ViewVC logotype

Diff of /code/trunk/ChangeLog

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

revision 13 by nigel, Sat Feb 24 21:38:21 2007 UTC revision 779 by ph10, Fri Dec 2 10:39:32 2011 UTC
# Line 1  Line 1 
1  ChangeLog for PCRE  ChangeLog for PCRE
2  ------------------  ------------------
4    Version 8.21
5    ------------
7    1.  Updating the JIT compiler.
9    2.  JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
10        are added as well.
12    3.  Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
13        PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
14        calling _pcre_jit_exec. Some extra comments are added.
16    4.  Mark settings inside atomic groups that do not contain any capturing
17        parentheses, for example, (?>a(*:m)), were not being passed out. This bug
18        was introduced by change 18 for 8.20.
20    5.  Supporting of \x, \U and \u in JavaScript compatibility mode based on the
21        ECMA-262 standard.
23    6.  Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
24        erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
25        This bug was probably introduced by change 9 of 8.13.
27    7.  While fixing 6 above, I noticed that a number of other items were being
28        incorrectly rejected as "not fixed length". This arose partly because newer
29        opcodes had not been added to the fixed-length checking code. I have (a)
30        corrected the bug and added tests for these items, and (b) arranged for an
31        error to occur if an unknown opcode is encountered while checking for fixed
32        length instead of just assuming "not fixed length". The items that were
33        rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
34        (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
35        repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
37    8.  A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
38        being incorrectly compiled and would have given unpredicatble results.
40    9.  A possessively repeated subpattern with minimum repeat count greater than
41        one behaved incorrectly. For example, (A){2,}+ behaved as if it was
42        (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
43        the first (A) could occur when it should not.
45    10. Add a cast and remove a redundant test from the code.
47    11. JIT should use pcre_malloc/pcre_free for allocation.
49    12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
50        best practice nowadays, and helps with cross-compiling. (If the exec_prefix
51        is anything other than /usr, -L is still shown).
53    13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
55    14. Perl does not support \N without a following name in a [] class; PCRE now
56        also gives an error.
58    15. If a forward reference was repeated with an upper limit of around 2000,
59        it caused the error "internal error: overran compiling workspace". The
60        maximum number of forward references (including repeats) was limited by the
61        internal workspace, and dependent on the LINK_SIZE. The code has been
62        rewritten so that the workspace expands (via pcre_malloc) if necessary, and
63        the default depends on LINK_SIZE. There is a new upper limit (for safety)
64        of around 200,000 forward references. While doing this, I also speeded up
65        the filling in of repeated forward references.
67    16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
68        incorrectly expecting the subject to contain another "a" after the start.
70    17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
71        in the match, the SKIP should be ignored. This was not happening; instead
72        the SKIP was being treated as NOMATCH. For patterns such as
73        /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
74        tested.
76    18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
77        now much more compatible with Perl, in particular in cases where the result
78        is a non-match for a non-anchored pattern. For example, if
79        /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
80        "m", where previously it did not return a name. A side effect of this
81        change is that for partial matches, the last encountered mark name is
82        returned, as for non matches. A number of tests that were previously not
83        Perl-compatible have been moved into the Perl-compatible test files. The
84        refactoring has had the pleasing side effect of removing one argument from
85        the match() function, thus reducing its stack requirements.
87    19. If the /S+ option was used in pcretest to study a pattern using JIT,
88        subsequent uses of /S (without +) incorrectly behaved like /S+.
90    21. Retrieve executable code size support for the JIT compiler and fixing
91        some warnings.
93    22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
94        not work when the shorter character appeared right at the end of the
95        subject string.
97    23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
98        systems.
101    Version 8.20 21-Oct-2011
102    ------------------------
104    1.  Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
105        a POSIX class. After further experiments with Perl, which convinced me that
106        Perl has bugs and confusions, a closing square bracket is no longer allowed
107        in a POSIX name. This bug also affected patterns with classes that started
108        with full stops.
110    2.  If a pattern such as /(a)b|ac/ is matched against "ac", there is no
111        captured substring, but while checking the failing first alternative,
112        substring 1 is temporarily captured. If the output vector supplied to
113        pcre_exec() was not big enough for this capture, the yield of the function
114        was still zero ("insufficient space for captured substrings"). This cannot
115        be totally fixed without adding another stack variable, which seems a lot
116        of expense for a edge case. However, I have improved the situation in cases
117        such as /(a)(b)x|abc/ matched against "abc", where the return code
118        indicates that fewer than the maximum number of slots in the ovector have
119        been set.
121    3.  Related to (2) above: when there are more back references in a pattern than
122        slots in the output vector, pcre_exec() uses temporary memory during
123        matching, and copies in the captures as far as possible afterwards. It was
124        using the entire output vector, but this conflicts with the specification
125        that only 2/3 is used for passing back captured substrings. Now it uses
126        only the first 2/3, for compatibility. This is, of course, another edge
127        case.
129    4.  Zoltan Herczeg's just-in-time compiler support has been integrated into the
130        main code base, and can be used by building with --enable-jit. When this is
131        done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
132        runtime --no-jit option is given.
134    5.  When the number of matches in a pcre_dfa_exec() run exactly filled the
135        ovector, the return from the function was zero, implying that there were
136        other matches that did not fit. The correct "exactly full" value is now
137        returned.
139    6.  If a subpattern that was called recursively or as a subroutine contained
140        (*PRUNE) or any other control that caused it to give a non-standard return,
141        invalid errors such as "Error -26 (nested recursion at the same subject
142        position)" or even infinite loops could occur.
144    7.  If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
145        computing the minimum length on reaching *ACCEPT, and so ended up with the
146        wrong value of 1 rather than 0. Further investigation indicates that
147        computing a minimum subject length in the presence of *ACCEPT is difficult
148        (think back references, subroutine calls), and so I have changed the code
149        so that no minimum is registered for a pattern that contains *ACCEPT.
151    8.  If (*THEN) was present in the first (true) branch of a conditional group,
152        it was not handled as intended. [But see 16 below.]
154    9.  Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
155        Sheri Pierce.
157    10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
158        the first byte in a match must be "a".
160    11. Change 17 for 8.13 increased the recursion depth for patterns like
161        /a(?:.)*?a/ drastically. I've improved things by remembering whether a
162        pattern contains any instances of (*THEN). If it does not, the old
163        optimizations are restored. It would be nice to do this on a per-group
164        basis, but at the moment that is not feasible.
166    12. In some environments, the output of pcretest -C is CRLF terminated. This
167        broke RunTest's code that checks for the link size. A single white space
168        character after the value is now allowed for.
170    13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
171        For "fr", it uses the Windows-specific input and output files.
173    14. If (*THEN) appeared in a group that was called recursively or as a
174        subroutine, it did not work as intended. [But see next item.]
176    15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
177        pattern fragments (but not containing any | characters). If A and B are
178        matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
179        was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
180        D. In other words, Perl considers parentheses that do not contain any |
181        characters to be part of a surrounding alternative, whereas PCRE was
182        treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
183        differently. PCRE now behaves in the same way as Perl, except in the case
184        of subroutine/recursion calls such as (?1) which have in any case always
185        been different (but PCRE had them first :-).
187    16. Related to 15 above: Perl does not treat the | in a conditional group as
188        creating alternatives. Such a group is treated in the same way as an
189        ordinary group without any | characters when processing (*THEN). PCRE has
190        been changed to match Perl's behaviour.
192    17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
193        RunGrepTest script failed.
195    18. Change 22 for version 13 caused atomic groups to use more stack. This is
196        inevitable for groups that contain captures, but it can lead to a lot of
197        stack use in large patterns. The old behaviour has been restored for atomic
198        groups that do not contain any capturing parentheses.
200    19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
201        suppress the check for a minimum subject length at run time. (If it was
202        given to pcre_exec() or pcre_dfa_exec() it did work.)
204    20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
205        fail to work when decoding hex characters in data strings in EBCDIC
206        environments.
208    21. It appears that in at least one Mac OS environment, the isxdigit() function
209        is implemented as a macro that evaluates to its argument more than once,
210        contravening the C 90 Standard (I haven't checked a later standard). There
211        was an instance in pcretest which caused it to go wrong when processing
212        \x{...} escapes in subject strings. The has been rewritten to avoid using
213        things like p++ in the argument of isxdigit().
216    Version 8.13 16-Aug-2011
217    ------------------------
219    1.  The Unicode data tables have been updated to Unicode 6.0.0.
221    2.  Two minor typos in pcre_internal.h have been fixed.
223    3.  Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
224        pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
225        in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
227    4.  There were a number of related bugs in the code for matching backrefences
228        caselessly in UTF-8 mode when codes for the characters concerned were
229        different numbers of bytes. For example, U+023A and U+2C65 are an upper
230        and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
231        (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
232        code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
233        2-byte code at the end of the subject (it thought there wasn't enough data
234        left).
236    5.  Comprehensive information about what went wrong is now returned by
237        pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
238        as the output vector has at least 2 elements. The offset of the start of
239        the failing character and a reason code are placed in the vector.
241    6.  When the UTF-8 string check fails for pcre_compile(), the offset that is
242        now returned is for the first byte of the failing character, instead of the
243        last byte inspected. This is an incompatible change, but I hope it is small
244        enough not to be a problem. It makes the returned offset consistent with
245        pcre_exec() and pcre_dfa_exec().
247    7.  pcretest now gives a text phrase as well as the error number when
248        pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
249        failure, the offset and reason code are output.
251    8.  When \R was used with a maximizing quantifier it failed to skip backwards
252        over a \r\n pair if the subsequent match failed. Instead, it just skipped
253        back over a single character (\n). This seems wrong (because it treated the
254        two characters as a single entity when going forwards), conflicts with the
255        documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
256        behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
257        has been changed.
259    9.  Some internal refactoring has changed the processing so that the handling
260        of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
261        time (the PCRE_DOTALL option was changed this way some time ago: version
262        7.7 change 16). This has made it possible to abolish the OP_OPT op code,
263        which was always a bit of a fudge. It also means that there is one less
264        argument for the match() function, which reduces its stack requirements
265        slightly. This change also fixes an incompatibility with Perl: the pattern
266        (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
268    10. More internal refactoring has drastically reduced the number of recursive
269        calls to match() for possessively repeated groups such as (abc)++ when
270        using pcre_exec().
272    11. While implementing 10, a number of bugs in the handling of groups were
273        discovered and fixed:
275        (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
276        (a|)*(?1) gave a compile-time internal error.
277        ((a|)+)+  did not notice that the outer group could match an empty string.
278        (^a|^)+   was not marked as anchored.
279        (.*a|.*)+ was not marked as matching at start or after a newline.
281    12. Yet more internal refactoring has removed another argument from the match()
282        function. Special calls to this function are now indicated by setting a
283        value in a variable in the "match data" data block.
285    13. Be more explicit in pcre_study() instead of relying on "default" for
286        opcodes that mean there is no starting character; this means that when new
287        ones are added and accidentally left out of pcre_study(), testing should
288        pick them up.
290    14. The -s option of pcretest has been documented for ages as being an old
291        synonym of -m (show memory usage). I have changed it to mean "force study
292        for every regex", that is, assume /S for every regex. This is similar to -i
293        and -d etc. It's slightly incompatible, but I'm hoping nobody is still
294        using it. It makes it easier to run collections of tests with and without
295        study enabled, and thereby test pcre_study() more easily. All the standard
296        tests are now run with and without -s (but some patterns can be marked as
297        "never study" - see 20 below).
299    15. When (*ACCEPT) was used in a subpattern that was called recursively, the
300        restoration of the capturing data to the outer values was not happening
301        correctly.
303    16. If a recursively called subpattern ended with (*ACCEPT) and matched an
304        empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
305        pattern had matched an empty string, and so incorrectly returned a no
306        match.
308    17. There was optimizing code for the last branch of non-capturing parentheses,
309        and also for the obeyed branch of a conditional subexpression, which used
310        tail recursion to cut down on stack usage. Unfortunately, now that there is
311        the possibility of (*THEN) occurring in these branches, tail recursion is
312        no longer possible because the return has to be checked for (*THEN). These
313        two optimizations have therefore been removed. [But see 8.20/11 above.]
315    18. If a pattern containing \R was studied, it was assumed that \R always
316        matched two bytes, thus causing the minimum subject length to be
317        incorrectly computed because \R can also match just one byte.
319    19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
320        was incorrectly computed.
322    20. If /S is present twice on a test pattern in pcretest input, it now
323        *disables* studying, thereby overriding the use of -s on the command line
324        (see 14 above). This is necessary for one or two tests to keep the output
325        identical in both cases.
327    21. When (*ACCEPT) was used in an assertion that matched an empty string and
328        PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
330    22. When an atomic group that contained a capturing parenthesis was
331        successfully matched, but the branch in which it appeared failed, the
332        capturing was not being forgotten if a higher numbered group was later
333        captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
334        group 1 to "a", when in fact it should be unset. This applied to multi-
335        branched capturing and non-capturing groups, repeated or not, and also to
336        positive assertions (capturing in negative assertions does not happen
337        in PCRE) and also to nested atomic groups.
339    23. Add the ++ qualifier feature to pcretest, to show the remainder of the
340        subject after a captured substring, to make it easier to tell which of a
341        number of identical substrings has been captured.
343    24. The way atomic groups are processed by pcre_exec() has been changed so that
344        if they are repeated, backtracking one repetition now resets captured
345        values correctly. For example, if ((?>(a+)b)+aabab) is matched against
346        "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
347        "aaa". Previously, it would have been "a". As part of this code
348        refactoring, the way recursive calls are handled has also been changed.
350    25. If an assertion condition captured any substrings, they were not passed
351        back unless some other capturing happened later. For example, if
352        (?(?=(a))a) was matched against "a", no capturing was returned.
354    26. When studying a pattern that contained subroutine calls or assertions,
355        the code for finding the minimum length of a possible match was handling
356        direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
357        group 1 called group 2 while simultaneously a separate group 2 called group
358        1). A stack overflow occurred in this case. I have fixed this by limiting
359        the recursion depth to 10.
361    27. Updated RunTest.bat in the distribution to the version supplied by Tom
362        Fortmann. This supports explicit test numbers on the command line, and has
363        argument validation and error reporting.
365    28. An instance of \X with an unlimited repeat could fail if at any point the
366        first character it looked at was a mark character.
368    29. Some minor code refactoring concerning Unicode properties and scripts
369        should reduce the stack requirement of match() slightly.
371    30. Added the '=' option to pcretest to check the setting of unused capturing
372        slots at the end of the pattern, which are documented as being -1, but are
373        not included in the return count.
375    31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
376        compiled something random. Now it gives a compile-time error (as does
377        Perl).
379    32. A *MARK encountered during the processing of a positive assertion is now
380        recorded and passed back (compatible with Perl).
382    33. If --only-matching or --colour was set on a pcregrep call whose pattern
383        had alternative anchored branches, the search for a second match in a line
384        was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
385        matched the line "0102" twice. The same bug affected patterns that started
386        with a backwards assertion. For example /\b01|\b02/ also matched "0102"
387        twice.
389    34. Previously, PCRE did not allow quantification of assertions. However, Perl
390        does, and because of capturing effects, quantifying parenthesized
391        assertions may at times be useful. Quantifiers are now allowed for
392        parenthesized assertions.
394    35. A minor code tidy in pcre_compile() when checking options for \R usage.
396    36. \g was being checked for fancy things in a character class, when it should
397        just be a literal "g".
399    37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
400        appearance of a nested POSIX class supersedes an apparent external class.
401        For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
402        unescaped square brackets may also appear as part of class names. For
403        example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
404        more like Perl. (But see 8.20/1 above.)
406    38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
407        was because it thought it was \N{name}, which is not supported).
409    39. Add minix to OS list not supporting the -S option in pcretest.
411    40. PCRE tries to detect cases of infinite recursion at compile time, but it
412        cannot analyze patterns in sufficient detail to catch mutual recursions
413        such as ((?1))((?2)). There is now a runtime test that gives an error if a
414        subgroup is called recursively as a subpattern for a second time at the
415        same position in the subject string. In previous releases this might have
416        been caught by the recursion limit, or it might have run out of stack.
418    41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
419        happen only once. PCRE was, however incorrectly giving a compile time error
420        "recursive call could loop indefinitely" because it cannot analyze the
421        pattern in sufficient detail. The compile time test no longer happens when
422        PCRE is compiling a conditional subpattern, but actual runaway loops are
423        now caught at runtime (see 40 above).
425    42. It seems that Perl allows any characters other than a closing parenthesis
426        to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
427        has been changed to be the same.
429    43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
430        as not to get warnings when autogen.sh is called. Also changed
431        AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
433    44. To help people who use pcregrep to scan files containing exceedingly long
434        lines, the following changes have been made:
436        (a) The default value of the buffer size parameter has been increased from
437            8K to 20K. (The actual buffer used is three times this size.)
439        (b) The default can be changed by ./configure --with-pcregrep-bufsize when
440            PCRE is built.
442        (c) A --buffer-size=n option has been added to pcregrep, to allow the size
443            to be set at run time.
445        (d) Numerical values in pcregrep options can be followed by K or M, for
446            example --buffer-size=50K.
448        (e) If a line being scanned overflows pcregrep's buffer, an error is now
449            given and the return code is set to 2.
451    45. Add a pointer to the latest mark to the callout data block.
453    46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
454        partial match of an empty string instead of no match. This was specific to
455        the use of ".".
457    47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
458        complete match instead of a partial match. This bug was dependent on both
459        the PCRE_UTF8 and PCRE_DOTALL options being set.
461    48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
462        starting byte set, because \b was not being ignored.
465    Version 8.12 15-Jan-2011
466    ------------------------
468    1.  Fixed some typos in the markup of the man pages, and wrote a script that
469        checks for such things as part of the documentation building process.
471    2.  On a big-endian 64-bit system, pcregrep did not correctly process the
472        --match-limit and --recursion-limit options (added for 8.11). In
473        particular, this made one of the standard tests fail. (The integer value
474        went into the wrong half of a long int.)
476    3.  If the --colour option was given to pcregrep with -v (invert match), it
477        did strange things, either producing crazy output, or crashing. It should,
478        of course, ignore a request for colour when reporting lines that do not
479        match.
481    4.  Another pcregrep bug caused similar problems if --colour was specified with
482        -M (multiline) and the pattern match finished with a line ending.
484    5.  In pcregrep, when a pattern that ended with a literal newline sequence was
485        matched in multiline mode, the following line was shown as part of the
486        match. This seems wrong, so I have changed it.
488    6.  Another pcregrep bug in multiline mode, when --colour was specified, caused
489        the check for further matches in the same line (so they could be coloured)
490        to overrun the end of the current line. If another match was found, it was
491        incorrectly shown (and then shown again when found in the next line).
493    7.  If pcregrep was compiled under Windows, there was a reference to the
494        function pcregrep_exit() before it was defined. I am assuming this was
495        the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
496        reported by a user. I've moved the definition above the reference.
499    Version 8.11 10-Dec-2010
500    ------------------------
502    1.  (*THEN) was not working properly if there were untried alternatives prior
503        to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
504        backtracked to try for "b" instead of moving to the next alternative branch
505        at the same level (in this case, to look for "c"). The Perl documentation
506        is clear that when (*THEN) is backtracked onto, it goes to the "next
507        alternative in the innermost enclosing group".
509    2.  (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
510        such as   (A(*COMMIT)B(*THEN)C|D)  any failure after matching A should
511        result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
512        (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
513        (*THEN).
515    3.  If \s appeared in a character class, it removed the VT character from
516        the class, even if it had been included by some previous item, for example
517        in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
518        of \s, but is part of the POSIX "space" class.)
520    4.  A partial match never returns an empty string (because you can always
521        match an empty string at the end of the subject); however the checking for
522        an empty string was starting at the "start of match" point. This has been
523        changed to the "earliest inspected character" point, because the returned
524        data for a partial match starts at this character. This means that, for
525        example, /(?<=abc)def/ gives a partial match for the subject "abc"
526        (previously it gave "no match").
528    5.  Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
529        of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
530        previously a full match would be given. However, setting PCRE_PARTIAL_HARD
531        has an implication that the given string is incomplete (because a partial
532        match is preferred over a full match). For this reason, these items now
533        give a partial match in this situation. [Aside: previously, the one case
534        /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
535        match rather than a full match, which was wrong by the old rules, but is
536        now correct.]
538    6.  There was a bug in the handling of #-introduced comments, recognized when
539        PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
540        If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
541        UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
542        scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
543        but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
544        places in pcre_compile().
546    7.  Related to (6) above, when pcre_compile() was skipping #-introduced
547        comments when looking ahead for named forward references to subpatterns,
548        the only newline sequence it recognized was NL. It now handles newlines
549        according to the set newline convention.
551    8.  SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
552        former, but used strtoul(), whereas pcretest avoided strtoul() but did not
553        cater for a lack of strerror(). These oversights have been fixed.
555    9.  Added --match-limit and --recursion-limit to pcregrep.
557    10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
559    11. When the -o option was used, pcregrep was setting a return code of 1, even
560        when matches were found, and --line-buffered was not being honoured.
562    12. Added an optional parentheses number to the -o and --only-matching options
563        of pcregrep.
565    13. Imitating Perl's /g action for multiple matches is tricky when the pattern
566        can match an empty string. The code to do it in pcretest and pcredemo
567        needed fixing:
569        (a) When the newline convention was "crlf", pcretest got it wrong, skipping
570            only one byte after an empty string match just before CRLF (this case
571            just got forgotten; "any" and "anycrlf" were OK).
573        (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
574            mode when an empty string match preceded an ASCII character followed by
575            a non-ASCII character. (The code for advancing by one character rather
576            than one byte was nonsense.)
578        (c) The pcredemo.c sample program did not have any code at all to handle
579            the cases when CRLF is a valid newline sequence.
581    14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
582        as a starting offset was within the subject string. There is now a new
583        error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
584        negative or greater than the length of the string. In order to test this,
585        pcretest is extended to allow the setting of negative starting offsets.
587    15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
588        starting offset points to the beginning of a UTF-8 character was
589        unnecessarily clumsy. I tidied it up.
591    16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
592        bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
594    17. Nobody had reported that the --include_dir option, which was added in
595        release 7.7 should have been called --include-dir (hyphen, not underscore)
596        for compatibility with GNU grep. I have changed it to --include-dir, but
597        left --include_dir as an undocumented synonym, and the same for
598        --exclude-dir, though that is not available in GNU grep, at least as of
599        release 2.5.4.
601    18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
602        characters from a string of bytes) have been redefined so as not to use
603        loops, in order to improve performance in some environments. At the same
604        time, I abstracted some of the common code into auxiliary macros to save
605        repetition (this should not affect the compiled code).
607    19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
608        compile-time error is now given if \c is not followed by an ASCII
609        character, that is, a byte less than 128. (In EBCDIC mode, the code is
610        different, and any byte value is allowed.)
612    20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
613        START_OPTIMIZE option, which is now allowed at compile time - but just
614        passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
615        to pcregrep and other applications that have no direct access to PCRE
616        options. The new /Y option in pcretest sets this option when calling
617        pcre_compile().
619    21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
620        back references. Groups containing recursive back references were forced to
621        be atomic by that change, but in the case of named groups, the amount of
622        memory required was incorrectly computed, leading to "Failed: internal
623        error: code overflow". This has been fixed.
625    22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
626        pcretest.c, to avoid build problems in some Borland environments.
629    Version 8.10 25-Jun-2010
630    ------------------------
632    1.  Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
633        THEN.
635    2.  (*ACCEPT) was not working when inside an atomic group.
637    3.  Inside a character class, \B is treated as a literal by default, but
638        faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
639        causes the error). The code is unchanged, but I tidied the documentation.
641    4.  Inside a character class, PCRE always treated \R and \X as literals,
642        whereas Perl faults them if its -w option is set. I have changed PCRE so
643        that it faults them when PCRE_EXTRA is set.
645    5.  Added support for \N, which always matches any character other than
646        newline. (It is the same as "." when PCRE_DOTALL is not set.)
648    6.  When compiling pcregrep with newer versions of gcc which may have
649        FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
650        declared with attribute warn_unused_result" were given. Just casting the
651        result to (void) does not stop the warnings; a more elaborate fudge is
652        needed. I've used a macro to implement this.
654    7.  Minor change to pcretest.c to avoid a compiler warning.
656    8.  Added four artifical Unicode properties to help with an option to make
657        \s etc use properties (see next item). The new properties are: Xan
658        (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
660    9.  Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
661        use Unicode properties. (*UCP) at the start of a pattern can be used to set
662        this option. Modified pcretest to add /W to test this facility. Added
663        REG_UCP to make it available via the POSIX interface.
665    10. Added --line-buffered to pcregrep.
667    11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
668        studied, and the match started with a letter with a code point greater than
669        127 whose first byte was different to the first byte of the other case of
670        the letter, the other case of this starting letter was not recognized
671        (#976).
673    12. If a pattern that was studied started with a repeated Unicode property
674        test, for example, \p{Nd}+, there was the theoretical possibility of
675        setting up an incorrect bitmap of starting bytes, but fortunately it could
676        not have actually happened in practice until change 8 above was made (it
677        added property types that matched character-matching opcodes).
679    13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
680        possible starting bytes for non-anchored patterns.
682    14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
683        \R, and also a number of cases that involve Unicode properties, both
684        explicit and implicit when PCRE_UCP is set.
686    15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
687        input, it could crash or give wrong results if characters with values
688        greater than 0xc0 were present in the subject string. (Detail: it assumed
689        UTF-8 input when processing these items.)
691    16. Added a lot of (int) casts to avoid compiler warnings in systems where
692        size_t is 64-bit (#991).
694    17. Added a check for running out of memory when PCRE is compiled with
695        --disable-stack-for-recursion (#990).
697    18. If the last data line in a file for pcretest does not have a newline on
698        the end, a newline was missing in the output.
700    19. The default pcre_chartables.c file recognizes only ASCII characters (values
701        less than 128) in its various bitmaps. However, there is a facility for
702        generating tables according to the current locale when PCRE is compiled. It
703        turns out that in some environments, 0x85 and 0xa0, which are Unicode space
704        characters, are recognized by isspace() and therefore were getting set in
705        these tables, and indeed these tables seem to approximate to ISO 8859. This
706        caused a problem in UTF-8 mode when pcre_study() was used to create a list
707        of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
708        which of course cannot start UTF-8 characters. I have changed the code so
709        that only real ASCII characters (less than 128) and the correct starting
710        bytes for UTF-8 encodings are set for characters greater than 127 when in
711        UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
712        altogether.)
714    20. Added the /T option to pcretest so as to be able to run tests with non-
715        standard character tables, thus making it possible to include the tests
716        used for 19 above in the standard set of tests.
718    21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
719        reference to a subpattern the other side of a comment that contains an
720        opening parenthesis caused either an internal compiling error, or a
721        reference to the wrong subpattern.
724    Version 8.02 19-Mar-2010
725    ------------------------
727    1.  The Unicode data tables have been updated to Unicode 5.2.0.
729    2.  Added the option --libs-cpp to pcre-config, but only when C++ support is
730        configured.
732    3.  Updated the licensing terms in the pcregexp.pas file, as agreed with the
733        original author of that file, following a query about its status.
735    4.  On systems that do not have stdint.h (e.g. Solaris), check for and include
736        inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
738    5.  A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
739        quantifier applied to a forward-referencing subroutine call, could compile
740        incorrect code or give the error "internal error: previously-checked
741        referenced subpattern not found".
743    6.  Both MS Visual Studio and Symbian OS have problems with initializing
744        variables to point to external functions. For these systems, therefore,
745        pcre_malloc etc. are now initialized to local functions that call the
746        relevant global functions.
748    7.  There were two entries missing in the vectors called coptable and poptable
749        in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
750        I've fixed the data, and added a kludgy way of testing at compile time that
751        the lengths are correct (equal to the number of opcodes).
753    8.  Following on from 7, I added a similar kludge to check the length of the
754        eint vector in pcreposix.c.
756    9.  Error texts for pcre_compile() are held as one long string to avoid too
757        much relocation at load time. To find a text, the string is searched,
758        counting zeros. There was no check for running off the end of the string,
759        which could happen if a new error number was added without updating the
760        string.
762    10. \K gave a compile-time error if it appeared in a lookbehind assersion.
764    11. \K was not working if it appeared in an atomic group or in a group that
765        was called as a "subroutine", or in an assertion. Perl 5.11 documents that
766        \K is "not well defined" if used in an assertion. PCRE now accepts it if
767        the assertion is positive, but not if it is negative.
769    12. Change 11 fortuitously reduced the size of the stack frame used in the
770        "match()" function of pcre_exec.c by one pointer. Forthcoming
771        implementation of support for (*MARK) will need an extra pointer on the
772        stack; I have reserved it now, so that the stack frame size does not
773        decrease.
775    13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
776        item in branch that calls a recursion is a subroutine call - as in the
777        second branch in the above example - was incorrectly given the compile-
778        time error "recursive call could loop indefinitely" because pcre_compile()
779        was not correctly checking the subroutine for matching a non-empty string.
781    14. The checks for overrunning compiling workspace could trigger after an
782        overrun had occurred. This is a "should never occur" error, but it can be
783        triggered by pathological patterns such as hundreds of nested parentheses.
784        The checks now trigger 100 bytes before the end of the workspace.
786    15. Fix typo in configure.ac: "srtoq" should be "strtoq".
789    Version 8.01 19-Jan-2010
790    ------------------------
792    1.  If a pattern contained a conditional subpattern with only one branch (in
793        particular, this includes all (*DEFINE) patterns), a call to pcre_study()
794        computed the wrong minimum data length (which is of course zero for such
795        subpatterns). This could cause incorrect "no match" results.
797    2.  For patterns such as (?i)a(?-i)b|c where an option setting at the start of
798        the pattern is reset in the first branch, pcre_compile() failed with
799        "internal error: code overflow at offset...". This happened only when
800        the reset was to the original external option setting. (An optimization
801        abstracts leading options settings into an external setting, which was the
802        cause of this.)
804    3.  A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
805        of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
806        assertion pattern did not match (meaning that the assertion was true), it
807        was incorrectly treated as false if the SKIP had been reached during the
808        matching. This also applied to assertions used as conditions.
810    4.  If an item that is not supported by pcre_dfa_exec() was encountered in an
811        assertion subpattern, including such a pattern used as a condition,
812        unpredictable results occurred, instead of the error return
815    5.  The C++ GlobalReplace function was not working like Perl for the special
816        situation when an empty string is matched. It now does the fancy magic
817        stuff that is necessary.
819    6.  In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
820        removed. (These were left over from very, very early versions of PCRE.)
822    7.  Some cosmetic changes to the code to make life easier when compiling it
823        as part of something else:
825        (a) Change DEBUG to PCRE_DEBUG.
827        (b) In pcre_compile(), rename the member of the "branch_chain" structure
828            called "current" as "current_branch", to prevent a collision with the
829            Linux macro when compiled as a kernel module.
831        (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
832            prevent a collision with the Linux macro when compiled as a kernel
833            module.
835    8.  In pcre_compile() there are some checks for integer overflows that used to
836        cast potentially large values to (double). This has been changed to that
837        when building, a check for int64_t is made, and if it is found, it is used
838        instead, thus avoiding the use of floating point arithmetic. (There is no
839        other use of FP in PCRE.) If int64_t is not found, the fallback is to
840        double.
842    9.  Added two casts to avoid signed/unsigned warnings from VS Studio Express
843        2005 (difference between two addresses compared to an unsigned value).
845    10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
846        custom one, because of the following reported problem in Windows:
848          - libbz2 uses the Pascal calling convention (WINAPI) for the functions
849              under Win32.
850          - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
851              therefore missing the function definition.
852          - The compiler thus generates a "C" signature for the test function.
853          - The linker fails to find the "C" function.
854          - PCRE fails to configure if asked to do so against libbz2.
856    11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
857        messages were output:
859          Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
860          rerunning libtoolize, to keep the correct libtool macros in-tree.
861          Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
863        I have done both of these things.
865    12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
866        most of the time, it *can* run out if it is given a pattern that contains a
867        runaway infinite recursion. I updated the discussion in the pcrestack man
868        page.
870    13. Now that we have gone to the x.xx style of version numbers, the minor
871        version may start with zero. Using 08 or 09 is a bad idea because users
872        might check the value of PCRE_MINOR in their code, and 08 or 09 may be
873        interpreted as invalid octal numbers. I've updated the previous comment in
874        configure.ac, and also added a check that gives an error if 08 or 09 are
875        used.
877    14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
878        causing partial matching to fail when the end of the subject matched \W
879        in a UTF-8 pattern where \W was quantified with a minimum of 3.
881    15. There were some discrepancies between the declarations in pcre_internal.h
882        of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
883        their definitions. The declarations used "const uschar *" and the
884        definitions used USPTR. Even though USPTR is normally defined as "const
885        unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
886        reported that: "This difference in casting confuses some C++ compilers, for
887        example, SunCC recognizes above declarations as different functions and
888        generates broken code for hbpcre." I have changed the declarations to use
889        USPTR.
891    16. GNU libtool is named differently on some systems. The autogen.sh script now
892        tries several variants such as glibtoolize (MacOSX) and libtoolize1x
893        (FreeBSD).
895    17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
896        (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
897        comment: "Figure out how to create a longlong from a string: strtoll and
898        equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
899        instance, but it only takes 2 args instead of 3!"
901    18. A subtle bug concerned with back references has been fixed by a change of
902        specification, with a corresponding code fix. A pattern such as
903        ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
904        refers, was giving matches when it shouldn't. For example, xa=xaaa would
905        match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
906        same bug. Such groups have to be quantified to be useful, or contained
907        inside another quantified group. (If there's no repetition, the reference
908        can never match.) The problem arises because, having left the group and
909        moved on to the rest of the pattern, a later failure that backtracks into
910        the group uses the captured value from the final iteration of the group
911        rather than the correct earlier one. I have fixed this in PCRE by forcing
912        any group that contains a reference to itself to be an atomic group; that
913        is, there cannot be any backtracking into it once it has completed. This is
914        similar to recursive and subroutine calls.
917    Version 8.00 19-Oct-09
918    ----------------------
920    1.  The table for translating pcre_compile() error codes into POSIX error codes
921        was out-of-date, and there was no check on the pcre_compile() error code
922        being within the table. This could lead to an OK return being given in
923        error.
925    2.  Changed the call to open a subject file in pcregrep from fopen(pathname,
926        "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
927        in a Windows environment.
929    3.  The pcregrep --count option prints the count for each file even when it is
930        zero, as does GNU grep. However, pcregrep was also printing all files when
931        --files-with-matches was added. Now, when both options are given, it prints
932        counts only for those files that have at least one match. (GNU grep just
933        prints the file name in this circumstance, but including the count seems
934        more useful - otherwise, why use --count?) Also ensured that the
935        combination -clh just lists non-zero counts, with no names.
937    4.  The long form of the pcregrep -F option was incorrectly implemented as
938        --fixed_strings instead of --fixed-strings. This is an incompatible change,
939        but it seems right to fix it, and I didn't think it was worth preserving
940        the old behaviour.
942    5.  The command line items --regex=pattern and --regexp=pattern were not
943        recognized by pcregrep, which required --regex pattern or --regexp pattern
944        (with a space rather than an '='). The man page documented the '=' forms,
945        which are compatible with GNU grep; these now work.
947    6.  No libpcreposix.pc file was created for pkg-config; there was just
948        libpcre.pc and libpcrecpp.pc. The omission has been rectified.
950    7.  Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
951        when UCP support is not needed, by modifying the Python script that
952        generates it from Unicode data files. This should not matter if the module
953        is correctly used as a library, but I received one complaint about 50K of
954        unwanted data. My guess is that the person linked everything into his
955        program rather than using a library. Anyway, it does no harm.
957    8.  A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
958        was a minimum greater than 1 for a wide character in a possessive
959        repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
960        which had an unlimited repeat of a nested, fixed maximum repeat of a wide
961        character. Chaos in the form of incorrect output or a compiling loop could
962        result.
964    9.  The restrictions on what a pattern can contain when partial matching is
965        requested for pcre_exec() have been removed. All patterns can now be
966        partially matched by this function. In addition, if there are at least two
967        slots in the offset vector, the offset of the earliest inspected character
968        for the match and the offset of the end of the subject are set in them when
969        PCRE_ERROR_PARTIAL is returned.
971    10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
972        synonymous with PCRE_PARTIAL, for backwards compatibility, and
973        PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
974        and may be more useful for multi-segment matching.
976    11. Partial matching with pcre_exec() is now more intuitive. A partial match
977        used to be given if ever the end of the subject was reached; now it is
978        given only if matching could not proceed because another character was
979        needed. This makes a difference in some odd cases such as Z(*FAIL) with the
980        string "Z", which now yields "no match" instead of "partial match". In the
981        case of pcre_dfa_exec(), "no match" is given if every matching path for the
982        final character ended with (*FAIL).
984    12. Restarting a match using pcre_dfa_exec() after a partial match did not work
985        if the pattern had a "must contain" character that was already found in the
986        earlier partial match, unless partial matching was again requested. For
987        example, with the pattern /dog.(body)?/, the "must contain" character is
988        "g". If the first part-match was for the string "dog", restarting with
989        "sbody" failed. This bug has been fixed.
991    13. The string returned by pcre_dfa_exec() after a partial match has been
992        changed so that it starts at the first inspected character rather than the
993        first character of the match. This makes a difference only if the pattern
994        starts with a lookbehind assertion or \b or \B (\K is not supported by
995        pcre_dfa_exec()). It's an incompatible change, but it makes the two
996        matching functions compatible, and I think it's the right thing to do.
998    14. Added a pcredemo man page, created automatically from the pcredemo.c file,
999        so that the demonstration program is easily available in environments where
1000        PCRE has not been installed from source.
1002    15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1003        libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1004        library.
1006    16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1007        It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1008        is not the first non-POSIX option to be added. Clearly some people find
1009        these options useful.
1011    17. If a caller to the POSIX matching function regexec() passes a non-zero
1012        value for nmatch with a NULL value for pmatch, the value of
1013        nmatch is forced to zero.
1015    18. RunGrepTest did not have a test for the availability of the -u option of
1016        the diff command, as RunTest does. It now checks in the same way as
1017        RunTest, and also checks for the -b option.
1019    19. If an odd number of negated classes containing just a single character
1020        interposed, within parentheses, between a forward reference to a named
1021        subpattern and the definition of the subpattern, compilation crashed with
1022        an internal error, complaining that it could not find the referenced
1023        subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1024        [The bug was that it was starting one character too far in when skipping
1025        over the character class, thus treating the ] as data rather than
1026        terminating the class. This meant it could skip too much.]
1028    20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1029        /g option in pcretest when the pattern contains \K, which makes it possible
1030        to have an empty string match not at the start, even when the pattern is
1031        anchored. Updated pcretest and pcredemo to use this option.
1033    21. If the maximum number of capturing subpatterns in a recursion was greater
1034        than the maximum at the outer level, the higher number was returned, but
1035        with unset values at the outer level. The correct (outer level) value is
1036        now given.
1038    22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1039        PCRE did not set those parentheses (unlike Perl). I have now found a way to
1040        make it do so. The string so far is captured, making this feature
1041        compatible with Perl.
1043    23. The tests have been re-organized, adding tests 11 and 12, to make it
1044        possible to check the Perl 5.10 features against Perl 5.10.
1046    24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1047        pattern matches a fixed length string. PCRE did not allow this; now it
1048        does. Neither allows recursion.
1050    25. I finally figured out how to implement a request to provide the minimum
1051        length of subject string that was needed in order to match a given pattern.
1052        (It was back references and recursion that I had previously got hung up
1053        on.) This code has now been added to pcre_study(); it finds a lower bound
1054        to the length of subject needed. It is not necessarily the greatest lower
1055        bound, but using it to avoid searching strings that are too short does give
1056        some useful speed-ups. The value is available to calling programs via
1057        pcre_fullinfo().
1059    26. While implementing 25, I discovered to my embarrassment that pcretest had
1060        not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1061        study optimizations had never been tested with that matching function.
1062        Oops. What is worse, even when it was passed study data, there was a bug in
1063        pcre_dfa_exec() that meant it never actually used it. Double oops. There
1064        were also very few tests of studied patterns with pcre_dfa_exec().
1066    27. If (?| is used to create subpatterns with duplicate numbers, they are now
1067        allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1068        on the other side of the coin, they are no longer allowed to have different
1069        names, because these cannot be distinguished in PCRE, and this has caused
1070        confusion. (This is a difference from Perl.)
1072    28. When duplicate subpattern names are present (necessarily with different
1073        numbers, as required by 27 above), and a test is made by name in a
1074        conditional pattern, either for a subpattern having been matched, or for
1075        recursion in such a pattern, all the associated numbered subpatterns are
1076        tested, and the overall condition is true if the condition is true for any
1077        one of them. This is the way Perl works, and is also more like the way
1078        testing by number works.
1081    Version 7.9 11-Apr-09
1082    ---------------------
1084    1.  When building with support for bzlib/zlib (pcregrep) and/or readline
1085        (pcretest), all targets were linked against these libraries. This included
1086        libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1087        libraries. This caused unwanted dependencies to be created. This problem
1088        has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1089        pcretest is linked with readline.
1091    2.  The "typedef int BOOL" in pcre_internal.h that was included inside the
1092        "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1093        moved outside it again, because FALSE and TRUE are already defined in AIX,
1094        but BOOL is not.
1096    3.  The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1097        PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1099    4.  The pcregrep documentation said spaces were inserted as well as colons (or
1100        hyphens) following file names and line numbers when outputting matching
1101        lines. This is not true; no spaces are inserted. I have also clarified the
1102        wording for the --colour (or --color) option.
1104    5.  In pcregrep, when --colour was used with -o, the list of matching strings
1105        was not coloured; this is different to GNU grep, so I have changed it to be
1106        the same.
1108    6.  When --colo(u)r was used in pcregrep, only the first matching substring in
1109        each matching line was coloured. Now it goes on to look for further matches
1110        of any of the test patterns, which is the same behaviour as GNU grep.
1112    7.  A pattern that could match an empty string could cause pcregrep to loop; it
1113        doesn't make sense to accept an empty string match in pcregrep, so I have
1114        locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1115        seems to be how GNU grep behaves.
1117    8.  The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1118        start or after a newline", because the conditional assertion was not being
1119        correctly handled. The rule now is that both the assertion and what follows
1120        in the first alternative must satisfy the test.
1122    9.  If auto-callout was enabled in a pattern with a conditional group whose
1123        condition was an assertion, PCRE could crash during matching, both with
1124        pcre_exec() and pcre_dfa_exec().
1126    10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1127        used for matching.
1129    11. Unicode property support in character classes was not working for
1130        characters (bytes) greater than 127 when not in UTF-8 mode.
1132    12. Added the -M command line option to pcretest.
1134    14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1136    15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1138    16. Added comments and documentation about mis-use of no_arg in the C++
1139        wrapper.
1141    17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1142        from Martin Jerabek that uses macro names for all relevant character and
1143        string constants.
1145    18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1146        SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1147        SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1148        these, but not everybody uses configure.
1150    19. A conditional group that had only one branch was not being correctly
1151        recognized as an item that could match an empty string. This meant that an
1152        enclosing group might also not be so recognized, causing infinite looping
1153        (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1154        with the subject "ab", where knowledge that the repeated group can match
1155        nothing is needed in order to break the loop.
1157    20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1158        exec(), but without supplying a callout function, matching went wrong.
1160    21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1161        leak if the size of the offset vector was greater than 30. When the vector
1162        is smaller, the saved offsets during recursion go onto a local stack
1163        vector, but for larger vectors malloc() is used. It was failing to free
1164        when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1165        error, in fact).
1167    22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1168        heapframe that is used only when UTF-8 support is enabled. This caused no
1169        problem, but was untidy.
1171    23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1172        CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1173        included within another project.
1175    24. Steven Van Ingelgem's patches to add more options to the CMake support,
1176        slightly modified by me:
1178          (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1179              not building pcregrep.
1181          (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1182              if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1184    25. Forward references, both numeric and by name, in patterns that made use of
1185        duplicate group numbers, could behave incorrectly or give incorrect errors,
1186        because when scanning forward to find the reference group, PCRE was not
1187        taking into account the duplicate group numbers. A pattern such as
1188        ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1190    26. Changed a few more instances of "const unsigned char *" to USPTR, making
1191        the feature of a custom pointer more persuasive (as requested by a user).
1193    27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1194        pcretest.c, inside #ifndefs, because it seems they are sometimes already
1195        pre-defined.
1197    28. Added support for (*UTF8) at the start of a pattern.
1199    29. Arrange for flags added by the "release type" setting in CMake to be shown
1200        in the configuration summary.
1203    Version 7.8 05-Sep-08
1204    ---------------------
1206    1.  Replaced UCP searching code with optimized version as implemented for Ad
1207        Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1208        stage table and inline lookup instead of a function, giving speed ups of 2
1209        to 5 times on some simple patterns that I tested. Permission was given to
1210        distribute the MultiStage2.py script that generates the tables (it's not in
1211        the tarball, but is in the Subversion repository).
1213    2.  Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1214        scripts.
1216    3.  Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1217        a group with a zero qualifier. The result of the study could be incorrect,
1218        or the function might crash, depending on the pattern.
1220    4.  Caseless matching was not working for non-ASCII characters in back
1221        references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1222        It now works when Unicode Property Support is available.
1224    5.  In pcretest, an escape such as \x{de} in the data was always generating
1225        a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1226        non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1227        truncation.
1229    6.  Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1231    7.  Added two (int) casts to pcregrep when printing the difference of two
1232        pointers, in case they are 64-bit values.
1234    8.  Added comments about Mac OS X stack usage to the pcrestack man page and to
1235        test 2 if it fails.
1237    9.  Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1238        and a #define of that name to empty if it is not externally set. This is to
1239        allow users of MSVC to set it if necessary.
1241    10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1242        the convenience functions in the pcre_get.c source file.
1244    11. An option change at the start of a pattern that had top-level alternatives
1245        could cause overwriting and/or a crash. This command provoked a crash in
1246        some environments:
1248          printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1250        This potential security problem was recorded as CVE-2008-2371.
1252    12. For a pattern where the match had to start at the beginning or immediately
1253        after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1254        pcre_dfa_exec() could read past the end of the passed subject if there was
1255        no match. To help with detecting such bugs (e.g. with valgrind), I modified
1256        pcretest so that it places the subject at the end of its malloc-ed buffer.
1258    13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1259        exec() might read past the end of the data buffer in UTF-8 mode.
1261    14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1262        the data contained the byte 0x85 as part of a UTF-8 character within its
1263        first line. This applied both to normal and DFA matching.
1265    15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1266        /^[^d]*?$/8 failed to match "abc".
1268    16. Added a missing copyright notice to pcrecpp_internal.h.
1270    17. Make it more clear in the documentation that values returned from
1271        pcre_exec() in ovector are byte offsets, not character counts.
1273    18. Tidied a few places to stop certain compilers from issuing warnings.
1275    19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1276        supplied by Stefan Weber. I made a further small update for 7.8 because
1277        there is a change of source arrangements: the pcre_searchfuncs.c module is
1278        replaced by pcre_ucd.c.
1281    Version 7.7 07-May-08
1282    ---------------------
1284    1.  Applied Craig's patch to sort out a long long problem: "If we can't convert
1285        a string to a long long, pretend we don't even have a long long." This is
1286        done by checking for the strtoq, strtoll, and _strtoi64 functions.
1288    2.  Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1289        pre-7.6 versions, which defined a global no_arg variable instead of putting
1290        it in the RE class. (See also #8 below.)
1292    3.  Remove a line of dead code, identified by coverity and reported by Nuno
1293        Lopes.
1295    4.  Fixed two related pcregrep bugs involving -r with --include or --exclude:
1297        (1) The include/exclude patterns were being applied to the whole pathnames
1298            of files, instead of just to the final components.
1300        (2) If there was more than one level of directory, the subdirectories were
1301            skipped unless they satisfied the include/exclude conditions. This is
1302            inconsistent with GNU grep (and could even be seen as contrary to the
1303            pcregrep specification - which I improved to make it absolutely clear).
1304            The action now is always to scan all levels of directory, and just
1305            apply the include/exclude patterns to regular files.
1307    5.  Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1308        --exclude_dir in the tests to avoid scanning .svn directories.
1310    6.  Applied Craig's patch to the QuoteMeta function so that it escapes the
1311        NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1312        doesn't support NULs in patterns.
1314    7.  Added some missing "const"s to declarations of static tables in
1315        pcre_compile.c and pcre_dfa_exec.c.
1317    8.  Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1318        caused by fix #2  above. (Subsequently also a second patch to fix the
1319        first patch. And a third patch - this was a messy problem.)
1321    9.  Applied Craig's patch to remove the use of push_back().
1323    10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1324        matching function regexec().
1326    11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1327        which, however, unlike Perl's \g{...}, are subroutine calls, not back
1328        references. PCRE supports relative numbers with this syntax (I don't think
1329        Oniguruma does).
1331    12. Previously, a group with a zero repeat such as (...){0} was completely
1332        omitted from the compiled regex. However, this means that if the group
1333        was called as a subroutine from elsewhere in the pattern, things went wrong
1334        (an internal error was given). Such groups are now left in the compiled
1335        pattern, with a new opcode that causes them to be skipped at execution
1336        time.
1338    13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1339        to the way PCRE behaves:
1341        (a) A lone ] character is dis-allowed (Perl treats it as data).
1343        (b) A back reference to an unmatched subpattern matches an empty string
1344            (Perl fails the current match path).
1346        (c) A data ] in a character class must be notated as \] because if the
1347            first data character in a class is ], it defines an empty class. (In
1348            Perl it is not possible to have an empty class.) The empty class []
1349            never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1350            The negative empty class [^] matches any one character, independently
1351            of the DOTALL setting.
1353    14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1354        non-existent subpattern following a character class starting with ']' and
1355        containing () gave an internal compiling error instead of "reference to
1356        non-existent subpattern". Fortunately, when the pattern did exist, the
1357        compiled code was correct. (When scanning forwards to check for the
1358        existencd of the subpattern, it was treating the data ']' as terminating
1359        the class, so got the count wrong. When actually compiling, the reference
1360        was subsequently set up correctly.)
1362    15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1363        it was being rejected as not supported by pcre_dfa_exec(), even though
1364        other assertions are supported. I have made pcre_dfa_exec() support
1365        (*FAIL).
1367    16. The implementation of 13c above involved the invention of a new opcode,
1368        OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1369        cannot be changed at match time, I realized I could make a small
1370        improvement to matching performance by compiling OP_ALLANY instead of
1371        OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1372        on the OP_ANY path.
1374    17. Compiling pcretest on Windows with readline support failed without the
1375        following two fixes: (1) Make the unistd.h include conditional on
1376        HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1378    18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1379        ncurses library to be included for pcretest when ReadLine support is
1380        requested, but also to allow for it to be overridden. This patch came from
1381        Daniel Bergström.
1383    19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1384        as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1385        any errors with the current Unicode tables. Thanks to Peter Kankowski for
1386        spotting this.
1389    Version 7.6 28-Jan-08
1390    ---------------------
1392    1.  A character class containing a very large number of characters with
1393        codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1394        overflow.
1396    2.  Patch to cut out the "long long" test in pcrecpp_unittest when
1397        HAVE_LONG_LONG is not defined.
1399    3.  Applied Christian Ehrlicher's patch to update the CMake build files to
1400        bring them up to date and include new features. This patch includes:
1402        - Fixed PH's badly added libz and libbz2 support.
1403        - Fixed a problem with static linking.
1404        - Added pcredemo. [But later removed - see 7 below.]
1405        - Fixed dftables problem and added an option.
1406        - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1407            HAVE_LONG_LONG.
1408        - Added readline support for pcretest.
1409        - Added an listing of the option settings after cmake has run.
1411    4.  A user submitted a patch to Makefile that makes it easy to create
1412        "pcre.dll" under mingw when using Configure/Make. I added stuff to
1413        Makefile.am that cause it to include this special target, without
1414        affecting anything else. Note that the same mingw target plus all
1415        the other distribution libraries and programs are now supported
1416        when configuring with CMake (see 6 below) instead of with
1417        Configure/Make.
1419    5.  Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1420        This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1421        exported in the Windows port". It has not yet been confirmed that the patch
1422        solves the problem, but it does no harm.
1424    6.  Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1425        NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1426        with CMake, and also correct the comment about stack recursion.
1428    7.  Remove the automatic building of pcredemo from the ./configure system and
1429        from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1430        of a program that users should build themselves after PCRE is installed, so
1431        building it automatically is not really right. What is more, it gave
1432        trouble in some build environments.
1434    8.  Further tidies to CMakeLists.txt from Sheri and Christian.
1437    Version 7.5 10-Jan-08
1438    ---------------------
1440    1.  Applied a patch from Craig: "This patch makes it possible to 'ignore'
1441        values in parens when parsing an RE using the C++ wrapper."
1443    2.  Negative specials like \S did not work in character classes in UTF-8 mode.
1444        Characters greater than 255 were excluded from the class instead of being
1445        included.
1447    3.  The same bug as (2) above applied to negated POSIX classes such as
1448        [:^space:].
1450    4.  PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1451        defined or documented. It seems to have been a typo for PCRE_STATIC, so
1452        I have changed it.
1454    5.  The construct (?&) was not diagnosed as a syntax error (it referenced the
1455        first named subpattern) and a construct such as (?&a) would reference the
1456        first named subpattern whose name started with "a" (in other words, the
1457        length check was missing). Both these problems are fixed. "Subpattern name
1458        expected" is now given for (?&) (a zero-length name), and this patch also
1459        makes it give the same error for \k'' (previously it complained that that
1460        was a reference to a non-existent subpattern).
1462    6.  The erroneous patterns (?+-a) and (?-+a) give different error messages;
1463        this is right because (?- can be followed by option settings as well as by
1464        digits. I have, however, made the messages clearer.
1466    7.  Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1467        than the number used in the conditional) now cause a compile-time error.
1468        This is actually not compatible with Perl, which accepts such patterns, but
1469        treats the conditional as always being FALSE (as PCRE used to), but it
1470        seems to me that giving a diagnostic is better.
1472    8.  Change "alphameric" to the more common word "alphanumeric" in comments
1473        and messages.
1475    9.  Fix two occurrences of "backslash" in comments that should have been
1476        "backspace".
1478    10. Remove two redundant lines of code that can never be obeyed (their function
1479        was moved elsewhere).
1481    11. The program that makes PCRE's Unicode character property table had a bug
1482        which caused it to generate incorrect table entries for sequences of
1483        characters that have the same character type, but are in different scripts.
1484        It amalgamated them into a single range, with the script of the first of
1485        them. In other words, some characters were in the wrong script. There were
1486        thirteen such cases, affecting characters in the following ranges:
1488          U+002b0 - U+002c1
1489          U+0060c - U+0060d
1490          U+0061e - U+00612
1491          U+0064b - U+0065e
1492          U+0074d - U+0076d
1493          U+01800 - U+01805
1494          U+01d00 - U+01d77
1495          U+01d9b - U+01dbf
1496          U+0200b - U+0200f
1497          U+030fc - U+030fe
1498          U+03260 - U+0327f
1499          U+0fb46 - U+0fbb1
1500          U+10450 - U+1049d
1502    12. The -o option (show only the matching part of a line) for pcregrep was not
1503        compatible with GNU grep in that, if there was more than one match in a
1504        line, it showed only the first of them. It now behaves in the same way as
1505        GNU grep.
1507    13. If the -o and -v options were combined for pcregrep, it printed a blank
1508        line for every non-matching line. GNU grep prints nothing, and pcregrep now
1509        does the same. The return code can be used to tell if there were any
1510        non-matching lines.
1512    14. Added --file-offsets and --line-offsets to pcregrep.
1514    15. The pattern (?=something)(?R) was not being diagnosed as a potentially
1515        infinitely looping recursion. The bug was that positive lookaheads were not
1516        being skipped when checking for a possible empty match (negative lookaheads
1517        and both kinds of lookbehind were skipped).
1519    16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
1520        inclusion of <windows.h> to before rather than after the definition of
1521        INVALID_FILE_ATTRIBUTES (patch from David Byron).
1523    17. Specifying a possessive quantifier with a specific limit for a Unicode
1524        character property caused pcre_compile() to compile bad code, which led at
1525        runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
1526        are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
1527        caused the error; without that there was no problem.
1529    18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
1531    19. Added --enable-pcretest-libreadline.
1533    20. In pcrecpp.cc, the variable 'count' was incremented twice in
1534        RE::GlobalReplace(). As a result, the number of replacements returned was
1535        double what it should be. I removed one of the increments, but Craig sent a
1536        later patch that removed the other one (the right fix) and added unit tests
1537        that check the return values (which was not done before).
1539    21. Several CMake things:
1541        (1) Arranged that, when cmake is used on Unix, the libraries end up with
1542            the names libpcre and libpcreposix, not just pcre and pcreposix.
1544        (2) The above change means that pcretest and pcregrep are now correctly
1545            linked with the newly-built libraries, not previously installed ones.
1549    22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
1550        crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
1551        UTF-8 newline character). The key issue is that the pattern starts .*;
1552        this means that the match must be either at the beginning, or after a
1553        newline. The bug was in the code for advancing after a failed match and
1554        checking that the new position followed a newline. It was not taking
1555        account of UTF-8 characters correctly.
1557    23. PCRE was behaving differently from Perl in the way it recognized POSIX
1558        character classes. PCRE was not treating the sequence [:...:] as a
1559        character class unless the ... were all letters. Perl, however, seems to
1560        allow any characters between [: and :], though of course it rejects as
1561        unknown any "names" that contain non-letters, because all the known class
1562        names consist only of letters. Thus, Perl gives an error for [[:1234:]],
1563        for example, whereas PCRE did not - it did not recognize a POSIX character
1564        class. This seemed a bit dangerous, so the code has been changed to be
1565        closer to Perl. The behaviour is not identical to Perl, because PCRE will
1566        diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
1567        treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
1568        Perl does, and where it didn't before.
1570    24. Rewrite so as to remove the single use of %n from pcregrep because in some
1571        Windows environments %n is disabled by default.
1574    Version 7.4 21-Sep-07
1575    ---------------------
1577    1.  Change 7.3/28 was implemented for classes by looking at the bitmap. This
1578        means that a class such as [\s] counted as "explicit reference to CR or
1579        LF". That isn't really right - the whole point of the change was to try to
1580        help when there was an actual mention of one of the two characters. So now
1581        the change happens only if \r or \n (or a literal CR or LF) character is
1582        encountered.
1584    2.  The 32-bit options word was also used for 6 internal flags, but the numbers
1585        of both had grown to the point where there were only 3 bits left.
1586        Fortunately, there was spare space in the data structure, and so I have
1587        moved the internal flags into a new 16-bit field to free up more option
1588        bits.
1590    3.  The appearance of (?J) at the start of a pattern set the DUPNAMES option,
1591        but did not set the internal JCHANGED flag - either of these is enough to
1592        control the way the "get" function works - but the PCRE_INFO_JCHANGED
1593        facility is supposed to tell if (?J) was ever used, so now (?J) at the
1594        start sets both bits.
1596    4.  Added options (at build time, compile time, exec time) to change \R from
1597        matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
1599    5.  doc/pcresyntax.html was missing from the distribution.
1601    6.  Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
1602        compatibility, even though it is no longer used.
1604    7.  Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
1605        strtoull to pcrecpp.cc to select the available functions in WIN32 when the
1606        windows.h file is present (where different names are used). [This was
1607        reversed later after testing - see 16 below.]
1609    8.  Changed all #include <config.h> to #include "config.h". There were also
1610        some further <pcre.h> cases that I changed to "pcre.h".
1612    9.  When pcregrep was used with the --colour option, it missed the line ending
1613        sequence off the lines that it output.
1615    10. It was pointed out to me that arrays of string pointers cause lots of
1616        relocations when a shared library is dynamically loaded. A technique of
1617        using a single long string with a table of offsets can drastically reduce
1618        these. I have refactored PCRE in four places to do this. The result is
1619        dramatic:
1621          Originally:                          290
1622          After changing UCP table:            187
1623          After changing error message table:   43
1624          After changing table of "verbs"       36
1625          After changing table of Posix names   22
1627        Thanks to the folks working on Gregex for glib for this insight.
1629    11. --disable-stack-for-recursion caused compiling to fail unless -enable-
1630        unicode-properties was also set.
1632    12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
1634    13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
1635        checked only for CRLF.
1637    14. Added casts to pcretest.c to avoid compiler warnings.
1639    15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
1641    16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
1642        and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
1643        entirely. This removes changes made in 7 above.
1645    17. The CMake files have been updated, and there is now more information about
1646        building with CMake in the NON-UNIX-USE document.
1649    Version 7.3 28-Aug-07
1650    ---------------------
1652     1. In the rejigging of the build system that eventually resulted in 7.1, the
1653        line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
1654        brackets there is not right, since it causes compilers to look for an
1655        installed pcre.h, not the version that is in the source that is being
1656        compiled (which of course may be different). I have changed it back to:
1658          #include "pcre.h"
1660        I have a vague recollection that the change was concerned with compiling in
1661        different directories, but in the new build system, that is taken care of
1662        by the VPATH setting the Makefile.
1664     2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
1665        when the subject happened to end in the byte 0x85 (e.g. if the last
1666        character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
1667        characters but of course it shouldn't be taken as a newline when it is part
1668        of another character. The bug was that, for an unlimited repeat of . in
1669        not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
1670        characters when looking for a newline.
1672     3. A small performance improvement in the DOTALL UTF-8 mode .* case.
1674     4. Debugging: adjusted the names of opcodes for different kinds of parentheses
1675        in debug output.
1677     5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
1678        long printing in the pcrecpp unittest when running under MinGW.
1680     6. ESC_K was left out of the EBCDIC table.
1682     7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
1683        parentheses; I made it 1000, which seemed large enough. Unfortunately, the
1684        limit also applies to "virtual nesting" when a pattern is recursive, and in
1685        this case 1000 isn't so big. I have been able to remove this limit at the
1686        expense of backing off one optimization in certain circumstances. Normally,
1687        when pcre_exec() would call its internal match() function recursively and
1688        immediately return the result unconditionally, it uses a "tail recursion"
1689        feature to save stack. However, when a subpattern that can match an empty
1690        string has an unlimited repetition quantifier, it no longer makes this
1691        optimization. That gives it a stack frame in which to save the data for
1692        checking that an empty string has been matched. Previously this was taken
1693        from the 1000-entry workspace that had been reserved. So now there is no
1694        explicit limit, but more stack is used.
1696     8. Applied Daniel's patches to solve problems with the import/export magic
1697        syntax that is required for Windows, and which was going wrong for the
1698        pcreposix and pcrecpp parts of the library. These were overlooked when this
1699        problem was solved for the main library.
1701     9. There were some crude static tests to avoid integer overflow when computing
1702        the size of patterns that contain repeated groups with explicit upper
1703        limits. As the maximum quantifier is 65535, the maximum group length was
1704        set at 30,000 so that the product of these two numbers did not overflow a
1705        32-bit integer. However, it turns out that people want to use groups that
1706        are longer than 30,000 bytes (though not repeat them that many times).
1707        Change 7.0/17 (the refactoring of the way the pattern size is computed) has
1708        made it possible to implement the integer overflow checks in a much more
1709        dynamic way, which I have now done. The artificial limitation on group
1710        length has been removed - we now have only the limit on the total length of
1711        the compiled pattern, which depends on the LINK_SIZE setting.
1713    10. Fixed a bug in the documentation for get/copy named substring when
1714        duplicate names are permitted. If none of the named substrings are set, the
1715        functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
1716        empty string.
1718    11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
1719        instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
1720        because the ] is interpreted as the first data character and the
1721        terminating ] is not found. PCRE has been made compatible with Perl in this
1722        regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
1723        cause memory overwriting.
1725    10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
1726        string has been matched (to stop an infinite loop). It was not recognizing
1727        a conditional subpattern that could match an empty string if that
1728        subpattern was within another subpattern. For example, it looped when
1729        trying to match  (((?(1)X|))*)  but it was OK with  ((?(1)X|)*)  where the
1730        condition was not nested. This bug has been fixed.
1732    12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
1733        past the start of the subject in the presence of bytes with the top bit
1734        set, for example "\x8aBCD".
1736    13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
1737        (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
1739    14. Optimized (?!) to (*FAIL).
1741    15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
1742        This restricts code points to be within the range 0 to 0x10FFFF, excluding
1743        the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
1744        full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
1745        does: it's just the validity check that is more restrictive.
1747    16. Inserted checks for integer overflows during escape sequence (backslash)
1748        processing, and also fixed erroneous offset values for syntax errors during
1749        backslash processing.
1751    17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
1752        for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
1754    18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
1755        caused an overrun.
1757    19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
1758        something other than just ASCII characters) inside a group that had an
1759        unlimited repeat caused a loop at compile time (while checking to see
1760        whether the group could match an empty string).
1762    20. Debugging a pattern containing \p or \P could cause a crash. For example,
1763        [\P{Any}] did so. (Error in the code for printing property names.)
1765    21. An orphan \E inside a character class could cause a crash.
1767    22. A repeated capturing bracket such as (A)? could cause a wild memory
1768        reference during compilation.
1770    23. There are several functions in pcre_compile() that scan along a compiled
1771        expression for various reasons (e.g. to see if it's fixed length for look
1772        behind). There were bugs in these functions when a repeated \p or \P was
1773        present in the pattern. These operators have additional parameters compared
1774        with \d, etc, and these were not being taken into account when moving along
1775        the compiled data. Specifically:
1777        (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
1778            length.
1780        (b) An item such as \pL+ within a repeated group could cause crashes or
1781            loops.
1783        (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
1784            "reference to non-existent subpattern" error.
1786        (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
1788    24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
1789        characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
1791    25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
1793    26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
1794        character were causing crashes (broken optimization).
1796    27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
1797        \p or \P) caused a compile-time loop.
1799    28. More problems have arisen in unanchored patterns when CRLF is a valid line
1800        break. For example, the unstudied pattern [\r\n]A does not match the string
1801        "\r\nA" because change 7.0/46 below moves the current point on by two
1802        characters after failing to match at the start. However, the pattern \nA
1803        *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
1804        the same is true. There doesn't seem any very clean way out of this, but
1805        what I have chosen to do makes the common cases work: PCRE now takes note
1806        of whether there can be an explicit match for \r or \n anywhere in the
1807        pattern, and if so, 7.0/46 no longer applies. As part of this change,
1808        there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
1809        pattern has explicit CR or LF references.
1811    29. Added (*CR) etc for changing newline setting at start of pattern.
1814    Version 7.2 19-Jun-07
1815    ---------------------
1817     1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
1818        which is apparently normally available under Windows.
1820     2. Re-jig the pcregrep tests with different newline settings in an attempt
1821        to make them independent of the local environment's newline setting.
1823     3. Add code to configure.ac to remove -g from the CFLAGS default settings.
1825     4. Some of the "internals" tests were previously cut out when the link size
1826        was not 2, because the output contained actual offsets. The recent new
1827        "Z" feature of pcretest means that these can be cut out, making the tests
1828        usable with all link sizes.
1830     5. Implemented Stan Switzer's goto replacement for longjmp() when not using
1831        stack recursion. This gives a massive performance boost under BSD, but just
1832        a small improvement under Linux. However, it saves one field in the frame
1833        in all cases.
1835     6. Added more features from the forthcoming Perl 5.10:
1837        (a) (?-n) (where n is a string of digits) is a relative subroutine or
1838            recursion call. It refers to the nth most recently opened parentheses.
1840        (b) (?+n) is also a relative subroutine call; it refers to the nth next
1841            to be opened parentheses.
1843        (c) Conditions that refer to capturing parentheses can be specified
1844            relatively, for example, (?(-2)... or (?(+3)...
1846        (d) \K resets the start of the current match so that everything before
1847            is not part of it.
1849        (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
1851        (f) \g{name} is another synonym - part of Perl 5.10's unification of
1852            reference syntax.
1854        (g) (?| introduces a group in which the numbering of parentheses in each
1855            alternative starts with the same number.
1857        (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
1859     7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
1862     8. A pattern such as  (.*(.)?)*  caused pcre_exec() to fail by either not
1863        terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
1864        for detecting groups that can match an empty string.
1866     9. A pattern with a very large number of alternatives (more than several
1867        hundred) was running out of internal workspace during the pre-compile
1868        phase, where pcre_compile() figures out how much memory will be needed. A
1869        bit of new cunning has reduced the workspace needed for groups with
1870        alternatives. The 1000-alternative test pattern now uses 12 bytes of
1871        workspace instead of running out of the 4096 that are available.
1873    10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
1875    11. Applied patch from Google to remove an optimization that didn't quite work.
1876        The report of the bug said:
1878          pcrecpp::RE("a*").FullMatch("aaa") matches, while
1879          pcrecpp::RE("a*?").FullMatch("aaa") does not, and
1880          pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
1882    12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
1883        it matched the wrong number of bytes.
1886    Version 7.1 24-Apr-07
1887    ---------------------
1889     1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
1890        that is more "standard", making use of automake and other Autotools. There
1891        is some re-arrangement of the files and adjustment of comments consequent
1892        on this.
1894     2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
1895        for recursive directory scanning broke on some systems because the files
1896        are not scanned in any specific order and on different systems the order
1897        was different. A call to "sort" has been inserted into RunGrepTest for the
1898        approprate test as a short-term fix. In the longer term there may be an
1899        alternative.
1901     3. I had an email from Eric Raymond about problems translating some of PCRE's
1902        man pages to HTML (despite the fact that I distribute HTML pages, some
1903        people do their own conversions for various reasons). The problems
1904        concerned the use of low-level troff macros .br and .in. I have therefore
1905        removed all such uses from the man pages (some were redundant, some could
1906        be replaced by .nf/.fi pairs). The 132html script that I use to generate
1907        HTML has been updated to handle .nf/.fi and to complain if it encounters
1908        .br or .in.
1910     4. Updated comments in configure.ac that get placed in config.h.in and also
1911        arranged for config.h to be included in the distribution, with the name
1912        config.h.generic, for the benefit of those who have to compile without
1913        Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
1915     5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
1916        Weber: (1) pcre_internal.h was missing some function renames; (2) updated
1917        makevp.bat for the current PCRE, using the additional files
1918        makevp_c.txt, makevp_l.txt, and pcregexp.pas.
1920     6. A Windows user reported a minor discrepancy with test 2, which turned out
1921        to be caused by a trailing space on an input line that had got lost in his
1922        copy. The trailing space was an accident, so I've just removed it.
1924     7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
1925        that is needed.
1927     8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
1928        as "const" (a) because they are and (b) because it helps the PHP
1929        maintainers who have recently made a script to detect big data structures
1930        in the php code that should be moved to the .rodata section. I remembered
1931        to update Builducptable as well, so it won't revert if ucptable.h is ever
1932        re-created.
1934     9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
1935        pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
1936        order to be able to cut out the UTF-8 tables in the latter when UTF-8
1937        support is not required. This saves 1.5-2K of code, which is important in
1938        some applications.
1940        Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
1941        so as not to refer to the tables, even though these functions will never be
1942        called when UTF-8 support is disabled. Otherwise there are problems with a
1943        shared library.
1945    10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
1947        (a) It was defining its arguments as char * instead of void *.
1949        (b) It was assuming that all moves were upwards in memory; this was true
1950            a long time ago when I wrote it, but is no longer the case.
1952        The emulated memove() is provided for those environments that have neither
1953        memmove() nor bcopy(). I didn't think anyone used it these days, but that
1954        is clearly not the case, as these two bugs were recently reported.
1956    11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
1957        and Detrail to create the HTML documentation, the .txt form of the man
1958        pages, and it removes trailing spaces from listed files. It also creates
1959        pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
1960        case, it wraps all the #defines with #ifndefs. This script should be run
1961        before "make dist".
1963    12. Fixed two fairly obscure bugs concerned with quantified caseless matching
1964        with Unicode property support.
1966        (a) For a maximizing quantifier, if the two different cases of the
1967            character were of different lengths in their UTF-8 codings (there are
1968            some cases like this - I found 11), and the matching function had to
1969            back up over a mixture of the two cases, it incorrectly assumed they
1970            were both the same length.
1972        (b) When PCRE was configured to use the heap rather than the stack for
1973            recursion during matching, it was not correctly preserving the data for
1974            the other case of a UTF-8 character when checking ahead for a match
1975            while processing a minimizing repeat. If the check also involved
1976            matching a wide character, but failed, corruption could cause an
1977            erroneous result when trying to check for a repeat of the original
1978            character.
1980    13. Some tidying changes to the testing mechanism:
1982        (a) The RunTest script now detects the internal link size and whether there
1983            is UTF-8 and UCP support by running ./pcretest -C instead of relying on
1984            values substituted by "configure". (The RunGrepTest script already did
1985            this for UTF-8.) The configure.ac script no longer substitutes the
1986            relevant variables.
1988        (b) The debugging options /B and /D in pcretest show the compiled bytecode
1989            with length and offset values. This means that the output is different
1990            for different internal link sizes. Test 2 is skipped for link sizes
1991            other than 2 because of this, bypassing the problem. Unfortunately,
1992            there was also a test in test 3 (the locale tests) that used /B and
1993            failed for link sizes other than 2. Rather than cut the whole test out,
1994            I have added a new /Z option to pcretest that replaces the length and
1995            offset values with spaces. This is now used to make test 3 independent
1996            of link size. (Test 2 will be tidied up later.)
1998    14. If erroroffset was passed as NULL to pcre_compile, it provoked a
1999        segmentation fault instead of returning the appropriate error message.
2001    15. In multiline mode when the newline sequence was set to "any", the pattern
2002        ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2003        This doesn't seem right; it now treats the CRLF combination as the line
2004        ending, and so does not match in that case. It's only a pattern such as ^$
2005        that would hit this one: something like ^ABC$ would have failed after \r
2006        and then tried again after \r\n.
2008    16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2009        in an attempt to make files that differ only in their line terminators
2010        compare equal. This works on Linux.
2012    17. Under certain error circumstances pcregrep might try to free random memory
2013        as it exited. This is now fixed, thanks to valgrind.
2015    19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2016        "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2017        was because its rules for how to advance for /g after matching an empty
2018        string at the end of a line did not allow for this case. They now check for
2019        it specially.
2021    20. pcretest is supposed to handle patterns and data of any length, by
2022        extending its buffers when necessary. It was getting this wrong when the
2023        buffer for a data line had to be extended.
2025    21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2026        CRLF as a newline sequence.
2028    22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2029        out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2030        I have nevertheless tidied it up.
2032    23. Added some casts to kill warnings from HP-UX ia64 compiler.
2034    24. Added a man page for pcre-config.
2037    Version 7.0 19-Dec-06
2038    ---------------------
2040     1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2041        moving to gcc 4.1.1.
2043     2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2044        sys/time.h, which is documented as needed for this function. It doesn't
2045        seem to matter on Linux, but it showed up on some releases of OS X.
2047     3. It seems that there are systems where bytes whose values are greater than
2048        127 match isprint() in the "C" locale. The "C" locale should be the
2049        default when a C program starts up. In most systems, only ASCII printing
2050        characters match isprint(). This difference caused the output from pcretest
2051        to vary, making some of the tests fail. I have changed pcretest so that:
2053        (a) When it is outputting text in the compiled version of a pattern, bytes
2054            other than 32-126 are always shown as hex escapes.
2056        (b) When it is outputting text that is a matched part of a subject string,
2057            it does the same, unless a different locale has been set for the match
2058            (using the /L modifier). In this case, it uses isprint() to decide.
2060     4. Fixed a major bug that caused incorrect computation of the amount of memory
2061        required for a compiled pattern when options that changed within the
2062        pattern affected the logic of the preliminary scan that determines the
2063        length. The relevant options are -x, and -i in UTF-8 mode. The result was
2064        that the computed length was too small. The symptoms of this bug were
2065        either the PCRE error "internal error: code overflow" from pcre_compile(),
2066        or a glibc crash with a message such as "pcretest: free(): invalid next
2067        size (fast)". Examples of patterns that provoked this bug (shown in
2068        pcretest format) are:
2070          /(?-x: )/x
2071          /(?x)(?-x: \s*#\s*)/
2072          /((?i)[\x{c0}])/8
2073          /(?i:[\x{c0}])/8
2075        HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2076        is now done differently.
2078     5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2079        wrapper classes; (b) implement a new function in the C++ scanner that is
2080        more efficient than the old way of doing things because it avoids levels of
2081        recursion in the regex matching; (c) add a paragraph to the documentation
2082        for the FullMatch() function.
2084     6. The escape sequence \n was being treated as whatever was defined as
2085        "newline". Not only was this contrary to the documentation, which states
2086        that \n is character 10 (hex 0A), but it also went horribly wrong when
2087        "newline" was defined as CRLF. This has been fixed.
2089     7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2090        was being set to -1 for the "end of line" case (supposedly a value that no
2091        character can have). Though this value is never used (the check for end of
2092        line is "zero bytes in current character"), it caused compiler complaints.
2093        I've changed it to 0xffffffff.
2095     8. In pcre_version.c, the version string was being built by a sequence of
2096        C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2097        string (as it is for production releases) called a macro with an empty
2098        argument. The C standard says the result of this is undefined. The gcc
2099        compiler treats it as an empty string (which was what was wanted) but it is
2100        reported that Visual C gives an error. The source has been hacked around to
2101        avoid this problem.
2103     9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2104        builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2105        instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2106        of them did).
2108    10. Originally, pcretest opened its input and output without "b"; then I was
2109        told that "b" was needed in some environments, so it was added for release
2110        5.0 to both the input and output. (It makes no difference on Unix-like
2111        systems.) Later I was told that it is wrong for the input on Windows. I've
2112        now abstracted the modes into two macros, to make it easier to fiddle with
2113        them, and removed "b" from the input mode under Windows.
2115    11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2117    12. Added -help and --help to pcretest as an official way of being reminded
2118        of the options.
2120    13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2121        and pcrecpp.cc because they annoy compilers at high warning levels.
2123    14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2125    15. Fixed an occurrence of == in configure.ac that should have been = (shell
2126        scripts are not C programs :-) and which was not noticed because it works
2127        on Linux.
2129    16. pcretest is supposed to handle any length of pattern and data line (as one
2130        line or as a continued sequence of lines) by extending its input buffer if
2131        necessary. This feature was broken for very long pattern lines, leading to
2132        a string of junk being passed to pcre_compile() if the pattern was longer
2133        than about 50K.
2135    17. I have done a major re-factoring of the way pcre_compile() computes the
2136        amount of memory needed for a compiled pattern. Previously, there was code
2137        that made a preliminary scan of the pattern in order to do this. That was
2138        OK when PCRE was new, but as the facilities have expanded, it has become
2139        harder and harder to keep it in step with the real compile phase, and there
2140        have been a number of bugs (see for example, 4 above). I have now found a
2141        cunning way of running the real compile function in a "fake" mode that
2142        enables it to compute how much memory it would need, while actually only
2143        ever using a few hundred bytes of working memory and without too many
2144        tests of the mode. This should make future maintenance and development
2145        easier. A side effect of this work is that the limit of 200 on the nesting
2146        depth of parentheses has been removed (though this was never a serious
2147        limitation, I suspect). However, there is a downside: pcre_compile() now
2148        runs more slowly than before (30% or more, depending on the pattern). I
2149        hope this isn't a big issue. There is no effect on runtime performance.
2151    18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2152        newline (only possible for the last line of a file) and it was a
2153        pattern that set a locale (followed by /Lsomething), pcretest crashed.
2155    19. Added additional timing features to pcretest. (1) The -tm option now times
2156        matching only, not compiling. (2) Both -t and -tm can be followed, as a
2157        separate command line item, by a number that specifies the number of
2158        repeats to use when timing. The default is 50000; this gives better
2159        precision, but takes uncomfortably long for very large patterns.
2161    20. Extended pcre_study() to be more clever in cases where a branch of a
2162        subpattern has no definite first character. For example, (a*|b*)[cd] would
2163        previously give no result from pcre_study(). Now it recognizes that the
2164        first character must be a, b, c, or d.
2166    21. There was an incorrect error "recursive call could loop indefinitely" if
2167        a subpattern (or the entire pattern) that was being tested for matching an
2168        empty string contained only one non-empty item after a nested subpattern.
2169        For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2170        incorrectly, because the \d was being skipped in the check.
2172    22. The pcretest program now has a new pattern option /B and a command line
2173        option -b, which is equivalent to adding /B to every pattern. This causes
2174        it to show the compiled bytecode, without the additional information that
2175        -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2176        is the same as /B/I).
2178    23. A new optimization is now able automatically to treat some sequences such
2179        as a*b as a*+b. More specifically, if something simple (such as a character
2180        or a simple class like \d) has an unlimited quantifier, and is followed by
2181        something that cannot possibly match the quantified thing, the quantifier
2182        is automatically "possessified".
2184    24. A recursive reference to a subpattern whose number was greater than 39
2185        went wrong under certain circumstances in UTF-8 mode. This bug could also
2186        have affected the operation of pcre_study().
2188    25. Realized that a little bit of performance could be had by replacing
2189        (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2191    26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2193    27. Possessive quantifiers such as a++ were previously implemented by turning
2194        them into atomic groups such as ($>a+). Now they have their own opcodes,
2195        which improves performance. This includes the automatically created ones
2196        from 23 above.
2198    28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2199        lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2200        the first matched character to be a colon. This applied both to named and
2201        numbered groups.
2203    29. The ucpinternal.h header file was missing its idempotency #ifdef.
2205    30. I was sent a "project" file called libpcre.a.dev which I understand makes
2206        building PCRE on Windows easier, so I have included it in the distribution.
2208    31. There is now a check in pcretest against a ridiculously large number being
2209        returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2210        loop, the loop is abandoned.
2212    32. Forward references to subpatterns in conditions such as (?(2)...) where
2213        subpattern 2 is defined later cause pcre_compile() to search forwards in
2214        the pattern for the relevant set of parentheses. This search went wrong
2215        when there were unescaped parentheses in a character class, parentheses
2216        escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2218    33. "Subroutine" calls and backreferences were previously restricted to
2219        referencing subpatterns earlier in the regex. This restriction has now
2220        been removed.
2222    34. Added a number of extra features that are going to be in Perl 5.10. On the
2223        whole, these are just syntactic alternatives for features that PCRE had
2224        previously implemented using the Python syntax or my own invention. The
2225        other formats are all retained for compatibility.
2227        (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2228            as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2229            also .NET compatible.
2231        (b) A recursion or subroutine call to a named group can now be defined as
2232            (?&name) as well as (?P>name).
2234        (c) A backreference to a named group can now be defined as \k<name> or
2235            \k'name' as well as (?P=name). The new forms, as well as being in Perl
2236            5.10, are also .NET compatible.
2238        (d) A conditional reference to a named group can now use the syntax
2239            (?(<name>) or (?('name') as well as (?(name).
2241        (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2242            groups (named and numbered) that are never evaluated inline, but can be
2243            called as "subroutines" from elsewhere. In effect, the DEFINE condition
2244            is always false. There may be only one alternative in such a group.
2246        (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2247            as the simple (?(R). The condition is true only if the most recent
2248            recursion is that of the given number or name. It does not search out
2249            through the entire recursion stack.
2251        (g) The escape \gN or \g{N} has been added, where N is a positive or
2252            negative number, specifying an absolute or relative reference.
2254    35. Tidied to get rid of some further signed/unsigned compiler warnings and
2255        some "unreachable code" warnings.
2257    36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2258        things, this adds five new scripts.
2260    37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2261        There were also incompatibilities regarding the handling of \Q..\E inside
2262        character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2263        hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2265    38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2266        matches an empty string, and forcibly breaks the loop. There were bugs in
2267        this code in non-simple cases. For a pattern such as  ^(a()*)*  matched
2268        against  aaaa  the result was just "a" rather than "aaaa", for example. Two
2269        separate and independent bugs (that affected different cases) have been
2270        fixed.
2272    39. Refactored the code to abolish the use of different opcodes for small
2273        capturing bracket numbers. This is a tidy that I avoided doing when I
2274        removed the limit on the number of capturing brackets for 3.5 back in 2001.
2275        The new approach is not only tidier, it makes it possible to reduce the
2276        memory needed to fix the previous bug (38).
2278    40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2279        sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2280        processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2281        mode.
2283    41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2284        report.
2286    42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2287        copy construction and assignment in the C++ wrapper.
2289    43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2290        couple of bugs that could have given wrong results in the "--newline=crlf"
2291        case.
2293    44. Added a number of casts and did some reorganization of signed/unsigned int
2294        variables following suggestions from Dair Grant. Also renamed the variable
2295        "this" as "item" because it is a C++ keyword.
2297    45. Arranged for dftables to add
2299          #include "pcre_internal.h"
2301        to pcre_chartables.c because without it, gcc 4.x may remove the array
2302        definition from the final binary if PCRE is built into a static library and
2303        dead code stripping is activated.
2305    46. For an unanchored pattern, if a match attempt fails at the start of a
2306        newline sequence, and the newline setting is CRLF or ANY, and the next two
2307        characters are CRLF, advance by two characters instead of one.
2310    Version 6.7 04-Jul-06
2311    ---------------------
2313     1. In order to handle tests when input lines are enormously long, pcretest has
2314        been re-factored so that it automatically extends its buffers when
2315        necessary. The code is crude, but this _is_ just a test program. The
2316        default size has been increased from 32K to 50K.
2318     2. The code in pcre_study() was using the value of the re argument before
2319        testing it for NULL. (Of course, in any sensible call of the function, it
2320        won't be NULL.)
2322     3. The memmove() emulation function in pcre_internal.h, which is used on
2323        systems that lack both memmove() and bcopy() - that is, hardly ever -
2324        was missing a "static" storage class specifier.
2326     4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2327        containing an extended class (one that cannot be represented by a bitmap
2328        because it contains high-valued characters or Unicode property items, e.g.
2329        [\pZ]). Almost always one would set UTF-8 mode when processing such a
2330        pattern, but PCRE should not loop if you do not (it no longer does).
2331        [Detail: two cases were found: (a) a repeated subpattern containing an
2332        extended class; (b) a recursive reference to a subpattern that followed a
2333        previous extended class. It wasn't skipping over the extended class
2334        correctly when UTF-8 mode was not set.]
2336     5. A negated single-character class was not being recognized as fixed-length
2337        in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2338        compile error "lookbehind assertion is not fixed length".
2340     6. The RunPerlTest auxiliary script was showing an unexpected difference
2341        between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2342        write a Perl script that can interpret lines of an input file either as
2343        byte characters or as UTF-8, which is what "perltest" was being required to
2344        do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2345        can't do is switch easily at run time between having the "use utf8;" pragma
2346        or not. In the end, I fudged it by using the RunPerlTest script to insert
2347        "use utf8;" explicitly for the UTF-8 tests.
2349     7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2350        the end of the subject string, contrary to the documentation and to what
2351        Perl does. This was true of both matching functions. Now it matches only at
2352        the start of the subject and immediately after *internal* newlines.
2354     8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2355        a pointer to an int instead of a pointer to an unsigned long int. This
2356        caused problems on 64-bit systems.
2358     9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2359        instance of the 'standard' template library not being so standard".
2361    10. There was no check on the number of named subpatterns nor the maximum
2362        length of a subpattern name. The product of these values is used to compute
2363        the size of the memory block for a compiled pattern. By supplying a very
2364        long subpattern name and a large number of named subpatterns, the size
2365        computation could be caused to overflow. This is now prevented by limiting
2366        the length of names to 32 characters, and the number of named subpatterns
2367        to 10,000.
2369    11. Subpatterns that are repeated with specific counts have to be replicated in
2370        the compiled pattern. The size of memory for this was computed from the
2371        length of the subpattern and the repeat count. The latter is limited to
2372        65535, but there was no limit on the former, meaning that integer overflow
2373        could in principle occur. The compiled length of a repeated subpattern is
2374        now limited to 30,000 bytes in order to prevent this.
2376    12. Added the optional facility to have named substrings with the same name.
2378    13. Added the ability to use a named substring as a condition, using the
2379        Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2380        are numbers (not recommended). Forward references are permitted.
2382    14. Added forward references in named backreferences (if you see what I mean).
2384    15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2385        pattern could run off the end of the subject. For example, the pattern
2386        "(?s)(.{1,5})"8 did this with the subject "ab".
2388    16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2389        PCRE_CASELESS was set when matching characters that were quantified with ?
2390        or *.
2392    17. A character class other than a single negated character that had a minimum
2393        but no maximum quantifier - for example [ab]{6,} - was not handled
2394        correctly by pce_dfa_exec(). It would match only one character.
2396    18. A valid (though odd) pattern that looked like a POSIX character
2397        class but used an invalid character after [ (for example [[,abc,]]) caused
2398        pcre_compile() to give the error "Failed: internal error: code overflow" or
2399        in some cases to crash with a glibc free() error. This could even happen if
2400        the pattern terminated after [[ but there just happened to be a sequence of
2401        letters, a binary zero, and a closing ] in the memory that followed.
2403    19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2404        over the years. Originally (before any Unicode support), just the bottom 8
2405        bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2406        output from "man perlunicode" includes this:
2408          The regular expression compiler produces polymorphic opcodes.  That
2409          is, the pattern adapts to the data and automatically switches to
2410          the Unicode character scheme when presented with Unicode data--or
2411          instead uses a traditional byte scheme when presented with byte
2412          data.
2414        Sadly, a wide octal escape does not cause a switch, and in a string with
2415        no other multibyte characters, these octal escapes are treated as before.
2416        Thus, in Perl, the pattern  /\500/ actually matches \100 but the pattern
2417        /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2418        Unicode string.
2420        I have not perpetrated such confusion in PCRE. Up till now, it took just
2421        the bottom 8 bits, as in old Perl. I have now made octal escapes with
2422        values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2423        translate to the appropriate multibyte character.
2425    29. Applied some refactoring to reduce the number of warnings from Microsoft
2426        and Borland compilers. This has included removing the fudge introduced
2427        seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2428        a warning about an unused variable.
2430    21. PCRE has not included VT (character 0x0b) in the set of whitespace
2431        characters since release 4.0, because Perl (from release 5.004) does not.
2432        [Or at least, is documented not to: some releases seem to be in conflict
2433        with the documentation.] However, when a pattern was studied with
2434        pcre_study() and all its branches started with \s, PCRE still included VT
2435        as a possible starting character. Of course, this did no harm; it just
2436        caused an unnecessary match attempt.
2438    22. Removed a now-redundant internal flag bit that recorded the fact that case
2439        dependency changed within the pattern. This was once needed for "required
2440        byte" processing, but is no longer used. This recovers a now-scarce options
2441        bit. Also moved the least significant internal flag bit to the most-
2442        significant bit of the word, which was not previously used (hangover from
2443        the days when it was an int rather than a uint) to free up another bit for
2444        the future.
2446    23. Added support for CRLF line endings as well as CR and LF. As well as the
2447        default being selectable at build time, it can now be changed at runtime
2448        via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2449        specify that it is scanning data with non-default line endings.
2451    24. Changed the definition of CXXLINK to make it agree with the definition of
2452        LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2454    25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2455        recursions. This makes a big different to stack usage for some patterns.
2457    26. If a subpattern containing a named recursion or subroutine reference such
2458        as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2459        the space required for the compiled pattern went wrong and gave too small a
2460        value. Depending on the environment, this could lead to "Failed: internal
2461        error: code overflow at offset 49" or "glibc detected double free or
2462        corruption" errors.
2464    27. Applied patches from Google (a) to support the new newline modes and (b) to
2465        advance over multibyte UTF-8 characters in GlobalReplace.
2467    28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2468        difference for some implementation of PCRE in some Windows version.
2470    29. Added some extra testing facilities to pcretest:
2472        \q<number>   in a data line sets the "match limit" value
2473        \Q<number>   in a data line sets the "match recursion limt" value
2474        -S <number>  sets the stack size, where <number> is in megabytes
2476        The -S option isn't available for Windows.
2479    Version 6.6 06-Feb-06
2480    ---------------------
2482     1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2483        in pcreposix.h. I have copied the definition from pcre.h.
2485     2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2486        because pcre.h is no longer a built file.
2488     3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2489        not normally included in the compiled code.
2492    Version 6.5 01-Feb-06
2493    ---------------------
2495     1. When using the partial match feature with pcre_dfa_exec(), it was not
2496        anchoring the second and subsequent partial matches at the new starting
2497        point. This could lead to incorrect results. For example, with the pattern
2498        /1234/, partially matching against "123" and then "a4" gave a match.
2500     2. Changes to pcregrep:
2502        (a) All non-match returns from pcre_exec() were being treated as failures
2503            to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
2504            error message is output. Some extra information is given for the
2505            PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
2506            probably the only errors that are likely to be caused by users (by
2507            specifying a regex that has nested indefinite repeats, for instance).
2508            If there are more than 20 of these errors, pcregrep is abandoned.
2510        (b) A binary zero was treated as data while matching, but terminated the
2511            output line if it was written out. This has been fixed: binary zeroes
2512            are now no different to any other data bytes.
2514        (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
2515            used to set a locale for matching. The --locale=xxxx long option has
2516            been added (no short equivalent) to specify a locale explicitly on the
2517            pcregrep command, overriding the environment variables.
2519        (d) When -B was used with -n, some line numbers in the output were one less
2520            than they should have been.
2522        (e) Added the -o (--only-matching) option.
2524        (f) If -A or -C was used with -c (count only), some lines of context were
2525            accidentally printed for the final match.
2527        (g) Added the -H (--with-filename) option.
2529        (h) The combination of options -rh failed to suppress file names for files
2530            that were found from directory arguments.
2532        (i) Added the -D (--devices) and -d (--directories) options.
2534        (j) Added the -F (--fixed-strings) option.
2536        (k) Allow "-" to be used as a file name for -f as well as for a data file.
2538        (l) Added the --colo(u)r option.
2540        (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
2541            is not present by default.
2543     3. A nasty bug was discovered in the handling of recursive patterns, that is,
2544        items such as (?R) or (?1), when the recursion could match a number of
2545        alternatives. If it matched one of the alternatives, but subsequently,
2546        outside the recursion, there was a failure, the code tried to back up into
2547        the recursion. However, because of the way PCRE is implemented, this is not
2548        possible, and the result was an incorrect result from the match.
2550        In order to prevent this happening, the specification of recursion has
2551        been changed so that all such subpatterns are automatically treated as
2552        atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
2554     4. I had overlooked the fact that, in some locales, there are characters for
2555        which isalpha() is true but neither isupper() nor islower() are true. In
2556        the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
2557        and ordfeminine) are like this. This affected the treatment of \w and \W
2558        when they appeared in character classes, but not when they appeared outside
2559        a character class. The bit map for "word" characters is now created
2560        separately from the results of isalnum() instead of just taking it from the
2561        upper, lower, and digit maps. (Plus the underscore character, of course.)
2563     5. The above bug also affected the handling of POSIX character classes such as
2564        [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
2565        permanent tables. Instead, the bit maps for such a class were previously
2566        created as the appropriate unions of the upper, lower, and digit bitmaps.
2567        Now they are created by subtraction from the [[:word:]] class, which has
2568        its own bitmap.
2570     6. The [[:blank:]] character class matches horizontal, but not vertical space.
2571        It is created by subtracting the vertical space characters (\x09, \x0a,
2572        \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
2573        subtraction was done in the overall bitmap for a character class, meaning
2574        that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
2575        be recognized. This bug has been fixed.
2577     7. Patches from the folks at Google:
2579          (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
2580          real life, but is still worth protecting against".
2582          (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
2583          regular expressions".
2585          (c) pcre_scanner.cc: avoid use of std::count() because not all systems
2586          have it.
2588          (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
2589          "configure" and the latter not, in order to fix a problem somebody had
2590          with compiling the Arg class on HP-UX.
2592          (e) Improve the error-handling of the C++ wrapper a little bit.
2594          (f) New tests for checking recursion limiting.
2596     8. The pcre_memmove() function, which is used only if the environment does not
2597        have a standard memmove() function (and is therefore rarely compiled),
2598        contained two bugs: (a) use of int instead of size_t, and (b) it was not
2599        returning a result (though PCRE never actually uses the result).
2601     9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
2602        large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
2603        returned instead of calling malloc() with an overflowing number that would
2604        most likely cause subsequent chaos.
2606    10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
2608    11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
2609        with this option is matched, the nmatch and pmatch options of regexec() are
2610        ignored.
2612    12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
2613        provided in case anyone wants to the the POSIX interface with UTF-8
2614        strings.
2616    13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
2617        C++ linking (needed for some HP-UX environments).
2619    14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
2620        (unused parameter) and in the pcre_printint() function (omitted "default"
2621        switch label when the default is to do nothing).
2623    15. Added some code to make it possible, when PCRE is compiled as a C++
2624        library, to replace subject pointers for pcre_exec() with a smart pointer
2625        class, thus making it possible to process discontinuous strings.
2627    16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
2628        much the same function. They were added by different people who were trying
2629        to make PCRE easy to compile on non-Unix systems. It has been suggested
2630        that PCRE_EXPORT be abolished now that there is more automatic apparatus
2631        for compiling on Windows systems. I have therefore replaced it with
2632        PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
2633        defaults to "extern" for C or "extern C" for C++, which works fine on
2634        Unix-like systems. It is now possible to override the value of PCRE_DATA_
2635        SCOPE with something explicit in config.h. In addition:
2637        (a) pcreposix.h still had just "extern" instead of either of these macros;
2638            I have replaced it with PCRE_DATA_SCOPE.
2640        (b) Functions such as _pcre_xclass(), which are internal to the library,
2641            but external in the C sense, all had PCRE_EXPORT in their definitions.
2642            This is apparently wrong for the Windows case, so I have removed it.
2643            (It makes no difference on Unix-like systems.)
2645    17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
2646        of recursive calls to match(). This is different to MATCH_LIMIT because
2647        that limits the total number of calls to match(), not all of which increase
2648        the depth of recursion. Limiting the recursion depth limits the amount of
2649        stack (or heap if NO_RECURSE is set) that is used. The default can be set
2650        when PCRE is compiled, and changed at run time. A patch from Google adds
2651        this functionality to the C++ interface.
2653    18. Changes to the handling of Unicode character properties:
2655        (a) Updated the table to Unicode 4.1.0.
2657        (b) Recognize characters that are not in the table as "Cn" (undefined).
2659        (c) I revised the way the table is implemented to a much improved format
2660            which includes recognition of ranges. It now supports the ranges that
2661            are defined in UnicodeData.txt, and it also amalgamates other
2662            characters into ranges. This has reduced the number of entries in the
2663            table from around 16,000 to around 3,000, thus reducing its size
2664            considerably. I realized I did not need to use a tree structure after
2665            all - a binary chop search is just as efficient. Having reduced the
2666            number of entries, I extended their size from 6 bytes to 8 bytes to
2667            allow for more data.
2669        (d) Added support for Unicode script names via properties such as \p{Han}.
2671    19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
2672        matching that character.
2674    20. When matching a repeated Unicode property with a minimum greater than zero,
2675        (for example \pL{2,}), PCRE could look past the end of the subject if it
2676        reached it while seeking the minimum number of characters. This could
2677        happen only if some of the characters were more than one byte long, because
2678        there is a check for at least the minimum number of bytes.
2680    21. Refactored the implementation of \p and \P so as to be more general, to
2681        allow for more different types of property in future. This has changed the
2682        compiled form incompatibly. Anybody with saved compiled patterns that use
2683        \p or \P will have to recompile them.
2685    22. Added "Any" and "L&" to the supported property types.
2687    23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
2688        but give a compile time error if the value is greater than 0xff.
2690    24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
2691        accidentally not being installed or uninstalled.
2693    25. The pcre.h file was built from pcre.h.in, but the only changes that were
2694        made were to insert the current release number. This seemed silly, because
2695        it made things harder for people building PCRE on systems that don't run
2696        "configure". I have turned pcre.h into a distributed file, no longer built
2697        by "configure", with the version identification directly included. There is
2698        no longer a pcre.h.in file.
2700        However, this change necessitated a change to the pcre-config script as
2701        well. It is built from pcre-config.in, and one of the substitutions was the
2702        release number. I have updated configure.ac so that ./configure now finds
2703        the release number by grepping pcre.h.
2705    26. Added the ability to run the tests under valgrind.
2708    Version 6.4 05-Sep-05
2709    ---------------------
2711     1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
2712        "--" to be printed when multiple files were scanned, even when none of the
2713        -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
2714        consider it to be a bug, and have restored the previous behaviour.
2716     2. A couple of code tidies to get rid of compiler warnings.
2718     3. The pcretest program used to cheat by referring to symbols in the library
2719        whose names begin with _pcre_. These are internal symbols that are not
2720        really supposed to be visible externally, and in some environments it is
2721        possible to suppress them. The cheating is now confined to including
2722        certain files from the library's source, which is a bit cleaner.
2724     4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
2725        file's purpose clearer.
2727     5. Reorganized pcre_ucp_findchar().
2730    Version 6.3 15-Aug-05
2731    ---------------------
2733     1. The file libpcre.pc.in did not have general read permission in the tarball.
2735     2. There were some problems when building without C++ support:
2737        (a) If C++ support was not built, "make install" and "make test" still
2738            tried to test it.
2740        (b) There were problems when the value of CXX was explicitly set. Some
2741            changes have been made to try to fix these, and ...
2743        (c) --disable-cpp can now be used to explicitly disable C++ support.
2745        (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
2746            backslash in a target when C++ was disabled. This confuses some
2747            versions of "make", apparently. Using an intermediate variable solves
2748            this. (Same for CPP_LOBJ.)
2750     3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
2751        (non-Windows) now includes $(CFLAGS) because these flags are sometimes
2752        necessary on certain architectures.
2754     4. Added a setting of -export-symbols-regex to the link command to remove
2755        those symbols that are exported in the C sense, but actually are local
2756        within the library, and not documented. Their names all begin with
2757        "_pcre_". This is not a perfect job, because (a) we have to except some
2758        symbols that pcretest ("illegally") uses, and (b) the facility isn't always
2759        available (and never for static libraries). I have made a note to try to
2760        find a way round (a) in the future.
2763    Version 6.2 01-Aug-05
2764    ---------------------
2766     1. There was no test for integer overflow of quantifier values. A construction
2767        such as {1111111111111111} would give undefined results. What is worse, if
2768        a minimum quantifier for a parenthesized subpattern overflowed and became
2769        negative, the calculation of the memory size went wrong. This could have
2770        led to memory overwriting.
2772     2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
2774     3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
2775        operating environments where this matters.
2777     4. Applied Giuseppe Maxia's patch to add additional features for controlling
2778        PCRE options from within the C++ wrapper.
2780     5. Named capturing subpatterns were not being correctly counted when a pattern
2781        was compiled. This caused two problems: (a) If there were more than 100
2782        such subpatterns, the calculation of the memory needed for the whole
2783        compiled pattern went wrong, leading to an overflow error. (b) Numerical
2784        back references of the form \12, where the number was greater than 9, were
2785        not recognized as back references, even though there were sufficient
2786        previous subpatterns.
2788     6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
2789        versions of gcc, e.g. 2.95.4.
2792    Version 6.1 21-Jun-05
2793    ---------------------
2795     1. There was one reference to the variable "posix" in pcretest.c that was not
2796        surrounded by "#if !defined NOPOSIX".
2798     2. Make it possible to compile pcretest without DFA support, UTF8 support, or
2799        the cross-check on the old pcre_info() function, for the benefit of the
2800        cut-down version of PCRE that is currently imported into Exim.
2802     3. A (silly) pattern starting with (?i)(?-i) caused an internal space
2803        allocation error. I've done the easy fix, which wastes 2 bytes for sensible
2804        patterns that start (?i) but I don't think that matters. The use of (?i) is
2805        just an example; this all applies to the other options as well.
2807     4. Since libtool seems to echo the compile commands it is issuing, the output
2808        from "make" can be reduced a bit by putting "@" in front of each libtool
2809        compile command.
2811     5. Patch from the folks at Google for configure.in to be a bit more thorough
2812        in checking for a suitable C++ installation before trying to compile the
2813        C++ stuff. This should fix a reported problem when a compiler was present,
2814        but no suitable headers.
2816     6. The man pages all had just "PCRE" as their title. I have changed them to
2817        be the relevant file name. I have also arranged that these names are
2818        retained in the file doc/pcre.txt, which is a concatenation in text format
2819        of all the man pages except the little individual ones for each function.
2821     7. The NON-UNIX-USE file had not been updated for the different set of source
2822        files that come with release 6. I also added a few comments about the C++
2823        wrapper.
2826    Version 6.0 07-Jun-05
2827    ---------------------
2829     1. Some minor internal re-organization to help with my DFA experiments.
2831     2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
2832        didn't matter for the library itself when fully configured, but did matter
2833        when compiling without UCP support, or within Exim, where the ucp files are
2834        not imported.
2836     3. Refactoring of the library code to split up the various functions into
2837        different source modules. The addition of the new DFA matching code (see
2838        below) to a single monolithic source would have made it really too
2839        unwieldy, quite apart from causing all the code to be include in a
2840        statically linked application, when only some functions are used. This is
2841        relevant even without the DFA addition now that patterns can be compiled in
2842        one application and matched in another.
2844        The downside of splitting up is that there have to be some external
2845        functions and data tables that are used internally in different modules of
2846        the library but which are not part of the API. These have all had their
2847        names changed to start with "_pcre_" so that they are unlikely to clash
2848        with other external names.
2850     4. Added an alternate matching function, pcre_dfa_exec(), which matches using
2851        a different (DFA) algorithm. Although it is slower than the original
2852        function, it does have some advantages for certain types of matching
2853        problem.
2855     5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
2856        including restarting after a partial match.
2858     6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
2859        defined when compiling for Windows was sent to me. I have put it into the
2860        code, though I have no means of testing or verifying it.
2862     7. Added the pcre_refcount() auxiliary function.
2864     8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
2865        match before or at the first newline in the subject string. In pcretest,
2866        the /f option on a pattern can be used to set this.
2868     9. A repeated \w when used in UTF-8 mode with characters greater than 256
2869        would behave wrongly. This has been present in PCRE since release 4.0.
2871    10. A number of changes to the pcregrep command:
2873        (a) Refactored how -x works; insert ^(...)$ instead of setting
2874            PCRE_ANCHORED and checking the length, in preparation for adding
2875            something similar for -w.
2877        (b) Added the -w (match as a word) option.
2879        (c) Refactored the way lines are read and buffered so as to have more
2880            than one at a time available.
2882        (d) Implemented a pcregrep test script.
2884        (e) Added the -M (multiline match) option. This allows patterns to match
2885            over several lines of the subject. The buffering ensures that at least
2886            8K, or the rest of the document (whichever is the shorter) is available
2887            for matching (and similarly the previous 8K for lookbehind assertions).
2889        (f) Changed the --help output so that it now says
2891              -w, --word-regex(p)
2893            instead of two lines, one with "regex" and the other with "regexp"
2894            because that confused at least one person since the short forms are the
2895            same. (This required a bit of code, as the output is generated
2896            automatically from a table. It wasn't just a text change.)
2898        (g) -- can be used to terminate pcregrep options if the next thing isn't an
2899            option but starts with a hyphen. Could be a pattern or a path name
2900            starting with a hyphen, for instance.
2902        (h) "-" can be given as a file name to represent stdin.
2904        (i) When file names are being printed, "(standard input)" is used for
2905            the standard input, for compatibility with GNU grep. Previously
2906            "<stdin>" was used.
2908        (j) The option --label=xxx can be used to supply a name to be used for
2909            stdin when file names are being printed. There is no short form.
2911        (k) Re-factored the options decoding logic because we are going to add
2912            two more options that take data. Such options can now be given in four
2913            different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
2915        (l) Added the -A, -B, and -C options for requesting that lines of context
2916            around matches be printed.
2918        (m) Added the -L option to print the names of files that do not contain
2919            any matching lines, that is, the complement of -l.
2921        (n) The return code is 2 if any file cannot be opened, but pcregrep does
2922            continue to scan other files.
2924        (o) The -s option was incorrectly implemented. For compatibility with other
2925            greps, it now suppresses the error message for a non-existent or non-
2926            accessible file (but not the return code). There is a new option called
2927            -q that suppresses the output of matching lines, which was what -s was
2928            previously doing.
2930        (p) Added --include and --exclude options to specify files for inclusion
2931            and exclusion when recursing.
2933    11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
2934        Hopefully, it now does.
2936    12. Missing cast in pcre_study().
2938    13. Added an "uninstall" target to the makefile.
2940    14. Replaced "extern" in the function prototypes in Makefile.in with
2941        "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
2942        world, but is set differently for Windows.
2944    15. Added a second compiling function called pcre_compile2(). The only
2945        difference is that it has an extra argument, which is a pointer to an
2946        integer error code. When there is a compile-time failure, this is set
2947        non-zero, in addition to the error test pointer being set to point to an
2948        error message. The new argument may be NULL if no error number is required
2949        (but then you may as well call pcre_compile(), which is now just a
2950        wrapper). This facility is provided because some applications need a
2951        numeric error indication, but it has also enabled me to tidy up the way
2952        compile-time errors are handled in the POSIX wrapper.
2954    16. Added VPATH=.libs to the makefile; this should help when building with one
2955        prefix path and installing with another. (Or so I'm told by someone who
2956        knows more about this stuff than I do.)
2958    17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
2959        passes PCRE_DOTALL to the pcre_compile() function, making the "." character
2960        match everything, including newlines. This is not POSIX-compatible, but
2961        somebody wanted the feature. From pcretest it can be activated by using
2962        both the P and the s flags.
2964    18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
2966    19. libpcre.pc was being incorrectly installed as executable.
2968    20. A couple of places in pcretest check for end-of-line by looking for '\n';
2969        it now also looks for '\r' so that it will work unmodified on Windows.
2971    21. Added Google's contributed C++ wrapper to the distribution.
2973    22. Added some untidy missing memory free() calls in pcretest, to keep
2974        Electric Fence happy when testing.
2978    Version 5.0 13-Sep-04
2979    ---------------------
2981     1. Internal change: literal characters are no longer packed up into items
2982        containing multiple characters in a single byte-string. Each character
2983        is now matched using a separate opcode. However, there may be more than one
2984        byte in the character in UTF-8 mode.
2986     2. The pcre_callout_block structure has two new fields: pattern_position and
2987        next_item_length. These contain the offset in the pattern to the next match
2988        item, and its length, respectively.
2990     3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
2991        insertion of callouts before each pattern item. Added the /C option to
2992        pcretest to make use of this.
2994     4. On the advice of a Windows user, the lines
2996          #if defined(_WIN32) || defined(WIN32)
2997          _setmode( _fileno( stdout ), 0x8000 );
2998          #endif  /* defined(_WIN32) || defined(WIN32) */
3000        have been added to the source of pcretest. This apparently does useful
3001        magic in relation to line terminators.
3003     5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3004        for the benefit of those environments where the "b" makes a difference.
3006     6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3007        to know about it. I have put a hack into configure.in that adds in code
3008        to set GCC=yes if CC=icc. This seems to end up at a point in the
3009        generated configure script that is early enough to affect the setting of
3010        compiler options, which is what is needed, but I have no means of testing
3011        whether it really works. (The user who reported this had patched the
3012        generated configure script, which of course I cannot do.)
3014        LATER: After change 22 below (new libtool files), the configure script
3015        seems to know about icc (and also ecc). Therefore, I have commented out
3016        this hack in configure.in.
3018     7. Added support for pkg-config (2 patches were sent in).
3020     8. Negated POSIX character classes that used a combination of internal tables
3021        were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3022        [[:^ascii]]. Typically, they would match almost any characters. The other
3023        POSIX classes were not broken in this way.
3025     9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3026        to find the match, as PCRE was deluded into thinking that the match had to
3027        start at the start point or following a newline. The same bug applied to
3028        patterns with negative forward assertions or any backward assertions
3029        preceding ".*" at the start, unless the pattern required a fixed first
3030        character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3032    10. In UTF-8 mode, when moving forwards in the subject after a failed match
3033        starting at the last subject character, bytes beyond the end of the subject
3034        string were read.
3036    11. Renamed the variable "class" as "classbits" to make life easier for C++
3037        users. (Previously there was a macro definition, but it apparently wasn't
3038        enough.)
3040    12. Added the new field "tables" to the extra data so that tables can be passed
3041        in at exec time, or the internal tables can be re-selected. This allows
3042        a compiled regex to be saved and re-used at a later time by a different
3043        program that might have everything at different addresses.
3045    13. Modified the pcre-config script so that, when run on Solaris, it shows a
3046        -R library as well as a -L library.
3048    14. The debugging options of pcretest (-d on the command line or D on a
3049        pattern) showed incorrect output for anything following an extended class
3050        that contained multibyte characters and which was followed by a quantifier.
3052    15. Added optional support for general category Unicode character properties
3053        via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3054        support. It adds about 90K to the size of the library. The meanings of the
3055        inbuilt class escapes such as \d and \s have NOT been changed.
3057    16. Updated pcredemo.c to include calls to free() to release the memory for the
3058        compiled pattern.
3060    17. The generated file chartables.c was being created in the source directory
3061        instead of in the building directory. This caused the build to fail if the
3062        source directory was different from the building directory, and was
3063        read-only.
3065    18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3066        file. No doubt somebody will tell me if they don't make sense... Also added
3067        Dan Mooney's comments about building on OpenVMS.
3069    19. Added support for partial matching via the PCRE_PARTIAL option for
3070        pcre_exec() and the \P data escape in pcretest.
3072    20. Extended pcretest with 3 new pattern features:
3074        (i)   A pattern option of the form ">rest-of-line" causes pcretest to
3075              write the compiled pattern to the file whose name is "rest-of-line".
3076              This is a straight binary dump of the data, with the saved pointer to
3077              the character tables forced to be NULL. The study data, if any, is
3078              written too. After writing, pcretest reads a new pattern.
3080        (ii)  If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3081              compiled pattern from the given file. There must not be any
3082              occurrences of "<" in the file name (pretty unlikely); if there are,
3083              pcretest will instead treat the initial "<" as a pattern delimiter.
3084              After reading in the pattern, pcretest goes on to read data lines as
3085              usual.
3087        (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3088              and 16-bit fields in a compiled pattern, to simulate a pattern that
3089              was compiled on a host of opposite endianness.
3091    21. The pcre-exec() function can now cope with patterns that were compiled on
3092        hosts of opposite endianness, with this restriction:
3094          As for any compiled expression that is saved and used later, the tables
3095          pointer field cannot be preserved; the extra_data field in the arguments
3096          to pcre_exec() should be used to pass in a tables address if a value
3097          other than the default internal tables were used at compile time.
3099    22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3100        now diagnosed as an error. Previously, most of the time, a negative number
3101        would have been treated as zero, but if in addition "ovector" was passed as
3102        NULL, a crash could occur.
3104    23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3105        new versions from the libtool 1.5 distribution (the last one is a copy of
3106        a file called libtool.m4). This seems to have fixed the need to patch
3107        "configure" to support Darwin 1.3 (which I used to do). However, I still
3108        had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3109        workstation).
3111    24. Changed the PCRE licence to be the more standard "BSD" licence.
3114    Version 4.5 01-Dec-03
3115    ---------------------
3117     1. There has been some re-arrangement of the code for the match() function so
3118        that it can be compiled in a version that does not call itself recursively.
3119        Instead, it keeps those local variables that need separate instances for
3120        each "recursion" in a frame on the heap, and gets/frees frames whenever it
3121        needs to "recurse". Keeping track of where control must go is done by means
3122        of setjmp/longjmp. The whole thing is implemented by a set of macros that
3123        hide most of the details from the main code, and operates only if
3124        NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3125        "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3126        operating.
3128        To make it easier for callers to provide specially tailored get/free
3129        functions for this usage, two new functions, pcre_stack_malloc, and
3130        pcre_stack_free, are used. They are always called in strict stacking order,
3131        and the size of block requested is always the same.
3133        The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3134        PCRE has been compiled to use the stack or the heap for recursion. The
3135        -C option of pcretest uses this to show which version is compiled.
3137        A new data escape \S, is added to pcretest; it causes the amounts of store
3138        obtained and freed by both kinds of malloc/free at match time to be added
3139        to the output.
3141     2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3142        what's available on my current Linux desktop machine.
3144     3. When matching a UTF-8 string, the test for a valid string at the start has
3145        been extended. If start_offset is not zero, PCRE now checks that it points
3146        to a byte that is the start of a UTF-8 character. If not, it returns
3147        PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3148        this is necessary because there may be backward assertions in the pattern.
3149        When matching the same subject several times, it may save resources to use
3150        PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3152     4. The code for checking the validity of UTF-8 strings has been tightened so
3153        that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3154        containing "overlong sequences".
3156     5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3157        I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3158        should have been "&", but it just so happened that all the cases this let
3159        through by mistake were picked up later in the function.
3161     6. I had used a variable called "isblank" - this is a C99 function, causing
3162        some compilers to warn. To avoid this, I renamed it (as "blankclass").
3164     7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3165        prompting; (b) run "./pcretest /dev/null" at the start of the test script
3166        so the version is shown; (c) stop "make test" echoing "./RunTest".
3168     8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3170     9. The prototype for memmove() for systems that don't have it was using
3171        size_t, but the inclusion of the header that defines size_t was later. I've
3172        moved the #includes for the C headers earlier to avoid this.
3174    10. Added some adjustments to the code to make it easier to compiler on certain
3175        special systems:
3177          (a) Some "const" qualifiers were missing.
3178          (b) Added the macro EXPORT before all exported functions; by default this
3179              is defined to be empty.
3180          (c) Changed the dftables auxiliary program (that builds chartables.c) so
3181              that it reads its output file name as an argument instead of writing
3182              to the standard output and assuming this can be redirected.
3184    11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3185        class containing characters with values greater than 255, PCRE compilation
3186        went into a loop.
3188    12. A recursive reference to a subpattern that was within another subpattern
3189        that had a minimum quantifier of zero caused PCRE to crash. For example,
3190        (x(y(?2))z)? provoked this bug with a subject that got as far as the
3191        recursion. If the recursively-called subpattern itself had a zero repeat,
3192        that was OK.
3194    13. In pcretest, the buffer for reading a data line was set at 30K, but the
3195        buffer into which it was copied (for escape processing) was still set at
3196        1024, so long lines caused crashes.
3198    14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3199        "internal error: code overflow...". This applied to any character class
3200        that was followed by a possessive quantifier.
3202    15. Modified the Makefile to add libpcre.la as a prerequisite for
3203        libpcreposix.la because I was told this is needed for a parallel build to
3204        work.
3206    16. If a pattern that contained .* following optional items at the start was
3207        studied, the wrong optimizing data was generated, leading to matching
3208        errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3209        matching string must start with a or b or c. The correct conclusion for
3210        this pattern is that a match can start with any character.
3213    Version 4.4 13-Aug-03
3214    ---------------------
3216     1. In UTF-8 mode, a character class containing characters with values between
3217        127 and 255 was not handled correctly if the compiled pattern was studied.
3218        In fixing this, I have also improved the studying algorithm for such
3219        classes (slightly).
3221     2. Three internal functions had redundant arguments passed to them. Removal
3222        might give a very teeny performance improvement.
3224     3. Documentation bug: the value of the capture_top field in a callout is *one
3225        more than* the number of the hightest numbered captured substring.
3227     4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3228        in incorrectly linking with a previously installed version. They now link
3229        explicitly with libpcre.la.
3231     5. configure.in no longer needs to recognize Cygwin specially.
3233     6. A problem in pcre.in for Windows platforms is fixed.
3235     7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3236        pcretest, it used to include the size of the study block as part of its
3237        output. Unfortunately, the structure contains a field that has a different
3238        size on different hardware architectures. This meant that the tests that
3239        showed this size failed. As the block is currently always of a fixed size,
3240        this information isn't actually particularly useful in pcretest output, so
3241        I have just removed it.
3243     8. Three pre-processor statements accidentally did not start in column 1.
3244        Sadly, there are *still* compilers around that complain, even though
3245        standard C has not required this for well over a decade. Sigh.
3247     9. In pcretest, the code for checking callouts passed small integers in the
3248        callout_data field, which is a void * field. However, some picky compilers
3249        complained about the casts involved for this on 64-bit systems. Now
3250        pcretest passes the address of the small integer instead, which should get
3251        rid of the warnings.
3253    10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3254        both compile and run time, and gives an error if an invalid UTF-8 sequence
3255        is found. There is a option for disabling this check in cases where the
3256        string is known to be correct and/or the maximum performance is wanted.
3258    11. In response to a bug report, I changed one line in Makefile.in from
3260            -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3261        to
3262            -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3264        to look similar to other lines, but I have no way of telling whether this
3265        is the right thing to do, as I do not use Windows. No doubt I'll get told
3266        if it's wrong...
3269    Version 4.3 21-May-03
3270    ---------------------
3272    1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3273       Makefile.
3275    2. Some refactoring to improve the quality of the code:
3277       (i)   The utf8_table... variables are now declared "const".
3279       (ii)  The code for \cx, which used the "case flipping" table to upper case
3280             lower case letters, now just substracts 32. This is ASCII-specific,
3281             but the whole concept of \cx is ASCII-specific, so it seems
3282             reasonable.
3284       (iii) PCRE was using its character types table to recognize decimal and
3285             hexadecimal digits in the pattern. This is silly, because it handles
3286             only 0-9, a-f, and A-F, but the character types table is locale-
3287             specific, which means strange things might happen. A private
3288             table is now used for this - though it costs 256 bytes, a table is
3289             much faster than multiple explicit tests. Of course, the standard
3290             character types table is still used for matching digits in subject
3291             strings against \d.
3293       (iv)  Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3294             ending in _t are). So I've renamed it as ESC_tee.
3296    3. The first argument for regexec() in the POSIX wrapper should have been
3297       defined as "const".
3299    4. Changed pcretest to use malloc() for its buffers so that they can be
3300       Electric Fenced for debugging.
3302    5. There were several places in the code where, in UTF-8 mode, PCRE would try
3303       to read one or more bytes before the start of the subject string. Often this
3304       had no effect on PCRE's behaviour, but in some circumstances it could
3305       provoke a segmentation fault.
3307    6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3308       to try to read one or more bytes before the start of the subject string.
3310    7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3311       UTF-8 support could misbehave in various ways if the subject string
3312       contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3313       area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3314       back over UTF-8 characters.)
3317    Version 4.2 14-Apr-03
3318    ---------------------
3320    1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3322    2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3323         [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3324         [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3325         [WIN_PREFIX]: new variable, "cyg" for Cygwin
3326         * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3327           and BUILD_EXEEXT
3328         Note: automatic setting of the BUILD variables is not yet working
3329         set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3330           compile-time but not at link-time
3331         [LINK]: use for linking executables only
3332         make different versions for Windows and non-Windows
3333         [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3334           libraries
3335         [LINK_FOR_BUILD]: new variable
3336         [OBJEXT]: use throughout
3337         [EXEEXT]: use throughout
3338         <winshared>: new target
3339         <wininstall>: new target
3340         <dftables.o>: use native compiler
3341         <dftables>: use native linker
3342         <install>: handle Windows platform correctly
3343         <clean>: ditto
3344         <check>: ditto
3345         copy DLL to top builddir before testing
3347       As part of these changes, -no-undefined was removed again. This was reported
3348       to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3349       in any case.
3351    3. Some tidies to get rid of compiler warnings:
3353       . In the match_data structure, match_limit was an unsigned long int, whereas
3354         match_call_count was an int. I've made them both unsigned long ints.
3356       . In pcretest the fact that a const uschar * doesn't automatically cast to
3357         a void * provoked a warning.
3359       . Turning on some more compiler warnings threw up some "shadow" variables
3360         and a few more missing casts.
3362    4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3363       option, a class that contained a single character with a value between 128
3364       and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3366    5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3367       option, a class that contained several characters, but with at least one
3368       whose value was between 128 and 255 caused PCRE to crash.
3371    Version 4.1 12-Mar-03
3372    ---------------------
3374    1. Compiling with gcc -pedantic found a couple of places where casts were
3375    needed, and a string in dftables.c that was longer than standard compilers are
3376    required to support.
3378    2. Compiling with Sun's compiler found a few more places where the code could
3379    be tidied up in order to avoid warnings.
3381    3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3382    first of these names is deprecated in the latest Autoconf in favour of the name
3383    CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3384    compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3385    analogy I have changed it to CFLAGS_FOR_BUILD.
3387    4. Added -no-undefined to the linking command in the Makefile, because this is
3388    apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3389    linking step for the pcreposix library.
3391    5. PCRE was failing to diagnose the case of two named groups with the same
3392    name.
3394    6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3395    literal character that is needed in the subject for a match, and scans along to
3396    ensure that it is present before embarking on the full matching process. This
3397    saves time in cases of nested unlimited repeats that are never going to match.
3398    Problem: the scan can take a lot of time if the subject is very long (e.g.
3399    megabytes), thus penalizing straightforward matches. It is now done only if the
3400    amount of subject to be scanned is less than 1000 bytes.
3402    7. A lesser problem with the same optimization is that it was recording the
3403    first character of an anchored pattern as "needed", thus provoking a search
3404    right along the subject, even when the first match of the pattern was going to
3405    fail. The "needed" character is now not set for anchored patterns, unless it
3406    follows something in the pattern that is of non-fixed length. Thus, it still
3407    fulfils its original purpose of finding quick non-matches in cases of nested
3408    unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3411    Version 4.0 17-Feb-03
3412    ---------------------
3414    1. If a comment in an extended regex that started immediately after a meta-item
3415    extended to the end of string, PCRE compiled incorrect data. This could lead to
3416    all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3418    2. Moved to autoconf 2.53 and libtool 1.4.2.
3420    3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3421    the special perltest8 script is no longer needed - all the tests can be run
3422    from a single perltest script.
3424    4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3425    by \s. It has now been removed in PCRE. This means it isn't recognized as
3426    whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3427    class [:space:] *does* include VT, thereby creating a mess.
3429    5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3430    space and tab.
3432    6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3433    its new features into the main test script, reducing the number of scripts.
3435    7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3436    were backward compatible, and made the (?i) apply to the whole pattern, as if
3437    /i were given. Now it behaves more logically, and applies the option setting
3438    only to what follows. PCRE has been changed to follow suit. However, if it
3439    finds options settings right at the start of the pattern, it extracts them into
3440    the global options, as before. Thus, they show up in the info data.
3442    8. Added support for the \Q...\E escape sequence. Characters in between are
3443    treated as literals. This is slightly different from Perl in that $ and @ are
3444    also handled as literals inside the quotes. In Perl, they will cause variable
3445    interpolation. Note the following examples:
3447        Pattern            PCRE matches      Perl matches
3449        \Qabc$xyz\E        abc$xyz           abc followed by the contents of $xyz
3450        \Qabc\$xyz\E       abc\$xyz          abc\$xyz
3451        \Qabc\E\$\Qxyz\E   abc$xyz           abc$xyz
3453    For compatibility with Perl, \Q...\E sequences are recognized inside character
3454    classes as well as outside them.
3456    9. Re-organized 3 code statements in pcretest to avoid "overflow in
3457    floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3458    (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3459    signed/unsigned warnings.
3461    10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3462    option for pcretest, so I've replaced it by a simple function that does just
3463    that job.
3465    11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3466    "pcregrep -".
3468    12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3469    Java package. This provides some syntactic sugar for simple cases of what my
3470    documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3471    as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3472    item, you can use this simplified notation. Note that only makes sense with
3473    greedy quantifiers. Consequently, the use of the possessive quantifier forces
3474    greediness, whatever the setting of the PCRE_UNGREEDY option.
3476    13. A change of greediness default within a pattern was not taking effect at
3477    the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3478    subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3479    was abstracted outside.
3481    14. PCRE now supports the \G assertion. It is true when the current matching
3482    position is at the start point of the match. This differs from \A when the
3483    starting offset is non-zero. Used with the /g option of pcretest (or similar
3484    code), it works in the same way as it does for Perl's /g option. If all
3485    alternatives of a regex begin with \G, the expression is anchored to the start
3486    match position, and the "anchored" flag is set in the compiled expression.
3488    15. Some bugs concerning the handling of certain option changes within patterns
3489    have been fixed. These applied to options other than (?ims). For example,
3490    "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3491    been the other way round. Some of this was related to change 7 above.
3493    16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3494    features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3495    and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
3496    POSIX classes only within a class (e.g. /[[:alpha:]]/).
3498    17. Added support for Perl's \C escape. This matches one byte, even in UTF8
3499    mode. Unlike ".", it always matches newline, whatever the setting of
3500    PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
3501    assertions. Perl allows it, but it doesn't (in general) work because it can't
3502    calculate the length of the lookbehind. At least, that's the case for Perl
3503    5.8.0 - I've been told they are going to document that it doesn't work in
3504    future.
3506    18. Added an error diagnosis for escapes that PCRE does not support: these are
3507    \L, \l, \N, \P, \p, \U, \u, and \X.
3509    19. Although correctly diagnosing a missing ']' in a character class, PCRE was
3510    reading past the end of the pattern in cases such as /[abcd/.
3512    20. PCRE was getting more memory than necessary for patterns with classes that
3513    contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
3515    21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
3516    compiling PCRE for use with Virtual Pascal.
3518    22. Small fix to the Makefile to make it work properly if the build is done
3519    outside the source tree.
3521    23. Added a new extension: a condition to go with recursion. If a conditional
3522    subpattern starts with (?(R) the "true" branch is used if recursion has
3523    happened, whereas the "false" branch is used only at the top level.
3525    24. When there was a very long string of literal characters (over 255 bytes
3526    without UTF support, over 250 bytes with UTF support), the computation of how
3527    much memory was required could be incorrect, leading to segfaults or other
3528    strange effects.
3530    25. PCRE was incorrectly assuming anchoring (either to start of subject or to
3531    start of line for a non-DOTALL pattern) when a pattern started with (.*) and
3532    there was a subsequent back reference to those brackets. This meant that, for
3533    example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
3534    possible to check for precisely this case. All we can do is abandon the
3535    optimization if .* occurs inside capturing brackets when there are any back
3536    references whatsoever. (See below for a better fix that came later.)
3538    26. The handling of the optimization for finding the first character of a
3539    non-anchored pattern, and for finding a character that is required later in the
3540    match were failing in some cases. This didn't break the matching; it just
3541    failed to optimize when it could. The way this is done has been re-implemented.
3543    27. Fixed typo in error message for invalid (?R item (it said "(?p").
3545    28. Added a new feature that provides some of the functionality that Perl
3546    provides with (?{...}). The facility is termed a "callout". The way it is done
3547    in PCRE is for the caller to provide an optional function, by setting
3548    pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
3549    global variable. By default it is unset, which disables all calling out. To get
3550    the function called, the regex must include (?C) at appropriate points. This
3551    is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
3552    This provides a means of identifying different callout points. When PCRE
3553    reaches such a point in the regex, if pcre_callout has been set, the external
3554    function is called. It is provided with data in a structure called
3555    pcre_callout_block, which is defined in pcre.h. If the function returns 0,
3556    matching continues; if it returns a non-zero value, the match at the current
3557    point fails. However, backtracking will occur if possible. [This was changed
3558    later and other features added - see item 49 below.]
3560    29. pcretest is upgraded to test the callout functionality. It provides a
3561    callout function that displays information. By default, it shows the start of
3562    the match and the current position in the text. There are some new data escapes
3563    to vary what happens:
3565        \C+         in addition, show current contents of captured substrings
3566        \C-         do not supply a callout function
3567        \C!n        return 1 when callout number n is reached
3568        \C!n!m      return 1 when callout number n is reached for the mth time
3570    30. If pcregrep was called with the -l option and just a single file name, it
3571    output "<stdin>" if a match was found, instead of the file name.
3573    31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
3574    slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
3575    pcre_exec(). This saves a malloc/free per call. The default value of
3576    POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
3577    when configuring.
3579    32. The default maximum size of a compiled pattern is 64K. There have been a
3580    few cases of people hitting this limit. The code now uses macros to handle the
3581    storing of links as offsets within the compiled pattern. It defaults to 2-byte
3582    links, but this can be changed to 3 or 4 bytes by --with-link-size when
3583    configuring. Tests 2 and 5 work only with 2-byte links because they output
3584    debugging information about compiled patterns.
3586    33. Internal code re-arrangements:
3588    (a) Moved the debugging function for printing out a compiled regex into
3589        its own source file (printint.c) and used #include to pull it into
3590        pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
3591        separate copies.
3593    (b) Defined the list of op-code names for debugging as a macro in
3594        internal.h so that it is next to the definition of the opcodes.
3596    (c) Defined a table of op-code lengths for simpler skipping along compiled