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

Diff of /code/trunk/ChangeLog

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

revision 740 by zherczeg, Mon Oct 31 06:10:14 2011 UTC revision 1667 by ph10, Thu Oct 13 16:00:48 2016 UTC
# Line 1  Line 1 
1  ChangeLog for PCRE  ChangeLog for PCRE
2  ------------------  ------------------
3    
4  Version 8.21  Note that the PCRE 8.xx series (PCRE1) is now in a bugfix-only state. All
5  ------------  development is happening in the PCRE2 10.xx series.
6    
7    Version 8.40 xx-xxxx-2016
8    -------------------------
9    
10    1.  Using -o with -M in pcregrep could cause unnecessary repeated output when
11        the match extended over a line boundary.
12    
13    2.  Applied Chris Wilson's second patch (Bugzilla #1681) to CMakeLists.txt for
14        MSVC static compilation, putting the first patch under a new option.
15    
16    3.  Fix register overwite in JIT when SSE2 acceleration is enabled.
17    
18    4.  Ignore "show all captures" (/=) for DFA matching.
19    
20    5.  Fix JIT unaligned accesses on x86. Patch by Marc Mutz.
21    
22    6.  In any wide-character mode (8-bit UTF or any 16-bit or 32-bit mode),
23        without PCRE_UCP set, a negative character type such as \D in a positive
24        class should cause all characters greater than 255 to match, whatever else
25        is in the class. There was a bug that caused this not to happen if a
26        Unicode property item was added to such a class, for example [\D\P{Nd}] or
27        [\W\pL].
28    
29    7.  When pcretest was outputing information from a callout, the caret indicator
30        for the current position in the subject line was incorrect if it was after
31        an escape sequence for a character whose code point was greater than
32        \x{ff}.
33    
34    8.  A pattern such as (?<RA>abc)(?(R)xyz) was incorrectly compiled such that
35        the conditional was interpreted as a reference to capturing group 1 instead
36        of a test for recursion. Any group whose name began with R was
37        misinterpreted in this way. (The reference interpretation should only
38        happen if the group's name is precisely "R".)
39    
40    
41    Version 8.39 14-June-2016
42    -------------------------
43    
44    1.  If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
45        an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
46        misbehaved. This bug was found by the LLVM fuzzer.
47    
48    2.  Similar to the above, if an isolated \E was present between an item and its
49        qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
50        bug was found by the LLVM fuzzer.
51    
52    3.  Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
53        working correctly in UCP mode.
54    
55    4.  The POSIX wrapper function regexec() crashed if the option REG_STARTEND
56        was set when the pmatch argument was NULL. It now returns REG_INVARG.
57    
58    5.  Allow for up to 32-bit numbers in the ordin() function in pcregrep.
59    
60    6.  An empty \Q\E sequence between an item and its qualifier caused
61        pcre_compile() to misbehave when auto callouts were enabled. This bug was
62        found by the LLVM fuzzer.
63    
64    7.  If a pattern that was compiled with PCRE_EXTENDED started with white
65        space or a #-type comment that was followed by (?-x), which turns off
66        PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
67        pcre_compile() assumed that (?-x) applied to the whole pattern and
68        consequently mis-compiled it. This bug was found by the LLVM fuzzer.
69    
70    8.  A call of pcre_copy_named_substring() for a named substring whose number
71        was greater than the space in the ovector could cause a crash.
72    
73    9.  Yet another buffer overflow bug involved duplicate named groups with a
74        group that reset capture numbers (compare 8.38/7 below). Once again, I have
75        just allowed for more memory, even if not needed. (A proper fix is
76        implemented in PCRE2, but it involves a lot of refactoring.)
77    
78    10. pcre_get_substring_list() crashed if the use of \K in a match caused the
79        start of the match to be earlier than the end.
80    
81    11. Migrating appropriate PCRE2 JIT improvements to PCRE.
82    
83    12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
84        assertion, caused pcretest to generate incorrect output, and also to read
85        uninitialized memory (detected by ASAN or valgrind).
86    
87    13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
88        nested set of parentheses of sufficient size caused an overflow of the
89        compiling workspace (which was diagnosed, but of course is not desirable).
90    
91    14. And yet another buffer overflow bug involving duplicate named groups, this
92        time nested, with a nested back reference. Yet again, I have just allowed
93        for more memory, because anything more needs all the refactoring that has
94        been done for PCRE2. An example pattern that provoked this bug is:
95        /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
96        registered as CVE-2016-1283.
97    
98    15. pcretest went into a loop if global matching was requested with an ovector
99        size less than 2. It now gives an error message. This bug was found by
100        afl-fuzz.
101    
102    16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
103        ("assertion expected") when (?(?C) was not followed by an opening
104        parenthesis.
105    
106    17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
107        actually affect anything, by sheer luck.
108    
109    18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
110        static compilation.
111    
112    19. Modified the RunTest script to incorporate a valgrind suppressions file so
113        that certain errors, provoked by the SSE2 instruction set when JIT is used,
114        are ignored.
115    
116    20. A racing condition is fixed in JIT reported by Mozilla.
117    
118    21. Minor code refactor to avoid "array subscript is below array bounds"
119        compiler warning.
120    
121    22. Minor code refactor to avoid "left shift of negative number" warning.
122    
123    23. Fix typo causing compile error when 16- or 32-bit JIT is compiled without
124        UCP support.
125    
126    24. Refactor to avoid compiler warnings in pcrecpp.cc.
127    
128    25. Refactor to fix a typo in pcre_jit_test.c
129    
130    26. Patch to support compiling pcrecpp.cc with Intel compiler.
131    
132    
133    Version 8.38 23-November-2015
134    -----------------------------
135    
136    1.  If a group that contained a recursive back reference also contained a
137        forward reference subroutine call followed by a non-forward-reference
138        subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
139        compile correct code, leading to undefined behaviour or an internally
140        detected error. This bug was discovered by the LLVM fuzzer.
141    
142    2.  Quantification of certain items (e.g. atomic back references) could cause
143        incorrect code to be compiled when recursive forward references were
144        involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
145        This bug was discovered by the LLVM fuzzer.
146    
147    3.  A repeated conditional group whose condition was a reference by name caused
148        a buffer overflow if there was more than one group with the given name.
149        This bug was discovered by the LLVM fuzzer.
150    
151    4.  A recursive back reference by name within a group that had the same name as
152        another group caused a buffer overflow. For example:
153        /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
154    
155    5.  A forward reference by name to a group whose number is the same as the
156        current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
157        a buffer overflow at compile time. This bug was discovered by the LLVM
158        fuzzer.
159    
160    6.  A lookbehind assertion within a set of mutually recursive subpatterns could
161        provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
162    
163    7.  Another buffer overflow bug involved duplicate named groups with a
164        reference between their definition, with a group that reset capture
165        numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
166        fixed by always allowing for more memory, even if not needed. (A proper fix
167        is implemented in PCRE2, but it involves more refactoring.)
168    
169    8.  There was no check for integer overflow in subroutine calls such as (?123).
170    
171    9.  The table entry for \l in EBCDIC environments was incorrect, leading to its
172        being treated as a literal 'l' instead of causing an error.
173    
174    10. There was a buffer overflow if pcre_exec() was called with an ovector of
175        size 1. This bug was found by american fuzzy lop.
176    
177    11. If a non-capturing group containing a conditional group that could match
178        an empty string was repeated, it was not identified as matching an empty
179        string itself. For example: /^(?:(?(1)x|)+)+$()/.
180    
181    12. In an EBCDIC environment, pcretest was mishandling the escape sequences
182        \a and \e in test subject lines.
183    
184    13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
185        instead of the EBCDIC value.
186    
187    14. The handling of \c in an EBCDIC environment has been revised so that it is
188        now compatible with the specification in Perl's perlebcdic page.
189    
190    15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
191        ASCII/Unicode. This has now been added to the list of characters that are
192        recognized as white space in EBCDIC.
193    
194    16. When PCRE was compiled without UCP support, the use of \p and \P gave an
195        error (correctly) when used outside a class, but did not give an error
196        within a class.
197    
198    17. \h within a class was incorrectly compiled in EBCDIC environments.
199    
200    18. A pattern with an unmatched closing parenthesis that contained a backward
201        assertion which itself contained a forward reference caused buffer
202        overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
203    
204    19. JIT should return with error when the compiled pattern requires more stack
205        space than the maximum.
206    
207    20. A possessively repeated conditional group that could match an empty string,
208        for example, /(?(R))*+/, was incorrectly compiled.
209    
210    21. Fix infinite recursion in the JIT compiler when certain patterns such as
211        /(?:|a|){100}x/ are analysed.
212    
213    22. Some patterns with character classes involving [: and \\ were incorrectly
214        compiled and could cause reading from uninitialized memory or an incorrect
215        error diagnosis.
216    
217    23. Pathological patterns containing many nested occurrences of [: caused
218        pcre_compile() to run for a very long time.
219    
220    24. A conditional group with only one branch has an implicit empty alternative
221        branch and must therefore be treated as potentially matching an empty
222        string.
223    
224    25. If (?R was followed by - or + incorrect behaviour happened instead of a
225        diagnostic.
226    
227    26. Arrange to give up on finding the minimum matching length for overly
228        complex patterns.
229    
230    27. Similar to (4) above: in a pattern with duplicated named groups and an
231        occurrence of (?| it is possible for an apparently non-recursive back
232        reference to become recursive if a later named group with the relevant
233        number is encountered. This could lead to a buffer overflow. Wen Guanxing
234        from Venustech ADLAB discovered this bug.
235    
236    28. If pcregrep was given the -q option with -c or -l, or when handling a
237        binary file, it incorrectly wrote output to stdout.
238    
239    29. The JIT compiler did not restore the control verb head in case of *THEN
240        control verbs. This issue was found by Karl Skomski with a custom LLVM
241        fuzzer.
242    
243    30. Error messages for syntax errors following \g and \k were giving inaccurate
244        offsets in the pattern.
245    
246    31. Added a check for integer overflow in conditions (?(<digits>) and
247        (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
248        fuzzer.
249    
250    32. Handling recursive references such as (?2) when the reference is to a group
251        later in the pattern uses code that is very hacked about and error-prone.
252        It has been re-written for PCRE2. Here in PCRE1, a check has been added to
253        give an internal error if it is obvious that compiling has gone wrong.
254    
255    33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
256        This issue was found by Karl Skomski with a custom LLVM fuzzer.
257    
258    34. The JIT compiler should restore the control chain for empty possessive
259        repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
260    
261    35. Match limit check added to JIT recursion. This issue was found by Karl
262        Skomski with a custom LLVM fuzzer.
263    
264    36. Yet another case similar to 27 above has been circumvented by an
265        unconditional allocation of extra memory. This issue is fixed "properly" in
266        PCRE2 by refactoring the way references are handled. Wen Guanxing
267        from Venustech ADLAB discovered this bug.
268    
269    37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
270        with a custom LLVM fuzzer.
271    
272    38. Fixed a corner case of range optimization in JIT.
273    
274    39. An incorrect error "overran compiling workspace" was given if there were
275        exactly enough group forward references such that the last one extended
276        into the workspace safety margin. The next one would have expanded the
277        workspace. The test for overflow was not including the safety margin.
278    
279    40. A match limit issue is fixed in JIT which was found by Karl Skomski
280        with a custom LLVM fuzzer.
281    
282    41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
283        work under Windows. (Why has it taken so long for anyone to notice?)
284    
285    42. In a character class such as [\W\p{Any}] where both a negative-type escape
286        ("not a word character") and a property escape were present, the property
287        escape was being ignored.
288    
289    43. Fix crash caused by very long (*MARK) or (*THEN) names.
290    
291    44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
292        by a single ASCII character in a class item, was incorrectly compiled in
293        UCP mode. The POSIX class got lost, but only if the single character
294        followed it.
295    
296    45. [:punct:] in UCP mode was matching some characters in the range 128-255
297        that should not have been matched.
298    
299    46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
300        class, all characters with code points greater than 255 are in the class.
301        When a Unicode property was also in the class (if PCRE_UCP is set, escapes
302        such as \w are turned into Unicode properties), wide characters were not
303        correctly handled, and could fail to match.
304    
305    
306    Version 8.37 28-April-2015
307    --------------------------
308    
309    1.  When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
310        for those parentheses to be closed with whatever has been captured so far.
311        However, it was failing to mark any other groups between the hightest
312        capture so far and the currrent group as "unset". Thus, the ovector for
313        those groups contained whatever was previously there. An example is the
314        pattern /(x)|((*ACCEPT))/ when matched against "abcd".
315    
316    2.  If an assertion condition was quantified with a minimum of zero (an odd
317        thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
318    
319    3.  If a pattern in pcretest input had the P (POSIX) modifier followed by an
320        unrecognized modifier, a crash could occur.
321    
322    4.  An attempt to do global matching in pcretest with a zero-length ovector
323        caused a crash.
324    
325    5.  Fixed a memory leak during matching that could occur for a subpattern
326        subroutine call (recursive or otherwise) if the number of captured groups
327        that had to be saved was greater than ten.
328    
329    6.  Catch a bad opcode during auto-possessification after compiling a bad UTF
330        string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
331        UTF with NO_UTF_CHECK is documented as having an undefined outcome.
332    
333    7.  A UTF pattern containing a "not" match of a non-ASCII character and a
334        subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
335    
336    8. When a pattern is compiled, it remembers the highest back reference so that
337       when matching, if the ovector is too small, extra memory can be obtained to
338       use instead. A conditional subpattern whose condition is a check on a
339       capture having happened, such as, for example in the pattern
340       /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
341       setting the highest backreference number. This mattered only if pcre_exec()
342       was called with an ovector that was too small to hold the capture, and there
343       was no other kind of back reference (a situation which is probably quite
344       rare). The effect of the bug was that the condition was always treated as
345       FALSE when the capture could not be consulted, leading to a incorrect
346       behaviour by pcre_exec(). This bug has been fixed.
347    
348    9. A reference to a duplicated named group (either a back reference or a test
349       for being set in a conditional) that occurred in a part of the pattern where
350       PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
351       to be incorrectly calculated, leading to overwriting.
352    
353    10. A mutually recursive set of back references such as (\2)(\1) caused a
354        segfault at study time (while trying to find the minimum matching length).
355        The infinite loop is now broken (with the minimum length unset, that is,
356        zero).
357    
358    11. If an assertion that was used as a condition was quantified with a minimum
359        of zero, matching went wrong. In particular, if the whole group had
360        unlimited repetition and could match an empty string, a segfault was
361        likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
362        assertions to be quantified, but not if they are being used as conditions,
363        so the above pattern is faulted by Perl. PCRE has now been changed so that
364        it also rejects such patterns.
365    
366    12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
367        failed to allow the zero-repeat case if pcre2_exec() was called with an
368        ovector too small to capture the group.
369    
370    13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
371        Red Hat Product Security:
372    
373        (a) A crash if /K and /F were both set with the option to save the compiled
374        pattern.
375    
376        (b) Another crash if the option to print captured substrings in a callout
377        was combined with setting a null ovector, for example \O\C+ as a subject
378        string.
379    
380    14. A pattern such as "((?2){0,1999}())?", which has a group containing a
381        forward reference repeated a large (but limited) number of times within a
382        repeated outer group that has a zero minimum quantifier, caused incorrect
383        code to be compiled, leading to the error "internal error:
384        previously-checked referenced subpattern not found" when an incorrect
385        memory address was read. This bug was reported as "heap overflow",
386        discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
387        CVE-2015-2325.
388    
389    23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
390        call within a group that also contained a recursive back reference caused
391        incorrect code to be compiled. This bug was reported as "heap overflow",
392        discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
393        number CVE-2015-2326.
394    
395    24. Computing the size of the JIT read-only data in advance has been a source
396        of various issues, and new ones are still appear unfortunately. To fix
397        existing and future issues, size computation is eliminated from the code,
398        and replaced by on-demand memory allocation.
399    
400    25. A pattern such as /(?i)[A-`]/, where characters in the other case are
401        adjacent to the end of the range, and the range contained characters with
402        more than one other case, caused incorrect behaviour when compiled in UTF
403        mode. In that example, the range a-j was left out of the class.
404    
405    26. Fix JIT compilation of conditional blocks, which assertion
406        is converted to (*FAIL). E.g: /(?(?!))/.
407    
408    27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
409        discovered by the LLVM fuzzer.
410    
411    28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
412        when this assertion was used as a condition, for example (?(?!)a|b). In
413        pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
414        error about an unsupported item.
415    
416    29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
417        possessification code could take exponential time to complete. A recursion
418        depth limit of 1000 has been imposed to limit the resources used by this
419        optimization.
420    
421    30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
422        such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
423        because \S ensures they are all in the class. The code for doing this was
424        interacting badly with the code for computing the amount of space needed to
425        compile the pattern, leading to a buffer overflow. This bug was discovered
426        by the LLVM fuzzer.
427    
428    31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
429        other kinds of group caused stack overflow at compile time. This bug was
430        discovered by the LLVM fuzzer.
431    
432    32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
433        between a subroutine call and its quantifier was incorrectly compiled,
434        leading to buffer overflow or other errors. This bug was discovered by the
435        LLVM fuzzer.
436    
437    33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
438        assertion after (?(. The code was failing to check the character after
439        (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
440        was discovered by the LLVM fuzzer.
441    
442    34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
443        a fixed maximum following a group that contains a subroutine reference was
444        incorrectly compiled and could trigger buffer overflow. This bug was
445        discovered by the LLVM fuzzer.
446    
447    35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
448        caused a stack overflow instead of the diagnosis of a non-fixed length
449        lookbehind assertion. This bug was discovered by the LLVM fuzzer.
450    
451    36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
452        (e.g. /(?<=\Ka)/) could make pcregrep loop.
453    
454    37. There was a similar problem to 36 in pcretest for global matches.
455    
456    38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
457        and a subsequent item in the pattern caused a non-match, backtracking over
458        the repeated \X did not stop, but carried on past the start of the subject,
459        causing reference to random memory and/or a segfault. There were also some
460        other cases where backtracking after \C could crash. This set of bugs was
461        discovered by the LLVM fuzzer.
462    
463    39. The function for finding the minimum length of a matching string could take
464        a very long time if mutual recursion was present many times in a pattern,
465        for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
466        method has been implemented. This infelicity was discovered by the LLVM
467        fuzzer.
468    
469    40. Static linking against the PCRE library using the pkg-config module was
470        failing on missing pthread symbols.
471    
472    
473    Version 8.36 26-September-2014
474    ------------------------------
475    
476    1.  Got rid of some compiler warnings in the C++ modules that were shown up by
477        -Wmissing-field-initializers and -Wunused-parameter.
478    
479    2.  The tests for quantifiers being too big (greater than 65535) were being
480        applied after reading the number, and stupidly assuming that integer
481        overflow would give a negative number. The tests are now applied as the
482        numbers are read.
483    
484    3.  Tidy code in pcre_exec.c where two branches that used to be different are
485        now the same.
486    
487    4.  The JIT compiler did not generate match limit checks for certain
488        bracketed expressions with quantifiers. This may lead to exponential
489        backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
490        issue should be resolved now.
491    
492    5.  Fixed an issue, which occures when nested alternatives are optimized
493        with table jumps.
494    
495    6.  Inserted two casts and changed some ints to size_t in the light of some
496        reported 64-bit compiler warnings (Bugzilla 1477).
497    
498    7.  Fixed a bug concerned with zero-minimum possessive groups that could match
499        an empty string, which sometimes were behaving incorrectly in the
500        interpreter (though correctly in the JIT matcher). This pcretest input is
501        an example:
502    
503          '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
504          NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
505    
506        the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
507        JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
508        for an empty string was breaking the inner loop and carrying on at a lower
509        level, when possessive repeated groups should always return to a higher
510        level as they have no backtrack points in them. The empty string test now
511        occurs at the outer level.
512    
513    8.  Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
514        ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
515    
516    9.  Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
517        doesn't).
518    
519    10. Change 8.34/15 introduced a bug that caused the amount of memory needed
520        to hold a pattern to be incorrectly computed (too small) when there were
521        named back references to duplicated names. This could cause "internal
522        error: code overflow" or "double free or corruption" or other memory
523        handling errors.
524    
525    11. When named subpatterns had the same prefixes, back references could be
526        confused. For example, in this pattern:
527    
528          /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
529    
530        the reference to 'Name' was incorrectly treated as a reference to a
531        duplicate name.
532    
533    12. A pattern such as /^s?c/mi8 where the optional character has more than
534        one "other case" was incorrectly compiled such that it would only try to
535        match starting at "c".
536    
537    13. When a pattern starting with \s was studied, VT was not included in the
538        list of possible starting characters; this should have been part of the
539        8.34/18 patch.
540    
541    14. If a character class started [\Qx]... where x is any character, the class
542        was incorrectly terminated at the ].
543    
544    15. If a pattern that started with a caseless match for a character with more
545        than one "other case" was studied, PCRE did not set up the starting code
546        unit bit map for the list of possible characters. Now it does. This is an
547        optimization improvement, not a bug fix.
548    
549    16. The Unicode data tables have been updated to Unicode 7.0.0.
550    
551    17. Fixed a number of memory leaks in pcregrep.
552    
553    18. Avoid a compiler warning (from some compilers) for a function call with
554        a cast that removes "const" from an lvalue by using an intermediate
555        variable (to which the compiler does not object).
556    
557    19. Incorrect code was compiled if a group that contained an internal recursive
558        back reference was optional (had quantifier with a minimum of zero). This
559        example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
560        caused segmentation faults because of stack overflows at compile time.
561    
562    20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
563        group that is quantified with an indefinite repeat, caused a compile-time
564        loop which used up all the system stack and provoked a segmentation fault.
565        This was not the same bug as 19 above.
566    
567    21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
568        Patch by Mike Frysinger.
569    
570    
571    Version 8.35 04-April-2014
572    --------------------------
573    
574    1.  A new flag is set, when property checks are present in an XCLASS.
575        When this flag is not set, PCRE can perform certain optimizations
576        such as studying these XCLASS-es.
577    
578    2.  The auto-possessification of character sets were improved: a normal
579        and an extended character set can be compared now. Furthermore
580        the JIT compiler optimizes more character set checks.
581    
582    3.  Got rid of some compiler warnings for potentially uninitialized variables
583        that show up only when compiled with -O2.
584    
585    4.  A pattern such as (?=ab\K) that uses \K in an assertion can set the start
586        of a match later then the end of the match. The pcretest program was not
587        handling the case sensibly - it was outputting from the start to the next
588        binary zero. It now reports this situation in a message, and outputs the
589        text from the end to the start.
590    
591    5.  Fast forward search is improved in JIT. Instead of the first three
592        characters, any three characters with fixed position can be searched.
593        Search order: first, last, middle.
594    
595    6.  Improve character range checks in JIT. Characters are read by an inprecise
596        function now, which returns with an unknown value if the character code is
597        above a certain threshold (e.g: 256). The only limitation is that the value
598        must be bigger than the threshold as well. This function is useful when
599        the characters above the threshold are handled in the same way.
600    
601    7.  The macros whose names start with RAWUCHAR are placeholders for a future
602        mode in which only the bottom 21 bits of 32-bit data items are used. To
603        make this more memorable for those maintaining the code, the names have
604        been changed to start with UCHAR21, and an extensive comment has been added
605        to their definition.
606    
607    8.  Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
608        to the export list in Makefile.am (they were accidentally omitted from the
609        8.34 tarball).
610    
611    9.  The informational output from pcretest used the phrase "starting byte set"
612        which is inappropriate for the 16-bit and 32-bit libraries. As the output
613        for "first char" and "need char" really means "non-UTF-char", I've changed
614        "byte" to "char", and slightly reworded the output. The documentation about
615        these values has also been (I hope) clarified.
616    
617    10. Another JIT related optimization: use table jumps for selecting the correct
618        backtracking path, when more than four alternatives are present inside a
619        bracket.
620    
621    11. Empty match is not possible, when the minimum length is greater than zero,
622        and there is no \K in the pattern. JIT should avoid empty match checks in
623        such cases.
624    
625    12. In a caseless character class with UCP support, when a character with more
626        than one alternative case was not the first character of a range, not all
627        the alternative cases were added to the class. For example, s and \x{17f}
628        are both alternative cases for S: the class [RST] was handled correctly,
629        but [R-T] was not.
630    
631    13. The configure.ac file always checked for pthread support when JIT was
632        enabled. This is not used in Windows, so I have put this test inside a
633        check for the presence of windows.h (which was already tested for).
634    
635    14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
636        The algorithm provides a way to skip certain starting offsets, and usually
637        faster than linear prefix searches.
638    
639    15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
640        as for 'fr_FR' and 'french'. For some reason, however, it then used the
641        Windows-specific input and output files, which have 'french' screwed in.
642        So this could never have worked. One of the problems with locales is that
643        they aren't always the same. I have now updated RunTest so that it checks
644        the output of the locale test (test 3) against three different output
645        files, and it allows the test to pass if any one of them matches. With luck
646        this should make the test pass on some versions of Solaris where it was
647        failing. Because of the uncertainty, the script did not used to stop if
648        test 3 failed; it now does. If further versions of a French locale ever
649        come to light, they can now easily be added.
650    
651    16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
652        there was a message during ./configure, but it did not stop. This now
653        provokes an error. The invalid example in README has been corrected.
654        If a value less than the minimum is given, the minimum value has always
655        been used, but now a warning is given.
656    
657    17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
658        was a bug in the test system, which is now fixed. Also, the list of various
659        configurations that are tested for each release did not have one with both
660        16/32 bits and --enable-bar-anycrlf. It now does.
661    
662    18. pcretest was missing "-C bsr" for displaying the \R default setting.
663    
664    19. Little endian PowerPC systems are supported now by the JIT compiler.
665    
666    20. The fast forward newline mechanism could enter to an infinite loop on
667        certain invalid UTF-8 input. Although we don't support these cases
668        this issue can be fixed by a performance optimization.
669    
670    21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
671        not take account if existing stack usage. There is now a new global
672        variable called pcre_stack_guard that can be set to point to an external
673        function to check stack availability. It is called at the start of
674        processing every parenthesized group.
675    
676    22. A typo in the code meant that in ungreedy mode the max/min qualifier
677        behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
678        not match "ab".
679    
680    23. When UTF was disabled, the JIT program reported some incorrect compile
681        errors. These messages are silenced now.
682    
683    24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
684        compiler.
685    
686    25. Change all the temporary files used in RunGrepTest to be different to those
687        used by RunTest so that the tests can be run simultaneously, for example by
688        "make -j check".
689    
690    
691    Version 8.34 15-December-2013
692    -----------------------------
693    
694    1.  Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
695        executable memory. Patch inspired by Carsten Klein.
696    
697    2.  ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
698        this macro is never tested and has no effect, because the work to support
699        coverage involves only compiling and linking options and special targets in
700        the Makefile. The comment in config.h implied that defining the macro would
701        enable coverage support, which is totally false. There was also support for
702        setting this macro in the CMake files (my fault, I just copied it from
703        configure). SUPPORT_GCOV has now been removed.
704    
705    3.  Make a small performance improvement in strlen16() and strlen32() in
706        pcretest.
707    
708    4.  Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
709        detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
710        these cases). There was also one in pcretest.c.
711    
712    5.  Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
713    
714    6.  In UTF mode, the code for checking whether a group could match an empty
715        string (which is used for indefinitely repeated groups to allow for
716        breaking an infinite loop) was broken when the group contained a repeated
717        negated single-character class with a character that occupied more than one
718        data item and had a minimum repetition of zero (for example, [^\x{100}]* in
719        UTF-8 mode). The effect was undefined: the group might or might not be
720        deemed as matching an empty string, or the program might have crashed.
721    
722    7.  The code for checking whether a group could match an empty string was not
723        recognizing that \h, \H, \v, \V, and \R must match a character.
724    
725    8.  Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
726        an empty string. If it can, pcretest shows this in its information output.
727    
728    9.  Fixed two related bugs that applied to Unicode extended grapheme clusters
729        that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
730        matched by pcre_exec() without using JIT:
731    
732        (a) If the rest of the pattern did not match after a maximal run of
733            grapheme clusters, the code for backing up to try with fewer of them
734            did not always back up over a full grapheme when characters that do not
735            have the modifier quality were involved, e.g. Hangul syllables.
736    
737        (b) If the match point in a subject started with modifier character, and
738            there was no match, the code could incorrectly back up beyond the match
739            point, and potentially beyond the first character in the subject,
740            leading to a segfault or an incorrect match result.
741    
742    10. A conditional group with an assertion condition could lead to PCRE
743        recording an incorrect first data item for a match if no other first data
744        item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
745        first data item, and therefore matched "ca" after "c" instead of at the
746        start.
747    
748    11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
749        bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
750    
751    12. The source of pcregrep now includes z/OS-specific code so that it can be
752        compiled for z/OS as part of the special z/OS distribution.
753    
754    13. Added the -T and -TM options to pcretest.
755    
756    14. The code in pcre_compile.c for creating the table of named capturing groups
757        has been refactored. Instead of creating the table dynamically during the
758        actual compiling pass, the information is remembered during the pre-compile
759        pass (on the stack unless there are more than 20 named groups, in which
760        case malloc() is used) and the whole table is created before the actual
761        compile happens. This has simplified the code (it is now nearly 150 lines
762        shorter) and prepared the way for better handling of references to groups
763        with duplicate names.
764    
765    15. A back reference to a named subpattern when there is more than one of the
766        same name now checks them in the order in which they appear in the pattern.
767        The first one that is set is used for the reference. Previously only the
768        first one was inspected. This change makes PCRE more compatible with Perl.
769    
770    16. Unicode character properties were updated from Unicode 6.3.0.
771    
772    17. The compile-time code for auto-possessification has been refactored, based
773        on a patch by Zoltan Herczeg. It now happens after instead of during
774        compilation. The code is cleaner, and more cases are handled. The option
775        PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
776        options in pcretest are provided to set it. It can also be set by
777        (*NO_AUTO_POSSESS) at the start of a pattern.
778    
779    18. The character VT has been added to the default ("C" locale) set of
780        characters that match \s and are generally treated as white space,
781        following this same change in Perl 5.18. There is now no difference between
782        "Perl space" and "POSIX space". Whether VT is treated as white space in
783        other locales depends on the locale.
784    
785    19. The code for checking named groups as conditions, either for being set or
786        for being recursed, has been refactored (this is related to 14 and 15
787        above). Processing unduplicated named groups should now be as fast at
788        numerical groups, and processing duplicated groups should be faster than
789        before.
790    
791    20. Two patches to the CMake build system, by Alexander Barkov:
792    
793          (1) Replace the "source" command by "." in CMakeLists.txt because
794              "source" is a bash-ism.
795    
796          (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
797              without these the CMake build does not work on Solaris.
798    
799    21. Perl has changed its handling of \8 and \9. If there is no previously
800        encountered capturing group of those numbers, they are treated as the
801        literal characters 8 and 9 instead of a binary zero followed by the
802        literals. PCRE now does the same.
803    
804    22. Following Perl, added \o{} to specify codepoints in octal, making it
805        possible to specify values greater than 0777 and also making them
806        unambiguous.
807    
808    23. Perl now gives an error for missing closing braces after \x{... instead of
809        treating the string as literal. PCRE now does the same.
810    
811    24. RunTest used to grumble if an inappropriate test was selected explicitly,
812        but just skip it when running all tests. This make it awkward to run ranges
813        of tests when one of them was inappropriate. Now it just skips any
814        inappropriate tests, as it always did when running all tests.
815    
816    25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
817        character types such as \d or \w, too many callouts were inserted, and the
818        data that they returned was rubbish.
819    
820    26. In UCP mode, \s was not matching two of the characters that Perl matches,
821        namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
822        were matched by \h. The code has now been refactored so that the lists of
823        the horizontal and vertical whitespace characters used for \h and \v (which
824        are defined only in one place) are now also used for \s.
825    
826    27. Add JIT support for the 64 bit TileGX architecture.
827        Patch by Jiong Wang (Tilera Corporation).
828    
829    28. Possessive quantifiers for classes (both explicit and automatically
830        generated) now use special opcodes instead of wrapping in ONCE brackets.
831    
832    29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
833        (because it's meaningless), this was not happening when PCRE_CASELESS was
834        set. Not wrong, but inefficient.
835    
836    30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
837        properties for \w, \d, etc) is present in a test regex. Otherwise if the
838        test contains no characters greater than 255, Perl doesn't realise it
839        should be using Unicode semantics.
840    
841    31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
842        [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
843        does in Unicode mode.
844    
845    32. Added the "forbid" facility to pcretest so that putting tests into the
846        wrong test files can sometimes be quickly detected.
847    
848    33. There is now a limit (default 250) on the depth of nesting of parentheses.
849        This limit is imposed to control the amount of system stack used at compile
850        time. It can be changed at build time by --with-parens-nest-limit=xxx or
851        the equivalent in CMake.
852    
853    34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
854        errors. Perl warns for these when in warning mode, but PCRE has no facility
855        for giving warnings.
856    
857    35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
858        However, this was not working for (?!) because it is optimized to (*FAIL),
859        for which PCRE does not allow quantifiers. The optimization is now disabled
860        when a quantifier follows (?!). I can't see any use for this, but it makes
861        things uniform.
862    
863    36. Perl no longer allows group names to start with digits, so I have made this
864        change also in PCRE. It simplifies the code a bit.
865    
866    37. In extended mode, Perl ignores spaces before a + that indicates a
867        possessive quantifier. PCRE allowed a space before the quantifier, but not
868        before the possessive +. It now does.
869    
870    38. The use of \K (reset reported match start) within a repeated possessive
871        group such as (a\Kb)*+ was not working.
872    
873    40. Document that the same character tables must be used at compile time and
874        run time, and that the facility to pass tables to pcre_exec() and
875        pcre_dfa_exec() is for use only with saved/restored patterns.
876    
877    41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
878        features for Builds with MSVC:
879    
880        1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
881           builds). The .rc files can be used to set FileDescription and many other
882           attributes.
883    
884        2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
885           This allows higher-level build scripts which want .pdb files to avoid
886           hard-coding the exact files needed."
887    
888    42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
889        mean "start of word" and "end of word", respectively, as a transition aid.
890    
891    43. A minimizing repeat of a class containing codepoints greater than 255 in
892        non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
893        compiled to use the heap for recursion.
894    
895    44. Got rid of some compiler warnings for unused variables when UTF but not UCP
896        is configured.
897    
898    
899    Version 8.33 28-May-2013
900    ------------------------
901    
902    1.  Added 'U' to some constants that are compared to unsigned integers, to
903        avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
904        variables that are added to signed variables, to ensure the result is
905        signed and can be negated.
906    
907    2.  Applied patch by Daniel Richard G for quashing MSVC warnings to the
908        CMake config files.
909    
910    3.  Revise the creation of config.h.generic so that all boolean macros are
911        #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
912        overriding via -D on the command line possible.
913    
914    4.  Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
915        to unsigned int is reported to make a quite noticeable speed difference in
916        a specific Windows environment. Testing on Linux did also appear to show
917        some benefit (and it is clearly not harmful). Also fixed the definition of
918        Xop which should be unsigned.
919    
920    5.  Related to (4), changing the definition of the intermediate variable cc
921        in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
922        improvements.
923    
924    6.  Fix forward search in JIT when link size is 3 or greater. Also removed some
925        unnecessary spaces.
926    
927    7.  Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
928        and later.
929    
930    8.  Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
931    
932    9.  Optimizing fast_forward_start_bits in JIT.
933    
934    10. Adding support for callouts in JIT, and fixing some issues revealed
935        during this work. Namely:
936    
937        (a) Unoptimized capturing brackets incorrectly reset on backtrack.
938    
939        (b) Minimum length was not checked before the matching is started.
940    
941    11. The value of capture_last that is passed to callouts was incorrect in some
942        cases when there was a capture on one path that was subsequently abandoned
943        after a backtrack. Also, the capture_last value is now reset after a
944        recursion, since all captures are also reset in this case.
945    
946    12. The interpreter no longer returns the "too many substrings" error in the
947        case when an overflowing capture is in a branch that is subsequently
948        abandoned after a backtrack.
949    
950    13. In the pathological case when an offset vector of size 2 is used, pcretest
951        now prints out the matched string after a yield of 0 or 1.
952    
953    14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
954        Only supported by the JIT compiler at the moment.
955    
956    15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
957    
958    16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
959        offset of the starting point of the matching process, provided the offsets
960        vector is large enough.
961    
962    17. The \A escape now records a lookbehind value of 1, though its execution
963        does not actually inspect the previous character. This is to ensure that,
964        in partial multi-segment matching, at least one character from the old
965        segment is retained when a new segment is processed. Otherwise, if there
966        are no lookbehinds in the pattern, \A might match incorrectly at the start
967        of a new segment.
968    
969    18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
970    
971    19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
972        gives some modest performance improvement in 8-bit mode.
973    
974    20. Added the PCRE-specific property \p{Xuc} for matching characters that can
975        be expressed in certain programming languages using Universal Character
976        Names.
977    
978    21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
979        which points out that "non characters" are not "characters that may not
980        appear in Unicode strings" but rather "characters that are reserved for
981        internal use and have only local meaning".
982    
983    22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
984        there was a conditional group that depended on an assertion, if the
985        assertion was false, the callout that immediately followed the alternation
986        in the condition was skipped when pcre_exec() was used for matching.
987    
988    23. Allow an explicit callout to be inserted before an assertion that is the
989        condition for a conditional group, for compatibility with automatic
990        callouts, which always insert a callout at this point.
991    
992    24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
993        confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
994    
995    25. (*PRUNE) is now supported by the JIT compiler.
996    
997    26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
998    
999    27. Fix the case where there are two or more SKIPs with arguments that may be
1000        ignored.
1001    
1002    28. (*SKIP) is now supported by the JIT compiler.
1003    
1004    29. (*THEN) is now supported by the JIT compiler.
1005    
1006    30. Update RunTest with additional test selector options.
1007    
1008    31. The way PCRE handles backtracking verbs has been changed in two ways.
1009    
1010        (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
1011        SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
1012        backtracking. In some cases this makes it more Perl-compatible, but Perl's
1013        rather obscure rules do not always do the same thing.
1014    
1015        (2) Previously, backtracking verbs were confined within assertions. This is
1016        no longer the case for positive assertions, except for (*ACCEPT). Again,
1017        this sometimes improves Perl compatibility, and sometimes does not.
1018    
1019    32. A number of tests that were in test 2 because Perl did things differently
1020        have been moved to test 1, because either Perl or PCRE has changed, and
1021        these tests are now compatible.
1022    
1023    32. Backtracking control verbs are now handled in the same way in JIT and
1024        interpreter.
1025    
1026    33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
1027        contained a forward subroutine reference caused a compile error.
1028    
1029    34. Auto-detect and optimize limited repetitions in JIT.
1030    
1031    35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
1032        blocking (*UTF) etc.
1033    
1034    36. In the interpreter, maximizing pattern repetitions for characters and
1035        character types now use tail recursion, which reduces stack usage.
1036    
1037    37. The value of the max lookbehind was not correctly preserved if a compiled
1038        and saved regex was reloaded on a host of different endianness.
1039    
1040    38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
1041        of the compiled pattern block, expand the flags field from 16 to 32 bits
1042        because it was almost full.
1043    
1044    39. Try madvise first before posix_madvise.
1045    
1046    40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
1047        with a pattern such as ^$. It has taken 4 years for anybody to notice! The
1048        original change locked out all matches of empty strings. This has been
1049        changed so that one match of an empty string per line is recognized.
1050        Subsequent searches on the same line (for colouring or for --only-matching,
1051        for example) do not recognize empty strings.
1052    
1053    41. Applied a user patch to fix a number of spelling mistakes in comments.
1054    
1055    42. Data lines longer than 65536 caused pcretest to crash.
1056    
1057    43. Clarified the data type for length and startoffset arguments for pcre_exec
1058        and pcre_dfa_exec in the function-specific man pages, where they were
1059        explicitly stated to be in bytes, never having been updated. I also added
1060        some clarification to the pcreapi man page.
1061    
1062    44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
1063        a segmentation fault.
1064    
1065    
1066    Version 8.32 30-November-2012
1067    -----------------------------
1068    
1069    1.  Improved JIT compiler optimizations for first character search and single
1070        character iterators.
1071    
1072    2.  Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
1073        Patch by Daniel Richard G.
1074    
1075    3.  Single character iterator optimizations in the JIT compiler.
1076    
1077    4.  Improved JIT compiler optimizations for character ranges.
1078    
1079    5.  Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1080        Reported by Giuseppe D'Angelo.
1081    
1082    6.  The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1083        of a line, was being set incorrectly in cases where .* appeared inside
1084        atomic brackets at the start of a pattern, or where there was a subsequent
1085        *PRUNE or *SKIP.
1086    
1087    7.  Improved instruction cache flush for POWER/PowerPC.
1088        Patch by Daniel Richard G.
1089    
1090    8.  Fixed a number of issues in pcregrep, making it more compatible with GNU
1091        grep:
1092    
1093        (a) There is now no limit to the number of patterns to be matched.
1094    
1095        (b) An error is given if a pattern is too long.
1096    
1097        (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1098            are now supported.
1099    
1100        (d) --exclude-from and --include-from (multiple use) have been added.
1101    
1102        (e) Exclusions and inclusions now apply to all files and directories, not
1103            just to those obtained from scanning a directory recursively.
1104    
1105        (f) Multiple uses of -f and --file-list are now supported.
1106    
1107        (g) In a Windows environment, the default for -d has been changed from
1108            "read" (the GNU grep default) to "skip", because otherwise the presence
1109            of a directory in the file list provokes an error.
1110    
1111        (h) The documentation has been revised and clarified in places.
1112    
1113    9.  Improve the matching speed of capturing brackets.
1114    
1115    10. Changed the meaning of \X so that it now matches a Unicode extended
1116        grapheme cluster.
1117    
1118    11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1119        out POSIX threads when JIT support is configured.
1120    
1121    12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1122    
1123    13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1124        structure could go wrong in environments where size_t is not the same size
1125        as int.
1126    
1127    14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1128        set.
1129    
1130    15. The EBCDIC support had decayed; later updates to the code had included
1131        explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1132        general tidy up of EBCDIC-related issues, and the documentation was also
1133        not quite right. There is now a test that can be run on ASCII systems to
1134        check some of the EBCDIC-related things (but is it not a full test).
1135    
1136    16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1137        in a small tidy to the code.
1138    
1139    17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1140    
1141    18. If the --only-matching (-o) option in pcregrep is specified multiple
1142        times, each one causes appropriate output. For example, -o1 -o2 outputs the
1143        substrings matched by the 1st and 2nd capturing parentheses. A separating
1144        string can be specified by --om-separator (default empty).
1145    
1146    19. Improving the first n character searches.
1147    
1148    20. Turn case lists for horizontal and vertical white space into macros so that
1149        they are defined only once.
1150    
1151    21. This set of changes together give more compatible Unicode case-folding
1152        behaviour for characters that have more than one other case when UCP
1153        support is available.
1154    
1155        (a) The Unicode property table now has offsets into a new table of sets of
1156            three or more characters that are case-equivalent. The MultiStage2.py
1157            script that generates these tables (the pcre_ucd.c file) now scans
1158            CaseFolding.txt instead of UnicodeData.txt for character case
1159            information.
1160    
1161        (b) The code for adding characters or ranges of characters to a character
1162            class has been abstracted into a generalized function that also handles
1163            case-independence. In UTF-mode with UCP support, this uses the new data
1164            to handle characters with more than one other case.
1165    
1166        (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1167            whose other case is greater than 256 are now correctly matched
1168            caselessly. Previously, the high codepoint matched the low one, but not
1169            vice versa.
1170    
1171        (d) The processing of \h, \H, \v, and \ in character classes now makes use
1172            of the new class addition function, using character lists defined as
1173            macros alongside the case definitions of 20 above.
1174    
1175        (e) Caseless back references now work with characters that have more than
1176            one other case.
1177    
1178        (f) General caseless matching of characters with more than one other case
1179            is supported.
1180    
1181    22. Unicode character properties were updated from Unicode 6.2.0
1182    
1183    23. Improved CMake support under Windows. Patch by Daniel Richard G.
1184    
1185    24. Add support for 32-bit character strings, and UTF-32
1186    
1187    25. Major JIT compiler update (code refactoring and bugfixing).
1188        Experimental Sparc 32 support is added.
1189    
1190    26. Applied a modified version of Daniel Richard G's patch to create
1191        pcre.h.generic and config.h.generic by "make" instead of in the
1192        PrepareRelease script.
1193    
1194    27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1195        pcre_compile.c when checking for a zero character.
1196    
1197    28. Introducing a native interface for JIT. Through this interface, the compiled
1198        machine code can be directly executed. The purpose of this interface is to
1199        provide fast pattern matching, so several sanity checks are not performed.
1200        However, feature tests are still performed. The new interface provides
1201        1.4x speedup compared to the old one.
1202    
1203    29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1204        the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1205        was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1206    
1207    30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1208        greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1209        the "old" RFC 2279). Instead, it ended up passing a negative length to
1210        pcre_exec().
1211    
1212    31. Add support for GCC's visibility feature to hide internal functions.
1213    
1214    32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1215        "unknown -C option" after outputting 0 or 1.
1216    
1217    33. There is now support for generating a code coverage report for the test
1218        suite in environments where gcc is the compiler and lcov is installed. This
1219        is mainly for the benefit of the developers.
1220    
1221    34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1222        unaddressable using valgrind annotations, allowing valgrind to detect
1223        invalid memory accesses. This is mainly for the benefit of the developers.
1224    
1225    25. (*UTF) can now be used to start a pattern in any of the three libraries.
1226    
1227    26. Give configure error if --enable-cpp but no C++ compiler found.
1228    
1229    
1230    Version 8.31 06-July-2012
1231    -------------------------
1232    
1233    1.  Fixing a wrong JIT test case and some compiler warnings.
1234    
1235    2.  Removed a bashism from the RunTest script.
1236    
1237    3.  Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1238        to unsigned type, result still unsigned" that was given by an MS compiler
1239        on encountering the code "-sizeof(xxx)".
1240    
1241    4.  Partial matching support is added to the JIT compiler.
1242    
1243    5.  Fixed several bugs concerned with partial matching of items that consist
1244        of more than one character:
1245    
1246        (a) /^(..)\1/ did not partially match "aba" because checking references was
1247            done on an "all or nothing" basis. This also applied to repeated
1248            references.
1249    
1250        (b) \R did not give a hard partial match if \r was found at the end of the
1251            subject.
1252    
1253        (c) \X did not give a hard partial match after matching one or more
1254            characters at the end of the subject.
1255    
1256        (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1257            a partial match for the string "\r".
1258    
1259        (e) When newline was set to CRLF, the metacharacter "." did not recognize
1260            a partial match for a CR character at the end of the subject string.
1261    
1262    6.  If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1263        running pcretest, the text "(JIT)" added to the output whenever JIT is
1264        actually used to run the match.
1265    
1266    7.  Individual JIT compile options can be set in pcretest by following -s+[+]
1267        or /S+[+] with a digit between 1 and 7.
1268    
1269    8.  OP_NOT now supports any UTF character not just single-byte ones.
1270    
1271    9.  (*MARK) control verb is now supported by the JIT compiler.
1272    
1273    10. The command "./RunTest list" lists the available tests without actually
1274        running any of them. (Because I keep forgetting what they all are.)
1275    
1276    11. Add PCRE_INFO_MAXLOOKBEHIND.
1277    
1278    12. Applied a (slightly modified) user-supplied patch that improves performance
1279        when the heap is used for recursion (compiled with --disable-stack-for-
1280        recursion). Instead of malloc and free for each heap frame each time a
1281        logical recursion happens, frames are retained on a chain and re-used where
1282        possible. This sometimes gives as much as 30% improvement.
1283    
1284    13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1285        call.
1286    
1287    14. As documented, (*COMMIT) is now confined to within a positive assertion.
1288    
1289    15. It is now possible to link pcretest with libedit as an alternative to
1290        libreadline.
1291    
1292    16. (*COMMIT) control verb is now supported by the JIT compiler.
1293    
1294    17. The Unicode data tables have been updated to Unicode 6.1.0.
1295    
1296    18. Added --file-list option to pcregrep.
1297    
1298    19. Added binary file support to pcregrep, including the -a, --binary-files,
1299        -I, and --text options.
1300    
1301    20. The madvise function is renamed for posix_madvise for QNX compatibility
1302        reasons. Fixed by Giuseppe D'Angelo.
1303    
1304    21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1305    
1306    22. Changed the option for creating symbolic links for 16-bit man pages from
1307        -s to -sf so that re-installing does not cause issues.
1308    
1309    23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1310    
1311    24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1312        matches in certain environments (the workspace was not being correctly
1313        retained). Also added to pcre_dfa_exec() a simple plausibility check on
1314        some of the workspace data at the beginning of a restart.
1315    
1316    25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1317        was not doing so when it should - probably a typo introduced by SVN 528
1318        (change 8.10/14).
1319    
1320    26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1321        \w+ when the character tables indicated that \x{c4} was a word character.
1322        There were several related cases, all because the tests for doing a table
1323        lookup were testing for characters less than 127 instead of 255.
1324    
1325    27. If a pattern contains capturing parentheses that are not used in a match,
1326        their slots in the ovector are set to -1. For those that are higher than
1327        any matched groups, this happens at the end of processing. In the case when
1328        there were back references that the ovector was too small to contain
1329        (causing temporary malloc'd memory to be used during matching), and the
1330        highest capturing number was not used, memory off the end of the ovector
1331        was incorrectly being set to -1. (It was using the size of the temporary
1332        memory instead of the true size.)
1333    
1334    28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1335        ovector size, it uses memory at the end of the block that it has got.
1336    
1337    29. Check for an overlong MARK name and give an error at compile time. The
1338        limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1339    
1340    30. JIT compiler update.
1341    
1342    31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1343        Rill for the patch.
1344    
1345    32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1346    
1347    33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1348    
1349    34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1350        SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1351        was enabled.
1352    
1353    35. Improve JIT code generation for greedy plus quantifier.
1354    
1355    36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1356        1 to "aa" instead of to an empty string. The bug affected repeated groups
1357        that could potentially match an empty string.
1358    
1359    37. Optimizing single character iterators in JIT.
1360    
1361    38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1362        the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1363        codepoints that are too big for the mode are faulted, and in a UTF mode,
1364        disallowed codepoints are also faulted.
1365    
1366    39. If PCRE was compiled with UTF support, in three places in the DFA
1367        matcher there was code that should only have been obeyed in UTF mode, but
1368        was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1369        processing when bytes with values greater than 127 were present. In 16-bit
1370        mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1371        both cases the values are those that cannot be the first data item in a UTF
1372        character. The three items that might have provoked this were recursions,
1373        possessively repeated groups, and atomic groups.
1374    
1375    40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1376        and pcregrep, because some OS require shared objects to be explicitly
1377        passed to ld, causing the link step to fail if they are not.
1378    
1379    41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1380        mode, patterns that started with \h* or \R* might be incorrectly matched.
1381    
1382    
1383    Version 8.30 04-February-2012
1384    -----------------------------
1385    
1386    1.  Renamed "isnumber" as "is_a_number" because in some Mac environments this
1387        name is defined in ctype.h.
1388    
1389    2.  Fixed a bug in fixed-length calculation for lookbehinds that would show up
1390        only in quite long subpatterns.
1391    
1392    3.  Removed the function pcre_info(), which has been obsolete and deprecated
1393        since it was replaced by pcre_fullinfo() in February 2000.
1394    
1395    4.  For a non-anchored pattern, if (*SKIP) was given with a name that did not
1396        match a (*MARK), and the match failed at the start of the subject, a
1397        reference to memory before the start of the subject could occur. This bug
1398        was introduced by fix 17 of release 8.21.
1399    
1400    5.  A reference to an unset group with zero minimum repetition was giving
1401        totally wrong answers (in non-JavaScript-compatibility mode). For example,
1402        /(another)?(\1?)test/ matched against "hello world test". This bug was
1403        introduced in release 8.13.
1404    
1405    6.  Add support for 16-bit character strings (a large amount of work involving
1406        many changes and refactorings).
1407    
1408    7.  RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1409        command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1410        from a file.
1411    
1412    8.  Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1413        rounding is not applied in this particular case).
1414    
1415    9.  The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1416        if they appear, or are escaped, in patterns.
1417    
1418    10. Get rid of a number of -Wunused-but-set-variable warnings.
1419    
1420    11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1421        "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1422        Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1423        also returns the mark "x". This bug applied to capturing parentheses,
1424        non-capturing parentheses, and atomic parentheses. It also applied to some
1425        assertions.
1426    
1427    12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1428        information out of configure.ac instead of relying on pcre.h.generic, which
1429        is not stored in the repository.
1430    
1431    13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1432        -lreadline.
1433    
1434    14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1435    
1436    15. Applied Graycode's patch to put the top-level frame on the stack rather
1437        than the heap when not using the stack for recursion. This gives a
1438        performance improvement in many cases when recursion is not deep.
1439    
1440    16. Experimental code added to "pcretest -C" to output the stack frame size.
1441    
1442    
1443    Version 8.21 12-Dec-2011
1444    ------------------------
1445    
1446  1.  Updating the JIT compiler.  1.  Updating the JIT compiler.
1447    
1448    2.  JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1449        are added as well.
1450    
1451    3.  Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1452        PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1453        calling _pcre_jit_exec. Some extra comments are added.
1454    
1455    4.  (*MARK) settings inside atomic groups that do not contain any capturing
1456        parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1457        was introduced by change 18 for 8.20.
1458    
1459    5.  Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1460        ECMA-262 standard.
1461    
1462    6.  Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1463        erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1464        This bug was probably introduced by change 9 of 8.13.
1465    
1466    7.  While fixing 6 above, I noticed that a number of other items were being
1467        incorrectly rejected as "not fixed length". This arose partly because newer
1468        opcodes had not been added to the fixed-length checking code. I have (a)
1469        corrected the bug and added tests for these items, and (b) arranged for an
1470        error to occur if an unknown opcode is encountered while checking for fixed
1471        length instead of just assuming "not fixed length". The items that were
1472        rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1473        (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1474        repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1475    
1476    8.  A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1477        being incorrectly compiled and would have given unpredicatble results.
1478    
1479    9.  A possessively repeated subpattern with minimum repeat count greater than
1480        one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1481        (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1482        the first (A) could occur when it should not.
1483    
1484    10. Add a cast and remove a redundant test from the code.
1485    
1486    11. JIT should use pcre_malloc/pcre_free for allocation.
1487    
1488    12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1489        best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1490        is anything other than /usr, -L is still shown).
1491    
1492    13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1493    
1494    14. Perl does not support \N without a following name in a [] class; PCRE now
1495        also gives an error.
1496    
1497    15. If a forward reference was repeated with an upper limit of around 2000,
1498        it caused the error "internal error: overran compiling workspace". The
1499        maximum number of forward references (including repeats) was limited by the
1500        internal workspace, and dependent on the LINK_SIZE. The code has been
1501        rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1502        the default depends on LINK_SIZE. There is a new upper limit (for safety)
1503        of around 200,000 forward references. While doing this, I also speeded up
1504        the filling in of repeated forward references.
1505    
1506    16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1507        incorrectly expecting the subject to contain another "a" after the start.
1508    
1509    17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1510        in the match, the SKIP should be ignored. This was not happening; instead
1511        the SKIP was being treated as NOMATCH. For patterns such as
1512        /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1513        tested.
1514    
1515    18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1516        now much more compatible with Perl, in particular in cases where the result
1517        is a non-match for a non-anchored pattern. For example, if
1518        /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1519        "m", where previously it did not return a name. A side effect of this
1520        change is that for partial matches, the last encountered mark name is
1521        returned, as for non matches. A number of tests that were previously not
1522        Perl-compatible have been moved into the Perl-compatible test files. The
1523        refactoring has had the pleasing side effect of removing one argument from
1524        the match() function, thus reducing its stack requirements.
1525    
1526    19. If the /S+ option was used in pcretest to study a pattern using JIT,
1527        subsequent uses of /S (without +) incorrectly behaved like /S+.
1528    
1529    21. Retrieve executable code size support for the JIT compiler and fixing
1530        some warnings.
1531    
1532    22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1533        not work when the shorter character appeared right at the end of the
1534        subject string.
1535    
1536    23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1537        systems.
1538    
1539    24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1540        output it when the /M option is used in pcretest.
1541    
1542    25. The CheckMan script was not being included in the distribution. Also, added
1543        an explicit "perl" to run Perl scripts from the PrepareRelease script
1544        because this is reportedly needed in Windows.
1545    
1546    26. If study data was being save in a file and studying had not found a set of
1547        "starts with" bytes for the pattern, the data written to the file (though
1548        never used) was taken from uninitialized memory and so caused valgrind to
1549        complain.
1550    
1551    27. Updated RunTest.bat as provided by Sheri Pierce.
1552    
1553    28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1554    
1555    29. Computation of memory usage for the table of capturing group names was
1556        giving an unnecessarily large value.
1557    
1558    
1559  Version 8.20 21-Oct-2011  Version 8.20 21-Oct-2011
1560  ------------------------  ------------------------
# Line 1021  Version 7.9 11-Apr-09 Line 2570  Version 7.9 11-Apr-09
2570  7.  A pattern that could match an empty string could cause pcregrep to loop; it  7.  A pattern that could match an empty string could cause pcregrep to loop; it
2571      doesn't make sense to accept an empty string match in pcregrep, so I have      doesn't make sense to accept an empty string match in pcregrep, so I have
2572      locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this      locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2573      seems to be how GNU grep behaves.      seems to be how GNU grep behaves. [But see later change 40 for release
2574        8.33.]
2575    
2576  8.  The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at  8.  The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2577      start or after a newline", because the conditional assertion was not being      start or after a newline", because the conditional assertion was not being
# Line 1264  Version 7.7 07-May-08 Line 2814  Version 7.7 07-May-08
2814      containing () gave an internal compiling error instead of "reference to      containing () gave an internal compiling error instead of "reference to
2815      non-existent subpattern". Fortunately, when the pattern did exist, the      non-existent subpattern". Fortunately, when the pattern did exist, the
2816      compiled code was correct. (When scanning forwards to check for the      compiled code was correct. (When scanning forwards to check for the
2817      existencd of the subpattern, it was treating the data ']' as terminating      existence of the subpattern, it was treating the data ']' as terminating
2818      the class, so got the count wrong. When actually compiling, the reference      the class, so got the count wrong. When actually compiling, the reference
2819      was subsequently set up correctly.)      was subsequently set up correctly.)
2820    

Legend:
Removed from v.740  
changed lines
  Added in v.1667

  ViewVC Help
Powered by ViewVC 1.1.5