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

Diff of /code/trunk/ChangeLog

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

revision 45 by nigel, Sat Feb 24 21:39:25 2007 UTC revision 1655 by ph10, Tue Jun 14 09:01:52 2016 UTC
# Line 1  Line 1 
1  ChangeLog for PCRE  ChangeLog for PCRE
2  ------------------  ------------------
3    
4    Note that the PCRE 8.xx series (PCRE1) is now in a bugfix-only state. All
5    development is happening in the PCRE2 10.xx series.
6    
7    Version 8.39 21-May-2016
8    ------------------------
9    
10    1.  If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
11        an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
12        misbehaved. This bug was found by the LLVM fuzzer.
13    
14    2.  Similar to the above, if an isolated \E was present between an item and its
15        qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
16        bug was found by the LLVM fuzzer.
17    
18    3.  Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
19        working correctly in UCP mode.
20    
21    4.  The POSIX wrapper function regexec() crashed if the option REG_STARTEND
22        was set when the pmatch argument was NULL. It now returns REG_INVARG.
23    
24    5.  Allow for up to 32-bit numbers in the ordin() function in pcregrep.
25    
26    6.  An empty \Q\E sequence between an item and its qualifier caused
27        pcre_compile() to misbehave when auto callouts were enabled. This bug was
28        found by the LLVM fuzzer.
29    
30    7.  If a pattern that was compiled with PCRE_EXTENDED started with white
31        space or a #-type comment that was followed by (?-x), which turns off
32        PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
33        pcre_compile() assumed that (?-x) applied to the whole pattern and
34        consequently mis-compiled it. This bug was found by the LLVM fuzzer.
35    
36    8.  An call of pcre_copy_named_substring() for a named substring whose number
37        was greater than the space in the ovector could cause a crash.
38    
39    9.  Yet another buffer overflow bug involved duplicate named groups with a
40        group that reset capture numbers (compare 8.38/7 below). Once again, I have
41        just allowed for more memory, even if not needed. (A proper fix is
42        implemented in PCRE2, but it involves a lot of refactoring.)
43    
44    10. pcre_get_substring_list() crashed if the use of \K in a match caused the
45        start of the match to be earlier than the end.
46    
47    11. Migrating appropriate PCRE2 JIT improvements to PCRE.
48    
49    12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
50        assertion, caused pcretest to generate incorrect output, and also to read
51        uninitialized memory (detected by ASAN or valgrind).
52    
53    13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
54        nested set of parentheses of sufficient size caused an overflow of the
55        compiling workspace (which was diagnosed, but of course is not desirable).
56    
57    14. And yet another buffer overflow bug involving duplicate named groups, this
58        time nested, with a nested back reference. Yet again, I have just allowed
59        for more memory, because anything more needs all the refactoring that has
60        been done for PCRE2. An example pattern that provoked this bug is:
61        /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
62        registered as CVE-2016-1283.
63    
64    15. pcretest went into a loop if global matching was requested with an ovector
65        size less than 2. It now gives an error message. This bug was found by
66        afl-fuzz.
67    
68    16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
69        ("assertion expected") when (?(?C) was not followed by an opening
70        parenthesis.
71    
72    17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
73        actually affect anything, by sheer luck.
74    
75    18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
76        static compilation.
77    
78    19. Modified the RunTest script to incorporate a valgrind suppressions file so
79        that certain errors, provoked by the SSE2 instruction set when JIT is used,
80        are ignored.
81    
82    20. A racing condition is fixed in JIT reported by Mozilla.
83    
84    21. Minor code refactor to avoid "array subscript is below array bounds"
85        compiler warning.
86    
87    22. Minor code refactor to avoid "left shift of negative number" warning.
88    
89    23. Fix typo causing compile error when 16- or 32-bit JIT is compiled without
90        UCP support.
91    
92    24. Refactor to avoid compiler warnings in pcrecpp.cc.
93    
94    25. Refactor to fix a typo in pcre_jit_test.c
95    
96    26. Patch to support compiling pcrecpp.cc with Intel compiler.
97    
98    
99    Version 8.38 23-November-2015
100    -----------------------------
101    
102    1.  If a group that contained a recursive back reference also contained a
103        forward reference subroutine call followed by a non-forward-reference
104        subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
105        compile correct code, leading to undefined behaviour or an internally
106        detected error. This bug was discovered by the LLVM fuzzer.
107    
108    2.  Quantification of certain items (e.g. atomic back references) could cause
109        incorrect code to be compiled when recursive forward references were
110        involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
111        This bug was discovered by the LLVM fuzzer.
112    
113    3.  A repeated conditional group whose condition was a reference by name caused
114        a buffer overflow if there was more than one group with the given name.
115        This bug was discovered by the LLVM fuzzer.
116    
117    4.  A recursive back reference by name within a group that had the same name as
118        another group caused a buffer overflow. For example:
119        /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
120    
121    5.  A forward reference by name to a group whose number is the same as the
122        current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
123        a buffer overflow at compile time. This bug was discovered by the LLVM
124        fuzzer.
125    
126    6.  A lookbehind assertion within a set of mutually recursive subpatterns could
127        provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
128    
129    7.  Another buffer overflow bug involved duplicate named groups with a
130        reference between their definition, with a group that reset capture
131        numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
132        fixed by always allowing for more memory, even if not needed. (A proper fix
133        is implemented in PCRE2, but it involves more refactoring.)
134    
135    8.  There was no check for integer overflow in subroutine calls such as (?123).
136    
137    9.  The table entry for \l in EBCDIC environments was incorrect, leading to its
138        being treated as a literal 'l' instead of causing an error.
139    
140    10. There was a buffer overflow if pcre_exec() was called with an ovector of
141        size 1. This bug was found by american fuzzy lop.
142    
143    11. If a non-capturing group containing a conditional group that could match
144        an empty string was repeated, it was not identified as matching an empty
145        string itself. For example: /^(?:(?(1)x|)+)+$()/.
146    
147    12. In an EBCDIC environment, pcretest was mishandling the escape sequences
148        \a and \e in test subject lines.
149    
150    13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
151        instead of the EBCDIC value.
152    
153    14. The handling of \c in an EBCDIC environment has been revised so that it is
154        now compatible with the specification in Perl's perlebcdic page.
155    
156    15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
157        ASCII/Unicode. This has now been added to the list of characters that are
158        recognized as white space in EBCDIC.
159    
160    16. When PCRE was compiled without UCP support, the use of \p and \P gave an
161        error (correctly) when used outside a class, but did not give an error
162        within a class.
163    
164    17. \h within a class was incorrectly compiled in EBCDIC environments.
165    
166    18. A pattern with an unmatched closing parenthesis that contained a backward
167        assertion which itself contained a forward reference caused buffer
168        overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
169    
170    19. JIT should return with error when the compiled pattern requires more stack
171        space than the maximum.
172    
173    20. A possessively repeated conditional group that could match an empty string,
174        for example, /(?(R))*+/, was incorrectly compiled.
175    
176    21. Fix infinite recursion in the JIT compiler when certain patterns such as
177        /(?:|a|){100}x/ are analysed.
178    
179    22. Some patterns with character classes involving [: and \\ were incorrectly
180        compiled and could cause reading from uninitialized memory or an incorrect
181        error diagnosis.
182    
183    23. Pathological patterns containing many nested occurrences of [: caused
184        pcre_compile() to run for a very long time.
185    
186    24. A conditional group with only one branch has an implicit empty alternative
187        branch and must therefore be treated as potentially matching an empty
188        string.
189    
190    25. If (?R was followed by - or + incorrect behaviour happened instead of a
191        diagnostic.
192    
193    26. Arrange to give up on finding the minimum matching length for overly
194        complex patterns.
195    
196    27. Similar to (4) above: in a pattern with duplicated named groups and an
197        occurrence of (?| it is possible for an apparently non-recursive back
198        reference to become recursive if a later named group with the relevant
199        number is encountered. This could lead to a buffer overflow. Wen Guanxing
200        from Venustech ADLAB discovered this bug.
201    
202    28. If pcregrep was given the -q option with -c or -l, or when handling a
203        binary file, it incorrectly wrote output to stdout.
204    
205    29. The JIT compiler did not restore the control verb head in case of *THEN
206        control verbs. This issue was found by Karl Skomski with a custom LLVM
207        fuzzer.
208    
209    30. Error messages for syntax errors following \g and \k were giving inaccurate
210        offsets in the pattern.
211    
212    31. Added a check for integer overflow in conditions (?(<digits>) and
213        (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
214        fuzzer.
215    
216    32. Handling recursive references such as (?2) when the reference is to a group
217        later in the pattern uses code that is very hacked about and error-prone.
218        It has been re-written for PCRE2. Here in PCRE1, a check has been added to
219        give an internal error if it is obvious that compiling has gone wrong.
220    
221    33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
222        This issue was found by Karl Skomski with a custom LLVM fuzzer.
223    
224    34. The JIT compiler should restore the control chain for empty possessive
225        repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
226    
227    35. Match limit check added to JIT recursion. This issue was found by Karl
228        Skomski with a custom LLVM fuzzer.
229    
230    36. Yet another case similar to 27 above has been circumvented by an
231        unconditional allocation of extra memory. This issue is fixed "properly" in
232        PCRE2 by refactoring the way references are handled. Wen Guanxing
233        from Venustech ADLAB discovered this bug.
234    
235    37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
236        with a custom LLVM fuzzer.
237    
238    38. Fixed a corner case of range optimization in JIT.
239    
240    39. An incorrect error "overran compiling workspace" was given if there were
241        exactly enough group forward references such that the last one extended
242        into the workspace safety margin. The next one would have expanded the
243        workspace. The test for overflow was not including the safety margin.
244    
245    40. A match limit issue is fixed in JIT which was found by Karl Skomski
246        with a custom LLVM fuzzer.
247    
248    41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
249        work under Windows. (Why has it taken so long for anyone to notice?)
250    
251    42. In a character class such as [\W\p{Any}] where both a negative-type escape
252        ("not a word character") and a property escape were present, the property
253        escape was being ignored.
254    
255    43. Fix crash caused by very long (*MARK) or (*THEN) names.
256    
257    44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
258        by a single ASCII character in a class item, was incorrectly compiled in
259        UCP mode. The POSIX class got lost, but only if the single character
260        followed it.
261    
262    45. [:punct:] in UCP mode was matching some characters in the range 128-255
263        that should not have been matched.
264    
265    46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
266        class, all characters with code points greater than 255 are in the class.
267        When a Unicode property was also in the class (if PCRE_UCP is set, escapes
268        such as \w are turned into Unicode properties), wide characters were not
269        correctly handled, and could fail to match.
270    
271    
272    Version 8.37 28-April-2015
273    --------------------------
274    
275    1.  When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
276        for those parentheses to be closed with whatever has been captured so far.
277        However, it was failing to mark any other groups between the hightest
278        capture so far and the currrent group as "unset". Thus, the ovector for
279        those groups contained whatever was previously there. An example is the
280        pattern /(x)|((*ACCEPT))/ when matched against "abcd".
281    
282    2.  If an assertion condition was quantified with a minimum of zero (an odd
283        thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
284    
285    3.  If a pattern in pcretest input had the P (POSIX) modifier followed by an
286        unrecognized modifier, a crash could occur.
287    
288    4.  An attempt to do global matching in pcretest with a zero-length ovector
289        caused a crash.
290    
291    5.  Fixed a memory leak during matching that could occur for a subpattern
292        subroutine call (recursive or otherwise) if the number of captured groups
293        that had to be saved was greater than ten.
294    
295    6.  Catch a bad opcode during auto-possessification after compiling a bad UTF
296        string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
297        UTF with NO_UTF_CHECK is documented as having an undefined outcome.
298    
299    7.  A UTF pattern containing a "not" match of a non-ASCII character and a
300        subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
301    
302    8. When a pattern is compiled, it remembers the highest back reference so that
303       when matching, if the ovector is too small, extra memory can be obtained to
304       use instead. A conditional subpattern whose condition is a check on a
305       capture having happened, such as, for example in the pattern
306       /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
307       setting the highest backreference number. This mattered only if pcre_exec()
308       was called with an ovector that was too small to hold the capture, and there
309       was no other kind of back reference (a situation which is probably quite
310       rare). The effect of the bug was that the condition was always treated as
311       FALSE when the capture could not be consulted, leading to a incorrect
312       behaviour by pcre_exec(). This bug has been fixed.
313    
314    9. A reference to a duplicated named group (either a back reference or a test
315       for being set in a conditional) that occurred in a part of the pattern where
316       PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
317       to be incorrectly calculated, leading to overwriting.
318    
319    10. A mutually recursive set of back references such as (\2)(\1) caused a
320        segfault at study time (while trying to find the minimum matching length).
321        The infinite loop is now broken (with the minimum length unset, that is,
322        zero).
323    
324    11. If an assertion that was used as a condition was quantified with a minimum
325        of zero, matching went wrong. In particular, if the whole group had
326        unlimited repetition and could match an empty string, a segfault was
327        likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
328        assertions to be quantified, but not if they are being used as conditions,
329        so the above pattern is faulted by Perl. PCRE has now been changed so that
330        it also rejects such patterns.
331    
332    12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
333        failed to allow the zero-repeat case if pcre2_exec() was called with an
334        ovector too small to capture the group.
335    
336    13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
337        Red Hat Product Security:
338    
339        (a) A crash if /K and /F were both set with the option to save the compiled
340        pattern.
341    
342        (b) Another crash if the option to print captured substrings in a callout
343        was combined with setting a null ovector, for example \O\C+ as a subject
344        string.
345    
346    14. A pattern such as "((?2){0,1999}())?", which has a group containing a
347        forward reference repeated a large (but limited) number of times within a
348        repeated outer group that has a zero minimum quantifier, caused incorrect
349        code to be compiled, leading to the error "internal error:
350        previously-checked referenced subpattern not found" when an incorrect
351        memory address was read. This bug was reported as "heap overflow",
352        discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
353        CVE-2015-2325.
354    
355    23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
356        call within a group that also contained a recursive back reference caused
357        incorrect code to be compiled. This bug was reported as "heap overflow",
358        discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
359        number CVE-2015-2326.
360    
361    24. Computing the size of the JIT read-only data in advance has been a source
362        of various issues, and new ones are still appear unfortunately. To fix
363        existing and future issues, size computation is eliminated from the code,
364        and replaced by on-demand memory allocation.
365    
366    25. A pattern such as /(?i)[A-`]/, where characters in the other case are
367        adjacent to the end of the range, and the range contained characters with
368        more than one other case, caused incorrect behaviour when compiled in UTF
369        mode. In that example, the range a-j was left out of the class.
370    
371    26. Fix JIT compilation of conditional blocks, which assertion
372        is converted to (*FAIL). E.g: /(?(?!))/.
373    
374    27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
375        discovered by the LLVM fuzzer.
376    
377    28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
378        when this assertion was used as a condition, for example (?(?!)a|b). In
379        pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
380        error about an unsupported item.
381    
382    29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
383        possessification code could take exponential time to complete. A recursion
384        depth limit of 1000 has been imposed to limit the resources used by this
385        optimization.
386    
387    30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
388        such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
389        because \S ensures they are all in the class. The code for doing this was
390        interacting badly with the code for computing the amount of space needed to
391        compile the pattern, leading to a buffer overflow. This bug was discovered
392        by the LLVM fuzzer.
393    
394    31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
395        other kinds of group caused stack overflow at compile time. This bug was
396        discovered by the LLVM fuzzer.
397    
398    32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
399        between a subroutine call and its quantifier was incorrectly compiled,
400        leading to buffer overflow or other errors. This bug was discovered by the
401        LLVM fuzzer.
402    
403    33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
404        assertion after (?(. The code was failing to check the character after
405        (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
406        was discovered by the LLVM fuzzer.
407    
408    34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
409        a fixed maximum following a group that contains a subroutine reference was
410        incorrectly compiled and could trigger buffer overflow. This bug was
411        discovered by the LLVM fuzzer.
412    
413    35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
414        caused a stack overflow instead of the diagnosis of a non-fixed length
415        lookbehind assertion. This bug was discovered by the LLVM fuzzer.
416    
417    36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
418        (e.g. /(?<=\Ka)/) could make pcregrep loop.
419    
420    37. There was a similar problem to 36 in pcretest for global matches.
421    
422    38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
423        and a subsequent item in the pattern caused a non-match, backtracking over
424        the repeated \X did not stop, but carried on past the start of the subject,
425        causing reference to random memory and/or a segfault. There were also some
426        other cases where backtracking after \C could crash. This set of bugs was
427        discovered by the LLVM fuzzer.
428    
429    39. The function for finding the minimum length of a matching string could take
430        a very long time if mutual recursion was present many times in a pattern,
431        for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
432        method has been implemented. This infelicity was discovered by the LLVM
433        fuzzer.
434    
435    40. Static linking against the PCRE library using the pkg-config module was
436        failing on missing pthread symbols.
437    
438    
439    Version 8.36 26-September-2014
440    ------------------------------
441    
442    1.  Got rid of some compiler warnings in the C++ modules that were shown up by
443        -Wmissing-field-initializers and -Wunused-parameter.
444    
445    2.  The tests for quantifiers being too big (greater than 65535) were being
446        applied after reading the number, and stupidly assuming that integer
447        overflow would give a negative number. The tests are now applied as the
448        numbers are read.
449    
450    3.  Tidy code in pcre_exec.c where two branches that used to be different are
451        now the same.
452    
453    4.  The JIT compiler did not generate match limit checks for certain
454        bracketed expressions with quantifiers. This may lead to exponential
455        backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
456        issue should be resolved now.
457    
458    5.  Fixed an issue, which occures when nested alternatives are optimized
459        with table jumps.
460    
461    6.  Inserted two casts and changed some ints to size_t in the light of some
462        reported 64-bit compiler warnings (Bugzilla 1477).
463    
464    7.  Fixed a bug concerned with zero-minimum possessive groups that could match
465        an empty string, which sometimes were behaving incorrectly in the
466        interpreter (though correctly in the JIT matcher). This pcretest input is
467        an example:
468    
469          '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
470          NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
471    
472        the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
473        JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
474        for an empty string was breaking the inner loop and carrying on at a lower
475        level, when possessive repeated groups should always return to a higher
476        level as they have no backtrack points in them. The empty string test now
477        occurs at the outer level.
478    
479    8.  Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
480        ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
481    
482    9.  Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
483        doesn't).
484    
485    10. Change 8.34/15 introduced a bug that caused the amount of memory needed
486        to hold a pattern to be incorrectly computed (too small) when there were
487        named back references to duplicated names. This could cause "internal
488        error: code overflow" or "double free or corruption" or other memory
489        handling errors.
490    
491    11. When named subpatterns had the same prefixes, back references could be
492        confused. For example, in this pattern:
493    
494          /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
495    
496        the reference to 'Name' was incorrectly treated as a reference to a
497        duplicate name.
498    
499    12. A pattern such as /^s?c/mi8 where the optional character has more than
500        one "other case" was incorrectly compiled such that it would only try to
501        match starting at "c".
502    
503    13. When a pattern starting with \s was studied, VT was not included in the
504        list of possible starting characters; this should have been part of the
505        8.34/18 patch.
506    
507    14. If a character class started [\Qx]... where x is any character, the class
508        was incorrectly terminated at the ].
509    
510    15. If a pattern that started with a caseless match for a character with more
511        than one "other case" was studied, PCRE did not set up the starting code
512        unit bit map for the list of possible characters. Now it does. This is an
513        optimization improvement, not a bug fix.
514    
515    16. The Unicode data tables have been updated to Unicode 7.0.0.
516    
517    17. Fixed a number of memory leaks in pcregrep.
518    
519    18. Avoid a compiler warning (from some compilers) for a function call with
520        a cast that removes "const" from an lvalue by using an intermediate
521        variable (to which the compiler does not object).
522    
523    19. Incorrect code was compiled if a group that contained an internal recursive
524        back reference was optional (had quantifier with a minimum of zero). This
525        example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
526        caused segmentation faults because of stack overflows at compile time.
527    
528    20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
529        group that is quantified with an indefinite repeat, caused a compile-time
530        loop which used up all the system stack and provoked a segmentation fault.
531        This was not the same bug as 19 above.
532    
533    21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
534        Patch by Mike Frysinger.
535    
536    
537    Version 8.35 04-April-2014
538    --------------------------
539    
540    1.  A new flag is set, when property checks are present in an XCLASS.
541        When this flag is not set, PCRE can perform certain optimizations
542        such as studying these XCLASS-es.
543    
544    2.  The auto-possessification of character sets were improved: a normal
545        and an extended character set can be compared now. Furthermore
546        the JIT compiler optimizes more character set checks.
547    
548    3.  Got rid of some compiler warnings for potentially uninitialized variables
549        that show up only when compiled with -O2.
550    
551    4.  A pattern such as (?=ab\K) that uses \K in an assertion can set the start
552        of a match later then the end of the match. The pcretest program was not
553        handling the case sensibly - it was outputting from the start to the next
554        binary zero. It now reports this situation in a message, and outputs the
555        text from the end to the start.
556    
557    5.  Fast forward search is improved in JIT. Instead of the first three
558        characters, any three characters with fixed position can be searched.
559        Search order: first, last, middle.
560    
561    6.  Improve character range checks in JIT. Characters are read by an inprecise
562        function now, which returns with an unknown value if the character code is
563        above a certain threshold (e.g: 256). The only limitation is that the value
564        must be bigger than the threshold as well. This function is useful when
565        the characters above the threshold are handled in the same way.
566    
567    7.  The macros whose names start with RAWUCHAR are placeholders for a future
568        mode in which only the bottom 21 bits of 32-bit data items are used. To
569        make this more memorable for those maintaining the code, the names have
570        been changed to start with UCHAR21, and an extensive comment has been added
571        to their definition.
572    
573    8.  Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
574        to the export list in Makefile.am (they were accidentally omitted from the
575        8.34 tarball).
576    
577    9.  The informational output from pcretest used the phrase "starting byte set"
578        which is inappropriate for the 16-bit and 32-bit libraries. As the output
579        for "first char" and "need char" really means "non-UTF-char", I've changed
580        "byte" to "char", and slightly reworded the output. The documentation about
581        these values has also been (I hope) clarified.
582    
583    10. Another JIT related optimization: use table jumps for selecting the correct
584        backtracking path, when more than four alternatives are present inside a
585        bracket.
586    
587    11. Empty match is not possible, when the minimum length is greater than zero,
588        and there is no \K in the pattern. JIT should avoid empty match checks in
589        such cases.
590    
591    12. In a caseless character class with UCP support, when a character with more
592        than one alternative case was not the first character of a range, not all
593        the alternative cases were added to the class. For example, s and \x{17f}
594        are both alternative cases for S: the class [RST] was handled correctly,
595        but [R-T] was not.
596    
597    13. The configure.ac file always checked for pthread support when JIT was
598        enabled. This is not used in Windows, so I have put this test inside a
599        check for the presence of windows.h (which was already tested for).
600    
601    14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
602        The algorithm provides a way to skip certain starting offsets, and usually
603        faster than linear prefix searches.
604    
605    15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
606        as for 'fr_FR' and 'french'. For some reason, however, it then used the
607        Windows-specific input and output files, which have 'french' screwed in.
608        So this could never have worked. One of the problems with locales is that
609        they aren't always the same. I have now updated RunTest so that it checks
610        the output of the locale test (test 3) against three different output
611        files, and it allows the test to pass if any one of them matches. With luck
612        this should make the test pass on some versions of Solaris where it was
613        failing. Because of the uncertainty, the script did not used to stop if
614        test 3 failed; it now does. If further versions of a French locale ever
615        come to light, they can now easily be added.
616    
617    16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
618        there was a message during ./configure, but it did not stop. This now
619        provokes an error. The invalid example in README has been corrected.
620        If a value less than the minimum is given, the minimum value has always
621        been used, but now a warning is given.
622    
623    17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
624        was a bug in the test system, which is now fixed. Also, the list of various
625        configurations that are tested for each release did not have one with both
626        16/32 bits and --enable-bar-anycrlf. It now does.
627    
628    18. pcretest was missing "-C bsr" for displaying the \R default setting.
629    
630    19. Little endian PowerPC systems are supported now by the JIT compiler.
631    
632    20. The fast forward newline mechanism could enter to an infinite loop on
633        certain invalid UTF-8 input. Although we don't support these cases
634        this issue can be fixed by a performance optimization.
635    
636    21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
637        not take account if existing stack usage. There is now a new global
638        variable called pcre_stack_guard that can be set to point to an external
639        function to check stack availability. It is called at the start of
640        processing every parenthesized group.
641    
642    22. A typo in the code meant that in ungreedy mode the max/min qualifier
643        behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
644        not match "ab".
645    
646    23. When UTF was disabled, the JIT program reported some incorrect compile
647        errors. These messages are silenced now.
648    
649    24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
650        compiler.
651    
652    25. Change all the temporary files used in RunGrepTest to be different to those
653        used by RunTest so that the tests can be run simultaneously, for example by
654        "make -j check".
655    
656    
657    Version 8.34 15-December-2013
658    -----------------------------
659    
660    1.  Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
661        executable memory. Patch inspired by Carsten Klein.
662    
663    2.  ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
664        this macro is never tested and has no effect, because the work to support
665        coverage involves only compiling and linking options and special targets in
666        the Makefile. The comment in config.h implied that defining the macro would
667        enable coverage support, which is totally false. There was also support for
668        setting this macro in the CMake files (my fault, I just copied it from
669        configure). SUPPORT_GCOV has now been removed.
670    
671    3.  Make a small performance improvement in strlen16() and strlen32() in
672        pcretest.
673    
674    4.  Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
675        detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
676        these cases). There was also one in pcretest.c.
677    
678    5.  Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
679    
680    6.  In UTF mode, the code for checking whether a group could match an empty
681        string (which is used for indefinitely repeated groups to allow for
682        breaking an infinite loop) was broken when the group contained a repeated
683        negated single-character class with a character that occupied more than one
684        data item and had a minimum repetition of zero (for example, [^\x{100}]* in
685        UTF-8 mode). The effect was undefined: the group might or might not be
686        deemed as matching an empty string, or the program might have crashed.
687    
688    7.  The code for checking whether a group could match an empty string was not
689        recognizing that \h, \H, \v, \V, and \R must match a character.
690    
691    8.  Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
692        an empty string. If it can, pcretest shows this in its information output.
693    
694    9.  Fixed two related bugs that applied to Unicode extended grapheme clusters
695        that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
696        matched by pcre_exec() without using JIT:
697    
698        (a) If the rest of the pattern did not match after a maximal run of
699            grapheme clusters, the code for backing up to try with fewer of them
700            did not always back up over a full grapheme when characters that do not
701            have the modifier quality were involved, e.g. Hangul syllables.
702    
703        (b) If the match point in a subject started with modifier character, and
704            there was no match, the code could incorrectly back up beyond the match
705            point, and potentially beyond the first character in the subject,
706            leading to a segfault or an incorrect match result.
707    
708    10. A conditional group with an assertion condition could lead to PCRE
709        recording an incorrect first data item for a match if no other first data
710        item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
711        first data item, and therefore matched "ca" after "c" instead of at the
712        start.
713    
714    11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
715        bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
716    
717    12. The source of pcregrep now includes z/OS-specific code so that it can be
718        compiled for z/OS as part of the special z/OS distribution.
719    
720    13. Added the -T and -TM options to pcretest.
721    
722    14. The code in pcre_compile.c for creating the table of named capturing groups
723        has been refactored. Instead of creating the table dynamically during the
724        actual compiling pass, the information is remembered during the pre-compile
725        pass (on the stack unless there are more than 20 named groups, in which
726        case malloc() is used) and the whole table is created before the actual
727        compile happens. This has simplified the code (it is now nearly 150 lines
728        shorter) and prepared the way for better handling of references to groups
729        with duplicate names.
730    
731    15. A back reference to a named subpattern when there is more than one of the
732        same name now checks them in the order in which they appear in the pattern.
733        The first one that is set is used for the reference. Previously only the
734        first one was inspected. This change makes PCRE more compatible with Perl.
735    
736    16. Unicode character properties were updated from Unicode 6.3.0.
737    
738    17. The compile-time code for auto-possessification has been refactored, based
739        on a patch by Zoltan Herczeg. It now happens after instead of during
740        compilation. The code is cleaner, and more cases are handled. The option
741        PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
742        options in pcretest are provided to set it. It can also be set by
743        (*NO_AUTO_POSSESS) at the start of a pattern.
744    
745    18. The character VT has been added to the default ("C" locale) set of
746        characters that match \s and are generally treated as white space,
747        following this same change in Perl 5.18. There is now no difference between
748        "Perl space" and "POSIX space". Whether VT is treated as white space in
749        other locales depends on the locale.
750    
751    19. The code for checking named groups as conditions, either for being set or
752        for being recursed, has been refactored (this is related to 14 and 15
753        above). Processing unduplicated named groups should now be as fast at
754        numerical groups, and processing duplicated groups should be faster than
755        before.
756    
757    20. Two patches to the CMake build system, by Alexander Barkov:
758    
759          (1) Replace the "source" command by "." in CMakeLists.txt because
760              "source" is a bash-ism.
761    
762          (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
763              without these the CMake build does not work on Solaris.
764    
765    21. Perl has changed its handling of \8 and \9. If there is no previously
766        encountered capturing group of those numbers, they are treated as the
767        literal characters 8 and 9 instead of a binary zero followed by the
768        literals. PCRE now does the same.
769    
770    22. Following Perl, added \o{} to specify codepoints in octal, making it
771        possible to specify values greater than 0777 and also making them
772        unambiguous.
773    
774    23. Perl now gives an error for missing closing braces after \x{... instead of
775        treating the string as literal. PCRE now does the same.
776    
777    24. RunTest used to grumble if an inappropriate test was selected explicitly,
778        but just skip it when running all tests. This make it awkward to run ranges
779        of tests when one of them was inappropriate. Now it just skips any
780        inappropriate tests, as it always did when running all tests.
781    
782    25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
783        character types such as \d or \w, too many callouts were inserted, and the
784        data that they returned was rubbish.
785    
786    26. In UCP mode, \s was not matching two of the characters that Perl matches,
787        namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
788        were matched by \h. The code has now been refactored so that the lists of
789        the horizontal and vertical whitespace characters used for \h and \v (which
790        are defined only in one place) are now also used for \s.
791    
792    27. Add JIT support for the 64 bit TileGX architecture.
793        Patch by Jiong Wang (Tilera Corporation).
794    
795    28. Possessive quantifiers for classes (both explicit and automatically
796        generated) now use special opcodes instead of wrapping in ONCE brackets.
797    
798    29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
799        (because it's meaningless), this was not happening when PCRE_CASELESS was
800        set. Not wrong, but inefficient.
801    
802    30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
803        properties for \w, \d, etc) is present in a test regex. Otherwise if the
804        test contains no characters greater than 255, Perl doesn't realise it
805        should be using Unicode semantics.
806    
807    31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
808        [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
809        does in Unicode mode.
810    
811    32. Added the "forbid" facility to pcretest so that putting tests into the
812        wrong test files can sometimes be quickly detected.
813    
814    33. There is now a limit (default 250) on the depth of nesting of parentheses.
815        This limit is imposed to control the amount of system stack used at compile
816        time. It can be changed at build time by --with-parens-nest-limit=xxx or
817        the equivalent in CMake.
818    
819    34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
820        errors. Perl warns for these when in warning mode, but PCRE has no facility
821        for giving warnings.
822    
823    35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
824        However, this was not working for (?!) because it is optimized to (*FAIL),
825        for which PCRE does not allow quantifiers. The optimization is now disabled
826        when a quantifier follows (?!). I can't see any use for this, but it makes
827        things uniform.
828    
829    36. Perl no longer allows group names to start with digits, so I have made this
830        change also in PCRE. It simplifies the code a bit.
831    
832    37. In extended mode, Perl ignores spaces before a + that indicates a
833        possessive quantifier. PCRE allowed a space before the quantifier, but not
834        before the possessive +. It now does.
835    
836    38. The use of \K (reset reported match start) within a repeated possessive
837        group such as (a\Kb)*+ was not working.
838    
839    40. Document that the same character tables must be used at compile time and
840        run time, and that the facility to pass tables to pcre_exec() and
841        pcre_dfa_exec() is for use only with saved/restored patterns.
842    
843    41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
844        features for Builds with MSVC:
845    
846        1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
847           builds). The .rc files can be used to set FileDescription and many other
848           attributes.
849    
850        2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
851           This allows higher-level build scripts which want .pdb files to avoid
852           hard-coding the exact files needed."
853    
854    42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
855        mean "start of word" and "end of word", respectively, as a transition aid.
856    
857    43. A minimizing repeat of a class containing codepoints greater than 255 in
858        non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
859        compiled to use the heap for recursion.
860    
861    44. Got rid of some compiler warnings for unused variables when UTF but not UCP
862        is configured.
863    
864    
865    Version 8.33 28-May-2013
866    ------------------------
867    
868    1.  Added 'U' to some constants that are compared to unsigned integers, to
869        avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
870        variables that are added to signed variables, to ensure the result is
871        signed and can be negated.
872    
873    2.  Applied patch by Daniel Richard G for quashing MSVC warnings to the
874        CMake config files.
875    
876    3.  Revise the creation of config.h.generic so that all boolean macros are
877        #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
878        overriding via -D on the command line possible.
879    
880    4.  Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
881        to unsigned int is reported to make a quite noticeable speed difference in
882        a specific Windows environment. Testing on Linux did also appear to show
883        some benefit (and it is clearly not harmful). Also fixed the definition of
884        Xop which should be unsigned.
885    
886    5.  Related to (4), changing the definition of the intermediate variable cc
887        in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
888        improvements.
889    
890    6.  Fix forward search in JIT when link size is 3 or greater. Also removed some
891        unnecessary spaces.
892    
893    7.  Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
894        and later.
895    
896    8.  Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
897    
898    9.  Optimizing fast_forward_start_bits in JIT.
899    
900    10. Adding support for callouts in JIT, and fixing some issues revealed
901        during this work. Namely:
902    
903        (a) Unoptimized capturing brackets incorrectly reset on backtrack.
904    
905        (b) Minimum length was not checked before the matching is started.
906    
907    11. The value of capture_last that is passed to callouts was incorrect in some
908        cases when there was a capture on one path that was subsequently abandoned
909        after a backtrack. Also, the capture_last value is now reset after a
910        recursion, since all captures are also reset in this case.
911    
912    12. The interpreter no longer returns the "too many substrings" error in the
913        case when an overflowing capture is in a branch that is subsequently
914        abandoned after a backtrack.
915    
916    13. In the pathological case when an offset vector of size 2 is used, pcretest
917        now prints out the matched string after a yield of 0 or 1.
918    
919    14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
920        Only supported by the JIT compiler at the moment.
921    
922    15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
923    
924    16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
925        offset of the starting point of the matching process, provided the offsets
926        vector is large enough.
927    
928    17. The \A escape now records a lookbehind value of 1, though its execution
929        does not actually inspect the previous character. This is to ensure that,
930        in partial multi-segment matching, at least one character from the old
931        segment is retained when a new segment is processed. Otherwise, if there
932        are no lookbehinds in the pattern, \A might match incorrectly at the start
933        of a new segment.
934    
935    18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
936    
937    19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
938        gives some modest performance improvement in 8-bit mode.
939    
940    20. Added the PCRE-specific property \p{Xuc} for matching characters that can
941        be expressed in certain programming languages using Universal Character
942        Names.
943    
944    21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
945        which points out that "non characters" are not "characters that may not
946        appear in Unicode strings" but rather "characters that are reserved for
947        internal use and have only local meaning".
948    
949    22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
950        there was a conditional group that depended on an assertion, if the
951        assertion was false, the callout that immediately followed the alternation
952        in the condition was skipped when pcre_exec() was used for matching.
953    
954    23. Allow an explicit callout to be inserted before an assertion that is the
955        condition for a conditional group, for compatibility with automatic
956        callouts, which always insert a callout at this point.
957    
958    24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
959        confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
960    
961    25. (*PRUNE) is now supported by the JIT compiler.
962    
963    26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
964    
965    27. Fix the case where there are two or more SKIPs with arguments that may be
966        ignored.
967    
968    28. (*SKIP) is now supported by the JIT compiler.
969    
970    29. (*THEN) is now supported by the JIT compiler.
971    
972    30. Update RunTest with additional test selector options.
973    
974    31. The way PCRE handles backtracking verbs has been changed in two ways.
975    
976        (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
977        SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
978        backtracking. In some cases this makes it more Perl-compatible, but Perl's
979        rather obscure rules do not always do the same thing.
980    
981        (2) Previously, backtracking verbs were confined within assertions. This is
982        no longer the case for positive assertions, except for (*ACCEPT). Again,
983        this sometimes improves Perl compatibility, and sometimes does not.
984    
985    32. A number of tests that were in test 2 because Perl did things differently
986        have been moved to test 1, because either Perl or PCRE has changed, and
987        these tests are now compatible.
988    
989    32. Backtracking control verbs are now handled in the same way in JIT and
990        interpreter.
991    
992    33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
993        contained a forward subroutine reference caused a compile error.
994    
995    34. Auto-detect and optimize limited repetitions in JIT.
996    
997    35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
998        blocking (*UTF) etc.
999    
1000    36. In the interpreter, maximizing pattern repetitions for characters and
1001        character types now use tail recursion, which reduces stack usage.
1002    
1003    37. The value of the max lookbehind was not correctly preserved if a compiled
1004        and saved regex was reloaded on a host of different endianness.
1005    
1006    38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
1007        of the compiled pattern block, expand the flags field from 16 to 32 bits
1008        because it was almost full.
1009    
1010    39. Try madvise first before posix_madvise.
1011    
1012    40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
1013        with a pattern such as ^$. It has taken 4 years for anybody to notice! The
1014        original change locked out all matches of empty strings. This has been
1015        changed so that one match of an empty string per line is recognized.
1016        Subsequent searches on the same line (for colouring or for --only-matching,
1017        for example) do not recognize empty strings.
1018    
1019    41. Applied a user patch to fix a number of spelling mistakes in comments.
1020    
1021    42. Data lines longer than 65536 caused pcretest to crash.
1022    
1023    43. Clarified the data type for length and startoffset arguments for pcre_exec
1024        and pcre_dfa_exec in the function-specific man pages, where they were
1025        explicitly stated to be in bytes, never having been updated. I also added
1026        some clarification to the pcreapi man page.
1027    
1028    44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
1029        a segmentation fault.
1030    
1031    
1032    Version 8.32 30-November-2012
1033    -----------------------------
1034    
1035    1.  Improved JIT compiler optimizations for first character search and single
1036        character iterators.
1037    
1038    2.  Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
1039        Patch by Daniel Richard G.
1040    
1041    3.  Single character iterator optimizations in the JIT compiler.
1042    
1043    4.  Improved JIT compiler optimizations for character ranges.
1044    
1045    5.  Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1046        Reported by Giuseppe D'Angelo.
1047    
1048    6.  The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1049        of a line, was being set incorrectly in cases where .* appeared inside
1050        atomic brackets at the start of a pattern, or where there was a subsequent
1051        *PRUNE or *SKIP.
1052    
1053    7.  Improved instruction cache flush for POWER/PowerPC.
1054        Patch by Daniel Richard G.
1055    
1056    8.  Fixed a number of issues in pcregrep, making it more compatible with GNU
1057        grep:
1058    
1059        (a) There is now no limit to the number of patterns to be matched.
1060    
1061        (b) An error is given if a pattern is too long.
1062    
1063        (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1064            are now supported.
1065    
1066        (d) --exclude-from and --include-from (multiple use) have been added.
1067    
1068        (e) Exclusions and inclusions now apply to all files and directories, not
1069            just to those obtained from scanning a directory recursively.
1070    
1071        (f) Multiple uses of -f and --file-list are now supported.
1072    
1073        (g) In a Windows environment, the default for -d has been changed from
1074            "read" (the GNU grep default) to "skip", because otherwise the presence
1075            of a directory in the file list provokes an error.
1076    
1077        (h) The documentation has been revised and clarified in places.
1078    
1079    9.  Improve the matching speed of capturing brackets.
1080    
1081    10. Changed the meaning of \X so that it now matches a Unicode extended
1082        grapheme cluster.
1083    
1084    11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1085        out POSIX threads when JIT support is configured.
1086    
1087    12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1088    
1089    13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1090        structure could go wrong in environments where size_t is not the same size
1091        as int.
1092    
1093    14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1094        set.
1095    
1096    15. The EBCDIC support had decayed; later updates to the code had included
1097        explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1098        general tidy up of EBCDIC-related issues, and the documentation was also
1099        not quite right. There is now a test that can be run on ASCII systems to
1100        check some of the EBCDIC-related things (but is it not a full test).
1101    
1102    16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1103        in a small tidy to the code.
1104    
1105    17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1106    
1107    18. If the --only-matching (-o) option in pcregrep is specified multiple
1108        times, each one causes appropriate output. For example, -o1 -o2 outputs the
1109        substrings matched by the 1st and 2nd capturing parentheses. A separating
1110        string can be specified by --om-separator (default empty).
1111    
1112    19. Improving the first n character searches.
1113    
1114    20. Turn case lists for horizontal and vertical white space into macros so that
1115        they are defined only once.
1116    
1117    21. This set of changes together give more compatible Unicode case-folding
1118        behaviour for characters that have more than one other case when UCP
1119        support is available.
1120    
1121        (a) The Unicode property table now has offsets into a new table of sets of
1122            three or more characters that are case-equivalent. The MultiStage2.py
1123            script that generates these tables (the pcre_ucd.c file) now scans
1124            CaseFolding.txt instead of UnicodeData.txt for character case
1125            information.
1126    
1127        (b) The code for adding characters or ranges of characters to a character
1128            class has been abstracted into a generalized function that also handles
1129            case-independence. In UTF-mode with UCP support, this uses the new data
1130            to handle characters with more than one other case.
1131    
1132        (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1133            whose other case is greater than 256 are now correctly matched
1134            caselessly. Previously, the high codepoint matched the low one, but not
1135            vice versa.
1136    
1137        (d) The processing of \h, \H, \v, and \ in character classes now makes use
1138            of the new class addition function, using character lists defined as
1139            macros alongside the case definitions of 20 above.
1140    
1141        (e) Caseless back references now work with characters that have more than
1142            one other case.
1143    
1144        (f) General caseless matching of characters with more than one other case
1145            is supported.
1146    
1147    22. Unicode character properties were updated from Unicode 6.2.0
1148    
1149    23. Improved CMake support under Windows. Patch by Daniel Richard G.
1150    
1151    24. Add support for 32-bit character strings, and UTF-32
1152    
1153    25. Major JIT compiler update (code refactoring and bugfixing).
1154        Experimental Sparc 32 support is added.
1155    
1156    26. Applied a modified version of Daniel Richard G's patch to create
1157        pcre.h.generic and config.h.generic by "make" instead of in the
1158        PrepareRelease script.
1159    
1160    27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1161        pcre_compile.c when checking for a zero character.
1162    
1163    28. Introducing a native interface for JIT. Through this interface, the compiled
1164        machine code can be directly executed. The purpose of this interface is to
1165        provide fast pattern matching, so several sanity checks are not performed.
1166        However, feature tests are still performed. The new interface provides
1167        1.4x speedup compared to the old one.
1168    
1169    29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1170        the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1171        was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1172    
1173    30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1174        greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1175        the "old" RFC 2279). Instead, it ended up passing a negative length to
1176        pcre_exec().
1177    
1178    31. Add support for GCC's visibility feature to hide internal functions.
1179    
1180    32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1181        "unknown -C option" after outputting 0 or 1.
1182    
1183    33. There is now support for generating a code coverage report for the test
1184        suite in environments where gcc is the compiler and lcov is installed. This
1185        is mainly for the benefit of the developers.
1186    
1187    34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1188        unaddressable using valgrind annotations, allowing valgrind to detect
1189        invalid memory accesses. This is mainly for the benefit of the developers.
1190    
1191    25. (*UTF) can now be used to start a pattern in any of the three libraries.
1192    
1193    26. Give configure error if --enable-cpp but no C++ compiler found.
1194    
1195    
1196    Version 8.31 06-July-2012
1197    -------------------------
1198    
1199    1.  Fixing a wrong JIT test case and some compiler warnings.
1200    
1201    2.  Removed a bashism from the RunTest script.
1202    
1203    3.  Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1204        to unsigned type, result still unsigned" that was given by an MS compiler
1205        on encountering the code "-sizeof(xxx)".
1206    
1207    4.  Partial matching support is added to the JIT compiler.
1208    
1209    5.  Fixed several bugs concerned with partial matching of items that consist
1210        of more than one character:
1211    
1212        (a) /^(..)\1/ did not partially match "aba" because checking references was
1213            done on an "all or nothing" basis. This also applied to repeated
1214            references.
1215    
1216        (b) \R did not give a hard partial match if \r was found at the end of the
1217            subject.
1218    
1219        (c) \X did not give a hard partial match after matching one or more
1220            characters at the end of the subject.
1221    
1222        (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1223            a partial match for the string "\r".
1224    
1225        (e) When newline was set to CRLF, the metacharacter "." did not recognize
1226            a partial match for a CR character at the end of the subject string.
1227    
1228    6.  If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1229        running pcretest, the text "(JIT)" added to the output whenever JIT is
1230        actually used to run the match.
1231    
1232    7.  Individual JIT compile options can be set in pcretest by following -s+[+]
1233        or /S+[+] with a digit between 1 and 7.
1234    
1235    8.  OP_NOT now supports any UTF character not just single-byte ones.
1236    
1237    9.  (*MARK) control verb is now supported by the JIT compiler.
1238    
1239    10. The command "./RunTest list" lists the available tests without actually
1240        running any of them. (Because I keep forgetting what they all are.)
1241    
1242    11. Add PCRE_INFO_MAXLOOKBEHIND.
1243    
1244    12. Applied a (slightly modified) user-supplied patch that improves performance
1245        when the heap is used for recursion (compiled with --disable-stack-for-
1246        recursion). Instead of malloc and free for each heap frame each time a
1247        logical recursion happens, frames are retained on a chain and re-used where
1248        possible. This sometimes gives as much as 30% improvement.
1249    
1250    13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1251        call.
1252    
1253    14. As documented, (*COMMIT) is now confined to within a positive assertion.
1254    
1255    15. It is now possible to link pcretest with libedit as an alternative to
1256        libreadline.
1257    
1258    16. (*COMMIT) control verb is now supported by the JIT compiler.
1259    
1260    17. The Unicode data tables have been updated to Unicode 6.1.0.
1261    
1262    18. Added --file-list option to pcregrep.
1263    
1264    19. Added binary file support to pcregrep, including the -a, --binary-files,
1265        -I, and --text options.
1266    
1267    20. The madvise function is renamed for posix_madvise for QNX compatibility
1268        reasons. Fixed by Giuseppe D'Angelo.
1269    
1270    21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1271    
1272    22. Changed the option for creating symbolic links for 16-bit man pages from
1273        -s to -sf so that re-installing does not cause issues.
1274    
1275    23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1276    
1277    24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1278        matches in certain environments (the workspace was not being correctly
1279        retained). Also added to pcre_dfa_exec() a simple plausibility check on
1280        some of the workspace data at the beginning of a restart.
1281    
1282    25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1283        was not doing so when it should - probably a typo introduced by SVN 528
1284        (change 8.10/14).
1285    
1286    26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1287        \w+ when the character tables indicated that \x{c4} was a word character.
1288        There were several related cases, all because the tests for doing a table
1289        lookup were testing for characters less than 127 instead of 255.
1290    
1291    27. If a pattern contains capturing parentheses that are not used in a match,
1292        their slots in the ovector are set to -1. For those that are higher than
1293        any matched groups, this happens at the end of processing. In the case when
1294        there were back references that the ovector was too small to contain
1295        (causing temporary malloc'd memory to be used during matching), and the
1296        highest capturing number was not used, memory off the end of the ovector
1297        was incorrectly being set to -1. (It was using the size of the temporary
1298        memory instead of the true size.)
1299    
1300    28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1301        ovector size, it uses memory at the end of the block that it has got.
1302    
1303    29. Check for an overlong MARK name and give an error at compile time. The
1304        limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1305    
1306    30. JIT compiler update.
1307    
1308    31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1309        Rill for the patch.
1310    
1311    32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1312    
1313    33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1314    
1315    34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1316        SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1317        was enabled.
1318    
1319    35. Improve JIT code generation for greedy plus quantifier.
1320    
1321    36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1322        1 to "aa" instead of to an empty string. The bug affected repeated groups
1323        that could potentially match an empty string.
1324    
1325    37. Optimizing single character iterators in JIT.
1326    
1327    38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1328        the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1329        codepoints that are too big for the mode are faulted, and in a UTF mode,
1330        disallowed codepoints are also faulted.
1331    
1332    39. If PCRE was compiled with UTF support, in three places in the DFA
1333        matcher there was code that should only have been obeyed in UTF mode, but
1334        was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1335        processing when bytes with values greater than 127 were present. In 16-bit
1336        mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1337        both cases the values are those that cannot be the first data item in a UTF
1338        character. The three items that might have provoked this were recursions,
1339        possessively repeated groups, and atomic groups.
1340    
1341    40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1342        and pcregrep, because some OS require shared objects to be explicitly
1343        passed to ld, causing the link step to fail if they are not.
1344    
1345    41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1346        mode, patterns that started with \h* or \R* might be incorrectly matched.
1347    
1348    
1349    Version 8.30 04-February-2012
1350    -----------------------------
1351    
1352    1.  Renamed "isnumber" as "is_a_number" because in some Mac environments this
1353        name is defined in ctype.h.
1354    
1355    2.  Fixed a bug in fixed-length calculation for lookbehinds that would show up
1356        only in quite long subpatterns.
1357    
1358    3.  Removed the function pcre_info(), which has been obsolete and deprecated
1359        since it was replaced by pcre_fullinfo() in February 2000.
1360    
1361    4.  For a non-anchored pattern, if (*SKIP) was given with a name that did not
1362        match a (*MARK), and the match failed at the start of the subject, a
1363        reference to memory before the start of the subject could occur. This bug
1364        was introduced by fix 17 of release 8.21.
1365    
1366    5.  A reference to an unset group with zero minimum repetition was giving
1367        totally wrong answers (in non-JavaScript-compatibility mode). For example,
1368        /(another)?(\1?)test/ matched against "hello world test". This bug was
1369        introduced in release 8.13.
1370    
1371    6.  Add support for 16-bit character strings (a large amount of work involving
1372        many changes and refactorings).
1373    
1374    7.  RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1375        command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1376        from a file.
1377    
1378    8.  Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1379        rounding is not applied in this particular case).
1380    
1381    9.  The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1382        if they appear, or are escaped, in patterns.
1383    
1384    10. Get rid of a number of -Wunused-but-set-variable warnings.
1385    
1386    11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1387        "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1388        Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1389        also returns the mark "x". This bug applied to capturing parentheses,
1390        non-capturing parentheses, and atomic parentheses. It also applied to some
1391        assertions.
1392    
1393    12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1394        information out of configure.ac instead of relying on pcre.h.generic, which
1395        is not stored in the repository.
1396    
1397    13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1398        -lreadline.
1399    
1400    14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1401    
1402    15. Applied Graycode's patch to put the top-level frame on the stack rather
1403        than the heap when not using the stack for recursion. This gives a
1404        performance improvement in many cases when recursion is not deep.
1405    
1406    16. Experimental code added to "pcretest -C" to output the stack frame size.
1407    
1408    
1409    Version 8.21 12-Dec-2011
1410    ------------------------
1411    
1412    1.  Updating the JIT compiler.
1413    
1414    2.  JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1415        are added as well.
1416    
1417    3.  Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1418        PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1419        calling _pcre_jit_exec. Some extra comments are added.
1420    
1421    4.  (*MARK) settings inside atomic groups that do not contain any capturing
1422        parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1423        was introduced by change 18 for 8.20.
1424    
1425    5.  Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1426        ECMA-262 standard.
1427    
1428    6.  Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1429        erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1430        This bug was probably introduced by change 9 of 8.13.
1431    
1432    7.  While fixing 6 above, I noticed that a number of other items were being
1433        incorrectly rejected as "not fixed length". This arose partly because newer
1434        opcodes had not been added to the fixed-length checking code. I have (a)
1435        corrected the bug and added tests for these items, and (b) arranged for an
1436        error to occur if an unknown opcode is encountered while checking for fixed
1437        length instead of just assuming "not fixed length". The items that were
1438        rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1439        (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1440        repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1441    
1442    8.  A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1443        being incorrectly compiled and would have given unpredicatble results.
1444    
1445    9.  A possessively repeated subpattern with minimum repeat count greater than
1446        one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1447        (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1448        the first (A) could occur when it should not.
1449    
1450    10. Add a cast and remove a redundant test from the code.
1451    
1452    11. JIT should use pcre_malloc/pcre_free for allocation.
1453    
1454    12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1455        best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1456        is anything other than /usr, -L is still shown).
1457    
1458    13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1459    
1460    14. Perl does not support \N without a following name in a [] class; PCRE now
1461        also gives an error.
1462    
1463    15. If a forward reference was repeated with an upper limit of around 2000,
1464        it caused the error "internal error: overran compiling workspace". The
1465        maximum number of forward references (including repeats) was limited by the
1466        internal workspace, and dependent on the LINK_SIZE. The code has been
1467        rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1468        the default depends on LINK_SIZE. There is a new upper limit (for safety)
1469        of around 200,000 forward references. While doing this, I also speeded up
1470        the filling in of repeated forward references.
1471    
1472    16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1473        incorrectly expecting the subject to contain another "a" after the start.
1474    
1475    17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1476        in the match, the SKIP should be ignored. This was not happening; instead
1477        the SKIP was being treated as NOMATCH. For patterns such as
1478        /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1479        tested.
1480    
1481    18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1482        now much more compatible with Perl, in particular in cases where the result
1483        is a non-match for a non-anchored pattern. For example, if
1484        /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1485        "m", where previously it did not return a name. A side effect of this
1486        change is that for partial matches, the last encountered mark name is
1487        returned, as for non matches. A number of tests that were previously not
1488        Perl-compatible have been moved into the Perl-compatible test files. The
1489        refactoring has had the pleasing side effect of removing one argument from
1490        the match() function, thus reducing its stack requirements.
1491    
1492    19. If the /S+ option was used in pcretest to study a pattern using JIT,
1493        subsequent uses of /S (without +) incorrectly behaved like /S+.
1494    
1495    21. Retrieve executable code size support for the JIT compiler and fixing
1496        some warnings.
1497    
1498    22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1499        not work when the shorter character appeared right at the end of the
1500        subject string.
1501    
1502    23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1503        systems.
1504    
1505    24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1506        output it when the /M option is used in pcretest.
1507    
1508    25. The CheckMan script was not being included in the distribution. Also, added
1509        an explicit "perl" to run Perl scripts from the PrepareRelease script
1510        because this is reportedly needed in Windows.
1511    
1512    26. If study data was being save in a file and studying had not found a set of
1513        "starts with" bytes for the pattern, the data written to the file (though
1514        never used) was taken from uninitialized memory and so caused valgrind to
1515        complain.
1516    
1517    27. Updated RunTest.bat as provided by Sheri Pierce.
1518    
1519    28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1520    
1521    29. Computation of memory usage for the table of capturing group names was
1522        giving an unnecessarily large value.
1523    
1524    
1525    Version 8.20 21-Oct-2011
1526    ------------------------
1527    
1528    1.  Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1529        a POSIX class. After further experiments with Perl, which convinced me that
1530        Perl has bugs and confusions, a closing square bracket is no longer allowed
1531        in a POSIX name. This bug also affected patterns with classes that started
1532        with full stops.
1533    
1534    2.  If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1535        captured substring, but while checking the failing first alternative,
1536        substring 1 is temporarily captured. If the output vector supplied to
1537        pcre_exec() was not big enough for this capture, the yield of the function
1538        was still zero ("insufficient space for captured substrings"). This cannot
1539        be totally fixed without adding another stack variable, which seems a lot
1540        of expense for a edge case. However, I have improved the situation in cases
1541        such as /(a)(b)x|abc/ matched against "abc", where the return code
1542        indicates that fewer than the maximum number of slots in the ovector have
1543        been set.
1544    
1545    3.  Related to (2) above: when there are more back references in a pattern than
1546        slots in the output vector, pcre_exec() uses temporary memory during
1547        matching, and copies in the captures as far as possible afterwards. It was
1548        using the entire output vector, but this conflicts with the specification
1549        that only 2/3 is used for passing back captured substrings. Now it uses
1550        only the first 2/3, for compatibility. This is, of course, another edge
1551        case.
1552    
1553    4.  Zoltan Herczeg's just-in-time compiler support has been integrated into the
1554        main code base, and can be used by building with --enable-jit. When this is
1555        done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1556        runtime --no-jit option is given.
1557    
1558    5.  When the number of matches in a pcre_dfa_exec() run exactly filled the
1559        ovector, the return from the function was zero, implying that there were
1560        other matches that did not fit. The correct "exactly full" value is now
1561        returned.
1562    
1563    6.  If a subpattern that was called recursively or as a subroutine contained
1564        (*PRUNE) or any other control that caused it to give a non-standard return,
1565        invalid errors such as "Error -26 (nested recursion at the same subject
1566        position)" or even infinite loops could occur.
1567    
1568    7.  If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1569        computing the minimum length on reaching *ACCEPT, and so ended up with the
1570        wrong value of 1 rather than 0. Further investigation indicates that
1571        computing a minimum subject length in the presence of *ACCEPT is difficult
1572        (think back references, subroutine calls), and so I have changed the code
1573        so that no minimum is registered for a pattern that contains *ACCEPT.
1574    
1575    8.  If (*THEN) was present in the first (true) branch of a conditional group,
1576        it was not handled as intended. [But see 16 below.]
1577    
1578    9.  Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1579        Sheri Pierce.
1580    
1581    10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1582        the first byte in a match must be "a".
1583    
1584    11. Change 17 for 8.13 increased the recursion depth for patterns like
1585        /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1586        pattern contains any instances of (*THEN). If it does not, the old
1587        optimizations are restored. It would be nice to do this on a per-group
1588        basis, but at the moment that is not feasible.
1589    
1590    12. In some environments, the output of pcretest -C is CRLF terminated. This
1591        broke RunTest's code that checks for the link size. A single white space
1592        character after the value is now allowed for.
1593    
1594    13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1595        For "fr", it uses the Windows-specific input and output files.
1596    
1597    14. If (*THEN) appeared in a group that was called recursively or as a
1598        subroutine, it did not work as intended. [But see next item.]
1599    
1600    15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1601        pattern fragments (but not containing any | characters). If A and B are
1602        matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1603        was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1604        D. In other words, Perl considers parentheses that do not contain any |
1605        characters to be part of a surrounding alternative, whereas PCRE was
1606        treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1607        differently. PCRE now behaves in the same way as Perl, except in the case
1608        of subroutine/recursion calls such as (?1) which have in any case always
1609        been different (but PCRE had them first :-).
1610    
1611    16. Related to 15 above: Perl does not treat the | in a conditional group as
1612        creating alternatives. Such a group is treated in the same way as an
1613        ordinary group without any | characters when processing (*THEN). PCRE has
1614        been changed to match Perl's behaviour.
1615    
1616    17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1617        RunGrepTest script failed.
1618    
1619    18. Change 22 for version 13 caused atomic groups to use more stack. This is
1620        inevitable for groups that contain captures, but it can lead to a lot of
1621        stack use in large patterns. The old behaviour has been restored for atomic
1622        groups that do not contain any capturing parentheses.
1623    
1624    19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1625        suppress the check for a minimum subject length at run time. (If it was
1626        given to pcre_exec() or pcre_dfa_exec() it did work.)
1627    
1628    20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1629        fail to work when decoding hex characters in data strings in EBCDIC
1630        environments.
1631    
1632    21. It appears that in at least one Mac OS environment, the isxdigit() function
1633        is implemented as a macro that evaluates to its argument more than once,
1634        contravening the C 90 Standard (I haven't checked a later standard). There
1635        was an instance in pcretest which caused it to go wrong when processing
1636        \x{...} escapes in subject strings. The has been rewritten to avoid using
1637        things like p++ in the argument of isxdigit().
1638    
1639    
1640    Version 8.13 16-Aug-2011
1641    ------------------------
1642    
1643    1.  The Unicode data tables have been updated to Unicode 6.0.0.
1644    
1645    2.  Two minor typos in pcre_internal.h have been fixed.
1646    
1647    3.  Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1648        pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1649        in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1650    
1651    4.  There were a number of related bugs in the code for matching backrefences
1652        caselessly in UTF-8 mode when codes for the characters concerned were
1653        different numbers of bytes. For example, U+023A and U+2C65 are an upper
1654        and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1655        (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1656        code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1657        2-byte code at the end of the subject (it thought there wasn't enough data
1658        left).
1659    
1660    5.  Comprehensive information about what went wrong is now returned by
1661        pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1662        as the output vector has at least 2 elements. The offset of the start of
1663        the failing character and a reason code are placed in the vector.
1664    
1665    6.  When the UTF-8 string check fails for pcre_compile(), the offset that is
1666        now returned is for the first byte of the failing character, instead of the
1667        last byte inspected. This is an incompatible change, but I hope it is small
1668        enough not to be a problem. It makes the returned offset consistent with
1669        pcre_exec() and pcre_dfa_exec().
1670    
1671    7.  pcretest now gives a text phrase as well as the error number when
1672        pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1673        failure, the offset and reason code are output.
1674    
1675    8.  When \R was used with a maximizing quantifier it failed to skip backwards
1676        over a \r\n pair if the subsequent match failed. Instead, it just skipped
1677        back over a single character (\n). This seems wrong (because it treated the
1678        two characters as a single entity when going forwards), conflicts with the
1679        documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1680        behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1681        has been changed.
1682    
1683    9.  Some internal refactoring has changed the processing so that the handling
1684        of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1685        time (the PCRE_DOTALL option was changed this way some time ago: version
1686        7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1687        which was always a bit of a fudge. It also means that there is one less
1688        argument for the match() function, which reduces its stack requirements
1689        slightly. This change also fixes an incompatibility with Perl: the pattern
1690        (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1691    
1692    10. More internal refactoring has drastically reduced the number of recursive
1693        calls to match() for possessively repeated groups such as (abc)++ when
1694        using pcre_exec().
1695    
1696    11. While implementing 10, a number of bugs in the handling of groups were
1697        discovered and fixed:
1698    
1699        (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1700        (a|)*(?1) gave a compile-time internal error.
1701        ((a|)+)+  did not notice that the outer group could match an empty string.
1702        (^a|^)+   was not marked as anchored.
1703        (.*a|.*)+ was not marked as matching at start or after a newline.
1704    
1705    12. Yet more internal refactoring has removed another argument from the match()
1706        function. Special calls to this function are now indicated by setting a
1707        value in a variable in the "match data" data block.
1708    
1709    13. Be more explicit in pcre_study() instead of relying on "default" for
1710        opcodes that mean there is no starting character; this means that when new
1711        ones are added and accidentally left out of pcre_study(), testing should
1712        pick them up.
1713    
1714    14. The -s option of pcretest has been documented for ages as being an old
1715        synonym of -m (show memory usage). I have changed it to mean "force study
1716        for every regex", that is, assume /S for every regex. This is similar to -i
1717        and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1718        using it. It makes it easier to run collections of tests with and without
1719        study enabled, and thereby test pcre_study() more easily. All the standard
1720        tests are now run with and without -s (but some patterns can be marked as
1721        "never study" - see 20 below).
1722    
1723    15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1724        restoration of the capturing data to the outer values was not happening
1725        correctly.
1726    
1727    16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1728        empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1729        pattern had matched an empty string, and so incorrectly returned a no
1730        match.
1731    
1732    17. There was optimizing code for the last branch of non-capturing parentheses,
1733        and also for the obeyed branch of a conditional subexpression, which used
1734        tail recursion to cut down on stack usage. Unfortunately, now that there is
1735        the possibility of (*THEN) occurring in these branches, tail recursion is
1736        no longer possible because the return has to be checked for (*THEN). These
1737        two optimizations have therefore been removed. [But see 8.20/11 above.]
1738    
1739    18. If a pattern containing \R was studied, it was assumed that \R always
1740        matched two bytes, thus causing the minimum subject length to be
1741        incorrectly computed because \R can also match just one byte.
1742    
1743    19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1744        was incorrectly computed.
1745    
1746    20. If /S is present twice on a test pattern in pcretest input, it now
1747        *disables* studying, thereby overriding the use of -s on the command line
1748        (see 14 above). This is necessary for one or two tests to keep the output
1749        identical in both cases.
1750    
1751    21. When (*ACCEPT) was used in an assertion that matched an empty string and
1752        PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1753    
1754    22. When an atomic group that contained a capturing parenthesis was
1755        successfully matched, but the branch in which it appeared failed, the
1756        capturing was not being forgotten if a higher numbered group was later
1757        captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1758        group 1 to "a", when in fact it should be unset. This applied to multi-
1759        branched capturing and non-capturing groups, repeated or not, and also to
1760        positive assertions (capturing in negative assertions does not happen
1761        in PCRE) and also to nested atomic groups.
1762    
1763    23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1764        subject after a captured substring, to make it easier to tell which of a
1765        number of identical substrings has been captured.
1766    
1767    24. The way atomic groups are processed by pcre_exec() has been changed so that
1768        if they are repeated, backtracking one repetition now resets captured
1769        values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1770        "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1771        "aaa". Previously, it would have been "a". As part of this code
1772        refactoring, the way recursive calls are handled has also been changed.
1773    
1774    25. If an assertion condition captured any substrings, they were not passed
1775        back unless some other capturing happened later. For example, if
1776        (?(?=(a))a) was matched against "a", no capturing was returned.
1777    
1778    26. When studying a pattern that contained subroutine calls or assertions,
1779        the code for finding the minimum length of a possible match was handling
1780        direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1781        group 1 called group 2 while simultaneously a separate group 2 called group
1782        1). A stack overflow occurred in this case. I have fixed this by limiting
1783        the recursion depth to 10.
1784    
1785    27. Updated RunTest.bat in the distribution to the version supplied by Tom
1786        Fortmann. This supports explicit test numbers on the command line, and has
1787        argument validation and error reporting.
1788    
1789    28. An instance of \X with an unlimited repeat could fail if at any point the
1790        first character it looked at was a mark character.
1791    
1792    29. Some minor code refactoring concerning Unicode properties and scripts
1793        should reduce the stack requirement of match() slightly.
1794    
1795    30. Added the '=' option to pcretest to check the setting of unused capturing
1796        slots at the end of the pattern, which are documented as being -1, but are
1797        not included in the return count.
1798    
1799    31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1800        compiled something random. Now it gives a compile-time error (as does
1801        Perl).
1802    
1803    32. A *MARK encountered during the processing of a positive assertion is now
1804        recorded and passed back (compatible with Perl).
1805    
1806    33. If --only-matching or --colour was set on a pcregrep call whose pattern
1807        had alternative anchored branches, the search for a second match in a line
1808        was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1809        matched the line "0102" twice. The same bug affected patterns that started
1810        with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1811        twice.
1812    
1813    34. Previously, PCRE did not allow quantification of assertions. However, Perl
1814        does, and because of capturing effects, quantifying parenthesized
1815        assertions may at times be useful. Quantifiers are now allowed for
1816        parenthesized assertions.
1817    
1818    35. A minor code tidy in pcre_compile() when checking options for \R usage.
1819    
1820    36. \g was being checked for fancy things in a character class, when it should
1821        just be a literal "g".
1822    
1823    37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1824        appearance of a nested POSIX class supersedes an apparent external class.
1825        For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1826        unescaped square brackets may also appear as part of class names. For
1827        example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1828        more like Perl. (But see 8.20/1 above.)
1829    
1830    38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1831        was because it thought it was \N{name}, which is not supported).
1832    
1833    39. Add minix to OS list not supporting the -S option in pcretest.
1834    
1835    40. PCRE tries to detect cases of infinite recursion at compile time, but it
1836        cannot analyze patterns in sufficient detail to catch mutual recursions
1837        such as ((?1))((?2)). There is now a runtime test that gives an error if a
1838        subgroup is called recursively as a subpattern for a second time at the
1839        same position in the subject string. In previous releases this might have
1840        been caught by the recursion limit, or it might have run out of stack.
1841    
1842    41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1843        happen only once. PCRE was, however incorrectly giving a compile time error
1844        "recursive call could loop indefinitely" because it cannot analyze the
1845        pattern in sufficient detail. The compile time test no longer happens when
1846        PCRE is compiling a conditional subpattern, but actual runaway loops are
1847        now caught at runtime (see 40 above).
1848    
1849    42. It seems that Perl allows any characters other than a closing parenthesis
1850        to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1851        has been changed to be the same.
1852    
1853    43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1854        as not to get warnings when autogen.sh is called. Also changed
1855        AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1856    
1857    44. To help people who use pcregrep to scan files containing exceedingly long
1858        lines, the following changes have been made:
1859    
1860        (a) The default value of the buffer size parameter has been increased from
1861            8K to 20K. (The actual buffer used is three times this size.)
1862    
1863        (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1864            PCRE is built.
1865    
1866        (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1867            to be set at run time.
1868    
1869        (d) Numerical values in pcregrep options can be followed by K or M, for
1870            example --buffer-size=50K.
1871    
1872        (e) If a line being scanned overflows pcregrep's buffer, an error is now
1873            given and the return code is set to 2.
1874    
1875    45. Add a pointer to the latest mark to the callout data block.
1876    
1877    46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1878        partial match of an empty string instead of no match. This was specific to
1879        the use of ".".
1880    
1881    47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1882        complete match instead of a partial match. This bug was dependent on both
1883        the PCRE_UTF8 and PCRE_DOTALL options being set.
1884    
1885    48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1886        starting byte set, because \b was not being ignored.
1887    
1888    
1889    Version 8.12 15-Jan-2011
1890    ------------------------
1891    
1892    1.  Fixed some typos in the markup of the man pages, and wrote a script that
1893        checks for such things as part of the documentation building process.
1894    
1895    2.  On a big-endian 64-bit system, pcregrep did not correctly process the
1896        --match-limit and --recursion-limit options (added for 8.11). In
1897        particular, this made one of the standard tests fail. (The integer value
1898        went into the wrong half of a long int.)
1899    
1900    3.  If the --colour option was given to pcregrep with -v (invert match), it
1901        did strange things, either producing crazy output, or crashing. It should,
1902        of course, ignore a request for colour when reporting lines that do not
1903        match.
1904    
1905    4.  Another pcregrep bug caused similar problems if --colour was specified with
1906        -M (multiline) and the pattern match finished with a line ending.
1907    
1908    5.  In pcregrep, when a pattern that ended with a literal newline sequence was
1909        matched in multiline mode, the following line was shown as part of the
1910        match. This seems wrong, so I have changed it.
1911    
1912    6.  Another pcregrep bug in multiline mode, when --colour was specified, caused
1913        the check for further matches in the same line (so they could be coloured)
1914        to overrun the end of the current line. If another match was found, it was
1915        incorrectly shown (and then shown again when found in the next line).
1916    
1917    7.  If pcregrep was compiled under Windows, there was a reference to the
1918        function pcregrep_exit() before it was defined. I am assuming this was
1919        the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1920        reported by a user. I've moved the definition above the reference.
1921    
1922    
1923    Version 8.11 10-Dec-2010
1924    ------------------------
1925    
1926    1.  (*THEN) was not working properly if there were untried alternatives prior
1927        to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1928        backtracked to try for "b" instead of moving to the next alternative branch
1929        at the same level (in this case, to look for "c"). The Perl documentation
1930        is clear that when (*THEN) is backtracked onto, it goes to the "next
1931        alternative in the innermost enclosing group".
1932    
1933    2.  (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1934        such as   (A(*COMMIT)B(*THEN)C|D)  any failure after matching A should
1935        result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1936        (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1937        (*THEN).
1938    
1939    3.  If \s appeared in a character class, it removed the VT character from
1940        the class, even if it had been included by some previous item, for example
1941        in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1942        of \s, but is part of the POSIX "space" class.)
1943    
1944    4.  A partial match never returns an empty string (because you can always
1945        match an empty string at the end of the subject); however the checking for
1946        an empty string was starting at the "start of match" point. This has been
1947        changed to the "earliest inspected character" point, because the returned
1948        data for a partial match starts at this character. This means that, for
1949        example, /(?<=abc)def/ gives a partial match for the subject "abc"
1950        (previously it gave "no match").
1951    
1952    5.  Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1953        of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1954        previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1955        has an implication that the given string is incomplete (because a partial
1956        match is preferred over a full match). For this reason, these items now
1957        give a partial match in this situation. [Aside: previously, the one case
1958        /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1959        match rather than a full match, which was wrong by the old rules, but is
1960        now correct.]
1961    
1962    6.  There was a bug in the handling of #-introduced comments, recognized when
1963        PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1964        If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1965        UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1966        scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1967        but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1968        places in pcre_compile().
1969    
1970    7.  Related to (6) above, when pcre_compile() was skipping #-introduced
1971        comments when looking ahead for named forward references to subpatterns,
1972        the only newline sequence it recognized was NL. It now handles newlines
1973        according to the set newline convention.
1974    
1975    8.  SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1976        former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1977        cater for a lack of strerror(). These oversights have been fixed.
1978    
1979    9.  Added --match-limit and --recursion-limit to pcregrep.
1980    
1981    10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1982    
1983    11. When the -o option was used, pcregrep was setting a return code of 1, even
1984        when matches were found, and --line-buffered was not being honoured.
1985    
1986    12. Added an optional parentheses number to the -o and --only-matching options
1987        of pcregrep.
1988    
1989    13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1990        can match an empty string. The code to do it in pcretest and pcredemo
1991        needed fixing:
1992    
1993        (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1994            only one byte after an empty string match just before CRLF (this case
1995            just got forgotten; "any" and "anycrlf" were OK).
1996    
1997        (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1998            mode when an empty string match preceded an ASCII character followed by
1999            a non-ASCII character. (The code for advancing by one character rather
2000            than one byte was nonsense.)
2001    
2002        (c) The pcredemo.c sample program did not have any code at all to handle
2003            the cases when CRLF is a valid newline sequence.
2004    
2005    14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
2006        as a starting offset was within the subject string. There is now a new
2007        error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
2008        negative or greater than the length of the string. In order to test this,
2009        pcretest is extended to allow the setting of negative starting offsets.
2010    
2011    15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
2012        starting offset points to the beginning of a UTF-8 character was
2013        unnecessarily clumsy. I tidied it up.
2014    
2015    16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
2016        bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
2017    
2018    17. Nobody had reported that the --include_dir option, which was added in
2019        release 7.7 should have been called --include-dir (hyphen, not underscore)
2020        for compatibility with GNU grep. I have changed it to --include-dir, but
2021        left --include_dir as an undocumented synonym, and the same for
2022        --exclude-dir, though that is not available in GNU grep, at least as of
2023        release 2.5.4.
2024    
2025    18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
2026        characters from a string of bytes) have been redefined so as not to use
2027        loops, in order to improve performance in some environments. At the same
2028        time, I abstracted some of the common code into auxiliary macros to save
2029        repetition (this should not affect the compiled code).
2030    
2031    19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
2032        compile-time error is now given if \c is not followed by an ASCII
2033        character, that is, a byte less than 128. (In EBCDIC mode, the code is
2034        different, and any byte value is allowed.)
2035    
2036    20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
2037        START_OPTIMIZE option, which is now allowed at compile time - but just
2038        passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
2039        to pcregrep and other applications that have no direct access to PCRE
2040        options. The new /Y option in pcretest sets this option when calling
2041        pcre_compile().
2042    
2043    21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
2044        back references. Groups containing recursive back references were forced to
2045        be atomic by that change, but in the case of named groups, the amount of
2046        memory required was incorrectly computed, leading to "Failed: internal
2047        error: code overflow". This has been fixed.
2048    
2049    22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
2050        pcretest.c, to avoid build problems in some Borland environments.
2051    
2052    
2053    Version 8.10 25-Jun-2010
2054    ------------------------
2055    
2056    1.  Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2057        THEN.
2058    
2059    2.  (*ACCEPT) was not working when inside an atomic group.
2060    
2061    3.  Inside a character class, \B is treated as a literal by default, but
2062        faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2063        causes the error). The code is unchanged, but I tidied the documentation.
2064    
2065    4.  Inside a character class, PCRE always treated \R and \X as literals,
2066        whereas Perl faults them if its -w option is set. I have changed PCRE so
2067        that it faults them when PCRE_EXTRA is set.
2068    
2069    5.  Added support for \N, which always matches any character other than
2070        newline. (It is the same as "." when PCRE_DOTALL is not set.)
2071    
2072    6.  When compiling pcregrep with newer versions of gcc which may have
2073        FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2074        declared with attribute warn_unused_result" were given. Just casting the
2075        result to (void) does not stop the warnings; a more elaborate fudge is
2076        needed. I've used a macro to implement this.
2077    
2078    7.  Minor change to pcretest.c to avoid a compiler warning.
2079    
2080    8.  Added four artifical Unicode properties to help with an option to make
2081        \s etc use properties (see next item). The new properties are: Xan
2082        (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2083    
2084    9.  Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2085        use Unicode properties. (*UCP) at the start of a pattern can be used to set
2086        this option. Modified pcretest to add /W to test this facility. Added
2087        REG_UCP to make it available via the POSIX interface.
2088    
2089    10. Added --line-buffered to pcregrep.
2090    
2091    11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2092        studied, and the match started with a letter with a code point greater than
2093        127 whose first byte was different to the first byte of the other case of
2094        the letter, the other case of this starting letter was not recognized
2095        (#976).
2096    
2097    12. If a pattern that was studied started with a repeated Unicode property
2098        test, for example, \p{Nd}+, there was the theoretical possibility of
2099        setting up an incorrect bitmap of starting bytes, but fortunately it could
2100        not have actually happened in practice until change 8 above was made (it
2101        added property types that matched character-matching opcodes).
2102    
2103    13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2104        possible starting bytes for non-anchored patterns.
2105    
2106    14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2107        \R, and also a number of cases that involve Unicode properties, both
2108        explicit and implicit when PCRE_UCP is set.
2109    
2110    15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2111        input, it could crash or give wrong results if characters with values
2112        greater than 0xc0 were present in the subject string. (Detail: it assumed
2113        UTF-8 input when processing these items.)
2114    
2115    16. Added a lot of (int) casts to avoid compiler warnings in systems where
2116        size_t is 64-bit (#991).
2117    
2118    17. Added a check for running out of memory when PCRE is compiled with
2119        --disable-stack-for-recursion (#990).
2120    
2121    18. If the last data line in a file for pcretest does not have a newline on
2122        the end, a newline was missing in the output.
2123    
2124    19. The default pcre_chartables.c file recognizes only ASCII characters (values
2125        less than 128) in its various bitmaps. However, there is a facility for
2126        generating tables according to the current locale when PCRE is compiled. It
2127        turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2128        characters, are recognized by isspace() and therefore were getting set in
2129        these tables, and indeed these tables seem to approximate to ISO 8859. This
2130        caused a problem in UTF-8 mode when pcre_study() was used to create a list
2131        of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2132        which of course cannot start UTF-8 characters. I have changed the code so
2133        that only real ASCII characters (less than 128) and the correct starting
2134        bytes for UTF-8 encodings are set for characters greater than 127 when in
2135        UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2136        altogether.)
2137    
2138    20. Added the /T option to pcretest so as to be able to run tests with non-
2139        standard character tables, thus making it possible to include the tests
2140        used for 19 above in the standard set of tests.
2141    
2142    21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2143        reference to a subpattern the other side of a comment that contains an
2144        opening parenthesis caused either an internal compiling error, or a
2145        reference to the wrong subpattern.
2146    
2147    
2148    Version 8.02 19-Mar-2010
2149    ------------------------
2150    
2151    1.  The Unicode data tables have been updated to Unicode 5.2.0.
2152    
2153    2.  Added the option --libs-cpp to pcre-config, but only when C++ support is
2154        configured.
2155    
2156    3.  Updated the licensing terms in the pcregexp.pas file, as agreed with the
2157        original author of that file, following a query about its status.
2158    
2159    4.  On systems that do not have stdint.h (e.g. Solaris), check for and include
2160        inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2161    
2162    5.  A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2163        quantifier applied to a forward-referencing subroutine call, could compile
2164        incorrect code or give the error "internal error: previously-checked
2165        referenced subpattern not found".
2166    
2167    6.  Both MS Visual Studio and Symbian OS have problems with initializing
2168        variables to point to external functions. For these systems, therefore,
2169        pcre_malloc etc. are now initialized to local functions that call the
2170        relevant global functions.
2171    
2172    7.  There were two entries missing in the vectors called coptable and poptable
2173        in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2174        I've fixed the data, and added a kludgy way of testing at compile time that
2175        the lengths are correct (equal to the number of opcodes).
2176    
2177    8.  Following on from 7, I added a similar kludge to check the length of the
2178        eint vector in pcreposix.c.
2179    
2180    9.  Error texts for pcre_compile() are held as one long string to avoid too
2181        much relocation at load time. To find a text, the string is searched,
2182        counting zeros. There was no check for running off the end of the string,
2183        which could happen if a new error number was added without updating the
2184        string.
2185    
2186    10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2187    
2188    11. \K was not working if it appeared in an atomic group or in a group that
2189        was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2190        \K is "not well defined" if used in an assertion. PCRE now accepts it if
2191        the assertion is positive, but not if it is negative.
2192    
2193    12. Change 11 fortuitously reduced the size of the stack frame used in the
2194        "match()" function of pcre_exec.c by one pointer. Forthcoming
2195        implementation of support for (*MARK) will need an extra pointer on the
2196        stack; I have reserved it now, so that the stack frame size does not
2197        decrease.
2198    
2199    13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2200        item in branch that calls a recursion is a subroutine call - as in the
2201        second branch in the above example - was incorrectly given the compile-
2202        time error "recursive call could loop indefinitely" because pcre_compile()
2203        was not correctly checking the subroutine for matching a non-empty string.
2204    
2205    14. The checks for overrunning compiling workspace could trigger after an
2206        overrun had occurred. This is a "should never occur" error, but it can be
2207        triggered by pathological patterns such as hundreds of nested parentheses.
2208        The checks now trigger 100 bytes before the end of the workspace.
2209    
2210    15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2211    
2212    
2213    Version 8.01 19-Jan-2010
2214    ------------------------
2215    
2216    1.  If a pattern contained a conditional subpattern with only one branch (in
2217        particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2218        computed the wrong minimum data length (which is of course zero for such
2219        subpatterns). This could cause incorrect "no match" results.
2220    
2221    2.  For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2222        the pattern is reset in the first branch, pcre_compile() failed with
2223        "internal error: code overflow at offset...". This happened only when
2224        the reset was to the original external option setting. (An optimization
2225        abstracts leading options settings into an external setting, which was the
2226        cause of this.)
2227    
2228    3.  A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2229        of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2230        assertion pattern did not match (meaning that the assertion was true), it
2231        was incorrectly treated as false if the SKIP had been reached during the
2232        matching. This also applied to assertions used as conditions.
2233    
2234    4.  If an item that is not supported by pcre_dfa_exec() was encountered in an
2235        assertion subpattern, including such a pattern used as a condition,
2236        unpredictable results occurred, instead of the error return
2237        PCRE_ERROR_DFA_UITEM.
2238    
2239    5.  The C++ GlobalReplace function was not working like Perl for the special
2240        situation when an empty string is matched. It now does the fancy magic
2241        stuff that is necessary.
2242    
2243    6.  In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2244        removed. (These were left over from very, very early versions of PCRE.)
2245    
2246    7.  Some cosmetic changes to the code to make life easier when compiling it
2247        as part of something else:
2248    
2249        (a) Change DEBUG to PCRE_DEBUG.
2250    
2251        (b) In pcre_compile(), rename the member of the "branch_chain" structure
2252            called "current" as "current_branch", to prevent a collision with the
2253            Linux macro when compiled as a kernel module.
2254    
2255        (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2256            prevent a collision with the Linux macro when compiled as a kernel
2257            module.
2258    
2259    8.  In pcre_compile() there are some checks for integer overflows that used to
2260        cast potentially large values to (double). This has been changed to that
2261        when building, a check for int64_t is made, and if it is found, it is used
2262        instead, thus avoiding the use of floating point arithmetic. (There is no
2263        other use of FP in PCRE.) If int64_t is not found, the fallback is to
2264        double.
2265    
2266    9.  Added two casts to avoid signed/unsigned warnings from VS Studio Express
2267        2005 (difference between two addresses compared to an unsigned value).
2268    
2269    10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2270        custom one, because of the following reported problem in Windows:
2271    
2272          - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2273              under Win32.
2274          - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2275              therefore missing the function definition.
2276          - The compiler thus generates a "C" signature for the test function.
2277          - The linker fails to find the "C" function.
2278          - PCRE fails to configure if asked to do so against libbz2.
2279    
2280    11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2281        messages were output:
2282    
2283          Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2284          rerunning libtoolize, to keep the correct libtool macros in-tree.
2285          Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2286    
2287        I have done both of these things.
2288    
2289    12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2290        most of the time, it *can* run out if it is given a pattern that contains a
2291        runaway infinite recursion. I updated the discussion in the pcrestack man
2292        page.
2293    
2294    13. Now that we have gone to the x.xx style of version numbers, the minor
2295        version may start with zero. Using 08 or 09 is a bad idea because users
2296        might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2297        interpreted as invalid octal numbers. I've updated the previous comment in
2298        configure.ac, and also added a check that gives an error if 08 or 09 are
2299        used.
2300    
2301    14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2302        causing partial matching to fail when the end of the subject matched \W
2303        in a UTF-8 pattern where \W was quantified with a minimum of 3.
2304    
2305    15. There were some discrepancies between the declarations in pcre_internal.h
2306        of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2307        their definitions. The declarations used "const uschar *" and the
2308        definitions used USPTR. Even though USPTR is normally defined as "const
2309        unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2310        reported that: "This difference in casting confuses some C++ compilers, for
2311        example, SunCC recognizes above declarations as different functions and
2312        generates broken code for hbpcre." I have changed the declarations to use
2313        USPTR.
2314    
2315    16. GNU libtool is named differently on some systems. The autogen.sh script now
2316        tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2317        (FreeBSD).
2318    
2319    17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2320        (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2321        comment: "Figure out how to create a longlong from a string: strtoll and
2322        equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2323        instance, but it only takes 2 args instead of 3!"
2324    
2325    18. A subtle bug concerned with back references has been fixed by a change of
2326        specification, with a corresponding code fix. A pattern such as
2327        ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2328        refers, was giving matches when it shouldn't. For example, xa=xaaa would
2329        match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2330        same bug. Such groups have to be quantified to be useful, or contained
2331        inside another quantified group. (If there's no repetition, the reference
2332        can never match.) The problem arises because, having left the group and
2333        moved on to the rest of the pattern, a later failure that backtracks into
2334        the group uses the captured value from the final iteration of the group
2335        rather than the correct earlier one. I have fixed this in PCRE by forcing
2336        any group that contains a reference to itself to be an atomic group; that
2337        is, there cannot be any backtracking into it once it has completed. This is
2338        similar to recursive and subroutine calls.
2339    
2340    
2341    Version 8.00 19-Oct-09
2342    ----------------------
2343    
2344    1.  The table for translating pcre_compile() error codes into POSIX error codes
2345        was out-of-date, and there was no check on the pcre_compile() error code
2346        being within the table. This could lead to an OK return being given in
2347        error.
2348    
2349    2.  Changed the call to open a subject file in pcregrep from fopen(pathname,
2350        "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2351        in a Windows environment.
2352    
2353    3.  The pcregrep --count option prints the count for each file even when it is
2354        zero, as does GNU grep. However, pcregrep was also printing all files when
2355        --files-with-matches was added. Now, when both options are given, it prints
2356        counts only for those files that have at least one match. (GNU grep just
2357        prints the file name in this circumstance, but including the count seems
2358        more useful - otherwise, why use --count?) Also ensured that the
2359        combination -clh just lists non-zero counts, with no names.
2360    
2361    4.  The long form of the pcregrep -F option was incorrectly implemented as
2362        --fixed_strings instead of --fixed-strings. This is an incompatible change,
2363        but it seems right to fix it, and I didn't think it was worth preserving
2364        the old behaviour.
2365    
2366    5.  The command line items --regex=pattern and --regexp=pattern were not
2367        recognized by pcregrep, which required --regex pattern or --regexp pattern
2368        (with a space rather than an '='). The man page documented the '=' forms,
2369        which are compatible with GNU grep; these now work.
2370    
2371    6.  No libpcreposix.pc file was created for pkg-config; there was just
2372        libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2373    
2374    7.  Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2375        when UCP support is not needed, by modifying the Python script that
2376        generates it from Unicode data files. This should not matter if the module
2377        is correctly used as a library, but I received one complaint about 50K of
2378        unwanted data. My guess is that the person linked everything into his
2379        program rather than using a library. Anyway, it does no harm.
2380    
2381    8.  A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2382        was a minimum greater than 1 for a wide character in a possessive
2383        repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2384        which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2385        character. Chaos in the form of incorrect output or a compiling loop could
2386        result.
2387    
2388    9.  The restrictions on what a pattern can contain when partial matching is
2389        requested for pcre_exec() have been removed. All patterns can now be
2390        partially matched by this function. In addition, if there are at least two
2391        slots in the offset vector, the offset of the earliest inspected character
2392        for the match and the offset of the end of the subject are set in them when
2393        PCRE_ERROR_PARTIAL is returned.
2394    
2395    10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2396        synonymous with PCRE_PARTIAL, for backwards compatibility, and
2397        PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2398        and may be more useful for multi-segment matching.
2399    
2400    11. Partial matching with pcre_exec() is now more intuitive. A partial match
2401        used to be given if ever the end of the subject was reached; now it is
2402        given only if matching could not proceed because another character was
2403        needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2404        string "Z", which now yields "no match" instead of "partial match". In the
2405        case of pcre_dfa_exec(), "no match" is given if every matching path for the
2406        final character ended with (*FAIL).
2407    
2408    12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2409        if the pattern had a "must contain" character that was already found in the
2410        earlier partial match, unless partial matching was again requested. For
2411        example, with the pattern /dog.(body)?/, the "must contain" character is
2412        "g". If the first part-match was for the string "dog", restarting with
2413        "sbody" failed. This bug has been fixed.
2414    
2415    13. The string returned by pcre_dfa_exec() after a partial match has been
2416        changed so that it starts at the first inspected character rather than the
2417        first character of the match. This makes a difference only if the pattern
2418        starts with a lookbehind assertion or \b or \B (\K is not supported by
2419        pcre_dfa_exec()). It's an incompatible change, but it makes the two
2420        matching functions compatible, and I think it's the right thing to do.
2421    
2422    14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2423        so that the demonstration program is easily available in environments where
2424        PCRE has not been installed from source.
2425    
2426    15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2427        libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2428        library.
2429    
2430    16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2431        It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2432        is not the first non-POSIX option to be added. Clearly some people find
2433        these options useful.
2434    
2435    17. If a caller to the POSIX matching function regexec() passes a non-zero
2436        value for nmatch with a NULL value for pmatch, the value of
2437        nmatch is forced to zero.
2438    
2439    18. RunGrepTest did not have a test for the availability of the -u option of
2440        the diff command, as RunTest does. It now checks in the same way as
2441        RunTest, and also checks for the -b option.
2442    
2443    19. If an odd number of negated classes containing just a single character
2444        interposed, within parentheses, between a forward reference to a named
2445        subpattern and the definition of the subpattern, compilation crashed with
2446        an internal error, complaining that it could not find the referenced
2447        subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2448        [The bug was that it was starting one character too far in when skipping
2449        over the character class, thus treating the ] as data rather than
2450        terminating the class. This meant it could skip too much.]
2451    
2452    20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2453        /g option in pcretest when the pattern contains \K, which makes it possible
2454        to have an empty string match not at the start, even when the pattern is
2455        anchored. Updated pcretest and pcredemo to use this option.
2456    
2457    21. If the maximum number of capturing subpatterns in a recursion was greater
2458        than the maximum at the outer level, the higher number was returned, but
2459        with unset values at the outer level. The correct (outer level) value is
2460        now given.
2461    
2462    22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2463        PCRE did not set those parentheses (unlike Perl). I have now found a way to
2464        make it do so. The string so far is captured, making this feature
2465        compatible with Perl.
2466    
2467    23. The tests have been re-organized, adding tests 11 and 12, to make it
2468        possible to check the Perl 5.10 features against Perl 5.10.
2469    
2470    24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2471        pattern matches a fixed length string. PCRE did not allow this; now it
2472        does. Neither allows recursion.
2473    
2474    25. I finally figured out how to implement a request to provide the minimum
2475        length of subject string that was needed in order to match a given pattern.
2476        (It was back references and recursion that I had previously got hung up
2477        on.) This code has now been added to pcre_study(); it finds a lower bound
2478        to the length of subject needed. It is not necessarily the greatest lower
2479        bound, but using it to avoid searching strings that are too short does give
2480        some useful speed-ups. The value is available to calling programs via
2481        pcre_fullinfo().
2482    
2483    26. While implementing 25, I discovered to my embarrassment that pcretest had
2484        not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2485        study optimizations had never been tested with that matching function.
2486        Oops. What is worse, even when it was passed study data, there was a bug in
2487        pcre_dfa_exec() that meant it never actually used it. Double oops. There
2488        were also very few tests of studied patterns with pcre_dfa_exec().
2489    
2490    27. If (?| is used to create subpatterns with duplicate numbers, they are now
2491        allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2492        on the other side of the coin, they are no longer allowed to have different
2493        names, because these cannot be distinguished in PCRE, and this has caused
2494        confusion. (This is a difference from Perl.)
2495    
2496    28. When duplicate subpattern names are present (necessarily with different
2497        numbers, as required by 27 above), and a test is made by name in a
2498        conditional pattern, either for a subpattern having been matched, or for
2499        recursion in such a pattern, all the associated numbered subpatterns are
2500        tested, and the overall condition is true if the condition is true for any
2501        one of them. This is the way Perl works, and is also more like the way
2502        testing by number works.
2503    
2504    
2505    Version 7.9 11-Apr-09
2506    ---------------------
2507    
2508    1.  When building with support for bzlib/zlib (pcregrep) and/or readline
2509        (pcretest), all targets were linked against these libraries. This included
2510        libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2511        libraries. This caused unwanted dependencies to be created. This problem
2512        has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2513        pcretest is linked with readline.
2514    
2515    2.  The "typedef int BOOL" in pcre_internal.h that was included inside the
2516        "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2517        moved outside it again, because FALSE and TRUE are already defined in AIX,
2518        but BOOL is not.
2519    
2520    3.  The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2521        PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2522    
2523    4.  The pcregrep documentation said spaces were inserted as well as colons (or
2524        hyphens) following file names and line numbers when outputting matching
2525        lines. This is not true; no spaces are inserted. I have also clarified the
2526        wording for the --colour (or --color) option.
2527    
2528    5.  In pcregrep, when --colour was used with -o, the list of matching strings
2529        was not coloured; this is different to GNU grep, so I have changed it to be
2530        the same.
2531    
2532    6.  When --colo(u)r was used in pcregrep, only the first matching substring in
2533        each matching line was coloured. Now it goes on to look for further matches
2534        of any of the test patterns, which is the same behaviour as GNU grep.
2535    
2536    7.  A pattern that could match an empty string could cause pcregrep to loop; it
2537        doesn't make sense to accept an empty string match in pcregrep, so I have
2538        locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2539        seems to be how GNU grep behaves. [But see later change 40 for release
2540        8.33.]
2541    
2542    8.  The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2543        start or after a newline", because the conditional assertion was not being
2544        correctly handled. The rule now is that both the assertion and what follows
2545        in the first alternative must satisfy the test.
2546    
2547    9.  If auto-callout was enabled in a pattern with a conditional group whose
2548        condition was an assertion, PCRE could crash during matching, both with
2549        pcre_exec() and pcre_dfa_exec().
2550    
2551    10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2552        used for matching.
2553    
2554    11. Unicode property support in character classes was not working for
2555        characters (bytes) greater than 127 when not in UTF-8 mode.
2556    
2557    12. Added the -M command line option to pcretest.
2558    
2559    14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2560    
2561    15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2562    
2563    16. Added comments and documentation about mis-use of no_arg in the C++
2564        wrapper.
2565    
2566    17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2567        from Martin Jerabek that uses macro names for all relevant character and
2568        string constants.
2569    
2570    18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2571        SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2572        SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2573        these, but not everybody uses configure.
2574    
2575    19. A conditional group that had only one branch was not being correctly
2576        recognized as an item that could match an empty string. This meant that an
2577        enclosing group might also not be so recognized, causing infinite looping
2578        (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2579        with the subject "ab", where knowledge that the repeated group can match
2580        nothing is needed in order to break the loop.
2581    
2582    20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2583        exec(), but without supplying a callout function, matching went wrong.
2584    
2585    21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2586        leak if the size of the offset vector was greater than 30. When the vector
2587        is smaller, the saved offsets during recursion go onto a local stack
2588        vector, but for larger vectors malloc() is used. It was failing to free
2589        when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2590        error, in fact).
2591    
2592    22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2593        heapframe that is used only when UTF-8 support is enabled. This caused no
2594        problem, but was untidy.
2595    
2596    23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2597        CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2598        included within another project.
2599    
2600    24. Steven Van Ingelgem's patches to add more options to the CMake support,
2601        slightly modified by me:
2602    
2603          (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2604              not building pcregrep.
2605    
2606          (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2607              if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2608    
2609    25. Forward references, both numeric and by name, in patterns that made use of
2610        duplicate group numbers, could behave incorrectly or give incorrect errors,
2611        because when scanning forward to find the reference group, PCRE was not
2612        taking into account the duplicate group numbers. A pattern such as
2613        ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2614    
2615    26. Changed a few more instances of "const unsigned char *" to USPTR, making
2616        the feature of a custom pointer more persuasive (as requested by a user).
2617    
2618    27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2619        pcretest.c, inside #ifndefs, because it seems they are sometimes already
2620        pre-defined.
2621    
2622    28. Added support for (*UTF8) at the start of a pattern.
2623    
2624    29. Arrange for flags added by the "release type" setting in CMake to be shown
2625        in the configuration summary.
2626    
2627    
2628    Version 7.8 05-Sep-08
2629    ---------------------
2630    
2631    1.  Replaced UCP searching code with optimized version as implemented for Ad
2632        Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2633        stage table and inline lookup instead of a function, giving speed ups of 2
2634        to 5 times on some simple patterns that I tested. Permission was given to
2635        distribute the MultiStage2.py script that generates the tables (it's not in
2636        the tarball, but is in the Subversion repository).
2637    
2638    2.  Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2639        scripts.
2640    
2641    3.  Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2642        a group with a zero qualifier. The result of the study could be incorrect,
2643        or the function might crash, depending on the pattern.
2644    
2645    4.  Caseless matching was not working for non-ASCII characters in back
2646        references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2647        It now works when Unicode Property Support is available.
2648    
2649    5.  In pcretest, an escape such as \x{de} in the data was always generating
2650        a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2651        non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2652        truncation.
2653    
2654    6.  Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2655    
2656    7.  Added two (int) casts to pcregrep when printing the difference of two
2657        pointers, in case they are 64-bit values.
2658    
2659    8.  Added comments about Mac OS X stack usage to the pcrestack man page and to
2660        test 2 if it fails.
2661    
2662    9.  Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2663        and a #define of that name to empty if it is not externally set. This is to
2664        allow users of MSVC to set it if necessary.
2665    
2666    10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2667        the convenience functions in the pcre_get.c source file.
2668    
2669    11. An option change at the start of a pattern that had top-level alternatives
2670        could cause overwriting and/or a crash. This command provoked a crash in
2671        some environments:
2672    
2673          printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2674    
2675        This potential security problem was recorded as CVE-2008-2371.
2676    
2677    12. For a pattern where the match had to start at the beginning or immediately
2678        after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2679        pcre_dfa_exec() could read past the end of the passed subject if there was
2680        no match. To help with detecting such bugs (e.g. with valgrind), I modified
2681        pcretest so that it places the subject at the end of its malloc-ed buffer.
2682    
2683    13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2684        exec() might read past the end of the data buffer in UTF-8 mode.
2685    
2686    14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2687        the data contained the byte 0x85 as part of a UTF-8 character within its
2688        first line. This applied both to normal and DFA matching.
2689    
2690    15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2691        /^[^d]*?$/8 failed to match "abc".
2692    
2693    16. Added a missing copyright notice to pcrecpp_internal.h.
2694    
2695    17. Make it more clear in the documentation that values returned from
2696        pcre_exec() in ovector are byte offsets, not character counts.
2697    
2698    18. Tidied a few places to stop certain compilers from issuing warnings.
2699    
2700    19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2701        supplied by Stefan Weber. I made a further small update for 7.8 because
2702        there is a change of source arrangements: the pcre_searchfuncs.c module is
2703        replaced by pcre_ucd.c.
2704    
2705    
2706    Version 7.7 07-May-08
2707    ---------------------
2708    
2709    1.  Applied Craig's patch to sort out a long long problem: "If we can't convert
2710        a string to a long long, pretend we don't even have a long long." This is
2711        done by checking for the strtoq, strtoll, and _strtoi64 functions.
2712    
2713    2.  Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2714        pre-7.6 versions, which defined a global no_arg variable instead of putting
2715        it in the RE class. (See also #8 below.)
2716    
2717    3.  Remove a line of dead code, identified by coverity and reported by Nuno
2718        Lopes.
2719    
2720    4.  Fixed two related pcregrep bugs involving -r with --include or --exclude:
2721    
2722        (1) The include/exclude patterns were being applied to the whole pathnames
2723            of files, instead of just to the final components.
2724    
2725        (2) If there was more than one level of directory, the subdirectories were
2726            skipped unless they satisfied the include/exclude conditions. This is
2727            inconsistent with GNU grep (and could even be seen as contrary to the
2728            pcregrep specification - which I improved to make it absolutely clear).
2729            The action now is always to scan all levels of directory, and just
2730            apply the include/exclude patterns to regular files.
2731    
2732    5.  Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2733        --exclude_dir in the tests to avoid scanning .svn directories.
2734    
2735    6.  Applied Craig's patch to the QuoteMeta function so that it escapes the
2736        NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2737        doesn't support NULs in patterns.
2738    
2739    7.  Added some missing "const"s to declarations of static tables in
2740        pcre_compile.c and pcre_dfa_exec.c.
2741    
2742    8.  Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2743        caused by fix #2  above. (Subsequently also a second patch to fix the
2744        first patch. And a third patch - this was a messy problem.)
2745    
2746    9.  Applied Craig's patch to remove the use of push_back().
2747    
2748    10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2749        matching function regexec().
2750    
2751    11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2752        which, however, unlike Perl's \g{...}, are subroutine calls, not back
2753        references. PCRE supports relative numbers with this syntax (I don't think
2754        Oniguruma does).
2755    
2756    12. Previously, a group with a zero repeat such as (...){0} was completely
2757        omitted from the compiled regex. However, this means that if the group
2758        was called as a subroutine from elsewhere in the pattern, things went wrong
2759        (an internal error was given). Such groups are now left in the compiled
2760        pattern, with a new opcode that causes them to be skipped at execution
2761        time.
2762    
2763    13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2764        to the way PCRE behaves:
2765    
2766        (a) A lone ] character is dis-allowed (Perl treats it as data).
2767    
2768        (b) A back reference to an unmatched subpattern matches an empty string
2769            (Perl fails the current match path).
2770    
2771        (c) A data ] in a character class must be notated as \] because if the
2772            first data character in a class is ], it defines an empty class. (In
2773            Perl it is not possible to have an empty class.) The empty class []
2774            never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2775            The negative empty class [^] matches any one character, independently
2776            of the DOTALL setting.
2777    
2778    14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2779        non-existent subpattern following a character class starting with ']' and
2780        containing () gave an internal compiling error instead of "reference to
2781        non-existent subpattern". Fortunately, when the pattern did exist, the
2782        compiled code was correct. (When scanning forwards to check for the
2783        existence of the subpattern, it was treating the data ']' as terminating
2784        the class, so got the count wrong. When actually compiling, the reference
2785        was subsequently set up correctly.)
2786    
2787    15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2788        it was being rejected as not supported by pcre_dfa_exec(), even though
2789        other assertions are supported. I have made pcre_dfa_exec() support
2790        (*FAIL).
2791    
2792    16. The implementation of 13c above involved the invention of a new opcode,
2793        OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2794        cannot be changed at match time, I realized I could make a small
2795        improvement to matching performance by compiling OP_ALLANY instead of
2796        OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2797        on the OP_ANY path.
2798    
2799    17. Compiling pcretest on Windows with readline support failed without the
2800        following two fixes: (1) Make the unistd.h include conditional on
2801        HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2802    
2803    18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2804        ncurses library to be included for pcretest when ReadLine support is
2805        requested, but also to allow for it to be overridden. This patch came from
2806        Daniel Bergström.
2807    
2808    19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2809        as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2810        any errors with the current Unicode tables. Thanks to Peter Kankowski for
2811        spotting this.
2812    
2813    
2814    Version 7.6 28-Jan-08
2815    ---------------------
2816    
2817    1.  A character class containing a very large number of characters with
2818        codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2819        overflow.
2820    
2821    2.  Patch to cut out the "long long" test in pcrecpp_unittest when
2822        HAVE_LONG_LONG is not defined.
2823    
2824    3.  Applied Christian Ehrlicher's patch to update the CMake build files to
2825        bring them up to date and include new features. This patch includes:
2826    
2827        - Fixed PH's badly added libz and libbz2 support.
2828        - Fixed a problem with static linking.
2829        - Added pcredemo. [But later removed - see 7 below.]
2830        - Fixed dftables problem and added an option.
2831        - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2832            HAVE_LONG_LONG.
2833        - Added readline support for pcretest.
2834        - Added an listing of the option settings after cmake has run.
2835    
2836    4.  A user submitted a patch to Makefile that makes it easy to create
2837        "pcre.dll" under mingw when using Configure/Make. I added stuff to
2838        Makefile.am that cause it to include this special target, without
2839        affecting anything else. Note that the same mingw target plus all
2840        the other distribution libraries and programs are now supported
2841        when configuring with CMake (see 6 below) instead of with
2842        Configure/Make.
2843    
2844    5.  Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2845        This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2846        exported in the Windows port". It has not yet been confirmed that the patch
2847        solves the problem, but it does no harm.
2848    
2849    6.  Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2850        NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2851        with CMake, and also correct the comment about stack recursion.
2852    
2853    7.  Remove the automatic building of pcredemo from the ./configure system and
2854        from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2855        of a program that users should build themselves after PCRE is installed, so
2856        building it automatically is not really right. What is more, it gave
2857        trouble in some build environments.
2858    
2859    8.  Further tidies to CMakeLists.txt from Sheri and Christian.
2860    
2861    
2862    Version 7.5 10-Jan-08
2863    ---------------------
2864    
2865    1.  Applied a patch from Craig: "This patch makes it possible to 'ignore'
2866        values in parens when parsing an RE using the C++ wrapper."
2867    
2868    2.  Negative specials like \S did not work in character classes in UTF-8 mode.
2869        Characters greater than 255 were excluded from the class instead of being
2870        included.
2871    
2872    3.  The same bug as (2) above applied to negated POSIX classes such as
2873        [:^space:].
2874    
2875    4.  PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2876        defined or documented. It seems to have been a typo for PCRE_STATIC, so
2877        I have changed it.
2878    
2879    5.  The construct (?&) was not diagnosed as a syntax error (it referenced the
2880        first named subpattern) and a construct such as (?&a) would reference the
2881        first named subpattern whose name started with "a" (in other words, the
2882        length check was missing). Both these problems are fixed. "Subpattern name
2883        expected" is now given for (?&) (a zero-length name), and this patch also
2884        makes it give the same error for \k'' (previously it complained that that
2885        was a reference to a non-existent subpattern).
2886    
2887    6.  The erroneous patterns (?+-a) and (?-+a) give different error messages;
2888        this is right because (?- can be followed by option settings as well as by
2889        digits. I have, however, made the messages clearer.
2890    
2891    7.  Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2892        than the number used in the conditional) now cause a compile-time error.
2893        This is actually not compatible with Perl, which accepts such patterns, but
2894        treats the conditional as always being FALSE (as PCRE used to), but it
2895        seems to me that giving a diagnostic is better.
2896    
2897    8.  Change "alphameric" to the more common word "alphanumeric" in comments
2898        and messages.
2899    
2900    9.  Fix two occurrences of "backslash" in comments that should have been
2901        "backspace".
2902    
2903    10. Remove two redundant lines of code that can never be obeyed (their function
2904        was moved elsewhere).
2905    
2906    11. The program that makes PCRE's Unicode character property table had a bug
2907        which caused it to generate incorrect table entries for sequences of
2908        characters that have the same character type, but are in different scripts.
2909        It amalgamated them into a single range, with the script of the first of
2910        them. In other words, some characters were in the wrong script. There were
2911        thirteen such cases, affecting characters in the following ranges:
2912    
2913          U+002b0 - U+002c1
2914          U+0060c - U+0060d
2915          U+0061e - U+00612
2916          U+0064b - U+0065e
2917          U+0074d - U+0076d
2918          U+01800 - U+01805
2919          U+01d00 - U+01d77
2920          U+01d9b - U+01dbf
2921          U+0200b - U+0200f
2922          U+030fc - U+030fe
2923          U+03260 - U+0327f
2924          U+0fb46 - U+0fbb1
2925          U+10450 - U+1049d
2926    
2927    12. The -o option (show only the matching part of a line) for pcregrep was not
2928        compatible with GNU grep in that, if there was more than one match in a
2929        line, it showed only the first of them. It now behaves in the same way as
2930        GNU grep.
2931    
2932    13. If the -o and -v options were combined for pcregrep, it printed a blank
2933        line for every non-matching line. GNU grep prints nothing, and pcregrep now
2934        does the same. The return code can be used to tell if there were any
2935        non-matching lines.
2936    
2937    14. Added --file-offsets and --line-offsets to pcregrep.
2938    
2939    15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2940        infinitely looping recursion. The bug was that positive lookaheads were not
2941        being skipped when checking for a possible empty match (negative lookaheads
2942        and both kinds of lookbehind were skipped).
2943    
2944    16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2945        inclusion of <windows.h> to before rather than after the definition of
2946        INVALID_FILE_ATTRIBUTES (patch from David Byron).
2947    
2948    17. Specifying a possessive quantifier with a specific limit for a Unicode
2949        character property caused pcre_compile() to compile bad code, which led at
2950        runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2951        are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2952        caused the error; without that there was no problem.
2953    
2954    18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2955    
2956    19. Added --enable-pcretest-libreadline.
2957    
2958    20. In pcrecpp.cc, the variable 'count' was incremented twice in
2959        RE::GlobalReplace(). As a result, the number of replacements returned was
2960        double what it should be. I removed one of the increments, but Craig sent a
2961        later patch that removed the other one (the right fix) and added unit tests
2962        that check the return values (which was not done before).
2963    
2964    21. Several CMake things:
2965    
2966        (1) Arranged that, when cmake is used on Unix, the libraries end up with
2967            the names libpcre and libpcreposix, not just pcre and pcreposix.
2968    
2969        (2) The above change means that pcretest and pcregrep are now correctly
2970            linked with the newly-built libraries, not previously installed ones.
2971    
2972        (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2973    
2974    22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2975        crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2976        UTF-8 newline character). The key issue is that the pattern starts .*;
2977        this means that the match must be either at the beginning, or after a
2978        newline. The bug was in the code for advancing after a failed match and
2979        checking that the new position followed a newline. It was not taking
2980        account of UTF-8 characters correctly.
2981    
2982    23. PCRE was behaving differently from Perl in the way it recognized POSIX
2983        character classes. PCRE was not treating the sequence [:...:] as a
2984        character class unless the ... were all letters. Perl, however, seems to
2985        allow any characters between [: and :], though of course it rejects as
2986        unknown any "names" that contain non-letters, because all the known class
2987        names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2988        for example, whereas PCRE did not - it did not recognize a POSIX character
2989        class. This seemed a bit dangerous, so the code has been changed to be
2990        closer to Perl. The behaviour is not identical to Perl, because PCRE will
2991        diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2992        treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2993        Perl does, and where it didn't before.
2994    
2995    24. Rewrite so as to remove the single use of %n from pcregrep because in some
2996        Windows environments %n is disabled by default.
2997    
2998    
2999    Version 7.4 21-Sep-07
3000    ---------------------
3001    
3002    1.  Change 7.3/28 was implemented for classes by looking at the bitmap. This
3003        means that a class such as [\s] counted as "explicit reference to CR or
3004        LF". That isn't really right - the whole point of the change was to try to
3005        help when there was an actual mention of one of the two characters. So now
3006        the change happens only if \r or \n (or a literal CR or LF) character is
3007        encountered.
3008    
3009    2.  The 32-bit options word was also used for 6 internal flags, but the numbers
3010        of both had grown to the point where there were only 3 bits left.
3011        Fortunately, there was spare space in the data structure, and so I have
3012        moved the internal flags into a new 16-bit field to free up more option
3013        bits.
3014    
3015    3.  The appearance of (?J) at the start of a pattern set the DUPNAMES option,
3016        but did not set the internal JCHANGED flag - either of these is enough to
3017        control the way the "get" function works - but the PCRE_INFO_JCHANGED
3018        facility is supposed to tell if (?J) was ever used, so now (?J) at the
3019        start sets both bits.
3020    
3021    4.  Added options (at build time, compile time, exec time) to change \R from
3022        matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
3023    
3024    5.  doc/pcresyntax.html was missing from the distribution.
3025    
3026    6.  Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
3027        compatibility, even though it is no longer used.
3028    
3029    7.  Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
3030        strtoull to pcrecpp.cc to select the available functions in WIN32 when the
3031        windows.h file is present (where different names are used). [This was
3032        reversed later after testing - see 16 below.]
3033    
3034    8.  Changed all #include <config.h> to #include "config.h". There were also
3035        some further <pcre.h> cases that I changed to "pcre.h".
3036    
3037    9.  When pcregrep was used with the --colour option, it missed the line ending
3038        sequence off the lines that it output.
3039    
3040    10. It was pointed out to me that arrays of string pointers cause lots of
3041        relocations when a shared library is dynamically loaded. A technique of
3042        using a single long string with a table of offsets can drastically reduce
3043        these. I have refactored PCRE in four places to do this. The result is
3044        dramatic:
3045    
3046          Originally:                          290
3047          After changing UCP table:            187
3048          After changing error message table:   43
3049          After changing table of "verbs"       36
3050          After changing table of Posix names   22
3051    
3052        Thanks to the folks working on Gregex for glib for this insight.
3053    
3054    11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3055        unicode-properties was also set.
3056    
3057    12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3058    
3059    13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3060        checked only for CRLF.
3061    
3062    14. Added casts to pcretest.c to avoid compiler warnings.
3063    
3064    15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3065    
3066    16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3067        and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3068        entirely. This removes changes made in 7 above.
3069    
3070    17. The CMake files have been updated, and there is now more information about
3071        building with CMake in the NON-UNIX-USE document.
3072    
3073    
3074    Version 7.3 28-Aug-07
3075    ---------------------
3076    
3077     1. In the rejigging of the build system that eventually resulted in 7.1, the
3078        line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3079        brackets there is not right, since it causes compilers to look for an
3080        installed pcre.h, not the version that is in the source that is being
3081        compiled (which of course may be different). I have changed it back to:
3082    
3083          #include "pcre.h"
3084    
3085        I have a vague recollection that the change was concerned with compiling in
3086        different directories, but in the new build system, that is taken care of
3087        by the VPATH setting the Makefile.
3088    
3089     2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3090        when the subject happened to end in the byte 0x85 (e.g. if the last
3091        character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3092        characters but of course it shouldn't be taken as a newline when it is part
3093        of another character. The bug was that, for an unlimited repeat of . in
3094        not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3095        characters when looking for a newline.
3096    
3097     3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3098    
3099     4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3100        in debug output.
3101    
3102     5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3103        long printing in the pcrecpp unittest when running under MinGW.
3104    
3105     6. ESC_K was left out of the EBCDIC table.
3106    
3107     7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3108        parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3109        limit also applies to "virtual nesting" when a pattern is recursive, and in
3110        this case 1000 isn't so big. I have been able to remove this limit at the
3111        expense of backing off one optimization in certain circumstances. Normally,
3112        when pcre_exec() would call its internal match() function recursively and
3113        immediately return the result unconditionally, it uses a "tail recursion"
3114        feature to save stack. However, when a subpattern that can match an empty
3115        string has an unlimited repetition quantifier, it no longer makes this
3116        optimization. That gives it a stack frame in which to save the data for
3117        checking that an empty string has been matched. Previously this was taken
3118        from the 1000-entry workspace that had been reserved. So now there is no
3119        explicit limit, but more stack is used.
3120    
3121     8. Applied Daniel's patches to solve problems with the import/export magic
3122        syntax that is required for Windows, and which was going wrong for the
3123        pcreposix and pcrecpp parts of the library. These were overlooked when this
3124        problem was solved for the main library.
3125    
3126     9. There were some crude static tests to avoid integer overflow when computing
3127        the size of patterns that contain repeated groups with explicit upper
3128        limits. As the maximum quantifier is 65535, the maximum group length was
3129        set at 30,000 so that the product of these two numbers did not overflow a
3130        32-bit integer. However, it turns out that people want to use groups that
3131        are longer than 30,000 bytes (though not repeat them that many times).
3132        Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3133        made it possible to implement the integer overflow checks in a much more
3134        dynamic way, which I have now done. The artificial limitation on group
3135        length has been removed - we now have only the limit on the total length of
3136        the compiled pattern, which depends on the LINK_SIZE setting.
3137    
3138    10. Fixed a bug in the documentation for get/copy named substring when
3139        duplicate names are permitted. If none of the named substrings are set, the
3140        functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3141        empty string.
3142    
3143    11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3144        instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3145        because the ] is interpreted as the first data character and the
3146        terminating ] is not found. PCRE has been made compatible with Perl in this
3147        regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3148        cause memory overwriting.
3149    
3150    10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3151        string has been matched (to stop an infinite loop). It was not recognizing
3152        a conditional subpattern that could match an empty string if that
3153        subpattern was within another subpattern. For example, it looped when
3154        trying to match  (((?(1)X|))*)  but it was OK with  ((?(1)X|)*)  where the
3155        condition was not nested. This bug has been fixed.
3156    
3157    12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3158        past the start of the subject in the presence of bytes with the top bit
3159        set, for example "\x8aBCD".
3160    
3161    13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3162        (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3163    
3164    14. Optimized (?!) to (*FAIL).
3165    
3166    15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3167        This restricts code points to be within the range 0 to 0x10FFFF, excluding
3168        the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3169        full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3170        does: it's just the validity check that is more restrictive.
3171    
3172    16. Inserted checks for integer overflows during escape sequence (backslash)
3173        processing, and also fixed erroneous offset values for syntax errors during
3174        backslash processing.
3175    
3176    17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3177        for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3178    
3179    18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3180        caused an overrun.
3181    
3182    19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3183        something other than just ASCII characters) inside a group that had an
3184        unlimited repeat caused a loop at compile time (while checking to see
3185        whether the group could match an empty string).
3186    
3187    20. Debugging a pattern containing \p or \P could cause a crash. For example,
3188        [\P{Any}] did so. (Error in the code for printing property names.)
3189    
3190    21. An orphan \E inside a character class could cause a crash.
3191    
3192    22. A repeated capturing bracket such as (A)? could cause a wild memory
3193        reference during compilation.
3194    
3195    23. There are several functions in pcre_compile() that scan along a compiled
3196        expression for various reasons (e.g. to see if it's fixed length for look
3197        behind). There were bugs in these functions when a repeated \p or \P was
3198        present in the pattern. These operators have additional parameters compared
3199        with \d, etc, and these were not being taken into account when moving along
3200        the compiled data. Specifically:
3201    
3202        (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3203            length.
3204    
3205        (b) An item such as \pL+ within a repeated group could cause crashes or
3206            loops.
3207    
3208        (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3209            "reference to non-existent subpattern" error.
3210    
3211        (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3212    
3213    24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3214        characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3215    
3216    25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3217    
3218    26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3219        character were causing crashes (broken optimization).
3220    
3221    27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3222        \p or \P) caused a compile-time loop.
3223    
3224    28. More problems have arisen in unanchored patterns when CRLF is a valid line
3225        break. For example, the unstudied pattern [\r\n]A does not match the string
3226        "\r\nA" because change 7.0/46 below moves the current point on by two
3227        characters after failing to match at the start. However, the pattern \nA
3228        *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3229        the same is true. There doesn't seem any very clean way out of this, but
3230        what I have chosen to do makes the common cases work: PCRE now takes note
3231        of whether there can be an explicit match for \r or \n anywhere in the
3232        pattern, and if so, 7.0/46 no longer applies. As part of this change,
3233        there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3234        pattern has explicit CR or LF references.
3235    
3236    29. Added (*CR) etc for changing newline setting at start of pattern.
3237    
3238    
3239    Version 7.2 19-Jun-07
3240    ---------------------
3241    
3242     1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3243        which is apparently normally available under Windows.
3244    
3245     2. Re-jig the pcregrep tests with different newline settings in an attempt
3246        to make them independent of the local environment's newline setting.
3247    
3248     3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3249    
3250     4. Some of the "internals" tests were previously cut out when the link size
3251        was not 2, because the output contained actual offsets. The recent new
3252        "Z" feature of pcretest means that these can be cut out, making the tests
3253        usable with all link sizes.
3254    
3255     5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3256        stack recursion. This gives a massive performance boost under BSD, but just
3257        a small improvement under Linux. However, it saves one field in the frame
3258        in all cases.
3259    
3260     6. Added more features from the forthcoming Perl 5.10:
3261    
3262        (a) (?-n) (where n is a string of digits) is a relative subroutine or
3263            recursion call. It refers to the nth most recently opened parentheses.
3264    
3265        (b) (?+n) is also a relative subroutine call; it refers to the nth next
3266            to be opened parentheses.
3267    
3268        (c) Conditions that refer to capturing parentheses can be specified
3269            relatively, for example, (?(-2)... or (?(+3)...
3270    
3271        (d) \K resets the start of the current match so that everything before
3272            is not part of it.
3273    
3274        (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3275    
3276        (f) \g{name} is another synonym - part of Perl 5.10's unification of
3277            reference syntax.
3278    
3279        (g) (?| introduces a group in which the numbering of parentheses in each
3280            alternative starts with the same number.
3281    
3282        (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3283    
3284     7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3285        PCRE_INFO_JCHANGED.
3286    
3287     8. A pattern such as  (.*(.)?)*  caused pcre_exec() to fail by either not
3288        terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3289        for detecting groups that can match an empty string.
3290    
3291     9. A pattern with a very large number of alternatives (more than several
3292        hundred) was running out of internal workspace during the pre-compile
3293        phase, where pcre_compile() figures out how much memory will be needed. A
3294        bit of new cunning has reduced the workspace needed for groups with
3295        alternatives. The 1000-alternative test pattern now uses 12 bytes of
3296        workspace instead of running out of the 4096 that are available.
3297    
3298    10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3299    
3300    11. Applied patch from Google to remove an optimization that didn't quite work.
3301        The report of the bug said:
3302    
3303          pcrecpp::RE("a*").FullMatch("aaa") matches, while
3304          pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3305          pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3306    
3307    12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3308        it matched the wrong number of bytes.
3309    
3310    
3311    Version 7.1 24-Apr-07
3312    ---------------------
3313    
3314     1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3315        that is more "standard", making use of automake and other Autotools. There
3316        is some re-arrangement of the files and adjustment of comments consequent
3317        on this.
3318    
3319     2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3320        for recursive directory scanning broke on some systems because the files
3321        are not scanned in any specific order and on different systems the order
3322        was different. A call to "sort" has been inserted into RunGrepTest for the
3323        approprate test as a short-term fix. In the longer term there may be an
3324        alternative.
3325    
3326     3. I had an email from Eric Raymond about problems translating some of PCRE's
3327        man pages to HTML (despite the fact that I distribute HTML pages, some
3328        people do their own conversions for various reasons). The problems
3329        concerned the use of low-level troff macros .br and .in. I have therefore
3330        removed all such uses from the man pages (some were redundant, some could
3331        be replaced by .nf/.fi pairs). The 132html script that I use to generate
3332        HTML has been updated to handle .nf/.fi and to complain if it encounters
3333        .br or .in.
3334    
3335     4. Updated comments in configure.ac that get placed in config.h.in and also
3336        arranged for config.h to be included in the distribution, with the name
3337        config.h.generic, for the benefit of those who have to compile without
3338        Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3339    
3340     5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3341        Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3342        makevp.bat for the current PCRE, using the additional files
3343        makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3344    
3345     6. A Windows user reported a minor discrepancy with test 2, which turned out
3346        to be caused by a trailing space on an input line that had got lost in his
3347        copy. The trailing space was an accident, so I've just removed it.
3348    
3349     7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3350        that is needed.
3351    
3352     8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3353        as "const" (a) because they are and (b) because it helps the PHP
3354        maintainers who have recently made a script to detect big data structures
3355        in the php code that should be moved to the .rodata section. I remembered
3356        to update Builducptable as well, so it won't revert if ucptable.h is ever
3357        re-created.
3358    
3359     9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3360        pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3361        order to be able to cut out the UTF-8 tables in the latter when UTF-8
3362        support is not required. This saves 1.5-2K of code, which is important in
3363        some applications.
3364    
3365        Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3366        so as not to refer to the tables, even though these functions will never be
3367        called when UTF-8 support is disabled. Otherwise there are problems with a
3368        shared library.
3369    
3370    10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3371    
3372        (a) It was defining its arguments as char * instead of void *.
3373    
3374        (b) It was assuming that all moves were upwards in memory; this was true
3375            a long time ago when I wrote it, but is no longer the case.
3376    
3377        The emulated memove() is provided for those environments that have neither
3378        memmove() nor bcopy(). I didn't think anyone used it these days, but that
3379        is clearly not the case, as these two bugs were recently reported.
3380    
3381    11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3382        and Detrail to create the HTML documentation, the .txt form of the man
3383        pages, and it removes trailing spaces from listed files. It also creates
3384        pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3385        case, it wraps all the #defines with #ifndefs. This script should be run
3386        before "make dist".
3387    
3388    12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3389        with Unicode property support.
3390    
3391        (a) For a maximizing quantifier, if the two different cases of the
3392            character were of different lengths in their UTF-8 codings (there are
3393            some cases like this - I found 11), and the matching function had to
3394            back up over a mixture of the two cases, it incorrectly assumed they
3395            were both the same length.
3396    
3397        (b) When PCRE was configured to use the heap rather than the stack for
3398            recursion during matching, it was not correctly preserving the data for
3399            the other case of a UTF-8 character when checking ahead for a match
3400            while processing a minimizing repeat. If the check also involved
3401            matching a wide character, but failed, corruption could cause an
3402            erroneous result when trying to check for a repeat of the original
3403            character.
3404    
3405    13. Some tidying changes to the testing mechanism:
3406    
3407        (a) The RunTest script now detects the internal link size and whether there
3408            is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3409            values substituted by "configure". (The RunGrepTest script already did
3410            this for UTF-8.) The configure.ac script no longer substitutes the
3411            relevant variables.
3412    
3413        (b) The debugging options /B and /D in pcretest show the compiled bytecode
3414            with length and offset values. This means that the output is different
3415            for different internal link sizes. Test 2 is skipped for link sizes
3416            other than 2 because of this, bypassing the problem. Unfortunately,
3417            there was also a test in test 3 (the locale tests) that used /B and
3418            failed for link sizes other than 2. Rather than cut the whole test out,
3419            I have added a new /Z option to pcretest that replaces the length and
3420            offset values with spaces. This is now used to make test 3 independent
3421            of link size. (Test 2 will be tidied up later.)
3422    
3423    14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3424        segmentation fault instead of returning the appropriate error message.
3425    
3426    15. In multiline mode when the newline sequence was set to "any", the pattern
3427        ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3428        This doesn't seem right; it now treats the CRLF combination as the line
3429        ending, and so does not match in that case. It's only a pattern such as ^$
3430        that would hit this one: something like ^ABC$ would have failed after \r
3431        and then tried again after \r\n.
3432    
3433    16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3434        in an attempt to make files that differ only in their line terminators
3435        compare equal. This works on Linux.
3436    
3437    17. Under certain error circumstances pcregrep might try to free random memory
3438        as it exited. This is now fixed, thanks to valgrind.
3439    
3440    19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3441        "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3442        was because its rules for how to advance for /g after matching an empty
3443        string at the end of a line did not allow for this case. They now check for
3444        it specially.
3445    
3446    20. pcretest is supposed to handle patterns and data of any length, by
3447        extending its buffers when necessary. It was getting this wrong when the
3448        buffer for a data line had to be extended.
3449    
3450    21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3451        CRLF as a newline sequence.
3452    
3453    22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3454        out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3455        I have nevertheless tidied it up.
3456    
3457    23. Added some casts to kill warnings from HP-UX ia64 compiler.
3458    
3459    24. Added a man page for pcre-config.
3460    
3461    
3462    Version 7.0 19-Dec-06
3463    ---------------------
3464    
3465     1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3466        moving to gcc 4.1.1.
3467    
3468     2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3469        sys/time.h, which is documented as needed for this function. It doesn't
3470        seem to matter on Linux, but it showed up on some releases of OS X.
3471    
3472     3. It seems that there are systems where bytes whose values are greater than
3473        127 match isprint() in the "C" locale. The "C" locale should be the
3474        default when a C program starts up. In most systems, only ASCII printing
3475        characters match isprint(). This difference caused the output from pcretest
3476        to vary, making some of the tests fail. I have changed pcretest so that:
3477    
3478        (a) When it is outputting text in the compiled version of a pattern, bytes
3479            other than 32-126 are always shown as hex escapes.
3480    
3481        (b) When it is outputting text that is a matched part of a subject string,
3482            it does the same, unless a different locale has been set for the match
3483            (using the /L modifier). In this case, it uses isprint() to decide.
3484    
3485     4. Fixed a major bug that caused incorrect computation of the amount of memory
3486        required for a compiled pattern when options that changed within the
3487        pattern affected the logic of the preliminary scan that determines the
3488        length. The relevant options are -x, and -i in UTF-8 mode. The result was
3489        that the computed length was too small. The symptoms of this bug were
3490        either the PCRE error "internal error: code overflow" from pcre_compile(),
3491        or a glibc crash with a message such as "pcretest: free(): invalid next
3492        size (fast)". Examples of patterns that provoked this bug (shown in
3493        pcretest format) are:
3494    
3495          /(?-x: )/x
3496          /(?x)(?-x: \s*#\s*)/
3497          /((?i)[\x{c0}])/8
3498          /(?i:[\x{c0}])/8
3499    
3500        HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3501        is now done differently.
3502    
3503     5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3504        wrapper classes; (b) implement a new function in the C++ scanner that is
3505        more efficient than the old way of doing things because it avoids levels of
3506        recursion in the regex matching; (c) add a paragraph to the documentation
3507        for the FullMatch() function.
3508    
3509     6. The escape sequence \n was being treated as whatever was defined as
3510        "newline". Not only was this contrary to the documentation, which states
3511        that \n is character 10 (hex 0A), but it also went horribly wrong when
3512        "newline" was defined as CRLF. This has been fixed.
3513    
3514     7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3515        was being set to -1 for the "end of line" case (supposedly a value that no
3516        character can have). Though this value is never used (the check for end of
3517        line is "zero bytes in current character"), it caused compiler complaints.
3518        I've changed it to 0xffffffff.
3519    
3520     8. In pcre_version.c, the version string was being built by a sequence of
3521        C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3522        string (as it is for production releases) called a macro with an empty
3523        argument. The C standard says the result of this is undefined. The gcc
3524        compiler treats it as an empty string (which was what was wanted) but it is
3525        reported that Visual C gives an error. The source has been hacked around to
3526        avoid this problem.
3527    
3528     9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3529        builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3530        instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3531        of them did).
3532    
3533    10. Originally, pcretest opened its input and output without "b"; then I was
3534        told that "b" was needed in some environments, so it was added for release
3535        5.0 to both the input and output. (It makes no difference on Unix-like
3536        systems.) Later I was told that it is wrong for the input on Windows. I've
3537        now abstracted the modes into two macros, to make it easier to fiddle with
3538        them, and removed "b" from the input mode under Windows.
3539    
3540    11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3541    
3542    12. Added -help and --help to pcretest as an official way of being reminded
3543        of the options.
3544    
3545    13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3546        and pcrecpp.cc because they annoy compilers at high warning levels.
3547    
3548    14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3549    
3550    15. Fixed an occurrence of == in configure.ac that should have been = (shell
3551        scripts are not C programs :-) and which was not noticed because it works
3552        on Linux.
3553    
3554    16. pcretest is supposed to handle any length of pattern and data line (as one
3555        line or as a continued sequence of lines) by extending its input buffer if
3556        necessary. This feature was broken for very long pattern lines, leading to
3557        a string of junk being passed to pcre_compile() if the pattern was longer
3558        than about 50K.
3559    
3560    17. I have done a major re-factoring of the way pcre_compile() computes the
3561        amount of memory needed for a compiled pattern. Previously, there was code
3562        that made a preliminary scan of the pattern in order to do this. That was
3563        OK when PCRE was new, but as the facilities have expanded, it has become
3564        harder and harder to keep it in step with the real compile phase, and there
3565        have been a number of bugs (see for example, 4 above). I have now found a
3566        cunning way of running the real compile function in a "fake" mode that
3567        enables it to compute how much memory it would need, while actually only
3568        ever using a few hundred bytes of working memory and without too many
3569        tests of the mode. This should make future maintenance and development
3570        easier. A side effect of this work is that the limit of 200 on the nesting
3571        depth of parentheses has been removed (though this was never a serious
3572        limitation, I suspect). However, there is a downside: pcre_compile() now
3573        runs more slowly than before (30% or more, depending on the pattern). I
3574        hope this isn't a big issue. There is no effect on runtime performance.
3575    
3576    18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3577        newline (only possible for the last line of a file) and it was a
3578        pattern that set a locale (followed by /Lsomething), pcretest crashed.
3579    
3580    19. Added additional timing features to pcretest. (1) The -tm option now times
3581        matching only, not compiling. (2) Both -t and -tm can be followed, as a
3582        separate command line item, by a number that specifies the number of
3583        repeats to use when timing. The default is 50000; this gives better
3584        precision, but takes uncomfortably long for very large patterns.
3585    
3586    20. Extended pcre_study() to be more clever in cases where a branch of a
3587        subpattern has no definite first character. For example, (a*|b*)[cd] would
3588        previously give no result from pcre_study(). Now it recognizes that the
3589        first character must be a, b, c, or d.