/[pcre]/code/trunk/ChangeLog
ViewVC logotype

Diff of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 319 by ph10, Mon Jan 28 15:27:45 2008 UTC revision 1338 by zherczeg, Fri Jun 14 07:54:36 2013 UTC
# Line 1  Line 1 
1  ChangeLog for PCRE  ChangeLog for PCRE
2  ------------------  ------------------
3    
4    Version 8.34 xx-xxxx-201x
5    --------------------------
6    
7    1.  Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
8        executable memory. Patch inspired by Carsten Klein.
9    
10    
11    Version 8.33 28-May-2013
12    --------------------------
13    
14    1.  Added 'U' to some constants that are compared to unsigned integers, to
15        avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
16        variables that are added to signed variables, to ensure the result is
17        signed and can be negated.
18    
19    2.  Applied patch by Daniel Richard G for quashing MSVC warnings to the
20        CMake config files.
21    
22    3.  Revise the creation of config.h.generic so that all boolean macros are
23        #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
24        overriding via -D on the command line possible.
25    
26    4.  Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
27        to unsigned int is reported to make a quite noticeable speed difference in
28        a specific Windows environment. Testing on Linux did also appear to show
29        some benefit (and it is clearly not harmful). Also fixed the definition of
30        Xop which should be unsigned.
31    
32    5.  Related to (4), changing the definition of the intermediate variable cc
33        in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
34        improvements.
35    
36    6.  Fix forward search in JIT when link size is 3 or greater. Also removed some
37        unnecessary spaces.
38    
39    7.  Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
40        and later.
41    
42    8.  Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
43    
44    9.  Optimizing fast_forward_start_bits in JIT.
45    
46    10. Adding support for callouts in JIT, and fixing some issues revealed
47        during this work. Namely:
48    
49        (a) Unoptimized capturing brackets incorrectly reset on backtrack.
50    
51        (b) Minimum length was not checked before the matching is started.
52    
53    11. The value of capture_last that is passed to callouts was incorrect in some
54        cases when there was a capture on one path that was subsequently abandoned
55        after a backtrack. Also, the capture_last value is now reset after a
56        recursion, since all captures are also reset in this case.
57    
58    12. The interpreter no longer returns the "too many substrings" error in the
59        case when an overflowing capture is in a branch that is subsequently
60        abandoned after a backtrack.
61    
62    13. In the pathological case when an offset vector of size 2 is used, pcretest
63        now prints out the matched string after a yield of 0 or 1.
64    
65    14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
66        Only supported by the JIT compiler at the moment.
67    
68    15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
69    
70    16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
71        offset of the starting point of the matching process, provided the offsets
72        vector is large enough.
73    
74    17. The \A escape now records a lookbehind value of 1, though its execution
75        does not actually inspect the previous character. This is to ensure that,
76        in partial multi-segment matching, at least one character from the old
77        segment is retained when a new segment is processed. Otherwise, if there
78        are no lookbehinds in the pattern, \A might match incorrectly at the start
79        of a new segment.
80    
81    18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
82    
83    19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
84        gives some modest performance improvement in 8-bit mode.
85    
86    20. Added the PCRE-specific property \p{Xuc} for matching characters that can
87        be expressed in certain programming languages using Universal Character
88        Names.
89    
90    21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
91        which points out that "non characters" are not "characters that may not
92        appear in Unicode strings" but rather "characters that are reserved for
93        internal use and have only local meaning".
94    
95    22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
96        there was a conditional group that depended on an assertion, if the
97        assertion was false, the callout that immediately followed the alternation
98        in the condition was skipped when pcre_exec() was used for matching.
99    
100    23. Allow an explicit callout to be inserted before an assertion that is the
101        condition for a conditional group, for compatibility with automatic
102        callouts, which always insert a callout at this point.
103    
104    24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
105        confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
106    
107    25. (*PRUNE) is now supported by the JIT compiler.
108    
109    26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
110    
111    27. Fix the case where there are two or more SKIPs with arguments that may be
112        ignored.
113    
114    28. (*SKIP) is now supported by the JIT compiler.
115    
116    29. (*THEN) is now supported by the JIT compiler.
117    
118    30. Update RunTest with additional test selector options.
119    
120    31. The way PCRE handles backtracking verbs has been changed in two ways.
121    
122        (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
123        SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
124        backtracking. In some cases this makes it more Perl-compatible, but Perl's
125        rather obscure rules do not always do the same thing.
126    
127        (2) Previously, backtracking verbs were confined within assertions. This is
128        no longer the case for positive assertions, except for (*ACCEPT). Again,
129        this sometimes improves Perl compatibility, and sometimes does not.
130    
131    32. A number of tests that were in test 2 because Perl did things differently
132        have been moved to test 1, because either Perl or PCRE has changed, and
133        these tests are now compatible.
134    
135    32. Backtracking control verbs are now handled in the same way in JIT and
136        interpreter.
137    
138    33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
139        contained a forward subroutine reference caused a compile error.
140    
141    34. Auto-detect and optimize limited repetitions in JIT.
142    
143    35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
144        blocking (*UTF) etc.
145    
146    36. In the interpreter, maximizing pattern repetitions for characters and
147        character types now use tail recursion, which reduces stack usage.
148    
149    37. The value of the max lookbehind was not correctly preserved if a compiled
150        and saved regex was reloaded on a host of different endianness.
151    
152    38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
153        of the compiled pattern block, expand the flags field from 16 to 32 bits
154        because it was almost full.
155    
156    39. Try madvise first before posix_madvise.
157    
158    40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
159        with a pattern such as ^$. It has taken 4 years for anybody to notice! The
160        original change locked out all matches of empty strings. This has been
161        changed so that one match of an empty string per line is recognized.
162        Subsequent searches on the same line (for colouring or for --only-matching,
163        for example) do not recognize empty strings.
164    
165    41. Applied a user patch to fix a number of spelling mistakes in comments.
166    
167    42. Data lines longer than 65536 caused pcretest to crash.
168    
169    43. Clarified the data type for length and startoffset arguments for pcre_exec
170        and pcre_dfa_exec in the function-specific man pages, where they were
171        explicitly stated to be in bytes, never having been updated. I also added
172        some clarification to the pcreapi man page.
173    
174    44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
175        a segmentation fault.
176    
177    
178    Version 8.32 30-November-2012
179    -----------------------------
180    
181    1.  Improved JIT compiler optimizations for first character search and single
182        character iterators.
183    
184    2.  Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
185        Patch by Daniel Richard G.
186    
187    3.  Single character iterator optimizations in the JIT compiler.
188    
189    4.  Improved JIT compiler optimizations for character ranges.
190    
191    5.  Rename the "leave" variable names to "quit" to improve WinCE compatibility.
192        Reported by Giuseppe D'Angelo.
193    
194    6.  The PCRE_STARTLINE bit, indicating that a match can occur only at the start
195        of a line, was being set incorrectly in cases where .* appeared inside
196        atomic brackets at the start of a pattern, or where there was a subsequent
197        *PRUNE or *SKIP.
198    
199    7.  Improved instruction cache flush for POWER/PowerPC.
200        Patch by Daniel Richard G.
201    
202    8.  Fixed a number of issues in pcregrep, making it more compatible with GNU
203        grep:
204    
205        (a) There is now no limit to the number of patterns to be matched.
206    
207        (b) An error is given if a pattern is too long.
208    
209        (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
210            are now supported.
211    
212        (d) --exclude-from and --include-from (multiple use) have been added.
213    
214        (e) Exclusions and inclusions now apply to all files and directories, not
215            just to those obtained from scanning a directory recursively.
216    
217        (f) Multiple uses of -f and --file-list are now supported.
218    
219        (g) In a Windows environment, the default for -d has been changed from
220            "read" (the GNU grep default) to "skip", because otherwise the presence
221            of a directory in the file list provokes an error.
222    
223        (h) The documentation has been revised and clarified in places.
224    
225    9.  Improve the matching speed of capturing brackets.
226    
227    10. Changed the meaning of \X so that it now matches a Unicode extended
228        grapheme cluster.
229    
230    11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
231        out POSIX threads when JIT support is configured.
232    
233    12. Added support for PCRE_STUDY_EXTRA_NEEDED.
234    
235    13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
236        structure could go wrong in environments where size_t is not the same size
237        as int.
238    
239    14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
240        set.
241    
242    15. The EBCDIC support had decayed; later updates to the code had included
243        explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
244        general tidy up of EBCDIC-related issues, and the documentation was also
245        not quite right. There is now a test that can be run on ASCII systems to
246        check some of the EBCDIC-related things (but is it not a full test).
247    
248    16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
249        in a small tidy to the code.
250    
251    17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
252    
253    18. If the --only-matching (-o) option in pcregrep is specified multiple
254        times, each one causes appropriate output. For example, -o1 -o2 outputs the
255        substrings matched by the 1st and 2nd capturing parentheses. A separating
256        string can be specified by --om-separator (default empty).
257    
258    19. Improving the first n character searches.
259    
260    20. Turn case lists for horizontal and vertical white space into macros so that
261        they are defined only once.
262    
263    21. This set of changes together give more compatible Unicode case-folding
264        behaviour for characters that have more than one other case when UCP
265        support is available.
266    
267        (a) The Unicode property table now has offsets into a new table of sets of
268            three or more characters that are case-equivalent. The MultiStage2.py
269            script that generates these tables (the pcre_ucd.c file) now scans
270            CaseFolding.txt instead of UnicodeData.txt for character case
271            information.
272    
273        (b) The code for adding characters or ranges of characters to a character
274            class has been abstracted into a generalized function that also handles
275            case-independence. In UTF-mode with UCP support, this uses the new data
276            to handle characters with more than one other case.
277    
278        (c) A bug that is fixed as a result of (b) is that codepoints less than 256
279            whose other case is greater than 256 are now correctly matched
280            caselessly. Previously, the high codepoint matched the low one, but not
281            vice versa.
282    
283        (d) The processing of \h, \H, \v, and \ in character classes now makes use
284            of the new class addition function, using character lists defined as
285            macros alongside the case definitions of 20 above.
286    
287        (e) Caseless back references now work with characters that have more than
288            one other case.
289    
290        (f) General caseless matching of characters with more than one other case
291            is supported.
292    
293    22. Unicode character properties were updated from Unicode 6.2.0
294    
295    23. Improved CMake support under Windows. Patch by Daniel Richard G.
296    
297    24. Add support for 32-bit character strings, and UTF-32
298    
299    25. Major JIT compiler update (code refactoring and bugfixing).
300        Experimental Sparc 32 support is added.
301    
302    26. Applied a modified version of Daniel Richard G's patch to create
303        pcre.h.generic and config.h.generic by "make" instead of in the
304        PrepareRelease script.
305    
306    27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
307        pcre_compile.c when checking for a zero character.
308    
309    28. Introducing a native interface for JIT. Through this interface, the compiled
310        machine code can be directly executed. The purpose of this interface is to
311        provide fast pattern matching, so several sanity checks are not performed.
312        However, feature tests are still performed. The new interface provides
313        1.4x speedup compared to the old one.
314    
315    29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
316        the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
317        was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
318    
319    30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
320        greater than 0x7fffffff (which cannot be represented in UTF-8, even under
321        the "old" RFC 2279). Instead, it ended up passing a negative length to
322        pcre_exec().
323    
324    31. Add support for GCC's visibility feature to hide internal functions.
325    
326    32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
327        "unknown -C option" after outputting 0 or 1.
328    
329    33. There is now support for generating a code coverage report for the test
330        suite in environments where gcc is the compiler and lcov is installed. This
331        is mainly for the benefit of the developers.
332    
333    34. If PCRE is built with --enable-valgrind, certain memory regions are marked
334        unaddressable using valgrind annotations, allowing valgrind to detect
335        invalid memory accesses. This is mainly for the benefit of the developers.
336    
337    25. (*UTF) can now be used to start a pattern in any of the three libraries.
338    
339    26. Give configure error if --enable-cpp but no C++ compiler found.
340    
341    
342    Version 8.31 06-July-2012
343    -------------------------
344    
345    1.  Fixing a wrong JIT test case and some compiler warnings.
346    
347    2.  Removed a bashism from the RunTest script.
348    
349    3.  Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
350        to unsigned type, result still unsigned" that was given by an MS compiler
351        on encountering the code "-sizeof(xxx)".
352    
353    4.  Partial matching support is added to the JIT compiler.
354    
355    5.  Fixed several bugs concerned with partial matching of items that consist
356        of more than one character:
357    
358        (a) /^(..)\1/ did not partially match "aba" because checking references was
359            done on an "all or nothing" basis. This also applied to repeated
360            references.
361    
362        (b) \R did not give a hard partial match if \r was found at the end of the
363            subject.
364    
365        (c) \X did not give a hard partial match after matching one or more
366            characters at the end of the subject.
367    
368        (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
369            a partial match for the string "\r".
370    
371        (e) When newline was set to CRLF, the metacharacter "." did not recognize
372            a partial match for a CR character at the end of the subject string.
373    
374    6.  If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
375        running pcretest, the text "(JIT)" added to the output whenever JIT is
376        actually used to run the match.
377    
378    7.  Individual JIT compile options can be set in pcretest by following -s+[+]
379        or /S+[+] with a digit between 1 and 7.
380    
381    8.  OP_NOT now supports any UTF character not just single-byte ones.
382    
383    9.  (*MARK) control verb is now supported by the JIT compiler.
384    
385    10. The command "./RunTest list" lists the available tests without actually
386        running any of them. (Because I keep forgetting what they all are.)
387    
388    11. Add PCRE_INFO_MAXLOOKBEHIND.
389    
390    12. Applied a (slightly modified) user-supplied patch that improves performance
391        when the heap is used for recursion (compiled with --disable-stack-for-
392        recursion). Instead of malloc and free for each heap frame each time a
393        logical recursion happens, frames are retained on a chain and re-used where
394        possible. This sometimes gives as much as 30% improvement.
395    
396    13. As documented, (*COMMIT) is now confined to within a recursive subpattern
397        call.
398    
399    14. As documented, (*COMMIT) is now confined to within a positive assertion.
400    
401    15. It is now possible to link pcretest with libedit as an alternative to
402        libreadline.
403    
404    16. (*COMMIT) control verb is now supported by the JIT compiler.
405    
406    17. The Unicode data tables have been updated to Unicode 6.1.0.
407    
408    18. Added --file-list option to pcregrep.
409    
410    19. Added binary file support to pcregrep, including the -a, --binary-files,
411        -I, and --text options.
412    
413    20. The madvise function is renamed for posix_madvise for QNX compatibility
414        reasons. Fixed by Giuseppe D'Angelo.
415    
416    21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
417    
418    22. Changed the option for creating symbolic links for 16-bit man pages from
419        -s to -sf so that re-installing does not cause issues.
420    
421    23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
422    
423    24. Fixed a very old bug in pcretest that caused errors with restarted DFA
424        matches in certain environments (the workspace was not being correctly
425        retained). Also added to pcre_dfa_exec() a simple plausibility check on
426        some of the workspace data at the beginning of a restart.
427    
428    25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
429        was not doing so when it should - probably a typo introduced by SVN 528
430        (change 8.10/14).
431    
432    26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
433        \w+ when the character tables indicated that \x{c4} was a word character.
434        There were several related cases, all because the tests for doing a table
435        lookup were testing for characters less than 127 instead of 255.
436    
437    27. If a pattern contains capturing parentheses that are not used in a match,
438        their slots in the ovector are set to -1. For those that are higher than
439        any matched groups, this happens at the end of processing. In the case when
440        there were back references that the ovector was too small to contain
441        (causing temporary malloc'd memory to be used during matching), and the
442        highest capturing number was not used, memory off the end of the ovector
443        was incorrectly being set to -1. (It was using the size of the temporary
444        memory instead of the true size.)
445    
446    28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
447        ovector size, it uses memory at the end of the block that it has got.
448    
449    29. Check for an overlong MARK name and give an error at compile time. The
450        limit is 255 for the 8-bit library and 65535 for the 16-bit library.
451    
452    30. JIT compiler update.
453    
454    31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
455        Rill for the patch.
456    
457    32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
458    
459    33. Variable renamings in the PCRE-JIT compiler. No functionality change.
460    
461    34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
462        SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
463        was enabled.
464    
465    35. Improve JIT code generation for greedy plus quantifier.
466    
467    36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
468        1 to "aa" instead of to an empty string. The bug affected repeated groups
469        that could potentially match an empty string.
470    
471    37. Optimizing single character iterators in JIT.
472    
473    38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
474        the same checks as \x{...} characters in non-JavaScript mode. Specifically,
475        codepoints that are too big for the mode are faulted, and in a UTF mode,
476        disallowed codepoints are also faulted.
477    
478    39. If PCRE was compiled with UTF support, in three places in the DFA
479        matcher there was code that should only have been obeyed in UTF mode, but
480        was being obeyed unconditionally. In 8-bit mode this could cause incorrect
481        processing when bytes with values greater than 127 were present. In 16-bit
482        mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
483        both cases the values are those that cannot be the first data item in a UTF
484        character. The three items that might have provoked this were recursions,
485        possessively repeated groups, and atomic groups.
486    
487    40. Ensure that libpcre is explicitly listed in the link commands for pcretest
488        and pcregrep, because some OS require shared objects to be explicitly
489        passed to ld, causing the link step to fail if they are not.
490    
491    41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
492        mode, patterns that started with \h* or \R* might be incorrectly matched.
493    
494    
495    Version 8.30 04-February-2012
496    -----------------------------
497    
498    1.  Renamed "isnumber" as "is_a_number" because in some Mac environments this
499        name is defined in ctype.h.
500    
501    2.  Fixed a bug in fixed-length calculation for lookbehinds that would show up
502        only in quite long subpatterns.
503    
504    3.  Removed the function pcre_info(), which has been obsolete and deprecated
505        since it was replaced by pcre_fullinfo() in February 2000.
506    
507    4.  For a non-anchored pattern, if (*SKIP) was given with a name that did not
508        match a (*MARK), and the match failed at the start of the subject, a
509        reference to memory before the start of the subject could occur. This bug
510        was introduced by fix 17 of release 8.21.
511    
512    5.  A reference to an unset group with zero minimum repetition was giving
513        totally wrong answers (in non-JavaScript-compatibility mode). For example,
514        /(another)?(\1?)test/ matched against "hello world test". This bug was
515        introduced in release 8.13.
516    
517    6.  Add support for 16-bit character strings (a large amount of work involving
518        many changes and refactorings).
519    
520    7.  RunGrepTest failed on msys because \r\n was replaced by whitespace when the
521        command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
522        from a file.
523    
524    8.  Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
525        rounding is not applied in this particular case).
526    
527    9.  The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
528        if they appear, or are escaped, in patterns.
529    
530    10. Get rid of a number of -Wunused-but-set-variable warnings.
531    
532    11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
533        "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
534        Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
535        also returns the mark "x". This bug applied to capturing parentheses,
536        non-capturing parentheses, and atomic parentheses. It also applied to some
537        assertions.
538    
539    12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
540        information out of configure.ac instead of relying on pcre.h.generic, which
541        is not stored in the repository.
542    
543    13. Applied Dmitry V. Levin's patch for a more portable method for linking with
544        -lreadline.
545    
546    14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
547    
548    15. Applied Graycode's patch to put the top-level frame on the stack rather
549        than the heap when not using the stack for recursion. This gives a
550        performance improvement in many cases when recursion is not deep.
551    
552    16. Experimental code added to "pcretest -C" to output the stack frame size.
553    
554    
555    Version 8.21 12-Dec-2011
556    ------------------------
557    
558    1.  Updating the JIT compiler.
559    
560    2.  JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
561        are added as well.
562    
563    3.  Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
564        PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
565        calling _pcre_jit_exec. Some extra comments are added.
566    
567    4.  (*MARK) settings inside atomic groups that do not contain any capturing
568        parentheses, for example, (?>a(*:m)), were not being passed out. This bug
569        was introduced by change 18 for 8.20.
570    
571    5.  Supporting of \x, \U and \u in JavaScript compatibility mode based on the
572        ECMA-262 standard.
573    
574    6.  Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
575        erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
576        This bug was probably introduced by change 9 of 8.13.
577    
578    7.  While fixing 6 above, I noticed that a number of other items were being
579        incorrectly rejected as "not fixed length". This arose partly because newer
580        opcodes had not been added to the fixed-length checking code. I have (a)
581        corrected the bug and added tests for these items, and (b) arranged for an
582        error to occur if an unknown opcode is encountered while checking for fixed
583        length instead of just assuming "not fixed length". The items that were
584        rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
585        (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
586        repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
587    
588    8.  A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
589        being incorrectly compiled and would have given unpredicatble results.
590    
591    9.  A possessively repeated subpattern with minimum repeat count greater than
592        one behaved incorrectly. For example, (A){2,}+ behaved as if it was
593        (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
594        the first (A) could occur when it should not.
595    
596    10. Add a cast and remove a redundant test from the code.
597    
598    11. JIT should use pcre_malloc/pcre_free for allocation.
599    
600    12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
601        best practice nowadays, and helps with cross-compiling. (If the exec_prefix
602        is anything other than /usr, -L is still shown).
603    
604    13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
605    
606    14. Perl does not support \N without a following name in a [] class; PCRE now
607        also gives an error.
608    
609    15. If a forward reference was repeated with an upper limit of around 2000,
610        it caused the error "internal error: overran compiling workspace". The
611        maximum number of forward references (including repeats) was limited by the
612        internal workspace, and dependent on the LINK_SIZE. The code has been
613        rewritten so that the workspace expands (via pcre_malloc) if necessary, and
614        the default depends on LINK_SIZE. There is a new upper limit (for safety)
615        of around 200,000 forward references. While doing this, I also speeded up
616        the filling in of repeated forward references.
617    
618    16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
619        incorrectly expecting the subject to contain another "a" after the start.
620    
621    17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
622        in the match, the SKIP should be ignored. This was not happening; instead
623        the SKIP was being treated as NOMATCH. For patterns such as
624        /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
625        tested.
626    
627    18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
628        now much more compatible with Perl, in particular in cases where the result
629        is a non-match for a non-anchored pattern. For example, if
630        /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
631        "m", where previously it did not return a name. A side effect of this
632        change is that for partial matches, the last encountered mark name is
633        returned, as for non matches. A number of tests that were previously not
634        Perl-compatible have been moved into the Perl-compatible test files. The
635        refactoring has had the pleasing side effect of removing one argument from
636        the match() function, thus reducing its stack requirements.
637    
638    19. If the /S+ option was used in pcretest to study a pattern using JIT,
639        subsequent uses of /S (without +) incorrectly behaved like /S+.
640    
641    21. Retrieve executable code size support for the JIT compiler and fixing
642        some warnings.
643    
644    22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
645        not work when the shorter character appeared right at the end of the
646        subject string.
647    
648    23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
649        systems.
650    
651    24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
652        output it when the /M option is used in pcretest.
653    
654    25. The CheckMan script was not being included in the distribution. Also, added
655        an explicit "perl" to run Perl scripts from the PrepareRelease script
656        because this is reportedly needed in Windows.
657    
658    26. If study data was being save in a file and studying had not found a set of
659        "starts with" bytes for the pattern, the data written to the file (though
660        never used) was taken from uninitialized memory and so caused valgrind to
661        complain.
662    
663    27. Updated RunTest.bat as provided by Sheri Pierce.
664    
665    28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
666    
667    29. Computation of memory usage for the table of capturing group names was
668        giving an unnecessarily large value.
669    
670    
671    Version 8.20 21-Oct-2011
672    ------------------------
673    
674    1.  Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
675        a POSIX class. After further experiments with Perl, which convinced me that
676        Perl has bugs and confusions, a closing square bracket is no longer allowed
677        in a POSIX name. This bug also affected patterns with classes that started
678        with full stops.
679    
680    2.  If a pattern such as /(a)b|ac/ is matched against "ac", there is no
681        captured substring, but while checking the failing first alternative,
682        substring 1 is temporarily captured. If the output vector supplied to
683        pcre_exec() was not big enough for this capture, the yield of the function
684        was still zero ("insufficient space for captured substrings"). This cannot
685        be totally fixed without adding another stack variable, which seems a lot
686        of expense for a edge case. However, I have improved the situation in cases
687        such as /(a)(b)x|abc/ matched against "abc", where the return code
688        indicates that fewer than the maximum number of slots in the ovector have
689        been set.
690    
691    3.  Related to (2) above: when there are more back references in a pattern than
692        slots in the output vector, pcre_exec() uses temporary memory during
693        matching, and copies in the captures as far as possible afterwards. It was
694        using the entire output vector, but this conflicts with the specification
695        that only 2/3 is used for passing back captured substrings. Now it uses
696        only the first 2/3, for compatibility. This is, of course, another edge
697        case.
698    
699    4.  Zoltan Herczeg's just-in-time compiler support has been integrated into the
700        main code base, and can be used by building with --enable-jit. When this is
701        done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
702        runtime --no-jit option is given.
703    
704    5.  When the number of matches in a pcre_dfa_exec() run exactly filled the
705        ovector, the return from the function was zero, implying that there were
706        other matches that did not fit. The correct "exactly full" value is now
707        returned.
708    
709    6.  If a subpattern that was called recursively or as a subroutine contained
710        (*PRUNE) or any other control that caused it to give a non-standard return,
711        invalid errors such as "Error -26 (nested recursion at the same subject
712        position)" or even infinite loops could occur.
713    
714    7.  If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
715        computing the minimum length on reaching *ACCEPT, and so ended up with the
716        wrong value of 1 rather than 0. Further investigation indicates that
717        computing a minimum subject length in the presence of *ACCEPT is difficult
718        (think back references, subroutine calls), and so I have changed the code
719        so that no minimum is registered for a pattern that contains *ACCEPT.
720    
721    8.  If (*THEN) was present in the first (true) branch of a conditional group,
722        it was not handled as intended. [But see 16 below.]
723    
724    9.  Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
725        Sheri Pierce.
726    
727    10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
728        the first byte in a match must be "a".
729    
730    11. Change 17 for 8.13 increased the recursion depth for patterns like
731        /a(?:.)*?a/ drastically. I've improved things by remembering whether a
732        pattern contains any instances of (*THEN). If it does not, the old
733        optimizations are restored. It would be nice to do this on a per-group
734        basis, but at the moment that is not feasible.
735    
736    12. In some environments, the output of pcretest -C is CRLF terminated. This
737        broke RunTest's code that checks for the link size. A single white space
738        character after the value is now allowed for.
739    
740    13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
741        For "fr", it uses the Windows-specific input and output files.
742    
743    14. If (*THEN) appeared in a group that was called recursively or as a
744        subroutine, it did not work as intended. [But see next item.]
745    
746    15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
747        pattern fragments (but not containing any | characters). If A and B are
748        matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
749        was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
750        D. In other words, Perl considers parentheses that do not contain any |
751        characters to be part of a surrounding alternative, whereas PCRE was
752        treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
753        differently. PCRE now behaves in the same way as Perl, except in the case
754        of subroutine/recursion calls such as (?1) which have in any case always
755        been different (but PCRE had them first :-).
756    
757    16. Related to 15 above: Perl does not treat the | in a conditional group as
758        creating alternatives. Such a group is treated in the same way as an
759        ordinary group without any | characters when processing (*THEN). PCRE has
760        been changed to match Perl's behaviour.
761    
762    17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
763        RunGrepTest script failed.
764    
765    18. Change 22 for version 13 caused atomic groups to use more stack. This is
766        inevitable for groups that contain captures, but it can lead to a lot of
767        stack use in large patterns. The old behaviour has been restored for atomic
768        groups that do not contain any capturing parentheses.
769    
770    19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
771        suppress the check for a minimum subject length at run time. (If it was
772        given to pcre_exec() or pcre_dfa_exec() it did work.)
773    
774    20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
775        fail to work when decoding hex characters in data strings in EBCDIC
776        environments.
777    
778    21. It appears that in at least one Mac OS environment, the isxdigit() function
779        is implemented as a macro that evaluates to its argument more than once,
780        contravening the C 90 Standard (I haven't checked a later standard). There
781        was an instance in pcretest which caused it to go wrong when processing
782        \x{...} escapes in subject strings. The has been rewritten to avoid using
783        things like p++ in the argument of isxdigit().
784    
785    
786    Version 8.13 16-Aug-2011
787    ------------------------
788    
789    1.  The Unicode data tables have been updated to Unicode 6.0.0.
790    
791    2.  Two minor typos in pcre_internal.h have been fixed.
792    
793    3.  Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
794        pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
795        in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
796    
797    4.  There were a number of related bugs in the code for matching backrefences
798        caselessly in UTF-8 mode when codes for the characters concerned were
799        different numbers of bytes. For example, U+023A and U+2C65 are an upper
800        and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
801        (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
802        code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
803        2-byte code at the end of the subject (it thought there wasn't enough data
804        left).
805    
806    5.  Comprehensive information about what went wrong is now returned by
807        pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
808        as the output vector has at least 2 elements. The offset of the start of
809        the failing character and a reason code are placed in the vector.
810    
811    6.  When the UTF-8 string check fails for pcre_compile(), the offset that is
812        now returned is for the first byte of the failing character, instead of the
813        last byte inspected. This is an incompatible change, but I hope it is small
814        enough not to be a problem. It makes the returned offset consistent with
815        pcre_exec() and pcre_dfa_exec().
816    
817    7.  pcretest now gives a text phrase as well as the error number when
818        pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
819        failure, the offset and reason code are output.
820    
821    8.  When \R was used with a maximizing quantifier it failed to skip backwards
822        over a \r\n pair if the subsequent match failed. Instead, it just skipped
823        back over a single character (\n). This seems wrong (because it treated the
824        two characters as a single entity when going forwards), conflicts with the
825        documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
826        behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
827        has been changed.
828    
829    9.  Some internal refactoring has changed the processing so that the handling
830        of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
831        time (the PCRE_DOTALL option was changed this way some time ago: version
832        7.7 change 16). This has made it possible to abolish the OP_OPT op code,
833        which was always a bit of a fudge. It also means that there is one less
834        argument for the match() function, which reduces its stack requirements
835        slightly. This change also fixes an incompatibility with Perl: the pattern
836        (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
837    
838    10. More internal refactoring has drastically reduced the number of recursive
839        calls to match() for possessively repeated groups such as (abc)++ when
840        using pcre_exec().
841    
842    11. While implementing 10, a number of bugs in the handling of groups were
843        discovered and fixed:
844    
845        (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
846        (a|)*(?1) gave a compile-time internal error.
847        ((a|)+)+  did not notice that the outer group could match an empty string.
848        (^a|^)+   was not marked as anchored.
849        (.*a|.*)+ was not marked as matching at start or after a newline.
850    
851    12. Yet more internal refactoring has removed another argument from the match()
852        function. Special calls to this function are now indicated by setting a
853        value in a variable in the "match data" data block.
854    
855    13. Be more explicit in pcre_study() instead of relying on "default" for
856        opcodes that mean there is no starting character; this means that when new
857        ones are added and accidentally left out of pcre_study(), testing should
858        pick them up.
859    
860    14. The -s option of pcretest has been documented for ages as being an old
861        synonym of -m (show memory usage). I have changed it to mean "force study
862        for every regex", that is, assume /S for every regex. This is similar to -i
863        and -d etc. It's slightly incompatible, but I'm hoping nobody is still
864        using it. It makes it easier to run collections of tests with and without
865        study enabled, and thereby test pcre_study() more easily. All the standard
866        tests are now run with and without -s (but some patterns can be marked as
867        "never study" - see 20 below).
868    
869    15. When (*ACCEPT) was used in a subpattern that was called recursively, the
870        restoration of the capturing data to the outer values was not happening
871        correctly.
872    
873    16. If a recursively called subpattern ended with (*ACCEPT) and matched an
874        empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
875        pattern had matched an empty string, and so incorrectly returned a no
876        match.
877    
878    17. There was optimizing code for the last branch of non-capturing parentheses,
879        and also for the obeyed branch of a conditional subexpression, which used
880        tail recursion to cut down on stack usage. Unfortunately, now that there is
881        the possibility of (*THEN) occurring in these branches, tail recursion is
882        no longer possible because the return has to be checked for (*THEN). These
883        two optimizations have therefore been removed. [But see 8.20/11 above.]
884    
885    18. If a pattern containing \R was studied, it was assumed that \R always
886        matched two bytes, thus causing the minimum subject length to be
887        incorrectly computed because \R can also match just one byte.
888    
889    19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
890        was incorrectly computed.
891    
892    20. If /S is present twice on a test pattern in pcretest input, it now
893        *disables* studying, thereby overriding the use of -s on the command line
894        (see 14 above). This is necessary for one or two tests to keep the output
895        identical in both cases.
896    
897    21. When (*ACCEPT) was used in an assertion that matched an empty string and
898        PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
899    
900    22. When an atomic group that contained a capturing parenthesis was
901        successfully matched, but the branch in which it appeared failed, the
902        capturing was not being forgotten if a higher numbered group was later
903        captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
904        group 1 to "a", when in fact it should be unset. This applied to multi-
905        branched capturing and non-capturing groups, repeated or not, and also to
906        positive assertions (capturing in negative assertions does not happen
907        in PCRE) and also to nested atomic groups.
908    
909    23. Add the ++ qualifier feature to pcretest, to show the remainder of the
910        subject after a captured substring, to make it easier to tell which of a
911        number of identical substrings has been captured.
912    
913    24. The way atomic groups are processed by pcre_exec() has been changed so that
914        if they are repeated, backtracking one repetition now resets captured
915        values correctly. For example, if ((?>(a+)b)+aabab) is matched against
916        "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
917        "aaa". Previously, it would have been "a". As part of this code
918        refactoring, the way recursive calls are handled has also been changed.
919    
920    25. If an assertion condition captured any substrings, they were not passed
921        back unless some other capturing happened later. For example, if
922        (?(?=(a))a) was matched against "a", no capturing was returned.
923    
924    26. When studying a pattern that contained subroutine calls or assertions,
925        the code for finding the minimum length of a possible match was handling
926        direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
927        group 1 called group 2 while simultaneously a separate group 2 called group
928        1). A stack overflow occurred in this case. I have fixed this by limiting
929        the recursion depth to 10.
930    
931    27. Updated RunTest.bat in the distribution to the version supplied by Tom
932        Fortmann. This supports explicit test numbers on the command line, and has
933        argument validation and error reporting.
934    
935    28. An instance of \X with an unlimited repeat could fail if at any point the
936        first character it looked at was a mark character.
937    
938    29. Some minor code refactoring concerning Unicode properties and scripts
939        should reduce the stack requirement of match() slightly.
940    
941    30. Added the '=' option to pcretest to check the setting of unused capturing
942        slots at the end of the pattern, which are documented as being -1, but are
943        not included in the return count.
944    
945    31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
946        compiled something random. Now it gives a compile-time error (as does
947        Perl).
948    
949    32. A *MARK encountered during the processing of a positive assertion is now
950        recorded and passed back (compatible with Perl).
951    
952    33. If --only-matching or --colour was set on a pcregrep call whose pattern
953        had alternative anchored branches, the search for a second match in a line
954        was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
955        matched the line "0102" twice. The same bug affected patterns that started
956        with a backwards assertion. For example /\b01|\b02/ also matched "0102"
957        twice.
958    
959    34. Previously, PCRE did not allow quantification of assertions. However, Perl
960        does, and because of capturing effects, quantifying parenthesized
961        assertions may at times be useful. Quantifiers are now allowed for
962        parenthesized assertions.
963    
964    35. A minor code tidy in pcre_compile() when checking options for \R usage.
965    
966    36. \g was being checked for fancy things in a character class, when it should
967        just be a literal "g".
968    
969    37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
970        appearance of a nested POSIX class supersedes an apparent external class.
971        For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
972        unescaped square brackets may also appear as part of class names. For
973        example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
974        more like Perl. (But see 8.20/1 above.)
975    
976    38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
977        was because it thought it was \N{name}, which is not supported).
978    
979    39. Add minix to OS list not supporting the -S option in pcretest.
980    
981    40. PCRE tries to detect cases of infinite recursion at compile time, but it
982        cannot analyze patterns in sufficient detail to catch mutual recursions
983        such as ((?1))((?2)). There is now a runtime test that gives an error if a
984        subgroup is called recursively as a subpattern for a second time at the
985        same position in the subject string. In previous releases this might have
986        been caught by the recursion limit, or it might have run out of stack.
987    
988    41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
989        happen only once. PCRE was, however incorrectly giving a compile time error
990        "recursive call could loop indefinitely" because it cannot analyze the
991        pattern in sufficient detail. The compile time test no longer happens when
992        PCRE is compiling a conditional subpattern, but actual runaway loops are
993        now caught at runtime (see 40 above).
994    
995    42. It seems that Perl allows any characters other than a closing parenthesis
996        to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
997        has been changed to be the same.
998    
999    43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1000        as not to get warnings when autogen.sh is called. Also changed
1001        AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1002    
1003    44. To help people who use pcregrep to scan files containing exceedingly long
1004        lines, the following changes have been made:
1005    
1006        (a) The default value of the buffer size parameter has been increased from
1007            8K to 20K. (The actual buffer used is three times this size.)
1008    
1009        (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1010            PCRE is built.
1011    
1012        (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1013            to be set at run time.
1014    
1015        (d) Numerical values in pcregrep options can be followed by K or M, for
1016            example --buffer-size=50K.
1017    
1018        (e) If a line being scanned overflows pcregrep's buffer, an error is now
1019            given and the return code is set to 2.
1020    
1021    45. Add a pointer to the latest mark to the callout data block.
1022    
1023    46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1024        partial match of an empty string instead of no match. This was specific to
1025        the use of ".".
1026    
1027    47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1028        complete match instead of a partial match. This bug was dependent on both
1029        the PCRE_UTF8 and PCRE_DOTALL options being set.
1030    
1031    48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1032        starting byte set, because \b was not being ignored.
1033    
1034    
1035    Version 8.12 15-Jan-2011
1036    ------------------------
1037    
1038    1.  Fixed some typos in the markup of the man pages, and wrote a script that
1039        checks for such things as part of the documentation building process.
1040    
1041    2.  On a big-endian 64-bit system, pcregrep did not correctly process the
1042        --match-limit and --recursion-limit options (added for 8.11). In
1043        particular, this made one of the standard tests fail. (The integer value
1044        went into the wrong half of a long int.)
1045    
1046    3.  If the --colour option was given to pcregrep with -v (invert match), it
1047        did strange things, either producing crazy output, or crashing. It should,
1048        of course, ignore a request for colour when reporting lines that do not
1049        match.
1050    
1051    4.  Another pcregrep bug caused similar problems if --colour was specified with
1052        -M (multiline) and the pattern match finished with a line ending.
1053    
1054    5.  In pcregrep, when a pattern that ended with a literal newline sequence was
1055        matched in multiline mode, the following line was shown as part of the
1056        match. This seems wrong, so I have changed it.
1057    
1058    6.  Another pcregrep bug in multiline mode, when --colour was specified, caused
1059        the check for further matches in the same line (so they could be coloured)
1060        to overrun the end of the current line. If another match was found, it was
1061        incorrectly shown (and then shown again when found in the next line).
1062    
1063    7.  If pcregrep was compiled under Windows, there was a reference to the
1064        function pcregrep_exit() before it was defined. I am assuming this was
1065        the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1066        reported by a user. I've moved the definition above the reference.
1067    
1068    
1069    Version 8.11 10-Dec-2010
1070    ------------------------
1071    
1072    1.  (*THEN) was not working properly if there were untried alternatives prior
1073        to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1074        backtracked to try for "b" instead of moving to the next alternative branch
1075        at the same level (in this case, to look for "c"). The Perl documentation
1076        is clear that when (*THEN) is backtracked onto, it goes to the "next
1077        alternative in the innermost enclosing group".
1078    
1079    2.  (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1080        such as   (A(*COMMIT)B(*THEN)C|D)  any failure after matching A should
1081        result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1082        (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1083        (*THEN).
1084    
1085    3.  If \s appeared in a character class, it removed the VT character from
1086        the class, even if it had been included by some previous item, for example
1087        in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1088        of \s, but is part of the POSIX "space" class.)
1089    
1090    4.  A partial match never returns an empty string (because you can always
1091        match an empty string at the end of the subject); however the checking for
1092        an empty string was starting at the "start of match" point. This has been
1093        changed to the "earliest inspected character" point, because the returned
1094        data for a partial match starts at this character. This means that, for
1095        example, /(?<=abc)def/ gives a partial match for the subject "abc"
1096        (previously it gave "no match").
1097    
1098    5.  Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1099        of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1100        previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1101        has an implication that the given string is incomplete (because a partial
1102        match is preferred over a full match). For this reason, these items now
1103        give a partial match in this situation. [Aside: previously, the one case
1104        /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1105        match rather than a full match, which was wrong by the old rules, but is
1106        now correct.]
1107    
1108    6.  There was a bug in the handling of #-introduced comments, recognized when
1109        PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1110        If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1111        UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1112        scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1113        but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1114        places in pcre_compile().
1115    
1116    7.  Related to (6) above, when pcre_compile() was skipping #-introduced
1117        comments when looking ahead for named forward references to subpatterns,
1118        the only newline sequence it recognized was NL. It now handles newlines
1119        according to the set newline convention.
1120    
1121    8.  SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1122        former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1123        cater for a lack of strerror(). These oversights have been fixed.
1124    
1125    9.  Added --match-limit and --recursion-limit to pcregrep.
1126    
1127    10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1128    
1129    11. When the -o option was used, pcregrep was setting a return code of 1, even
1130        when matches were found, and --line-buffered was not being honoured.
1131    
1132    12. Added an optional parentheses number to the -o and --only-matching options
1133        of pcregrep.
1134    
1135    13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1136        can match an empty string. The code to do it in pcretest and pcredemo
1137        needed fixing:
1138    
1139        (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1140            only one byte after an empty string match just before CRLF (this case
1141            just got forgotten; "any" and "anycrlf" were OK).
1142    
1143        (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1144            mode when an empty string match preceded an ASCII character followed by
1145            a non-ASCII character. (The code for advancing by one character rather
1146            than one byte was nonsense.)
1147    
1148        (c) The pcredemo.c sample program did not have any code at all to handle
1149            the cases when CRLF is a valid newline sequence.
1150    
1151    14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1152        as a starting offset was within the subject string. There is now a new
1153        error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1154        negative or greater than the length of the string. In order to test this,
1155        pcretest is extended to allow the setting of negative starting offsets.
1156    
1157    15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1158        starting offset points to the beginning of a UTF-8 character was
1159        unnecessarily clumsy. I tidied it up.
1160    
1161    16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1162        bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1163    
1164    17. Nobody had reported that the --include_dir option, which was added in
1165        release 7.7 should have been called --include-dir (hyphen, not underscore)
1166        for compatibility with GNU grep. I have changed it to --include-dir, but
1167        left --include_dir as an undocumented synonym, and the same for
1168        --exclude-dir, though that is not available in GNU grep, at least as of
1169        release 2.5.4.
1170    
1171    18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1172        characters from a string of bytes) have been redefined so as not to use
1173        loops, in order to improve performance in some environments. At the same
1174        time, I abstracted some of the common code into auxiliary macros to save
1175        repetition (this should not affect the compiled code).
1176    
1177    19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1178        compile-time error is now given if \c is not followed by an ASCII
1179        character, that is, a byte less than 128. (In EBCDIC mode, the code is
1180        different, and any byte value is allowed.)
1181    
1182    20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1183        START_OPTIMIZE option, which is now allowed at compile time - but just
1184        passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1185        to pcregrep and other applications that have no direct access to PCRE
1186        options. The new /Y option in pcretest sets this option when calling
1187        pcre_compile().
1188    
1189    21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1190        back references. Groups containing recursive back references were forced to
1191        be atomic by that change, but in the case of named groups, the amount of
1192        memory required was incorrectly computed, leading to "Failed: internal
1193        error: code overflow". This has been fixed.
1194    
1195    22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1196        pcretest.c, to avoid build problems in some Borland environments.
1197    
1198    
1199    Version 8.10 25-Jun-2010
1200    ------------------------
1201    
1202    1.  Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1203        THEN.
1204    
1205    2.  (*ACCEPT) was not working when inside an atomic group.
1206    
1207    3.  Inside a character class, \B is treated as a literal by default, but
1208        faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1209        causes the error). The code is unchanged, but I tidied the documentation.
1210    
1211    4.  Inside a character class, PCRE always treated \R and \X as literals,
1212        whereas Perl faults them if its -w option is set. I have changed PCRE so
1213        that it faults them when PCRE_EXTRA is set.
1214    
1215    5.  Added support for \N, which always matches any character other than
1216        newline. (It is the same as "." when PCRE_DOTALL is not set.)
1217    
1218    6.  When compiling pcregrep with newer versions of gcc which may have
1219        FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1220        declared with attribute warn_unused_result" were given. Just casting the
1221        result to (void) does not stop the warnings; a more elaborate fudge is
1222        needed. I've used a macro to implement this.
1223    
1224    7.  Minor change to pcretest.c to avoid a compiler warning.
1225    
1226    8.  Added four artifical Unicode properties to help with an option to make
1227        \s etc use properties (see next item). The new properties are: Xan
1228        (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1229    
1230    9.  Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1231        use Unicode properties. (*UCP) at the start of a pattern can be used to set
1232        this option. Modified pcretest to add /W to test this facility. Added
1233        REG_UCP to make it available via the POSIX interface.
1234    
1235    10. Added --line-buffered to pcregrep.
1236    
1237    11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1238        studied, and the match started with a letter with a code point greater than
1239        127 whose first byte was different to the first byte of the other case of
1240        the letter, the other case of this starting letter was not recognized
1241        (#976).
1242    
1243    12. If a pattern that was studied started with a repeated Unicode property
1244        test, for example, \p{Nd}+, there was the theoretical possibility of
1245        setting up an incorrect bitmap of starting bytes, but fortunately it could
1246        not have actually happened in practice until change 8 above was made (it
1247        added property types that matched character-matching opcodes).
1248    
1249    13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1250        possible starting bytes for non-anchored patterns.
1251    
1252    14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1253        \R, and also a number of cases that involve Unicode properties, both
1254        explicit and implicit when PCRE_UCP is set.
1255    
1256    15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1257        input, it could crash or give wrong results if characters with values
1258        greater than 0xc0 were present in the subject string. (Detail: it assumed
1259        UTF-8 input when processing these items.)
1260    
1261    16. Added a lot of (int) casts to avoid compiler warnings in systems where
1262        size_t is 64-bit (#991).
1263    
1264    17. Added a check for running out of memory when PCRE is compiled with
1265        --disable-stack-for-recursion (#990).
1266    
1267    18. If the last data line in a file for pcretest does not have a newline on
1268        the end, a newline was missing in the output.
1269    
1270    19. The default pcre_chartables.c file recognizes only ASCII characters (values
1271        less than 128) in its various bitmaps. However, there is a facility for
1272        generating tables according to the current locale when PCRE is compiled. It
1273        turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1274        characters, are recognized by isspace() and therefore were getting set in
1275        these tables, and indeed these tables seem to approximate to ISO 8859. This
1276        caused a problem in UTF-8 mode when pcre_study() was used to create a list
1277        of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1278        which of course cannot start UTF-8 characters. I have changed the code so
1279        that only real ASCII characters (less than 128) and the correct starting
1280        bytes for UTF-8 encodings are set for characters greater than 127 when in
1281        UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1282        altogether.)
1283    
1284    20. Added the /T option to pcretest so as to be able to run tests with non-
1285        standard character tables, thus making it possible to include the tests
1286        used for 19 above in the standard set of tests.
1287    
1288    21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1289        reference to a subpattern the other side of a comment that contains an
1290        opening parenthesis caused either an internal compiling error, or a
1291        reference to the wrong subpattern.
1292    
1293    
1294    Version 8.02 19-Mar-2010
1295    ------------------------
1296    
1297    1.  The Unicode data tables have been updated to Unicode 5.2.0.
1298    
1299    2.  Added the option --libs-cpp to pcre-config, but only when C++ support is
1300        configured.
1301    
1302    3.  Updated the licensing terms in the pcregexp.pas file, as agreed with the
1303        original author of that file, following a query about its status.
1304    
1305    4.  On systems that do not have stdint.h (e.g. Solaris), check for and include
1306        inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1307    
1308    5.  A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1309        quantifier applied to a forward-referencing subroutine call, could compile
1310        incorrect code or give the error "internal error: previously-checked
1311        referenced subpattern not found".
1312    
1313    6.  Both MS Visual Studio and Symbian OS have problems with initializing
1314        variables to point to external functions. For these systems, therefore,
1315        pcre_malloc etc. are now initialized to local functions that call the
1316        relevant global functions.
1317    
1318    7.  There were two entries missing in the vectors called coptable and poptable
1319        in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1320        I've fixed the data, and added a kludgy way of testing at compile time that
1321        the lengths are correct (equal to the number of opcodes).
1322    
1323    8.  Following on from 7, I added a similar kludge to check the length of the
1324        eint vector in pcreposix.c.
1325    
1326    9.  Error texts for pcre_compile() are held as one long string to avoid too
1327        much relocation at load time. To find a text, the string is searched,
1328        counting zeros. There was no check for running off the end of the string,
1329        which could happen if a new error number was added without updating the
1330        string.
1331    
1332    10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1333    
1334    11. \K was not working if it appeared in an atomic group or in a group that
1335        was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1336        \K is "not well defined" if used in an assertion. PCRE now accepts it if
1337        the assertion is positive, but not if it is negative.
1338    
1339    12. Change 11 fortuitously reduced the size of the stack frame used in the
1340        "match()" function of pcre_exec.c by one pointer. Forthcoming
1341        implementation of support for (*MARK) will need an extra pointer on the
1342        stack; I have reserved it now, so that the stack frame size does not
1343        decrease.
1344    
1345    13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1346        item in branch that calls a recursion is a subroutine call - as in the
1347        second branch in the above example - was incorrectly given the compile-
1348        time error "recursive call could loop indefinitely" because pcre_compile()
1349        was not correctly checking the subroutine for matching a non-empty string.
1350    
1351    14. The checks for overrunning compiling workspace could trigger after an
1352        overrun had occurred. This is a "should never occur" error, but it can be
1353        triggered by pathological patterns such as hundreds of nested parentheses.
1354        The checks now trigger 100 bytes before the end of the workspace.
1355    
1356    15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1357    
1358    
1359    Version 8.01 19-Jan-2010
1360    ------------------------
1361    
1362    1.  If a pattern contained a conditional subpattern with only one branch (in
1363        particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1364        computed the wrong minimum data length (which is of course zero for such
1365        subpatterns). This could cause incorrect "no match" results.
1366    
1367    2.  For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1368        the pattern is reset in the first branch, pcre_compile() failed with
1369        "internal error: code overflow at offset...". This happened only when
1370        the reset was to the original external option setting. (An optimization
1371        abstracts leading options settings into an external setting, which was the
1372        cause of this.)
1373    
1374    3.  A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1375        of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1376        assertion pattern did not match (meaning that the assertion was true), it
1377        was incorrectly treated as false if the SKIP had been reached during the
1378        matching. This also applied to assertions used as conditions.
1379    
1380    4.  If an item that is not supported by pcre_dfa_exec() was encountered in an
1381        assertion subpattern, including such a pattern used as a condition,
1382        unpredictable results occurred, instead of the error return
1383        PCRE_ERROR_DFA_UITEM.
1384    
1385    5.  The C++ GlobalReplace function was not working like Perl for the special
1386        situation when an empty string is matched. It now does the fancy magic
1387        stuff that is necessary.
1388    
1389    6.  In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1390        removed. (These were left over from very, very early versions of PCRE.)
1391    
1392    7.  Some cosmetic changes to the code to make life easier when compiling it
1393        as part of something else:
1394    
1395        (a) Change DEBUG to PCRE_DEBUG.
1396    
1397        (b) In pcre_compile(), rename the member of the "branch_chain" structure
1398            called "current" as "current_branch", to prevent a collision with the
1399            Linux macro when compiled as a kernel module.
1400    
1401        (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1402            prevent a collision with the Linux macro when compiled as a kernel
1403            module.
1404    
1405    8.  In pcre_compile() there are some checks for integer overflows that used to
1406        cast potentially large values to (double). This has been changed to that
1407        when building, a check for int64_t is made, and if it is found, it is used
1408        instead, thus avoiding the use of floating point arithmetic. (There is no
1409        other use of FP in PCRE.) If int64_t is not found, the fallback is to
1410        double.
1411    
1412    9.  Added two casts to avoid signed/unsigned warnings from VS Studio Express
1413        2005 (difference between two addresses compared to an unsigned value).
1414    
1415    10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1416        custom one, because of the following reported problem in Windows:
1417    
1418          - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1419              under Win32.
1420          - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1421              therefore missing the function definition.
1422          - The compiler thus generates a "C" signature for the test function.
1423          - The linker fails to find the "C" function.
1424          - PCRE fails to configure if asked to do so against libbz2.
1425    
1426    11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1427        messages were output:
1428    
1429          Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1430          rerunning libtoolize, to keep the correct libtool macros in-tree.
1431          Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1432    
1433        I have done both of these things.
1434    
1435    12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1436        most of the time, it *can* run out if it is given a pattern that contains a
1437        runaway infinite recursion. I updated the discussion in the pcrestack man
1438        page.
1439    
1440    13. Now that we have gone to the x.xx style of version numbers, the minor
1441        version may start with zero. Using 08 or 09 is a bad idea because users
1442        might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1443        interpreted as invalid octal numbers. I've updated the previous comment in
1444        configure.ac, and also added a check that gives an error if 08 or 09 are
1445        used.
1446    
1447    14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1448        causing partial matching to fail when the end of the subject matched \W
1449        in a UTF-8 pattern where \W was quantified with a minimum of 3.
1450    
1451    15. There were some discrepancies between the declarations in pcre_internal.h
1452        of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1453        their definitions. The declarations used "const uschar *" and the
1454        definitions used USPTR. Even though USPTR is normally defined as "const
1455        unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1456        reported that: "This difference in casting confuses some C++ compilers, for
1457        example, SunCC recognizes above declarations as different functions and
1458        generates broken code for hbpcre." I have changed the declarations to use
1459        USPTR.
1460    
1461    16. GNU libtool is named differently on some systems. The autogen.sh script now
1462        tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1463        (FreeBSD).
1464    
1465    17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1466        (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1467        comment: "Figure out how to create a longlong from a string: strtoll and
1468        equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1469        instance, but it only takes 2 args instead of 3!"
1470    
1471    18. A subtle bug concerned with back references has been fixed by a change of
1472        specification, with a corresponding code fix. A pattern such as
1473        ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1474        refers, was giving matches when it shouldn't. For example, xa=xaaa would
1475        match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1476        same bug. Such groups have to be quantified to be useful, or contained
1477        inside another quantified group. (If there's no repetition, the reference
1478        can never match.) The problem arises because, having left the group and
1479        moved on to the rest of the pattern, a later failure that backtracks into
1480        the group uses the captured value from the final iteration of the group
1481        rather than the correct earlier one. I have fixed this in PCRE by forcing
1482        any group that contains a reference to itself to be an atomic group; that
1483        is, there cannot be any backtracking into it once it has completed. This is
1484        similar to recursive and subroutine calls.
1485    
1486    
1487    Version 8.00 19-Oct-09
1488    ----------------------
1489    
1490    1.  The table for translating pcre_compile() error codes into POSIX error codes
1491        was out-of-date, and there was no check on the pcre_compile() error code
1492        being within the table. This could lead to an OK return being given in
1493        error.
1494    
1495    2.  Changed the call to open a subject file in pcregrep from fopen(pathname,
1496        "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1497        in a Windows environment.
1498    
1499    3.  The pcregrep --count option prints the count for each file even when it is
1500        zero, as does GNU grep. However, pcregrep was also printing all files when
1501        --files-with-matches was added. Now, when both options are given, it prints
1502        counts only for those files that have at least one match. (GNU grep just
1503        prints the file name in this circumstance, but including the count seems
1504        more useful - otherwise, why use --count?) Also ensured that the
1505        combination -clh just lists non-zero counts, with no names.
1506    
1507    4.  The long form of the pcregrep -F option was incorrectly implemented as
1508        --fixed_strings instead of --fixed-strings. This is an incompatible change,
1509        but it seems right to fix it, and I didn't think it was worth preserving
1510        the old behaviour.
1511    
1512    5.  The command line items --regex=pattern and --regexp=pattern were not
1513        recognized by pcregrep, which required --regex pattern or --regexp pattern
1514        (with a space rather than an '='). The man page documented the '=' forms,
1515        which are compatible with GNU grep; these now work.
1516    
1517    6.  No libpcreposix.pc file was created for pkg-config; there was just
1518        libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1519    
1520    7.  Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1521        when UCP support is not needed, by modifying the Python script that
1522        generates it from Unicode data files. This should not matter if the module
1523        is correctly used as a library, but I received one complaint about 50K of
1524        unwanted data. My guess is that the person linked everything into his
1525        program rather than using a library. Anyway, it does no harm.
1526    
1527    8.  A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1528        was a minimum greater than 1 for a wide character in a possessive
1529        repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1530        which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1531        character. Chaos in the form of incorrect output or a compiling loop could
1532        result.
1533    
1534    9.  The restrictions on what a pattern can contain when partial matching is
1535        requested for pcre_exec() have been removed. All patterns can now be
1536        partially matched by this function. In addition, if there are at least two
1537        slots in the offset vector, the offset of the earliest inspected character
1538        for the match and the offset of the end of the subject are set in them when
1539        PCRE_ERROR_PARTIAL is returned.
1540    
1541    10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1542        synonymous with PCRE_PARTIAL, for backwards compatibility, and
1543        PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1544        and may be more useful for multi-segment matching.
1545    
1546    11. Partial matching with pcre_exec() is now more intuitive. A partial match
1547        used to be given if ever the end of the subject was reached; now it is
1548        given only if matching could not proceed because another character was
1549        needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1550        string "Z", which now yields "no match" instead of "partial match". In the
1551        case of pcre_dfa_exec(), "no match" is given if every matching path for the
1552        final character ended with (*FAIL).
1553    
1554    12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1555        if the pattern had a "must contain" character that was already found in the
1556        earlier partial match, unless partial matching was again requested. For
1557        example, with the pattern /dog.(body)?/, the "must contain" character is
1558        "g". If the first part-match was for the string "dog", restarting with
1559        "sbody" failed. This bug has been fixed.
1560    
1561    13. The string returned by pcre_dfa_exec() after a partial match has been
1562        changed so that it starts at the first inspected character rather than the
1563        first character of the match. This makes a difference only if the pattern
1564        starts with a lookbehind assertion or \b or \B (\K is not supported by
1565        pcre_dfa_exec()). It's an incompatible change, but it makes the two
1566        matching functions compatible, and I think it's the right thing to do.
1567    
1568    14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1569        so that the demonstration program is easily available in environments where
1570        PCRE has not been installed from source.
1571    
1572    15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1573        libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1574        library.
1575    
1576    16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1577        It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1578        is not the first non-POSIX option to be added. Clearly some people find
1579        these options useful.
1580    
1581    17. If a caller to the POSIX matching function regexec() passes a non-zero
1582        value for nmatch with a NULL value for pmatch, the value of
1583        nmatch is forced to zero.
1584    
1585    18. RunGrepTest did not have a test for the availability of the -u option of
1586        the diff command, as RunTest does. It now checks in the same way as
1587        RunTest, and also checks for the -b option.
1588    
1589    19. If an odd number of negated classes containing just a single character
1590        interposed, within parentheses, between a forward reference to a named
1591        subpattern and the definition of the subpattern, compilation crashed with
1592        an internal error, complaining that it could not find the referenced
1593        subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1594        [The bug was that it was starting one character too far in when skipping
1595        over the character class, thus treating the ] as data rather than
1596        terminating the class. This meant it could skip too much.]
1597    
1598    20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1599        /g option in pcretest when the pattern contains \K, which makes it possible
1600        to have an empty string match not at the start, even when the pattern is
1601        anchored. Updated pcretest and pcredemo to use this option.
1602    
1603    21. If the maximum number of capturing subpatterns in a recursion was greater
1604        than the maximum at the outer level, the higher number was returned, but
1605        with unset values at the outer level. The correct (outer level) value is
1606        now given.
1607    
1608    22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1609        PCRE did not set those parentheses (unlike Perl). I have now found a way to
1610        make it do so. The string so far is captured, making this feature
1611        compatible with Perl.
1612    
1613    23. The tests have been re-organized, adding tests 11 and 12, to make it
1614        possible to check the Perl 5.10 features against Perl 5.10.
1615    
1616    24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1617        pattern matches a fixed length string. PCRE did not allow this; now it
1618        does. Neither allows recursion.
1619    
1620    25. I finally figured out how to implement a request to provide the minimum
1621        length of subject string that was needed in order to match a given pattern.
1622        (It was back references and recursion that I had previously got hung up
1623        on.) This code has now been added to pcre_study(); it finds a lower bound
1624        to the length of subject needed. It is not necessarily the greatest lower
1625        bound, but using it to avoid searching strings that are too short does give
1626        some useful speed-ups. The value is available to calling programs via
1627        pcre_fullinfo().
1628    
1629    26. While implementing 25, I discovered to my embarrassment that pcretest had
1630        not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1631        study optimizations had never been tested with that matching function.
1632        Oops. What is worse, even when it was passed study data, there was a bug in
1633        pcre_dfa_exec() that meant it never actually used it. Double oops. There
1634        were also very few tests of studied patterns with pcre_dfa_exec().
1635    
1636    27. If (?| is used to create subpatterns with duplicate numbers, they are now
1637        allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1638        on the other side of the coin, they are no longer allowed to have different
1639        names, because these cannot be distinguished in PCRE, and this has caused
1640        confusion. (This is a difference from Perl.)
1641    
1642    28. When duplicate subpattern names are present (necessarily with different
1643        numbers, as required by 27 above), and a test is made by name in a
1644        conditional pattern, either for a subpattern having been matched, or for
1645        recursion in such a pattern, all the associated numbered subpatterns are
1646        tested, and the overall condition is true if the condition is true for any
1647        one of them. This is the way Perl works, and is also more like the way
1648        testing by number works.
1649    
1650    
1651    Version 7.9 11-Apr-09
1652    ---------------------
1653    
1654    1.  When building with support for bzlib/zlib (pcregrep) and/or readline
1655        (pcretest), all targets were linked against these libraries. This included
1656        libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1657        libraries. This caused unwanted dependencies to be created. This problem
1658        has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1659        pcretest is linked with readline.
1660    
1661    2.  The "typedef int BOOL" in pcre_internal.h that was included inside the
1662        "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1663        moved outside it again, because FALSE and TRUE are already defined in AIX,
1664        but BOOL is not.
1665    
1666    3.  The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1667        PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1668    
1669    4.  The pcregrep documentation said spaces were inserted as well as colons (or
1670        hyphens) following file names and line numbers when outputting matching
1671        lines. This is not true; no spaces are inserted. I have also clarified the
1672        wording for the --colour (or --color) option.
1673    
1674    5.  In pcregrep, when --colour was used with -o, the list of matching strings
1675        was not coloured; this is different to GNU grep, so I have changed it to be
1676        the same.
1677    
1678    6.  When --colo(u)r was used in pcregrep, only the first matching substring in
1679        each matching line was coloured. Now it goes on to look for further matches
1680        of any of the test patterns, which is the same behaviour as GNU grep.
1681    
1682    7.  A pattern that could match an empty string could cause pcregrep to loop; it
1683        doesn't make sense to accept an empty string match in pcregrep, so I have
1684        locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1685        seems to be how GNU grep behaves. [But see later change 40 for release
1686        8.33.]
1687    
1688    8.  The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1689        start or after a newline", because the conditional assertion was not being
1690        correctly handled. The rule now is that both the assertion and what follows
1691        in the first alternative must satisfy the test.
1692    
1693    9.  If auto-callout was enabled in a pattern with a conditional group whose
1694        condition was an assertion, PCRE could crash during matching, both with
1695        pcre_exec() and pcre_dfa_exec().
1696    
1697    10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1698        used for matching.
1699    
1700    11. Unicode property support in character classes was not working for
1701        characters (bytes) greater than 127 when not in UTF-8 mode.
1702    
1703    12. Added the -M command line option to pcretest.
1704    
1705    14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1706    
1707    15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1708    
1709    16. Added comments and documentation about mis-use of no_arg in the C++
1710        wrapper.
1711    
1712    17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1713        from Martin Jerabek that uses macro names for all relevant character and
1714        string constants.
1715    
1716    18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1717        SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1718        SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1719        these, but not everybody uses configure.
1720    
1721    19. A conditional group that had only one branch was not being correctly
1722        recognized as an item that could match an empty string. This meant that an
1723        enclosing group might also not be so recognized, causing infinite looping
1724        (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1725        with the subject "ab", where knowledge that the repeated group can match
1726        nothing is needed in order to break the loop.
1727    
1728    20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1729        exec(), but without supplying a callout function, matching went wrong.
1730    
1731    21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1732        leak if the size of the offset vector was greater than 30. When the vector
1733        is smaller, the saved offsets during recursion go onto a local stack
1734        vector, but for larger vectors malloc() is used. It was failing to free
1735        when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1736        error, in fact).
1737    
1738    22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1739        heapframe that is used only when UTF-8 support is enabled. This caused no
1740        problem, but was untidy.
1741    
1742    23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1743        CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1744        included within another project.
1745    
1746    24. Steven Van Ingelgem's patches to add more options to the CMake support,
1747        slightly modified by me:
1748    
1749          (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1750              not building pcregrep.
1751    
1752          (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1753              if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1754    
1755    25. Forward references, both numeric and by name, in patterns that made use of
1756        duplicate group numbers, could behave incorrectly or give incorrect errors,
1757        because when scanning forward to find the reference group, PCRE was not
1758        taking into account the duplicate group numbers. A pattern such as
1759        ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1760    
1761    26. Changed a few more instances of "const unsigned char *" to USPTR, making
1762        the feature of a custom pointer more persuasive (as requested by a user).
1763    
1764    27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1765        pcretest.c, inside #ifndefs, because it seems they are sometimes already
1766        pre-defined.
1767    
1768    28. Added support for (*UTF8) at the start of a pattern.
1769    
1770    29. Arrange for flags added by the "release type" setting in CMake to be shown
1771        in the configuration summary.
1772    
1773    
1774    Version 7.8 05-Sep-08
1775    ---------------------
1776    
1777    1.  Replaced UCP searching code with optimized version as implemented for Ad
1778        Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1779        stage table and inline lookup instead of a function, giving speed ups of 2
1780        to 5 times on some simple patterns that I tested. Permission was given to
1781        distribute the MultiStage2.py script that generates the tables (it's not in
1782        the tarball, but is in the Subversion repository).
1783    
1784    2.  Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1785        scripts.
1786    
1787    3.  Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1788        a group with a zero qualifier. The result of the study could be incorrect,
1789        or the function might crash, depending on the pattern.
1790    
1791    4.  Caseless matching was not working for non-ASCII characters in back
1792        references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1793        It now works when Unicode Property Support is available.
1794    
1795    5.  In pcretest, an escape such as \x{de} in the data was always generating
1796        a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1797        non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1798        truncation.
1799    
1800    6.  Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1801    
1802    7.  Added two (int) casts to pcregrep when printing the difference of two
1803        pointers, in case they are 64-bit values.
1804    
1805    8.  Added comments about Mac OS X stack usage to the pcrestack man page and to
1806        test 2 if it fails.
1807    
1808    9.  Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1809        and a #define of that name to empty if it is not externally set. This is to
1810        allow users of MSVC to set it if necessary.
1811    
1812    10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1813        the convenience functions in the pcre_get.c source file.
1814    
1815    11. An option change at the start of a pattern that had top-level alternatives
1816        could cause overwriting and/or a crash. This command provoked a crash in
1817        some environments:
1818    
1819          printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1820    
1821        This potential security problem was recorded as CVE-2008-2371.
1822    
1823    12. For a pattern where the match had to start at the beginning or immediately
1824        after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1825        pcre_dfa_exec() could read past the end of the passed subject if there was
1826        no match. To help with detecting such bugs (e.g. with valgrind), I modified
1827        pcretest so that it places the subject at the end of its malloc-ed buffer.
1828    
1829    13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1830        exec() might read past the end of the data buffer in UTF-8 mode.
1831    
1832    14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1833        the data contained the byte 0x85 as part of a UTF-8 character within its
1834        first line. This applied both to normal and DFA matching.
1835    
1836    15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1837        /^[^d]*?$/8 failed to match "abc".
1838    
1839    16. Added a missing copyright notice to pcrecpp_internal.h.
1840    
1841    17. Make it more clear in the documentation that values returned from
1842        pcre_exec() in ovector are byte offsets, not character counts.
1843    
1844    18. Tidied a few places to stop certain compilers from issuing warnings.
1845    
1846    19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1847        supplied by Stefan Weber. I made a further small update for 7.8 because
1848        there is a change of source arrangements: the pcre_searchfuncs.c module is
1849        replaced by pcre_ucd.c.
1850    
1851    
1852    Version 7.7 07-May-08
1853    ---------------------
1854    
1855    1.  Applied Craig's patch to sort out a long long problem: "If we can't convert
1856        a string to a long long, pretend we don't even have a long long." This is
1857        done by checking for the strtoq, strtoll, and _strtoi64 functions.
1858    
1859    2.  Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1860        pre-7.6 versions, which defined a global no_arg variable instead of putting
1861        it in the RE class. (See also #8 below.)
1862    
1863    3.  Remove a line of dead code, identified by coverity and reported by Nuno
1864        Lopes.
1865    
1866    4.  Fixed two related pcregrep bugs involving -r with --include or --exclude:
1867    
1868        (1) The include/exclude patterns were being applied to the whole pathnames
1869            of files, instead of just to the final components.
1870    
1871        (2) If there was more than one level of directory, the subdirectories were
1872            skipped unless they satisfied the include/exclude conditions. This is
1873            inconsistent with GNU grep (and could even be seen as contrary to the
1874            pcregrep specification - which I improved to make it absolutely clear).
1875            The action now is always to scan all levels of directory, and just
1876            apply the include/exclude patterns to regular files.
1877    
1878    5.  Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1879        --exclude_dir in the tests to avoid scanning .svn directories.
1880    
1881    6.  Applied Craig's patch to the QuoteMeta function so that it escapes the
1882        NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1883        doesn't support NULs in patterns.
1884    
1885    7.  Added some missing "const"s to declarations of static tables in
1886        pcre_compile.c and pcre_dfa_exec.c.
1887    
1888    8.  Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1889        caused by fix #2  above. (Subsequently also a second patch to fix the
1890        first patch. And a third patch - this was a messy problem.)
1891    
1892    9.  Applied Craig's patch to remove the use of push_back().
1893    
1894    10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1895        matching function regexec().
1896    
1897    11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1898        which, however, unlike Perl's \g{...}, are subroutine calls, not back
1899        references. PCRE supports relative numbers with this syntax (I don't think
1900        Oniguruma does).
1901    
1902    12. Previously, a group with a zero repeat such as (...){0} was completely
1903        omitted from the compiled regex. However, this means that if the group
1904        was called as a subroutine from elsewhere in the pattern, things went wrong
1905        (an internal error was given). Such groups are now left in the compiled
1906        pattern, with a new opcode that causes them to be skipped at execution
1907        time.
1908    
1909    13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1910        to the way PCRE behaves:
1911    
1912        (a) A lone ] character is dis-allowed (Perl treats it as data).
1913    
1914        (b) A back reference to an unmatched subpattern matches an empty string
1915            (Perl fails the current match path).
1916    
1917        (c) A data ] in a character class must be notated as \] because if the
1918            first data character in a class is ], it defines an empty class. (In
1919            Perl it is not possible to have an empty class.) The empty class []
1920            never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1921            The negative empty class [^] matches any one character, independently
1922            of the DOTALL setting.
1923    
1924    14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1925        non-existent subpattern following a character class starting with ']' and
1926        containing () gave an internal compiling error instead of "reference to
1927        non-existent subpattern". Fortunately, when the pattern did exist, the
1928        compiled code was correct. (When scanning forwards to check for the
1929        existence of the subpattern, it was treating the data ']' as terminating
1930        the class, so got the count wrong. When actually compiling, the reference
1931        was subsequently set up correctly.)
1932    
1933    15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1934        it was being rejected as not supported by pcre_dfa_exec(), even though
1935        other assertions are supported. I have made pcre_dfa_exec() support
1936        (*FAIL).
1937    
1938    16. The implementation of 13c above involved the invention of a new opcode,
1939        OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1940        cannot be changed at match time, I realized I could make a small
1941        improvement to matching performance by compiling OP_ALLANY instead of
1942        OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1943        on the OP_ANY path.
1944    
1945    17. Compiling pcretest on Windows with readline support failed without the
1946        following two fixes: (1) Make the unistd.h include conditional on
1947        HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1948    
1949    18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1950        ncurses library to be included for pcretest when ReadLine support is
1951        requested, but also to allow for it to be overridden. This patch came from
1952        Daniel Bergström.
1953    
1954    19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1955        as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1956        any errors with the current Unicode tables. Thanks to Peter Kankowski for
1957        spotting this.
1958    
1959    
1960  Version 7.6 28-Jan-08  Version 7.6 28-Jan-08
1961  ---------------------  ---------------------
1962    

Legend:
Removed from v.319  
changed lines
  Added in v.1338

  ViewVC Help
Powered by ViewVC 1.1.5