. Negated POSIX character classes that used a combination of internal tables
3913 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3914 [[:^ascii]]. Typically, they would match almost any characters. The other
3915 POSIX classes were not broken in this way.
3917 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3918 to find the match, as PCRE was deluded into thinking that the match had to
3919 start at the start point or following a newline. The same bug applied to
3920 patterns with negative forward assertions or any backward assertions
3921 preceding ".*" at the start, unless the pattern required a fixed first
3922 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3924 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3925 starting at the last subject character, bytes beyond the end of the subject
3928 11. Renamed the variable "class" as "classbits" to make life easier for C++
3929 users. (Previously there was a macro definition, but it apparently wasn't
3932 12. Added the new field "tables" to the extra data so that tables can be passed
3933 in at exec time, or the internal tables can be re-selected. This allows
3934 a compiled regex to be saved and re-used at a later time by a different
3935 program that might have everything at different addresses.
3937 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3938 -R library as well as a -L library.
3940 14. The debugging options of pcretest (-d on the command line or D on a
3941 pattern) showed incorrect output for anything following an extended class
3942 that contained multibyte characters and which was followed by a quantifier.
3944 15. Added optional support for general category Unicode character properties
3945 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3946 support. It adds about 90K to the size of the library. The meanings of the
3947 inbuilt class escapes such as \d and \s have NOT been changed.
3949 16. Updated pcredemo.c to include calls to free() to release the memory for the
3952 17. The generated file chartables.c was being created in the source directory
3953 instead of in the building directory. This caused the build to fail if the
3954 source directory was different from the building directory, and was
3957 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3958 file. No doubt somebody will tell me if they don't make sense... Also added
3959 Dan Mooney's comments about building on OpenVMS.
3961 19. Added support for partial matching via the PCRE_PARTIAL option for
3962 pcre_exec() and the \P data escape in pcretest.
3964 20. Extended pcretest with 3 new pattern features:
3966 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3967 write the compiled pattern to the file whose name is "rest-of-line".
3968 This is a straight binary dump of the data, with the saved pointer to
3969 the character tables forced to be NULL. The study data, if any, is
3970 written too. After writing, pcretest reads a new pattern.
3972 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3973 compiled pattern from the given file. There must not be any
3974 occurrences of "<" in the file name (pretty unlikely); if there are,
3975 pcretest will instead treat the initial "<" as a pattern delimiter.
3976 After reading in the pattern, pcretest goes on to read data lines as
3979 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3980 and 16-bit fields in a compiled pattern, to simulate a pattern that
3981 was compiled on a host of opposite endianness.
3983 21. The pcre-exec() function can now cope with patterns that were compiled on
3984 hosts of opposite endianness, with this restriction:
3986 As for any compiled expression that is saved and used later, the tables
3987 pointer field cannot be preserved; the extra_data field in the arguments
3988 to pcre_exec() should be used to pass in a tables address if a value
3989 other than the default internal tables were used at compile time.
3991 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3992 now diagnosed as an error. Previously, most of the time, a negative number
3993 would have been treated as zero, but if in addition "ovector" was passed as
3994 NULL, a crash could occur.
3996 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3997 new versions from the libtool 1.5 distribution (the last one is a copy of
3998 a file called libtool.m4). This seems to have fixed the need to patch
3999 "configure" to support Darwin 1.3 (which I used to do). However, I still
4000 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4003 24. Changed the PCRE licence to be the more standard "BSD" licence.
4006 Version 4.5 01-Dec-03
4007 ---------------------
4009 1. There has been some re-arrangement of the code for the match() function so
4010 that it can be compiled in a version that does not call itself recursively.
4011 Instead, it keeps those local variables that need separate instances for
4012 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4013 needs to "recurse". Keeping track of where control must go is done by means
4014 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4015 hide most of the details from the main code, and operates only if
4016 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4017 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4020 To make it easier for callers to provide specially tailored get/free
4021 functions for this usage, two new functions, pcre_stack_malloc, and
4022 pcre_stack_free, are used. They are always called in strict stacking order,
4023 and the size of block requested is always the same.
4025 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4026 PCRE has been compiled to use the stack or the heap for recursion. The
4027 -C option of pcretest uses this to show which version is compiled.
4029 A new data escape \S, is added to pcretest; it causes the amounts of store
4030 obtained and freed by both kinds of malloc/free at match time to be added
4033 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4034 what's available on my current Linux desktop machine.
4036 3. When matching a UTF-8 string, the test for a valid string at the start has
4037 been extended. If start_offset is not zero, PCRE now checks that it points
4038 to a byte that is the start of a UTF-8 character. If not, it returns
4039 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4040 this is necessary because there may be backward assertions in the pattern.
4041 When matching the same subject several times, it may save resources to use
4042 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4044 4. The code for checking the validity of UTF-8 strings has been tightened so
4045 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4046 containing "overlong sequences".
4048 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4049 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4050 should have been "&", but it just so happened that all the cases this let
4051 through by mistake were picked up later in the function.
4053 6. I had used a variable called "isblank" - this is a C99 function, causing
4054 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4056 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4057 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4058 so the version is shown; (c) stop "make test" echoing "./RunTest".
4060 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4062 9. The prototype for memmove() for systems that don't have it was using
4063 size_t, but the inclusion of the header that defines size_t was later. I've
4064 moved the #includes for the C headers earlier to avoid this.
4066 10. Added some adjustments to the code to make it easier to compiler on certain
4069 (a) Some "const" qualifiers were missing.
4070 (b) Added the macro EXPORT before all exported functions; by default this
4071 is defined to be empty.
4072 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4073 that it reads its output file name as an argument instead of writing
4074 to the standard output and assuming this can be redirected.
4076 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4077 class containing characters with values greater than 255, PCRE compilation
4080 12. A recursive reference to a subpattern that was within another subpattern
4081 that had a minimum quantifier of zero caused PCRE to crash. For example,
4082 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4083 recursion. If the recursively-called subpattern itself had a zero repeat,
4086 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4087 buffer into which it was copied (for escape processing) was still set at
4088 1024, so long lines caused crashes.
4090 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4091 "internal error: code overflow...". This applied to any character class
4092 that was followed by a possessive quantifier.
4094 15. Modified the Makefile to add libpcre.la as a prerequisite for
4095 libpcreposix.la because I was told this is needed for a parallel build to
4098 16. If a pattern that contained .* following optional items at the start was
4099 studied, the wrong optimizing data was generated, leading to matching
4100 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4101 matching string must start with a or b or c. The correct conclusion for
4102 this pattern is that a match can start with any character.
4105 Version 4.4 13-Aug-03
4106 ---------------------
4108 1. In UTF-8 mode, a character class containing characters with values between
4109 127 and 255 was not handled correctly if the compiled pattern was studied.
4110 In fixing this, I have also improved the studying algorithm for such
4113 2. Three internal functions had redundant arguments passed to them. Removal
4114 might give a very teeny performance improvement.
4116 3. Documentation bug: the value of the capture_top field in a callout is *one
4117 more than* the number of the hightest numbered captured substring.
4119 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4120 in incorrectly linking with a previously installed version. They now link
4121 explicitly with libpcre.la.
4123 5. configure.in no longer needs to recognize Cygwin specially.
4125 6. A problem in pcre.in for Windows platforms is fixed.
4127 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4128 pcretest, it used to include the size of the study block as part of its
4129 output. Unfortunately, the structure contains a field that has a different
4130 size on different hardware architectures. This meant that the tests that
4131 showed this size failed. As the block is currently always of a fixed size,
4132 this information isn't actually particularly useful in pcretest output, so
4133 I have just removed it.
4135 8. Three pre-processor statements accidentally did not start in column 1.
4136 Sadly, there are *still* compilers around that complain, even though
4137 standard C has not required this for well over a decade. Sigh.
4139 9. In pcretest, the code for checking callouts passed small integers in the
4140 callout_data field, which is a void * field. However, some picky compilers
4141 complained about the casts involved for this on 64-bit systems. Now
4142 pcretest passes the address of the small integer instead, which should get
4143 rid of the warnings.
4145 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4146 both compile and run time, and gives an error if an invalid UTF-8 sequence
4147 is found. There is a option for disabling this check in cases where the
4148 string is known to be correct and/or the maximum performance is wanted.
4150 11. In response to a bug report, I changed one line in Makefile.in from
4152 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4154 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4156 to look similar to other lines, but I have no way of telling whether this
4157 is the right thing to do, as I do not use Windows. No doubt I'll get told
4161 Version 4.3 21-May-03
4162 ---------------------
4164 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4167 2. Some refactoring to improve the quality of the code:
4169 (i) The utf8_table... variables are now declared "const".
4171 (ii) The code for \cx, which used the "case flipping" table to upper case
4172 lower case letters, now just substracts 32. This is ASCII-specific,
4173 but the whole concept of \cx is ASCII-specific, so it seems
4176 (iii) PCRE was using its character types table to recognize decimal and
4177 hexadecimal digits in the pattern. This is silly, because it handles
4178 only 0-9, a-f, and A-F, but the character types table is locale-
4179 specific, which means strange things might happen. A private
4180 table is now used for this - though it costs 256 bytes, a table is
4181 much faster than multiple explicit tests. Of course, the standard
4182 character types table is still used for matching digits in subject
4185 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4186 ending in _t are). So I've renamed it as ESC_tee.
4188 3. The first argument for regexec() in the POSIX wrapper should have been
4191 4. Changed pcretest to use malloc() for its buffers so that they can be
4192 Electric Fenced for debugging.
4194 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4195 to read one or more bytes before the start of the subject string. Often this
4196 had no effect on PCRE's behaviour, but in some circumstances it could
4197 provoke a segmentation fault.
4199 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4200 to try to read one or more bytes before the start of the subject string.
4202 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4203 UTF-8 support could misbehave in various ways if the subject string
4204 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4205 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4206 back over UTF-8 characters.)
4209 Version 4.2 14-Apr-03
4210 ---------------------
4212 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4214 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4215 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4216 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4217 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4218 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4220 Note: automatic setting of the BUILD variables is not yet working
4221 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4222 compile-time but not at link-time
4223 [LINK]: use for linking executables only
4224 make different versions for Windows and non-Windows
4225 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4227 [LINK_FOR_BUILD]: new variable
4228 [OBJEXT]: use throughout
4229 [EXEEXT]: use throughout
4230 <winshared>: new target
4231 <wininstall>: new target
4232 <dftables.o>: use native compiler
4233 <dftables>: use native linker
4234 <install>: handle Windows platform correctly
4237 copy DLL to top builddir before testing
4239 As part of these changes, -no-undefined was removed again. This was reported
4240 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4243 3. Some tidies to get rid of compiler warnings:
4245 . In the match_data structure, match_limit was an unsigned long int, whereas
4246 match_call_count was an int. I've made them both unsigned long ints.
4248 . In pcretest the fact that a const uschar * doesn't automatically cast to
4249 a void * provoked a warning.
4251 . Turning on some more compiler warnings threw up some "shadow" variables
4252 and a few more missing casts.
4254 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4255 option, a class that contained a single character with a value between 128
4256 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4258 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4259 option, a class that contained several characters, but with at least one
4260 whose value was between 128 and 255 caused PCRE to crash.
4263 Version 4.1 12-Mar-03
4264 ---------------------
4266 1. Compiling with gcc -pedantic found a couple of places where casts were
4267 needed, and a string in dftables.c that was longer than standard compilers are
4268 required to support.
4270 2. Compiling with Sun's compiler found a few more places where the code could
4271 be tidied up in order to avoid warnings.
4273 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4274 first of these names is deprecated in the latest Autoconf in favour of the name
4275 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4276 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4277 analogy I have changed it to CFLAGS_FOR_BUILD.
4279 4. Added -no-undefined to the linking command in the Makefile, because this is
4280 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4281 linking step for the pcreposix library.
4283 5. PCRE was failing to diagnose the case of two named groups with the same
4286 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4287 literal character that is needed in the subject for a match, and scans along to
4288 ensure that it is present before embarking on the full matching process. This
4289 saves time in cases of nested unlimited repeats that are never going to match.
4290 Problem: the scan can take a lot of time if the subject is very long (e.g.
4291 megabytes), thus penalizing straightforward matches. It is now done only if the
4292 amount of subject to be scanned is less than 1000 bytes.
4294 7. A lesser problem with the same optimization is that it was recording the
4295 first character of an anchored pattern as "needed", thus provoking a search
4296 right along the subject, even when the first match of the pattern was going to
4297 fail. The "needed" character is now not set for anchored patterns, unless it
4298 follows something in the pattern that is of non-fixed length. Thus, it still
4299 fulfils its original purpose of finding quick non-matches in cases of nested
4300 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4303 Version 4.0 17-Feb-03
4304 ---------------------
4306 1. If a comment in an extended regex that started immediately after a meta-item
4307 extended to the end of string, PCRE compiled incorrect data. This could lead to
4308 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4310 2. Moved to autoconf 2.53 and libtool 1.4.2.
4312 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4313 the special perltest8 script is no longer needed - all the tests can be run
4314 from a single perltest script.
4316 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4317 by \s. It has now been removed in PCRE. This means it isn't recognized as
4318 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4319 class [:space:] *does* include VT, thereby creating a mess.
4321 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4324 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4325 its new features into the main test script, reducing the number of scripts.
4327 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4328 were backward compatible, and made the (?i) apply to the whole pattern, as if
4329 /i were given. Now it behaves more logically, and applies the option setting
4330 only to what follows. PCRE has been changed to follow suit. However, if it
4331 finds options settings right at the start of the pattern, it extracts them into
4332 the global options, as before. Thus, they show up in the info data.
4334 8. Added support for the \Q...\E escape sequence. Characters in between are
4335 treated as literals. This is slightly different from Perl in that $ and @ are
4336 also handled as literals inside the quotes. In Perl, they will cause variable
4337 interpolation. Note the following examples:
4339 Pattern PCRE matches Perl matches
4341 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4342 \Qabc\$xyz\E abc\$xyz abc\$xyz
4343 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4345 For compatibility with Perl, \Q...\E sequences are recognized inside character
4346 classes as well as outside them.
4348 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4349 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4350 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4351 signed/unsigned warnings.
4353 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4354 option for pcretest, so I've replaced it by a simple function that does just
4357 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4360 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4361 Java package. This provides some syntactic sugar for simple cases of what my
4362 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4363 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4364 item, you can use this simplified notation. Note that only makes sense with
4365 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4366 greediness, whatever the setting of the PCRE_UNGREEDY option.
4368 13. A change of greediness default within a pattern was not taking effect at
4369 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4370 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4371 was abstracted outside.
4373 14. PCRE now supports the \G assertion. It is true when the current matching
4374 position is at the start point of the match. This differs from \A when the
4375 starting offset is non-zero. Used with the /g option of pcretest (or similar
4376 code), it works in the same way as it does for Perl's /g option. If all
4377 alternatives of a regex begin with \G, the expression is anchored to the start
4378 match position, and the "anchored" flag is set in the compiled expression.
4380 15. Some bugs concerning the handling of certain option changes within patterns
4381 have been fixed. These applied to options other than (?ims). For example,
4382 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4383 been the other way round. Some of this was related to change 7 above.
4385 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4386 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4387 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4388 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4390 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4391 mode. Unlike ".", it always matches newline, whatever the setting of
4392 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4393 assertions. Perl allows it, but it doesn't (in general) work because it can't
4394 calculate the length of the lookbehind. At least, that's the case for Perl
4395 5.8.0 - I've been told they are going to document that it doesn't work in
4398 18. Added an error diagnosis for escapes that PCRE does not support: these are
4399 \L, \l, \N, \P, \p, \U, \u, and \X.
4401 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4402 reading past the end of the pattern in cases such as /[abcd/.
4404 20. PCRE was getting more memory than necessary for patterns with classes that
4405 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4407 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4408 compiling PCRE for use with Virtual Pascal.
4410 22. Small fix to the Makefile to make it work properly if the build is done
4411 outside the source tree.
4413 23. Added a new extension: a condition to go with recursion. If a conditional
4414 subpattern starts with (?(R) the "true" branch is used if recursion has
4415 happened, whereas the "false" branch is used only at the top level.
4417 24. When there was a very long string of literal characters (over 255 bytes
4418 without UTF support, over 250 bytes with UTF support), the computation of how
4419 much memory was required could be incorrect, leading to segfaults or other
4422 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4423 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4424 there was a subsequent back reference to those brackets. This meant that, for
4425 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4426 possible to check for precisely this case. All we can do is abandon the
4427 optimization if .* occurs inside capturing brackets when there are any back
4428 references whatsoever. (See below for a better fix that came later.)
4430 26. The handling of the optimization for finding the first character of a
4431 non-anchored pattern, and for finding a character that is required later in the
4432 match were failing in some cases. This didn't break the matching; it just
4433 failed to optimize when it could. The way this is done has been re-implemented.
4435 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4437 28. Added a new feature that provides some of the functionality that Perl
4438 provides with (?{...}). The facility is termed a "callout". The way it is done
4439 in PCRE is for the caller to provide an optional function, by setting
4440 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4441 global variable. By default it is unset, which disables all calling out. To get
4442 the function called, the regex must include (?C) at appropriate points. This
4443 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4444 This provides a means of identifying different callout points. When PCRE
4445 reaches such a point in the regex, if pcre_callout has been set, the external
4446 function is called. It is provided with data in a structure called
4447 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4448 matching continues; if it returns a non-zero value, the match at the current
4449 point fails. However, backtracking will occur if possible. [This was changed
4450 later and other features added - see item 49 below.]
4452 29. pcretest is upgraded to test the callout functionality. It provides a
4453 callout function that displays information. By default, it shows the start of
4454 the match and the current position in the text. There are some new data escapes
4455 to vary what happens:
4457 \C+ in addition, show current contents of captured substrings
4458 \C- do not supply a callout function
4459 \C!n return 1 when callout number n is reached
4460 \C!n!m return 1 when callout number n is reached for the mth time
4462 30. If pcregrep was called with the -l option and just a single file name, it
4463 output "<stdin>" if a match was found, instead of the file name.
4465 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4466 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4467 pcre_exec(). This saves a malloc/free per call. The default value of
4468 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4471 32. The default maximum size of a compiled pattern is 64K. There have been a
4472 few cases of people hitting this limit. The code now uses macros to handle the
4473 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4474 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4475 configuring. Tests 2 and 5 work only with 2-byte links because they output
4476 debugging information about compiled patterns.
4478 33. Internal code re-arrangements:
4480 (a) Moved the debugging function for printing out a compiled regex into
4481 its own source file (printint.c) and used #include to pull it into
4482 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4485 (b) Defined the list of op-code names for debugging as a macro in
4486 internal.h so that it is next to the definition of the opcodes.
4488 (c) Defined a table of op-code lengths for simpler skipping along compiled
4489 code. This is again a macro in internal.h so that it is next to the
4490 definition of the opcodes.
4492 34. Added support for recursive calls to individual subpatterns, along the
4493 lines of Robin Houston's patch (but implemented somewhat differently).
4495 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4496 allow it to read and process whole directories in Win32. This code was
4497 contributed by Lionel Fourquaux; it has not been tested by me.
4499 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4500 used to name a group. Names consist of alphanumerics and underscores, and must
4501 be unique. Back references use the syntax (?P=name) and recursive calls use
4502 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4503 numbers. The function pcre_fullinfo() can be used after compilation to extract
4504 a name/number map. There are three relevant calls:
4506 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4507 PCRE_INFO_NAMECOUNT yields the number of entries
4508 PCRE_INFO_NAMETABLE yields a pointer to the map.
4510 The map is a vector of fixed-size entries. The size of each entry depends on
4511 the length of the longest name used. The first two bytes of each entry are the
4512 group number, most significant byte first. There follows the corresponding
4513 name, zero terminated. The names are in alphabetical order.
4515 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4516 case instead of 255. Making it the same both with and without UTF-8 support
4517 means that the same test output works with both.
4519 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4520 calling malloc() with a zero argument.
4522 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4523 optimization. I've improved things by keeping a bitmap of backreferences with
4524 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4525 fact referenced, the optimization can be applied. It is unlikely that a
4526 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4527 the match to follow \n) will appear inside brackets with a number greater than
4528 31, but if it does, any back reference > 31 suppresses the optimization.
4530 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4531 of disabling numbered capturing parentheses. Any opening parenthesis that is
4532 not followed by ? behaves as if it were followed by ?: but named parentheses
4533 can still be used for capturing (and they will acquire numbers in the usual
4536 41. Redesigned the return codes from the match() function into yes/no/error so
4537 that errors can be passed back from deep inside the nested calls. A malloc
4538 failure while inside a recursive subpattern call now causes the
4539 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4541 42. It is now possible to set a limit on the number of times the match()
4542 function is called in a call to pcre_exec(). This facility makes it possible to
4543 limit the amount of recursion and backtracking, though not in a directly
4544 obvious way, because the match() function is used in a number of different
4545 circumstances. The count starts from zero for each position in the subject
4546 string (for non-anchored patterns). The default limit is, for compatibility, a
4547 large number, namely 10 000 000. You can change this in two ways:
4549 (a) When configuring PCRE before making, you can use --with-match-limit=n
4550 to set a default value for the compiled library.
4552 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4553 a different value is set. See 45 below.
4555 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4557 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4558 of things that can be changed at compile time. The first argument specifies
4559 what is wanted and the second points to where the information is to be placed.
4560 The current list of available information is:
4564 The output is an integer that is set to one if UTF-8 support is available;
4565 otherwise it is set to zero.
4569 The output is an integer that it set to the value of the code that is used for
4570 newline. It is either LF (10) or CR (13).
4572 PCRE_CONFIG_LINK_SIZE
4574 The output is an integer that contains the number of bytes used for internal
4575 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4577 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4579 The output is an integer that contains the threshold above which the POSIX
4580 interface uses malloc() for output vectors. See item 31 above.
4582 PCRE_CONFIG_MATCH_LIMIT
4584 The output is an unsigned integer that contains the default limit of the number
4585 of match() calls in a pcre_exec() execution. See 42 above.
4587 44. pcretest has been upgraded by the addition of the -C option. This causes it
4588 to extract all the available output from the new pcre_config() function, and to
4589 output it. The program then exits immediately.
4591 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4592 order to support additional features. One way would have been to define
4593 pcre_exec2() (for example) with extra arguments, but this would not have been
4594 extensible, and would also have required all calls to the original function to
4595 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4596 is used for passing in "extra" data from pcre_study().
4598 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4599 contains the following fields:
4601 flags a bitmap indicating which of the following fields are set
4602 study_data opaque data from pcre_study()
4603 match_limit a way of specifying a limit on match() calls for a specific
4605 callout_data data for callouts (see 49 below)
4607 The flag bits are also defined in pcre.h, and are
4609 PCRE_EXTRA_STUDY_DATA
4610 PCRE_EXTRA_MATCH_LIMIT
4611 PCRE_EXTRA_CALLOUT_DATA
4613 The pcre_study() function now returns one of these new pcre_extra blocks, with
4614 the actual study data pointed to by the study_data field, and the
4615 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4616 before. That is, this change is entirely upwards-compatible and requires no
4617 change to existing code.
4619 If you want to pass in additional data to pcre_exec(), you can either place it
4620 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4623 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4624 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4625 times with different match limits, until it finds the minimum value needed for
4626 pcre_exec() to complete. The value is then output. This can be instructive; for
4627 most simple matches the number is quite small, but for pathological cases it
4628 gets very large very quickly.
4630 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4631 returns the size of the data block pointed to by the study_data field in a
4632 pcre_extra block, that is, the value that was passed as the argument to
4633 pcre_malloc() when PCRE was getting memory in which to place the information
4634 created by pcre_study(). The fourth argument should point to a size_t variable.
4635 pcretest has been extended so that this information is shown after a successful
4636 pcre_study() call when information about the compiled regex is being displayed.
4638 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4639 because what follows is always an absolute path. (Later: it turns out that this
4640 is more than cosmetic for MinGW, because it doesn't like empty path
4643 49. Some changes have been made to the callout feature (see 28 above):
4645 (i) A callout function now has three choices for what it returns:
4647 0 => success, carry on matching
4648 > 0 => failure at this point, but backtrack if possible
4649 < 0 => serious error, return this value from pcre_exec()
4651 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4652 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4653 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4654 use by callout functions. It will never be used by PCRE itself.
4656 (ii) The pcre_extra structure (see 45 above) has a void * field called
4657 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4658 pcre_callout_block structure has a field of the same name. The contents of
4659 the field passed in the pcre_extra structure are passed to the callout
4660 function in the corresponding field in the callout block. This makes it
4661 easier to use the same callout-containing regex from multiple threads. For
4662 testing, the pcretest program has a new data escape
4664 \C*n pass the number n (may be negative) as callout_data
4666 If the callout function in pcretest receives a non-zero value as
4667 callout_data, it returns that value.
4669 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4670 there were some redundant $(CFLAGS) in commands that are now specified as
4671 $(LINK), which already includes $(CFLAGS).
4673 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4674 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4675 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4676 one-byte characters throughout. Note that case-insensitive matching applies
4677 only to characters whose values are less than 256. PCRE doesn't support the
4678 notion of cases for higher-valued characters.
4680 (i) A character class whose characters are all within 0-255 is handled as
4681 a bit map, and the map is inverted for negative classes. Previously, a
4682 character > 255 always failed to match such a class; however it should
4683 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4685 (ii) A negated character class with a single character < 255 is coded as
4686 "not this character" (OP_NOT). This wasn't working properly when the test
4687 character was multibyte, either singly or repeated.
4689 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4690 mode, for example: \x{100}{2,3}.
4692 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4693 singly or repeated) now correctly test multibyte characters. However,
4694 PCRE doesn't recognize any characters with values greater than 255 as
4695 digits, spaces, or word characters. Such characters always match \D, \S,
4696 and \W, and never match \d, \s, or \w.
4698 (v) Classes may now contain characters and character ranges with values
4699 greater than 255. For example: [ab\x{100}-\x{400}].
4701 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4704 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4705 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4706 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4709 53. The single man page has become too large. I have therefore split it up into
4710 a number of separate man pages. These also give rise to individual HTML pages;
4711 these are now put in a separate directory, and there is an index.html page that
4712 lists them all. Some hyperlinking between the pages has been installed.
4714 54. Added convenience functions for handling named capturing parentheses.
4716 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4717 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4718 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4721 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4722 calling configure. These values are used when compiling the dftables.c program
4723 which is run to generate the source of the default character tables. They
4724 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4725 you will need to set these values.
4727 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4730 Version 3.9 02-Jan-02
4731 ---------------------
4733 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4735 2. If --disable-static was given, the building process failed when trying to
4736 build pcretest and pcregrep. (For some reason it was using libtool to compile
4737 them, which is not right, as they aren't part of the library.)
4740 Version 3.8 18-Dec-01
4741 ---------------------
4743 1. The experimental UTF-8 code was completely screwed up. It was packing the
4744 bytes in the wrong order. How dumb can you get?
4747 Version 3.7 29-Oct-01
4748 ---------------------
4750 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4751 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4752 this didn't happen under Solaris 8, where I normally test things.
4754 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4755 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4756 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4757 this fix an hour or so after the initial 3.7 release.)
4760 Version 3.6 23-Oct-01
4761 ---------------------
4763 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4764 offsets passed as NULL with zero offset count.
4766 2. The config.guess and config.sub files had not been updated when I moved to
4767 the latest autoconf.
4770 Version 3.5 15-Aug-01
4771 ---------------------
4773 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4776 2. By using declared but undefined structures, we can avoid using "void"
4777 definitions in pcre.h while keeping the internal definitions of the structures
4780 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4781 user point of view, this means that both static and shared libraries are built
4782 by default, but this can be individually controlled. More of the work of
4783 handling this static/shared cases is now inside libtool instead of PCRE's make
4786 4. The pcretest utility is now installed along with pcregrep because it is
4787 useful for users (to test regexs) and by doing this, it automatically gets
4788 relinked by libtool. The documentation has been turned into a man page, so
4789 there are now .1, .txt, and .html versions in /doc.
4791 5. Upgrades to pcregrep:
4792 (i) Added long-form option names like gnu grep.
4793 (ii) Added --help to list all options with an explanatory phrase.
4794 (iii) Added -r, --recursive to recurse into sub-directories.
4795 (iv) Added -f, --file to read patterns from a file.
4797 6. pcre_exec() was referring to its "code" argument before testing that
4798 argument for NULL (and giving an error if it was NULL).
4800 7. Upgraded Makefile.in to allow for compiling in a different directory from
4801 the source directory.
4803 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4804 options bits, the pointer it was passed was to an int instead of to an unsigned
4805 long int. This mattered only on 64-bit systems.
4807 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4808 generated) instead of pcre.in, which it its source. Also made the same change
4809 in several of the .c files.
4811 10. A new release of gcc defines printf() as a macro, which broke pcretest
4812 because it had an ifdef in the middle of a string argument for printf(). Fixed
4813 by using separate calls to printf().
4815 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4816 script, to force use of CR or LF instead of \n in the source. On non-Unix
4817 systems, the value can be set in config.h.
4819 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4820 absolute limit. Changed the text of the error message to make this clear, and
4821 likewise updated the man page.
4823 13. The limit of 99 on the number of capturing subpatterns has been removed.
4824 The new limit is 65535, which I hope will not be a "real" limit.
4827 Version 3.4 22-Aug-00
4828 ---------------------
4830 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4832 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4835 Version 3.3 01-Aug-00
4836 ---------------------
4838 1. If an octal character was given, but the value was greater than \377, it
4839 was not getting masked to the least significant bits, as documented. This could
4840 lead to crashes in some systems.
4842 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4843 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4845 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4846 These just pass their arguments on to (pcre_free)(), but they are provided
4847 because some uses of PCRE bind it to non-C systems that can call its functions,
4848 but cannot call free() or pcre_free() directly.
4850 4. Add "make test" as a synonym for "make check". Corrected some comments in
4853 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4856 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4857 command called pgrep for grepping around the active processes.
4859 7. Added the beginnings of support for UTF-8 character strings.
4861 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4862 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4863 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4864 out for the ar command.)
4867 Version 3.2 12-May-00
4868 ---------------------
4870 This is purely a bug fixing release.
4872 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4873 of ZA. This was just one example of several cases that could provoke this bug,
4874 which was introduced by change 9 of version 2.00. The code for breaking
4875 infinite loops after an iteration that matches an empty string was't working
4878 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4879 when matched against abbab (for example). After matching an empty string, it
4880 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4881 caused it to match further down the string than it should.
4883 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4884 was there because it doesn't seem to be needed, and it causes trouble on some
4885 systems, as it is not a Standard C header. It has been removed.
4887 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4888 were reported on the Macintosh. The changes were from
4890 while ((c = *(++ptr)) != 0 && c != '\n');
4892 while ((c = *(++ptr)) != 0 && c != '\n') ;
4894 Totally extraordinary, but if that's what it takes...
4896 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4897 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4898 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4899 assumes the way PCRE uses memmove() (always moving upwards).
4901 6. PCRE is being used in one environment where strchr() is not available. There
4902 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4906 Version 3.1 09-Feb-00
4907 ---------------------
4909 The only change in this release is the fixing of some bugs in Makefile.in for
4910 the "install" target:
4912 (1) It was failing to install pcreposix.h.
4914 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4917 Version 3.0 01-Feb-00
4918 ---------------------
4920 1. Add support for the /+ modifier to perltest (to output $` like it does in
4923 2. Add support for the /g modifier to perltest.
4925 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4926 matches null strings.
4928 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4929 pattern. Perl treats empty patterns specially - it reuses the most recent
4930 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4933 5. The POSIX interface was broken in that it was just handing over the POSIX
4934 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4935 required a bigger vector, with some working space on the end. This means that
4936 the POSIX wrapper now has to get and free some memory, and copy the results.
4938 6. Added some simple autoconf support, placing the test data and the
4939 documentation in separate directories, re-organizing some of the
4940 information files, and making it build pcre-config (a GNU standard). Also added
4941 libtool support for building PCRE as a shared library, which is now the
4944 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4945 09 are not valid octal constants. Single digits will be used for minor values
4948 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4949 existing programs that set these in the POSIX interface can use PCRE without
4952 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4953 return all that pcre_info() returns, plus additional data. The pcre_info()
4954 function is retained for compatibility, but is considered to be obsolete.
4956 10. Added experimental recursion feature (?R) to handle one common case that
4957 Perl 5.6 will be able to do with (?p{...}).
4959 11. Added support for POSIX character classes like [:alpha:], which Perl is
4963 Version 2.08 31-Aug-99
4964 ----------------------
4966 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4967 trying to match at the startoffset position, but instead was moving forward to
4968 the next newline as if a previous match had failed.
4970 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4971 and could get into a loop if a null string was matched other than at the start
4974 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4975 be distinguished at compile time, and for completeness also added PCRE_DATE.
4977 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4978 in GnuWin32 environments.
4981 Version 2.07 29-Jul-99
4982 ----------------------
4984 1. The documentation is now supplied in plain text form and HTML as well as in
4985 the form of man page sources.
4987 2. C++ compilers don't like assigning (void *) values to other pointer types.
4988 In particular this affects malloc(). Although there is no problem in Standard
4989 C, I've put in casts to keep C++ compilers happy.
4991 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4992 should be (const char *).
4994 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4995 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4996 However, I haven't made this a standard facility. The documentation doesn't
4997 mention it, and the Makefile doesn't support it.
4999 5. The Makefile now contains an "install" target, with editable destinations at
5000 the top of the file. The pcretest program is not installed.
5002 6. pgrep -V now gives the PCRE version number and date.
5004 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5005 causing the entire string to be ignored, instead of just the last character.
5007 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5008 non-matching string, it can take a very, very long time, even for strings of
5009 quite modest length, because of the nested recursion. PCRE now does better in
5010 some of these cases. It does this by remembering the last required literal
5011 character in the pattern, and pre-searching the subject to ensure it is present
5012 before running the real match. In other words, it applies a heuristic to detect
5013 some types of certain failure quickly, and in the above example, if presented
5014 with a string that has no trailing " it gives "no match" very quickly.
5016 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5017 other alternatives are tried instead.
5020 Version 2.06 09-Jun-99
5021 ----------------------
5023 1. Change pcretest's output for amount of store used to show just the code
5024 space, because the remainder (the data block) varies in size between 32-bit and
5027 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5028 start matching at. This allows lookbehinds to work when searching for multiple
5029 occurrences in a string.
5031 3. Added additional options to pcretest for testing multiple occurrences:
5033 /+ outputs the rest of the string that follows a match
5034 /g loops for multiple occurrences, using the new startoffset argument
5035 /G loops for multiple occurrences by passing an incremented pointer
5037 4. PCRE wasn't doing the "first character" optimization for patterns starting
5038 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5039 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5040 the letter 'x'. On long subject strings, this gives a significant speed-up.
5043 Version 2.05 21-Apr-99
5044 ----------------------
5046 1. Changed the type of magic_number from int to long int so that it works
5047 properly on 16-bit systems.
5049 2. Fixed a bug which caused patterns starting with .* not to work correctly
5050 when the subject string contained newline characters. PCRE was assuming
5051 anchoring for such patterns in all cases, which is not correct because .* will
5052 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5053 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5054 must be retried after every newline in the subject.
5057 Version 2.04 18-Feb-99
5058 ----------------------
5060 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5061 computation of the store needed to hold the pattern was incorrect (too large).
5062 If such patterns were nested a few deep, this could multiply and become a real
5065 2. Added /M option to pcretest to show the memory requirement of a specific
5066 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5068 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5069 compiled in such a way that the backtracking after subsequent failure was
5070 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5071 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5074 Version 2.03 02-Feb-99
5075 ----------------------
5077 1. Fixed typo and small mistake in man page.
5079 2. Added 4th condition (GPL supersedes if conflict) and created separate
5080 LICENCE file containing the conditions.
5082 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5083 Perl, that is the internal \ allows the delimiter to be included in the
5084 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5085 the final delimiter, add \ to the end of the pattern (to test the error).
5087 4. Added the convenience functions for extracting substrings after a successful
5088 match. Updated pcretest to make it able to test these functions.
5091 Version 2.02 14-Jan-99
5092 ----------------------
5094 1. Initialized the working variables associated with each extraction so that
5095 their saving and restoring doesn't refer to uninitialized store.
5097 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5098 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5101 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5102 calls, and wasn't printing the correct value for compiling calls. Increased the
5103 default value of LOOPREPEAT, and the number of significant figures in the
5106 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5108 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5109 a building problem on Windows NT with a FAT file system.
5112 Version 2.01 21-Oct-98
5113 ----------------------
5115 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5116 to character tables built by pcre_maketables() in the current locale. If NULL
5117 is passed, the default tables are used.
5120 Version 2.00 24-Sep-98
5121 ----------------------
5123 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5126 2. Allow quantification of (?>) groups, and make it work correctly.
5128 3. The first character computation wasn't working for (?>) groups.
5130 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5131 end of the subject) and add 5.005's \z, which really does match only at the
5132 very end of the subject.
5134 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5136 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5137 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5138 localized options. All options to pcre_study() were also removed.
5140 7. Add other new features from 5.005:
5142 $(?<= positive lookbehind
5143 $(?<! negative lookbehind
5144 (?imsx-imsx) added the unsetting capability
5145 such a setting is global if at outer level; local otherwise
5146 (?imsx-imsx:) non-capturing groups with option setting
5147 (?(cond)re|re) conditional pattern matching
5149 A backreference to itself in a repeated group matches the previous
5152 8. General tidying up of studying (both automatic and via "study")
5153 consequential on the addition of new assertions.
5155 9. As in 5.005, unlimited repeated groups that could match an empty substring
5156 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5157 runtime if any iteration does actually match an empty substring.
5159 10. Include the RunTest script in the distribution.
5161 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5162 discrepancies, some of which were old and were also with respect to 5.004. They
5163 have now been fixed.
5166 Version 1.09 28-Apr-98
5167 ----------------------
5169 1. A negated single character class followed by a quantifier with a minimum
5170 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5171 program crashes, or just wrong answers. This did not apply to negated classes
5172 containing more than one character, or to minima other than one.
5175 Version 1.08 27-Mar-98
5176 ----------------------
5178 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5180 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5181 latter must appear before anything that relies on it in the pattern.
5184 Version 1.07 16-Feb-98
5185 ----------------------
5187 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5188 repeat of a potentially empty string).
5191 Version 1.06 23-Jan-98
5192 ----------------------
5194 1. Added Markus Oberhumer's little patches for C++.
5196 2. Literal strings longer than 255 characters were broken.
5199 Version 1.05 23-Dec-97
5200 ----------------------
5202 1. Negated character classes containing more than one character were failing if
5203 PCRE_CASELESS was set at run time.
5206 Version 1.04 19-Dec-97
5207 ----------------------
5209 1. Corrected the man page, where some "const" qualifiers had been omitted.
5211 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5214 3. Fixed memory leak which occurred when a regex with back references was
5215 matched with an offsets vector that wasn't big enough. The temporary memory
5216 that is used in this case wasn't being freed if the match failed.
5218 4. Tidied pcretest to ensure it frees memory that it gets.
5220 5. Temporary memory was being obtained in the case where the passed offsets
5221 vector was exactly big enough.
5223 6. Corrected definition of offsetof() from change 5 below.
5225 7. I had screwed up change 6 below and broken the rules for the use of
5226 setjmp(). Now fixed.
5229 Version 1.03 18-Dec-97
5230 ----------------------
5232 1. A erroneous regex with a missing opening parenthesis was correctly
5233 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5236 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5237 it was reported that one broken compiler failed on the former because "code" is
5238 also an independent variable.
5240 3. The erroneous regex a[]b caused an array overrun reference.
5242 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5243 fail on data ending with that character. (It was going on too far, and checking
5244 the next character, typically a binary zero.) This was specific to the
5245 optimized code for single-character negative classes.
5247 5. Added a contributed patch from the TIN world which does the following:
5249 + Add an undef for memmove, in case the the system defines a macro for it.
5251 + Add a definition of offsetof(), in case there isn't one. (I don't know
5252 the reason behind this - offsetof() is part of the ANSI standard - but
5255 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5256 most of the places where whitespace preceded '#'. I have given up and
5257 allowed the remaining 2 cases to be at the margin.
5259 + Rename some variables in pcre to eliminate shadowing. This seems very
5260 pedantic, but does no harm, of course.
5262 6. Moved the call to setjmp() into its own function, to get rid of warnings
5263 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5265 7. Constructs such as \d{8,} were compiling into the equivalent of
5266 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5267 outcome, but in this particular case used more store than had been allocated,
5268 which caused the bug to be discovered because it threw up an internal error.
5270 8. The debugging code in both pcre and pcretest for outputting the compiled
5271 form of a regex was going wrong in the case of back references followed by
5272 curly-bracketed repeats.
5275 Version 1.02 12-Dec-97
5276 ----------------------
5278 1. Typos in pcre.3 and comments in the source fixed.
5280 2. Applied a contributed patch to get rid of places where it used to remove
5281 'const' from variables, and fixed some signed/unsigned and uninitialized
5284 3. Added the "runtest" target to Makefile.
5286 4. Set default compiler flag to -O2 rather than just -O.
5289 Version 1.01 19-Nov-97
5290 ----------------------
5292 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5293 like /([ab]*)*/, that is, for classes with more than one character in them.
5295 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5296 as /((?>a*))*/ (a PCRE_EXTRA facility).
5299 Version 1.00 18-Nov-97
5300 ----------------------
5302 1. Added compile-time macros to support systems such as SunOS4 which don't have
5303 memmove() or strerror() but have other things that can be used instead.
5305 2. Arranged that "make clean" removes the executables.
5308 Version 0.99 27-Oct-97
5309 ----------------------
5311 1. Fixed bug in code for optimizing classes with only one character. It was
5312 initializing a 32-byte map regardless, which could cause it to run off the end
5313 of the memory it had got.
5315 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5318 Version 0.98 22-Oct-97
5319 ----------------------
5321 1. Fixed bug in code for handling temporary memory usage when there are more
5322 back references than supplied space in the ovector. This could cause segfaults.
5325 Version 0.97 21-Oct-97
5326 ----------------------
5328 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5330 2. Optimized negated single characters not to use a bit map.
5332 3. Brought error texts together as macro definitions; clarified some of them;
5333 fixed one that was wrong - it said "range out of order" when it meant "invalid
5336 4. Changed some char * arguments to const char *.
5338 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5340 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5344 Version 0.96 16-Oct-97
5345 ----------------------
5347 1. Added a simple "pgrep" utility to the distribution.
5349 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5350 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5351 where "ddd" means "one or more decimal digits".
5353 3. Fixed serious bug. If a pattern had a back reference, but the call to
5354 pcre_exec() didn't supply a large enough ovector to record the related
5355 identifying subpattern, the match always failed. PCRE now remembers the number
5356 of the largest back reference, and gets some temporary memory in which to save
5357 the offsets during matching if necessary, in order to ensure that
5358 backreferences always work.
5360 4. Increased the compatibility with Perl in a number of ways:
5362 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5363 to request this handling. The option can be set at compile or exec time.
5365 (b) $ matches before a terminating newline by default; an option
5366 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5367 mode). The option can be set at compile or exec time.
5369 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5370 the same as Perl's. If the decimal number it represents is less than 10
5371 or there aren't that many previous left capturing parentheses, an octal
5372 escape is read. Inside a character class, it's always an octal escape,
5373 even if it is a single digit.
5375 (d) An escaped but undefined alphabetic character is taken as a literal,
5376 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5379 (e) {0} is now permitted. (The previous item is removed from the compiled
5382 5. Changed all the names of code files so that the basic parts are no longer
5383 than 10 characters, and abolished the teeny "globals.c" file.
5385 6. Changed the handling of character classes; they are now done with a 32-byte
5388 7. Added the -d and /D options to pcretest to make it possible to look at the
5389 internals of compilation without having to recompile pcre.
5392 Version 0.95 23-Sep-97
5393 ----------------------
5395 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5396 \x20 at the start of a run of normal characters. These were being treated as
5397 real characters, instead of the source characters being re-checked.
5400 Version 0.94 18-Sep-97
5401 ----------------------
5403 1. The functions are now thread-safe, with the caveat that the global variables
5404 containing pointers to malloc() and free() or alternative functions are the
5405 same for all threads.
5407 2. Get pcre_study() to generate a bitmap of initial characters for non-
5408 anchored patterns when this is possible, and use it if passed to pcre_exec().
5411 Version 0.93 15-Sep-97
5412 ----------------------
5414 1. /(b)|(:+)/ was computing an incorrect first character.
5416 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5417 but not actually doing anything yet.
5419 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5420 as Perl does (e.g. [-az] or [az-]).
5422 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5423 all possible positions.
5425 5. Split up into different modules to avoid including unneeded functions in a
5426 compiled binary. However, compile and exec are still in one module. The "study"
5427 function is split off.
5429 6. The character tables are now in a separate module whose source is generated
5430 by an auxiliary program - but can then be edited by hand if required. There are
5431 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5432 toupper() in the code.
5434 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5435 make them global. Abolish the function for setting them, as the caller can now
5439 Version 0.92 11-Sep-97
5440 ----------------------
5442 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5443 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5445 2. Caseless matching was not working in character classes if the characters in
5446 the pattern were in upper case.
5448 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5450 4. Make PCRE_ANCHORED public and accept as a compile option.
5452 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5453 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5456 6. Give an error if bad option bits passed at compile or run time.
5458 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5459 pcretest to cause it to pass that flag.
5461 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5462 options, and the first character, if set.
5464 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5467 Version 0.91 10-Sep-97
5468 ----------------------
5470 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5471 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5473 2. PCRE was looping on encountering an indefinitely repeated back reference to
5474 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5475 Perl does - treats the match as successful.