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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 503 - (hide annotations) (download)
Sun Mar 7 17:35:52 2010 UTC (4 years, 7 months ago) by ph10
File size: 184429 byte(s)
Fix incorrect compile time error for certain types of recursive patterns.

1 nigel 3 ChangeLog for PCRE
2     ------------------
3    
4 ph10 491 Version 8.02 01-Mar-2010
5     ------------------------
6    
7     1. The Unicode data tables have been updated to Unicode 5.2.0.
8    
9 ph10 492 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
10     configured.
11 ph10 493
12     3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
13     original author of that file, following a query about its status.
14 ph10 494
15     4. On systems that do not have stdint.h (e.g. Solaris), check for and include
16     inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
17 ph10 496
18     5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
19     quantifier applied to a forward-referencing subroutine call, could compile
20     incorrect code or give the error "internal error: previously-checked
21     referenced subpattern not found".
22 ph10 497
23     6. Both MS Visual Studio and Symbian OS have problems with initializing
24     variables to point to external functions. For these systems, therefore,
25     pcre_malloc etc. are now initialized to local functions that call the
26     relevant global functions.
27 ph10 498
28     7. There were two entries missing in the vectors called coptable and poptable
29     in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
30     I've fixed the data, and added a kludgy way of testing at compile time that
31     the lengths are correct (equal to the number of opcodes).
32 ph10 499
33     8. Following on from 7, I added a similar kludge to check the length of the
34     eint vector in pcreposix.c.
35    
36     9. Error texts for pcre_compile() are held as one long string to avoid too
37     much relocation at load time. To find a text, the string is searched,
38     counting zeros. There was no check for running off the end of the string,
39     which could happen if a new error number was added without updating the
40     string.
41 ph10 500
42     10. \K gave a compile-time error if it appeared in a lookbehind assersion.
43    
44     11. \K was not working if it appeared in an atomic group or in a group that
45     was called as a "subroutine", or in an assertion. Perl 5.11 documents that
46     \K is "not well defined" if used in an assertion. PCRE now accepts it if
47     the assertion is positive, but not if it is negative.
48 ph10 501
49     12. Change 11 fortuitously reduced the size of the stack frame used in the
50     "match()" function of pcre_exec.c by one pointer. Forthcoming
51     implementation of support for (*MARK) will need an extra pointer on the
52     stack; I have reserved it now, so that the stack frame size does not
53     decrease.
54 ph10 503
55     13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
56     item in branch that calls a recursion is a subroutine call - as in the
57     second branch in the above example - was incorrectly given the compile-
58     time error "recursive call could loop indefinitely" because pcre_compile()
59     was not correctly checking the subroutine for matching a non-empty string.
60 ph10 491
61 ph10 492
62 ph10 489 Version 8.01 19-Jan-2010
63     ------------------------
64 ph10 471
65     1. If a pattern contained a conditional subpattern with only one branch (in
66 ph10 489 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
67 ph10 473 computed the wrong minimum data length (which is of course zero for such
68 ph10 489 subpatterns). This could cause incorrect "no match" results.
69 ph10 473
70     2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
71 ph10 472 the pattern is reset in the first branch, pcre_compile() failed with
72     "internal error: code overflow at offset...". This happened only when
73 ph10 473 the reset was to the original external option setting. (An optimization
74     abstracts leading options settings into an external setting, which was the
75 ph10 472 cause of this.)
76 ph10 471
77 ph10 473 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
78     of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
79     assertion pattern did not match (meaning that the assertion was true), it
80     was incorrectly treated as false if the SKIP had been reached during the
81     matching. This also applied to assertions used as conditions.
82 ph10 471
83 ph10 473 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
84     assertion subpattern, including such a pattern used as a condition,
85     unpredictable results occurred, instead of the error return
86     PCRE_ERROR_DFA_UITEM.
87 ph10 478
88 ph10 474 5. The C++ GlobalReplace function was not working like Perl for the special
89     situation when an empty string is matched. It now does the fancy magic
90 ph10 478 stuff that is necessary.
91    
92     6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
93 ph10 475 removed. (These were left over from very, very early versions of PCRE.)
94 ph10 478
95 ph10 475 7. Some cosmetic changes to the code to make life easier when compiling it
96     as part of something else:
97 ph10 478
98     (a) Change DEBUG to PCRE_DEBUG.
99    
100     (b) In pcre_compile(), rename the member of the "branch_chain" structure
101     called "current" as "current_branch", to prevent a collision with the
102 ph10 475 Linux macro when compiled as a kernel module.
103 ph10 478
104     (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
105     prevent a collision with the Linux macro when compiled as a kernel
106 ph10 475 module.
107 ph10 478
108 ph10 475 8. In pcre_compile() there are some checks for integer overflows that used to
109     cast potentially large values to (double). This has been changed to that
110 ph10 478 when building, a check for int64_t is made, and if it is found, it is used
111     instead, thus avoiding the use of floating point arithmetic. (There is no
112     other use of FP in PCRE.) If int64_t is not found, the fallback is to
113     double.
114    
115     9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
116 ph10 476 2005 (difference between two addresses compared to an unsigned value).
117 ph10 473
118 ph10 478 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
119     custom one, because of the following reported problem in Windows:
120 ph10 473
121 ph10 478 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
122 ph10 489 under Win32.
123 ph10 478 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
124 ph10 489 therefore missing the function definition.
125 ph10 478 - The compiler thus generates a "C" signature for the test function.
126     - The linker fails to find the "C" function.
127     - PCRE fails to configure if asked to do so against libbz2.
128 ph10 486
129 ph10 479 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
130     messages were output:
131 ph10 478
132 ph10 479 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
133     rerunning libtoolize, to keep the correct libtool macros in-tree.
134     Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
135 ph10 486
136 ph10 479 I have done both of these things.
137 ph10 486
138 ph10 480 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
139 ph10 486 most of the time, it *can* run out if it is given a pattern that contains a
140     runaway infinite recursion. I updated the discussion in the pcrestack man
141     page.
142    
143     13. Now that we have gone to the x.xx style of version numbers, the minor
144     version may start with zero. Using 08 or 09 is a bad idea because users
145     might check the value of PCRE_MINOR in their code, and 08 or 09 may be
146     interpreted as invalid octal numbers. I've updated the previous comment in
147     configure.ac, and also added a check that gives an error if 08 or 09 are
148 ph10 481 used.
149 ph10 486
150 ph10 482 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
151 ph10 483 causing partial matching to fail when the end of the subject matched \W
152     in a UTF-8 pattern where \W was quantified with a minimum of 3.
153 ph10 486
154 ph10 483 15. There were some discrepancies between the declarations in pcre_internal.h
155 ph10 486 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
156 ph10 483 their definitions. The declarations used "const uschar *" and the
157     definitions used USPTR. Even though USPTR is normally defined as "const
158     unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
159     reported that: "This difference in casting confuses some C++ compilers, for
160     example, SunCC recognizes above declarations as different functions and
161 ph10 486 generates broken code for hbpcre." I have changed the declarations to use
162 ph10 483 USPTR.
163 ph10 486
164     16. GNU libtool is named differently on some systems. The autogen.sh script now
165 ph10 484 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
166     (FreeBSD).
167 ph10 478
168 ph10 486 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
169     (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
170     comment: "Figure out how to create a longlong from a string: strtoll and
171     equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
172     instance, but it only takes 2 args instead of 3!"
173 ph10 479
174 ph10 488 18. A subtle bug concerned with back references has been fixed by a change of
175     specification, with a corresponding code fix. A pattern such as
176     ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
177     refers, was giving matches when it shouldn't. For example, xa=xaaa would
178     match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
179     same bug. Such groups have to be quantified to be useful, or contained
180     inside another quantified group. (If there's no repetition, the reference
181     can never match.) The problem arises because, having left the group and
182     moved on to the rest of the pattern, a later failure that backtracks into
183     the group uses the captured value from the final iteration of the group
184     rather than the correct earlier one. I have fixed this in PCRE by forcing
185     any group that contains a reference to itself to be an atomic group; that
186     is, there cannot be any backtracking into it once it has completed. This is
187     similar to recursive and subroutine calls.
188 ph10 486
189 ph10 488
190 ph10 469 Version 8.00 19-Oct-09
191 ph10 418 ----------------------
192    
193     1. The table for translating pcre_compile() error codes into POSIX error codes
194 ph10 461 was out-of-date, and there was no check on the pcre_compile() error code
195     being within the table. This could lead to an OK return being given in
196 ph10 418 error.
197 ph10 461
198     2. Changed the call to open a subject file in pcregrep from fopen(pathname,
199     "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
200     in a Windows environment.
201    
202 ph10 420 3. The pcregrep --count option prints the count for each file even when it is
203     zero, as does GNU grep. However, pcregrep was also printing all files when
204     --files-with-matches was added. Now, when both options are given, it prints
205     counts only for those files that have at least one match. (GNU grep just
206 ph10 461 prints the file name in this circumstance, but including the count seems
207     more useful - otherwise, why use --count?) Also ensured that the
208 ph10 420 combination -clh just lists non-zero counts, with no names.
209 ph10 461
210     4. The long form of the pcregrep -F option was incorrectly implemented as
211     --fixed_strings instead of --fixed-strings. This is an incompatible change,
212     but it seems right to fix it, and I didn't think it was worth preserving
213     the old behaviour.
214    
215     5. The command line items --regex=pattern and --regexp=pattern were not
216 ph10 422 recognized by pcregrep, which required --regex pattern or --regexp pattern
217 ph10 461 (with a space rather than an '='). The man page documented the '=' forms,
218 ph10 422 which are compatible with GNU grep; these now work.
219 ph10 461
220     6. No libpcreposix.pc file was created for pkg-config; there was just
221 ph10 423 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
222 ph10 461
223 ph10 425 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
224 ph10 461 when UCP support is not needed, by modifying the Python script that
225 ph10 425 generates it from Unicode data files. This should not matter if the module
226     is correctly used as a library, but I received one complaint about 50K of
227     unwanted data. My guess is that the person linked everything into his
228     program rather than using a library. Anyway, it does no harm.
229 ph10 461
230 ph10 426 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
231 ph10 461 was a minimum greater than 1 for a wide character in a possessive
232 ph10 431 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
233     which had an unlimited repeat of a nested, fixed maximum repeat of a wide
234     character. Chaos in the form of incorrect output or a compiling loop could
235     result.
236 ph10 461
237 ph10 426 9. The restrictions on what a pattern can contain when partial matching is
238 ph10 461 requested for pcre_exec() have been removed. All patterns can now be
239 ph10 426 partially matched by this function. In addition, if there are at least two
240 ph10 435 slots in the offset vector, the offset of the earliest inspected character
241     for the match and the offset of the end of the subject are set in them when
242 ph10 461 PCRE_ERROR_PARTIAL is returned.
243    
244 ph10 428 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
245     synonymous with PCRE_PARTIAL, for backwards compatibility, and
246     PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
247 ph10 462 and may be more useful for multi-segment matching.
248 ph10 461
249     11. Partial matching with pcre_exec() is now more intuitive. A partial match
250     used to be given if ever the end of the subject was reached; now it is
251     given only if matching could not proceed because another character was
252     needed. This makes a difference in some odd cases such as Z(*FAIL) with the
253     string "Z", which now yields "no match" instead of "partial match". In the
254     case of pcre_dfa_exec(), "no match" is given if every matching path for the
255     final character ended with (*FAIL).
256    
257 ph10 428 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
258 ph10 461 if the pattern had a "must contain" character that was already found in the
259 ph10 428 earlier partial match, unless partial matching was again requested. For
260     example, with the pattern /dog.(body)?/, the "must contain" character is
261     "g". If the first part-match was for the string "dog", restarting with
262 ph10 435 "sbody" failed. This bug has been fixed.
263 ph10 461
264     13. The string returned by pcre_dfa_exec() after a partial match has been
265     changed so that it starts at the first inspected character rather than the
266     first character of the match. This makes a difference only if the pattern
267     starts with a lookbehind assertion or \b or \B (\K is not supported by
268     pcre_dfa_exec()). It's an incompatible change, but it makes the two
269 ph10 435 matching functions compatible, and I think it's the right thing to do.
270 ph10 461
271 ph10 435 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
272 ph10 461 so that the demonstration program is easily available in environments where
273     PCRE has not been installed from source.
274    
275 ph10 435 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
276 ph10 430 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
277     library.
278 ph10 461
279 ph10 435 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
280 ph10 432 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
281 ph10 461 is not the first non-POSIX option to be added. Clearly some people find
282 ph10 433 these options useful.
283 ph10 461
284     17. If a caller to the POSIX matching function regexec() passes a non-zero
285 ph10 438 value for nmatch with a NULL value for pmatch, the value of
286 ph10 461 nmatch is forced to zero.
287    
288 ph10 437 18. RunGrepTest did not have a test for the availability of the -u option of
289 ph10 461 the diff command, as RunTest does. It now checks in the same way as
290 ph10 437 RunTest, and also checks for the -b option.
291 ph10 461
292 ph10 438 19. If an odd number of negated classes containing just a single character
293     interposed, within parentheses, between a forward reference to a named
294 ph10 461 subpattern and the definition of the subpattern, compilation crashed with
295     an internal error, complaining that it could not find the referenced
296 ph10 438 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
297 ph10 461 [The bug was that it was starting one character too far in when skipping
298     over the character class, thus treating the ] as data rather than
299     terminating the class. This meant it could skip too much.]
300    
301 ph10 442 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
302 ph10 461 /g option in pcretest when the pattern contains \K, which makes it possible
303 ph10 442 to have an empty string match not at the start, even when the pattern is
304 ph10 461 anchored. Updated pcretest and pcredemo to use this option.
305    
306 ph10 446 21. If the maximum number of capturing subpatterns in a recursion was greater
307 ph10 461 than the maximum at the outer level, the higher number was returned, but
308     with unset values at the outer level. The correct (outer level) value is
309 ph10 446 now given.
310 ph10 461
311 ph10 447 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
312     PCRE did not set those parentheses (unlike Perl). I have now found a way to
313     make it do so. The string so far is captured, making this feature
314     compatible with Perl.
315 ph10 461
316     23. The tests have been re-organized, adding tests 11 and 12, to make it
317 ph10 448 possible to check the Perl 5.10 features against Perl 5.10.
318 ph10 461
319 ph10 454 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
320 ph10 461 pattern matches a fixed length string. PCRE did not allow this; now it
321     does. Neither allows recursion.
322    
323     25. I finally figured out how to implement a request to provide the minimum
324     length of subject string that was needed in order to match a given pattern.
325     (It was back references and recursion that I had previously got hung up
326     on.) This code has now been added to pcre_study(); it finds a lower bound
327 ph10 455 to the length of subject needed. It is not necessarily the greatest lower
328     bound, but using it to avoid searching strings that are too short does give
329     some useful speed-ups. The value is available to calling programs via
330     pcre_fullinfo().
331 ph10 461
332 ph10 455 26. While implementing 25, I discovered to my embarrassment that pcretest had
333     not been passing the result of pcre_study() to pcre_dfa_exec(), so the
334     study optimizations had never been tested with that matching function.
335     Oops. What is worse, even when it was passed study data, there was a bug in
336     pcre_dfa_exec() that meant it never actually used it. Double oops. There
337     were also very few tests of studied patterns with pcre_dfa_exec().
338 ph10 461
339 ph10 457 27. If (?| is used to create subpatterns with duplicate numbers, they are now
340     allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
341     on the other side of the coin, they are no longer allowed to have different
342     names, because these cannot be distinguished in PCRE, and this has caused
343     confusion. (This is a difference from Perl.)
344 ph10 461
345     28. When duplicate subpattern names are present (necessarily with different
346     numbers, as required by 27 above), and a test is made by name in a
347     conditional pattern, either for a subpattern having been matched, or for
348     recursion in such a pattern, all the associated numbered subpatterns are
349 ph10 459 tested, and the overall condition is true if the condition is true for any
350     one of them. This is the way Perl works, and is also more like the way
351     testing by number works.
352 ph10 418
353 ph10 461
354 ph10 412 Version 7.9 11-Apr-09
355 ph10 376 ---------------------
356    
357 ph10 392 1. When building with support for bzlib/zlib (pcregrep) and/or readline
358 ph10 376 (pcretest), all targets were linked against these libraries. This included
359     libpcre, libpcreposix, and libpcrecpp, even though they do not use these
360     libraries. This caused unwanted dependencies to be created. This problem
361 ph10 392 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
362 ph10 376 pcretest is linked with readline.
363 ph10 392
364 ph10 376 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
365     "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
366     moved outside it again, because FALSE and TRUE are already defined in AIX,
367     but BOOL is not.
368 ph10 392
369     3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
370 ph10 409 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
371 ph10 392
372 ph10 376 4. The pcregrep documentation said spaces were inserted as well as colons (or
373     hyphens) following file names and line numbers when outputting matching
374 ph10 392 lines. This is not true; no spaces are inserted. I have also clarified the
375 ph10 376 wording for the --colour (or --color) option.
376 ph10 392
377 ph10 378 5. In pcregrep, when --colour was used with -o, the list of matching strings
378     was not coloured; this is different to GNU grep, so I have changed it to be
379     the same.
380 ph10 392
381 ph10 378 6. When --colo(u)r was used in pcregrep, only the first matching substring in
382 ph10 392 each matching line was coloured. Now it goes on to look for further matches
383     of any of the test patterns, which is the same behaviour as GNU grep.
384    
385     7. A pattern that could match an empty string could cause pcregrep to loop; it
386     doesn't make sense to accept an empty string match in pcregrep, so I have
387 ph10 379 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
388     seems to be how GNU grep behaves.
389 ph10 392
390 ph10 380 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
391 ph10 392 start or after a newline", because the conditional assertion was not being
392     correctly handled. The rule now is that both the assertion and what follows
393     in the first alternative must satisfy the test.
394    
395 ph10 405 9. If auto-callout was enabled in a pattern with a conditional group whose
396 ph10 399 condition was an assertion, PCRE could crash during matching, both with
397     pcre_exec() and pcre_dfa_exec().
398 ph10 392
399     10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
400     used for matching.
401    
402     11. Unicode property support in character classes was not working for
403 ph10 385 characters (bytes) greater than 127 when not in UTF-8 mode.
404 ph10 388
405 ph10 392 12. Added the -M command line option to pcretest.
406    
407 ph10 388 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
408 ph10 389
409     15. Added the PCRE_NO_START_OPTIMIZE match-time option.
410 ph10 390
411 ph10 392 16. Added comments and documentation about mis-use of no_arg in the C++
412 ph10 390 wrapper.
413 ph10 392
414 ph10 391 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
415 ph10 392 from Martin Jerabek that uses macro names for all relevant character and
416 ph10 391 string constants.
417 ph10 405
418 ph10 393 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
419     SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
420 ph10 405 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
421     these, but not everybody uses configure.
422    
423     19. A conditional group that had only one branch was not being correctly
424     recognized as an item that could match an empty string. This meant that an
425     enclosing group might also not be so recognized, causing infinite looping
426     (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
427     with the subject "ab", where knowledge that the repeated group can match
428 ph10 396 nothing is needed in order to break the loop.
429 ph10 405
430 ph10 397 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
431 ph10 405 exec(), but without supplying a callout function, matching went wrong.
432    
433     21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
434     leak if the size of the offset vector was greater than 30. When the vector
435     is smaller, the saved offsets during recursion go onto a local stack
436     vector, but for larger vectors malloc() is used. It was failing to free
437     when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
438 ph10 400 error, in fact).
439 ph10 405
440 ph10 402 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
441 ph10 405 heapframe that is used only when UTF-8 support is enabled. This caused no
442     problem, but was untidy.
443 ph10 376
444 ph10 405 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
445     CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
446     included within another project.
447 ph10 411
448 ph10 407 24. Steven Van Ingelgem's patches to add more options to the CMake support,
449     slightly modified by me:
450 ph10 411
451     (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
452     not building pcregrep.
453    
454 ph10 407 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
455 ph10 411 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
456    
457 ph10 408 25. Forward references, both numeric and by name, in patterns that made use of
458     duplicate group numbers, could behave incorrectly or give incorrect errors,
459     because when scanning forward to find the reference group, PCRE was not
460     taking into account the duplicate group numbers. A pattern such as
461     ^X(?3)(a)(?|(b)|(q))(Y) is an example.
462 ph10 411
463     26. Changed a few more instances of "const unsigned char *" to USPTR, making
464     the feature of a custom pointer more persuasive (as requested by a user).
465 ph10 416
466 ph10 411 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
467 ph10 416 pcretest.c, inside #ifndefs, because it seems they are sometimes already
468     pre-defined.
469 ph10 392
470 ph10 416 28. Added support for (*UTF8) at the start of a pattern.
471    
472 ph10 413 29. Arrange for flags added by the "release type" setting in CMake to be shown
473     in the configuration summary.
474 ph10 405
475 ph10 411
476 ph10 374 Version 7.8 05-Sep-08
477 ph10 349 ---------------------
478    
479     1. Replaced UCP searching code with optimized version as implemented for Ad
480     Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
481 ph10 351 stage table and inline lookup instead of a function, giving speed ups of 2
482     to 5 times on some simple patterns that I tested. Permission was given to
483     distribute the MultiStage2.py script that generates the tables (it's not in
484     the tarball, but is in the Subversion repository).
485 ph10 349
486 ph10 351 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
487     scripts.
488 ph10 358
489 ph10 353 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
490     a group with a zero qualifier. The result of the study could be incorrect,
491 ph10 358 or the function might crash, depending on the pattern.
492    
493     4. Caseless matching was not working for non-ASCII characters in back
494 ph10 354 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
495 ph10 358 It now works when Unicode Property Support is available.
496    
497 ph10 355 5. In pcretest, an escape such as \x{de} in the data was always generating
498     a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
499     non-UTF-8 mode. If the value is greater than 255, it gives a warning about
500 ph10 358 truncation.
501 ph10 349
502 ph10 358 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
503    
504 ph10 357 7. Added two (int) casts to pcregrep when printing the difference of two
505     pointers, in case they are 64-bit values.
506 ph10 371
507     8. Added comments about Mac OS X stack usage to the pcrestack man page and to
508     test 2 if it fails.
509    
510 ph10 359 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
511 ph10 371 and a #define of that name to empty if it is not externally set. This is to
512     allow users of MSVC to set it if necessary.
513    
514     10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
515 ph10 359 the convenience functions in the pcre_get.c source file.
516 ph10 371
517 ph10 360 11. An option change at the start of a pattern that had top-level alternatives
518     could cause overwriting and/or a crash. This command provoked a crash in
519 ph10 371 some environments:
520    
521     printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
522    
523 ph10 360 This potential security problem was recorded as CVE-2008-2371.
524 ph10 371
525 ph10 361 12. For a pattern where the match had to start at the beginning or immediately
526     after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
527 ph10 371 pcre_dfa_exec() could read past the end of the passed subject if there was
528 ph10 361 no match. To help with detecting such bugs (e.g. with valgrind), I modified
529     pcretest so that it places the subject at the end of its malloc-ed buffer.
530 ph10 371
531 ph10 364 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
532 ph10 371 exec() might read past the end of the data buffer in UTF-8 mode.
533    
534 ph10 364 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
535 ph10 371 the data contained the byte 0x85 as part of a UTF-8 character within its
536     first line. This applied both to normal and DFA matching.
537    
538 ph10 366 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
539 ph10 371 /^[^d]*?$/8 failed to match "abc".
540 ph10 351
541 ph10 371 16. Added a missing copyright notice to pcrecpp_internal.h.
542    
543     17. Make it more clear in the documentation that values returned from
544 ph10 368 pcre_exec() in ovector are byte offsets, not character counts.
545 ph10 357
546 ph10 371 18. Tidied a few places to stop certain compilers from issuing warnings.
547 ph10 368
548 ph10 373 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
549     supplied by Stefan Weber. I made a further small update for 7.8 because
550 ph10 374 there is a change of source arrangements: the pcre_searchfuncs.c module is
551     replaced by pcre_ucd.c.
552 ph10 371
553 ph10 373
554 ph10 347 Version 7.7 07-May-08
555 ph10 321 ---------------------
556    
557     1. Applied Craig's patch to sort out a long long problem: "If we can't convert
558 ph10 345 a string to a long long, pretend we don't even have a long long." This is
559 ph10 321 done by checking for the strtoq, strtoll, and _strtoi64 functions.
560 ph10 345
561 ph10 322 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
562     pre-7.6 versions, which defined a global no_arg variable instead of putting
563 ph10 332 it in the RE class. (See also #8 below.)
564 ph10 345
565     3. Remove a line of dead code, identified by coverity and reported by Nuno
566     Lopes.
567    
568 ph10 324 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
569 ph10 321
570 ph10 324 (1) The include/exclude patterns were being applied to the whole pathnames
571 ph10 345 of files, instead of just to the final components.
572    
573 ph10 324 (2) If there was more than one level of directory, the subdirectories were
574     skipped unless they satisfied the include/exclude conditions. This is
575     inconsistent with GNU grep (and could even be seen as contrary to the
576     pcregrep specification - which I improved to make it absolutely clear).
577     The action now is always to scan all levels of directory, and just
578     apply the include/exclude patterns to regular files.
579 ph10 345
580 ph10 325 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
581 ph10 345 --exclude_dir in the tests to avoid scanning .svn directories.
582    
583 ph10 326 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
584 ph10 345 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
585 ph10 326 doesn't support NULs in patterns.
586 ph10 345
587     7. Added some missing "const"s to declarations of static tables in
588     pcre_compile.c and pcre_dfa_exec.c.
589    
590 ph10 328 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
591 ph10 329 caused by fix #2 above. (Subsequently also a second patch to fix the
592 ph10 345 first patch. And a third patch - this was a messy problem.)
593 ph10 321
594 ph10 345 9. Applied Craig's patch to remove the use of push_back().
595    
596 ph10 332 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
597     matching function regexec().
598 ph10 345
599     11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
600     which, however, unlike Perl's \g{...}, are subroutine calls, not back
601 ph10 333 references. PCRE supports relative numbers with this syntax (I don't think
602     Oniguruma does).
603 ph10 345
604     12. Previously, a group with a zero repeat such as (...){0} was completely
605 ph10 335 omitted from the compiled regex. However, this means that if the group
606     was called as a subroutine from elsewhere in the pattern, things went wrong
607 ph10 345 (an internal error was given). Such groups are now left in the compiled
608     pattern, with a new opcode that causes them to be skipped at execution
609 ph10 335 time.
610 ph10 345
611 ph10 341 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
612     to the way PCRE behaves:
613 ph10 345
614 ph10 336 (a) A lone ] character is dis-allowed (Perl treats it as data).
615 ph10 345
616     (b) A back reference to an unmatched subpattern matches an empty string
617 ph10 336 (Perl fails the current match path).
618 ph10 345
619 ph10 341 (c) A data ] in a character class must be notated as \] because if the
620 ph10 345 first data character in a class is ], it defines an empty class. (In
621 ph10 341 Perl it is not possible to have an empty class.) The empty class []
622     never matches; it forces failure and is equivalent to (*FAIL) or (?!).
623     The negative empty class [^] matches any one character, independently
624     of the DOTALL setting.
625 ph10 345
626     14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
627 ph10 340 non-existent subpattern following a character class starting with ']' and
628     containing () gave an internal compiling error instead of "reference to
629     non-existent subpattern". Fortunately, when the pattern did exist, the
630 ph10 345 compiled code was correct. (When scanning forwards to check for the
631 ph10 340 existencd of the subpattern, it was treating the data ']' as terminating
632 ph10 345 the class, so got the count wrong. When actually compiling, the reference
633 ph10 340 was subsequently set up correctly.)
634 ph10 345
635 ph10 341 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
636 ph10 345 it was being rejected as not supported by pcre_dfa_exec(), even though
637     other assertions are supported. I have made pcre_dfa_exec() support
638     (*FAIL).
639    
640 ph10 342 16. The implementation of 13c above involved the invention of a new opcode,
641 ph10 345 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
642     cannot be changed at match time, I realized I could make a small
643     improvement to matching performance by compiling OP_ALLANY instead of
644     OP_ANY for "." when DOTALL was set, and then removing the runtime tests
645     on the OP_ANY path.
646    
647     17. Compiling pcretest on Windows with readline support failed without the
648 ph10 343 following two fixes: (1) Make the unistd.h include conditional on
649 ph10 345 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
650    
651 ph10 344 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
652 ph10 345 ncurses library to be included for pcretest when ReadLine support is
653     requested, but also to allow for it to be overridden. This patch came from
654     Daniel Bergström.
655 ph10 347
656 ph10 346 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
657 ph10 347 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
658     any errors with the current Unicode tables. Thanks to Peter Kankowski for
659     spotting this.
660 ph10 324
661 ph10 332
662 ph10 319 Version 7.6 28-Jan-08
663 ph10 300 ---------------------
664    
665 ph10 302 1. A character class containing a very large number of characters with
666     codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
667     overflow.
668 ph10 309
669     2. Patch to cut out the "long long" test in pcrecpp_unittest when
670     HAVE_LONG_LONG is not defined.
671    
672 ph10 303 3. Applied Christian Ehrlicher's patch to update the CMake build files to
673 ph10 304 bring them up to date and include new features. This patch includes:
674 ph10 309
675 ph10 304 - Fixed PH's badly added libz and libbz2 support.
676     - Fixed a problem with static linking.
677 ph10 312 - Added pcredemo. [But later removed - see 7 below.]
678 ph10 304 - Fixed dftables problem and added an option.
679     - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
680     HAVE_LONG_LONG.
681     - Added readline support for pcretest.
682 ph10 309 - Added an listing of the option settings after cmake has run.
683    
684 ph10 314 4. A user submitted a patch to Makefile that makes it easy to create
685     "pcre.dll" under mingw when using Configure/Make. I added stuff to
686     Makefile.am that cause it to include this special target, without
687     affecting anything else. Note that the same mingw target plus all
688     the other distribution libraries and programs are now supported
689     when configuring with CMake (see 6 below) instead of with
690     Configure/Make.
691 ph10 309
692 ph10 308 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
693     This is an attempt to solve the reported problem "pcrecpp::no_arg is not
694 ph10 309 exported in the Windows port". It has not yet been confirmed that the patch
695     solves the problem, but it does no harm.
696 ph10 313
697 ph10 311 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
698 ph10 319 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
699 ph10 314 with CMake, and also correct the comment about stack recursion.
700 ph10 313
701 ph10 312 7. Remove the automatic building of pcredemo from the ./configure system and
702 ph10 313 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
703     of a program that users should build themselves after PCRE is installed, so
704     building it automatically is not really right. What is more, it gave
705 ph10 312 trouble in some build environments.
706 ph10 300
707 ph10 319 8. Further tidies to CMakeLists.txt from Sheri and Christian.
708 ph10 308
709 ph10 319
710 ph10 298 Version 7.5 10-Jan-08
711 ph10 263 ---------------------
712    
713     1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
714     values in parens when parsing an RE using the C++ wrapper."
715 ph10 286
716 ph10 264 2. Negative specials like \S did not work in character classes in UTF-8 mode.
717     Characters greater than 255 were excluded from the class instead of being
718     included.
719 ph10 286
720     3. The same bug as (2) above applied to negated POSIX classes such as
721 ph10 265 [:^space:].
722 ph10 286
723 ph10 267 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
724     defined or documented. It seems to have been a typo for PCRE_STATIC, so
725 ph10 286 I have changed it.
726    
727 ph10 268 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
728 ph10 286 first named subpattern) and a construct such as (?&a) would reference the
729     first named subpattern whose name started with "a" (in other words, the
730 ph10 272 length check was missing). Both these problems are fixed. "Subpattern name
731     expected" is now given for (?&) (a zero-length name), and this patch also
732     makes it give the same error for \k'' (previously it complained that that
733     was a reference to a non-existent subpattern).
734 ph10 286
735 ph10 269 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
736 ph10 286 this is right because (?- can be followed by option settings as well as by
737 ph10 269 digits. I have, however, made the messages clearer.
738 ph10 286
739 ph10 270 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
740     than the number used in the conditional) now cause a compile-time error.
741     This is actually not compatible with Perl, which accepts such patterns, but
742     treats the conditional as always being FALSE (as PCRE used to), but it
743     seems to me that giving a diagnostic is better.
744 ph10 286
745 ph10 274 8. Change "alphameric" to the more common word "alphanumeric" in comments
746 ph10 275 and messages.
747 ph10 286
748     9. Fix two occurrences of "backslash" in comments that should have been
749     "backspace".
750    
751     10. Remove two redundant lines of code that can never be obeyed (their function
752     was moved elsewhere).
753    
754     11. The program that makes PCRE's Unicode character property table had a bug
755     which caused it to generate incorrect table entries for sequences of
756 ph10 277 characters that have the same character type, but are in different scripts.
757 ph10 286 It amalgamated them into a single range, with the script of the first of
758 ph10 277 them. In other words, some characters were in the wrong script. There were
759     thirteen such cases, affecting characters in the following ranges:
760 ph10 286
761 ph10 277 U+002b0 - U+002c1
762     U+0060c - U+0060d
763 ph10 286 U+0061e - U+00612
764 ph10 277 U+0064b - U+0065e
765     U+0074d - U+0076d
766     U+01800 - U+01805
767     U+01d00 - U+01d77
768     U+01d9b - U+01dbf
769     U+0200b - U+0200f
770     U+030fc - U+030fe
771     U+03260 - U+0327f
772     U+0fb46 - U+0fbb1
773     U+10450 - U+1049d
774 ph10 286
775 ph10 279 12. The -o option (show only the matching part of a line) for pcregrep was not
776     compatible with GNU grep in that, if there was more than one match in a
777     line, it showed only the first of them. It now behaves in the same way as
778     GNU grep.
779 ph10 286
780     13. If the -o and -v options were combined for pcregrep, it printed a blank
781     line for every non-matching line. GNU grep prints nothing, and pcregrep now
782     does the same. The return code can be used to tell if there were any
783     non-matching lines.
784    
785 ph10 289 14. Added --file-offsets and --line-offsets to pcregrep.
786    
787     15. The pattern (?=something)(?R) was not being diagnosed as a potentially
788 ph10 282 infinitely looping recursion. The bug was that positive lookaheads were not
789 ph10 286 being skipped when checking for a possible empty match (negative lookaheads
790     and both kinds of lookbehind were skipped).
791    
792 ph10 289 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
793 ph10 284 inclusion of <windows.h> to before rather than after the definition of
794     INVALID_FILE_ATTRIBUTES (patch from David Byron).
795 ph10 263
796 ph10 289 17. Specifying a possessive quantifier with a specific limit for a Unicode
797 ph10 286 character property caused pcre_compile() to compile bad code, which led at
798     runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
799     are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
800     caused the error; without that there was no problem.
801 ph10 263
802 ph10 289 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
803 ph10 286
804 ph10 289 19. Added --enable-pcretest-libreadline.
805 ph10 286
806 ph10 289 20. In pcrecpp.cc, the variable 'count' was incremented twice in
807 ph10 288 RE::GlobalReplace(). As a result, the number of replacements returned was
808 ph10 298 double what it should be. I removed one of the increments, but Craig sent a
809     later patch that removed the other one (the right fix) and added unit tests
810     that check the return values (which was not done before).
811 ph10 292
812 ph10 291 21. Several CMake things:
813 ph10 287
814 ph10 291 (1) Arranged that, when cmake is used on Unix, the libraries end up with
815 ph10 292 the names libpcre and libpcreposix, not just pcre and pcreposix.
816 ph10 288
817 ph10 292 (2) The above change means that pcretest and pcregrep are now correctly
818     linked with the newly-built libraries, not previously installed ones.
819 ph10 291
820 ph10 292 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
821 ph10 298
822     22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
823     crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
824     UTF-8 newline character). The key issue is that the pattern starts .*;
825     this means that the match must be either at the beginning, or after a
826     newline. The bug was in the code for advancing after a failed match and
827     checking that the new position followed a newline. It was not taking
828 ph10 294 account of UTF-8 characters correctly.
829 ph10 298
830     23. PCRE was behaving differently from Perl in the way it recognized POSIX
831     character classes. PCRE was not treating the sequence [:...:] as a
832     character class unless the ... were all letters. Perl, however, seems to
833     allow any characters between [: and :], though of course it rejects as
834     unknown any "names" that contain non-letters, because all the known class
835     names consist only of letters. Thus, Perl gives an error for [[:1234:]],
836     for example, whereas PCRE did not - it did not recognize a POSIX character
837     class. This seemed a bit dangerous, so the code has been changed to be
838     closer to Perl. The behaviour is not identical to Perl, because PCRE will
839     diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
840     treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
841     Perl does, and where it didn't before.
842    
843 ph10 296 24. Rewrite so as to remove the single use of %n from pcregrep because in some
844     Windows environments %n is disabled by default.
845 ph10 292
846    
847 ph10 260 Version 7.4 21-Sep-07
848 ph10 230 ---------------------
849    
850     1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
851 ph10 231 means that a class such as [\s] counted as "explicit reference to CR or
852     LF". That isn't really right - the whole point of the change was to try to
853     help when there was an actual mention of one of the two characters. So now
854     the change happens only if \r or \n (or a literal CR or LF) character is
855 ph10 230 encountered.
856 ph10 231
857     2. The 32-bit options word was also used for 6 internal flags, but the numbers
858     of both had grown to the point where there were only 3 bits left.
859     Fortunately, there was spare space in the data structure, and so I have
860     moved the internal flags into a new 16-bit field to free up more option
861 ph10 230 bits.
862 ph10 231
863     3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
864     but did not set the internal JCHANGED flag - either of these is enough to
865     control the way the "get" function works - but the PCRE_INFO_JCHANGED
866     facility is supposed to tell if (?J) was ever used, so now (?J) at the
867 ph10 230 start sets both bits.
868    
869 ph10 231 4. Added options (at build time, compile time, exec time) to change \R from
870     matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
871 ph10 230
872 ph10 243 5. doc/pcresyntax.html was missing from the distribution.
873    
874     6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
875 ph10 233 compatibility, even though it is no longer used.
876 ph10 243
877     7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
878 ph10 254 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
879 ph10 259 windows.h file is present (where different names are used). [This was
880 ph10 257 reversed later after testing - see 16 below.]
881 ph10 243
882     8. Changed all #include <config.h> to #include "config.h". There were also
883 ph10 236 some further <pcre.h> cases that I changed to "pcre.h".
884 ph10 243
885 ph10 239 9. When pcregrep was used with the --colour option, it missed the line ending
886 ph10 243 sequence off the lines that it output.
887    
888     10. It was pointed out to me that arrays of string pointers cause lots of
889     relocations when a shared library is dynamically loaded. A technique of
890     using a single long string with a table of offsets can drastically reduce
891     these. I have refactored PCRE in four places to do this. The result is
892 ph10 240 dramatic:
893 ph10 243
894 ph10 240 Originally: 290
895     After changing UCP table: 187
896 ph10 243 After changing error message table: 43
897 ph10 240 After changing table of "verbs" 36
898     After changing table of Posix names 22
899 ph10 243
900 ph10 240 Thanks to the folks working on Gregex for glib for this insight.
901 ph10 247
902 ph10 244 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
903 ph10 247 unicode-properties was also set.
904    
905 ph10 248 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
906 ph10 231
907 ph10 253 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
908     checked only for CRLF.
909 ph10 233
910 ph10 259 14. Added casts to pcretest.c to avoid compiler warnings.
911    
912 ph10 256 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
913 ph10 253
914 ph10 257 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
915 ph10 259 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
916     entirely. This removes changes made in 7 above.
917 ph10 256
918 ph10 261 17. The CMake files have been updated, and there is now more information about
919     building with CMake in the NON-UNIX-USE document.
920 ph10 257
921 ph10 261
922 ph10 228 Version 7.3 28-Aug-07
923 ph10 157 ---------------------
924    
925 ph10 189 1. In the rejigging of the build system that eventually resulted in 7.1, the
926     line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
927     brackets there is not right, since it causes compilers to look for an
928     installed pcre.h, not the version that is in the source that is being
929     compiled (which of course may be different). I have changed it back to:
930 ph10 197
931 ph10 189 #include "pcre.h"
932 ph10 197
933     I have a vague recollection that the change was concerned with compiling in
934     different directories, but in the new build system, that is taken care of
935     by the VPATH setting the Makefile.
936    
937 ph10 190 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
938     when the subject happened to end in the byte 0x85 (e.g. if the last
939     character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
940     characters but of course it shouldn't be taken as a newline when it is part
941     of another character. The bug was that, for an unlimited repeat of . in
942     not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
943     characters when looking for a newline.
944 ph10 197
945     3. A small performance improvement in the DOTALL UTF-8 mode .* case.
946    
947     4. Debugging: adjusted the names of opcodes for different kinds of parentheses
948     in debug output.
949    
950 ph10 201 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
951     long printing in the pcrecpp unittest when running under MinGW.
952 ph10 189
953 ph10 197 6. ESC_K was left out of the EBCDIC table.
954 ph10 189
955 ph10 197 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
956     parentheses; I made it 1000, which seemed large enough. Unfortunately, the
957     limit also applies to "virtual nesting" when a pattern is recursive, and in
958     this case 1000 isn't so big. I have been able to remove this limit at the
959     expense of backing off one optimization in certain circumstances. Normally,
960     when pcre_exec() would call its internal match() function recursively and
961     immediately return the result unconditionally, it uses a "tail recursion"
962     feature to save stack. However, when a subpattern that can match an empty
963     string has an unlimited repetition quantifier, it no longer makes this
964     optimization. That gives it a stack frame in which to save the data for
965     checking that an empty string has been matched. Previously this was taken
966     from the 1000-entry workspace that had been reserved. So now there is no
967     explicit limit, but more stack is used.
968 ph10 189
969 ph10 200 8. Applied Daniel's patches to solve problems with the import/export magic
970     syntax that is required for Windows, and which was going wrong for the
971     pcreposix and pcrecpp parts of the library. These were overlooked when this
972     problem was solved for the main library.
973 ph10 197
974 ph10 202 9. There were some crude static tests to avoid integer overflow when computing
975     the size of patterns that contain repeated groups with explicit upper
976     limits. As the maximum quantifier is 65535, the maximum group length was
977     set at 30,000 so that the product of these two numbers did not overflow a
978     32-bit integer. However, it turns out that people want to use groups that
979     are longer than 30,000 bytes (though not repeat them that many times).
980     Change 7.0/17 (the refactoring of the way the pattern size is computed) has
981     made it possible to implement the integer overflow checks in a much more
982     dynamic way, which I have now done. The artificial limitation on group
983     length has been removed - we now have only the limit on the total length of
984     the compiled pattern, which depends on the LINK_SIZE setting.
985 ph10 208
986     10. Fixed a bug in the documentation for get/copy named substring when
987     duplicate names are permitted. If none of the named substrings are set, the
988     functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
989     empty string.
990    
991     11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
992     instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
993     because the ] is interpreted as the first data character and the
994     terminating ] is not found. PCRE has been made compatible with Perl in this
995     regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
996     cause memory overwriting.
997    
998 ph10 206 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
999     string has been matched (to stop an infinite loop). It was not recognizing
1000 ph10 208 a conditional subpattern that could match an empty string if that
1001 ph10 206 subpattern was within another subpattern. For example, it looped when
1002 ph10 208 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
1003 ph10 206 condition was not nested. This bug has been fixed.
1004 ph10 208
1005 ph10 207 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
1006     past the start of the subject in the presence of bytes with the top bit
1007     set, for example "\x8aBCD".
1008 ph10 211
1009 ph10 210 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
1010     (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
1011 ph10 200
1012 ph10 211 14. Optimized (?!) to (*FAIL).
1013 ph10 202
1014 ph10 212 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
1015     This restricts code points to be within the range 0 to 0x10FFFF, excluding
1016     the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
1017     full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
1018     does: it's just the validity check that is more restrictive.
1019 ph10 220
1020     16. Inserted checks for integer overflows during escape sequence (backslash)
1021     processing, and also fixed erroneous offset values for syntax errors during
1022     backslash processing.
1023    
1024 ph10 214 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
1025 ph10 220 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
1026    
1027 ph10 215 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
1028     caused an overrun.
1029 ph10 220
1030     19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
1031     something other than just ASCII characters) inside a group that had an
1032     unlimited repeat caused a loop at compile time (while checking to see
1033     whether the group could match an empty string).
1034    
1035 ph10 217 20. Debugging a pattern containing \p or \P could cause a crash. For example,
1036     [\P{Any}] did so. (Error in the code for printing property names.)
1037 ph10 210
1038 ph10 220 21. An orphan \E inside a character class could cause a crash.
1039    
1040     22. A repeated capturing bracket such as (A)? could cause a wild memory
1041 ph10 218 reference during compilation.
1042 ph10 220
1043     23. There are several functions in pcre_compile() that scan along a compiled
1044     expression for various reasons (e.g. to see if it's fixed length for look
1045 ph10 218 behind). There were bugs in these functions when a repeated \p or \P was
1046 ph10 220 present in the pattern. These operators have additional parameters compared
1047     with \d, etc, and these were not being taken into account when moving along
1048 ph10 218 the compiled data. Specifically:
1049 ph10 220
1050     (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
1051     length.
1052    
1053     (b) An item such as \pL+ within a repeated group could cause crashes or
1054 ph10 218 loops.
1055 ph10 220
1056 ph10 218 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
1057 ph10 220 "reference to non-existent subpattern" error.
1058    
1059 ph10 221 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
1060    
1061 ph10 219 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
1062 ph10 220 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
1063 ph10 211
1064 ph10 222 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
1065 ph10 218
1066 ph10 227 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
1067 ph10 223 character were causing crashes (broken optimization).
1068 ph10 227
1069 ph10 224 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
1070 ph10 227 \p or \P) caused a compile-time loop.
1071    
1072 ph10 226 28. More problems have arisen in unanchored patterns when CRLF is a valid line
1073     break. For example, the unstudied pattern [\r\n]A does not match the string
1074     "\r\nA" because change 7.0/46 below moves the current point on by two
1075     characters after failing to match at the start. However, the pattern \nA
1076 ph10 227 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
1077     the same is true. There doesn't seem any very clean way out of this, but
1078     what I have chosen to do makes the common cases work: PCRE now takes note
1079     of whether there can be an explicit match for \r or \n anywhere in the
1080     pattern, and if so, 7.0/46 no longer applies. As part of this change,
1081     there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
1082     pattern has explicit CR or LF references.
1083 ph10 222
1084 ph10 227 29. Added (*CR) etc for changing newline setting at start of pattern.
1085 ph10 223
1086 ph10 227
1087 ph10 189 Version 7.2 19-Jun-07
1088     ---------------------
1089    
1090 ph10 157 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
1091     which is apparently normally available under Windows.
1092 ph10 159
1093     2. Re-jig the pcregrep tests with different newline settings in an attempt
1094     to make them independent of the local environment's newline setting.
1095 ph10 157
1096 ph10 160 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
1097 ph10 165
1098 ph10 161 4. Some of the "internals" tests were previously cut out when the link size
1099     was not 2, because the output contained actual offsets. The recent new
1100 ph10 165 "Z" feature of pcretest means that these can be cut out, making the tests
1101     usable with all link sizes.
1102    
1103 ph10 164 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
1104     stack recursion. This gives a massive performance boost under BSD, but just
1105 ph10 165 a small improvement under Linux. However, it saves one field in the frame
1106 ph10 164 in all cases.
1107 ph10 172
1108 ph10 166 6. Added more features from the forthcoming Perl 5.10:
1109 ph10 172
1110 ph10 166 (a) (?-n) (where n is a string of digits) is a relative subroutine or
1111     recursion call. It refers to the nth most recently opened parentheses.
1112 ph10 172
1113 ph10 166 (b) (?+n) is also a relative subroutine call; it refers to the nth next
1114 ph10 172 to be opened parentheses.
1115    
1116     (c) Conditions that refer to capturing parentheses can be specified
1117 ph10 167 relatively, for example, (?(-2)... or (?(+3)...
1118 ph10 172
1119 ph10 168 (d) \K resets the start of the current match so that everything before
1120 ph10 172 is not part of it.
1121    
1122 ph10 171 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
1123 ph10 172
1124 ph10 171 (f) \g{name} is another synonym - part of Perl 5.10's unification of
1125 ph10 172 reference syntax.
1126 ph10 182
1127 ph10 175 (g) (?| introduces a group in which the numbering of parentheses in each
1128 ph10 182 alternative starts with the same number.
1129 ph10 172
1130 ph10 182 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
1131    
1132 ph10 172 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
1133     PCRE_INFO_JCHANGED.
1134    
1135     8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
1136     terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
1137 ph10 170 for detecting groups that can match an empty string.
1138 ph10 159
1139 ph10 172 9. A pattern with a very large number of alternatives (more than several
1140     hundred) was running out of internal workspace during the pre-compile
1141     phase, where pcre_compile() figures out how much memory will be needed. A
1142     bit of new cunning has reduced the workspace needed for groups with
1143     alternatives. The 1000-alternative test pattern now uses 12 bytes of
1144     workspace instead of running out of the 4096 that are available.
1145 ph10 182
1146 ph10 176 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
1147 ph10 172
1148 ph10 179 11. Applied patch from Google to remove an optimization that didn't quite work.
1149     The report of the bug said:
1150 ph10 182
1151 ph10 179 pcrecpp::RE("a*").FullMatch("aaa") matches, while
1152     pcrecpp::RE("a*?").FullMatch("aaa") does not, and
1153 ph10 182 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
1154 ph10 185
1155 ph10 184 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
1156 ph10 185 it matched the wrong number of bytes.
1157 ph10 172
1158 ph10 179
1159 ph10 155 Version 7.1 24-Apr-07
1160 ph10 98 ---------------------
1161    
1162 ph10 111 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
1163 ph10 122 that is more "standard", making use of automake and other Autotools. There
1164 ph10 99 is some re-arrangement of the files and adjustment of comments consequent
1165     on this.
1166 ph10 111
1167     2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
1168     for recursive directory scanning broke on some systems because the files
1169     are not scanned in any specific order and on different systems the order
1170     was different. A call to "sort" has been inserted into RunGrepTest for the
1171     approprate test as a short-term fix. In the longer term there may be an
1172 ph10 100 alternative.
1173 ph10 111
1174 ph10 100 3. I had an email from Eric Raymond about problems translating some of PCRE's
1175 ph10 111 man pages to HTML (despite the fact that I distribute HTML pages, some
1176     people do their own conversions for various reasons). The problems
1177     concerned the use of low-level troff macros .br and .in. I have therefore
1178     removed all such uses from the man pages (some were redundant, some could
1179 ph10 113 be replaced by .nf/.fi pairs). The 132html script that I use to generate
1180     HTML has been updated to handle .nf/.fi and to complain if it encounters
1181     .br or .in.
1182 ph10 111
1183 ph10 100 4. Updated comments in configure.ac that get placed in config.h.in and also
1184 ph10 123 arranged for config.h to be included in the distribution, with the name
1185 ph10 111 config.h.generic, for the benefit of those who have to compile without
1186     Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
1187    
1188     5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
1189     Weber: (1) pcre_internal.h was missing some function renames; (2) updated
1190 ph10 127 makevp.bat for the current PCRE, using the additional files
1191 ph10 135 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
1192 ph10 111
1193     6. A Windows user reported a minor discrepancy with test 2, which turned out
1194     to be caused by a trailing space on an input line that had got lost in his
1195 ph10 102 copy. The trailing space was an accident, so I've just removed it.
1196 ph10 111
1197 ph10 104 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
1198 ph10 111 that is needed.
1199    
1200 ph10 105 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
1201     as "const" (a) because they are and (b) because it helps the PHP
1202     maintainers who have recently made a script to detect big data structures
1203 ph10 111 in the php code that should be moved to the .rodata section. I remembered
1204     to update Builducptable as well, so it won't revert if ucptable.h is ever
1205 ph10 105 re-created.
1206 ph10 111
1207     9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
1208 ph10 107 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
1209 ph10 111 order to be able to cut out the UTF-8 tables in the latter when UTF-8
1210     support is not required. This saves 1.5-2K of code, which is important in
1211 ph10 107 some applications.
1212 ph10 111
1213 ph10 110 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
1214 ph10 111 so as not to refer to the tables, even though these functions will never be
1215     called when UTF-8 support is disabled. Otherwise there are problems with a
1216     shared library.
1217    
1218 ph10 118 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
1219    
1220     (a) It was defining its arguments as char * instead of void *.
1221    
1222 ph10 123 (b) It was assuming that all moves were upwards in memory; this was true
1223     a long time ago when I wrote it, but is no longer the case.
1224    
1225 ph10 118 The emulated memove() is provided for those environments that have neither
1226 ph10 123 memmove() nor bcopy(). I didn't think anyone used it these days, but that
1227 ph10 118 is clearly not the case, as these two bugs were recently reported.
1228 ph10 123
1229 ph10 111 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
1230 ph10 123 and Detrail to create the HTML documentation, the .txt form of the man
1231     pages, and it removes trailing spaces from listed files. It also creates
1232     pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
1233     case, it wraps all the #defines with #ifndefs. This script should be run
1234 ph10 111 before "make dist".
1235 ph10 123
1236 ph10 115 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
1237     with Unicode property support.
1238 ph10 123
1239     (a) For a maximizing quantifier, if the two different cases of the
1240     character were of different lengths in their UTF-8 codings (there are
1241     some cases like this - I found 11), and the matching function had to
1242 ph10 115 back up over a mixture of the two cases, it incorrectly assumed they
1243     were both the same length.
1244 ph10 123
1245     (b) When PCRE was configured to use the heap rather than the stack for
1246     recursion during matching, it was not correctly preserving the data for
1247     the other case of a UTF-8 character when checking ahead for a match
1248     while processing a minimizing repeat. If the check also involved
1249 ph10 115 matching a wide character, but failed, corruption could cause an
1250     erroneous result when trying to check for a repeat of the original
1251     character.
1252 ph10 123
1253 ph10 116 13. Some tidying changes to the testing mechanism:
1254 ph10 98
1255 ph10 116 (a) The RunTest script now detects the internal link size and whether there
1256     is UTF-8 and UCP support by running ./pcretest -C instead of relying on
1257 ph10 123 values substituted by "configure". (The RunGrepTest script already did
1258     this for UTF-8.) The configure.ac script no longer substitutes the
1259     relevant variables.
1260    
1261 ph10 116 (b) The debugging options /B and /D in pcretest show the compiled bytecode
1262     with length and offset values. This means that the output is different
1263     for different internal link sizes. Test 2 is skipped for link sizes
1264     other than 2 because of this, bypassing the problem. Unfortunately,
1265     there was also a test in test 3 (the locale tests) that used /B and
1266 ph10 123 failed for link sizes other than 2. Rather than cut the whole test out,
1267     I have added a new /Z option to pcretest that replaces the length and
1268     offset values with spaces. This is now used to make test 3 independent
1269 ph10 122 of link size. (Test 2 will be tidied up later.)
1270 ph10 123
1271     14. If erroroffset was passed as NULL to pcre_compile, it provoked a
1272 ph10 122 segmentation fault instead of returning the appropriate error message.
1273 ph10 134
1274 ph10 131 15. In multiline mode when the newline sequence was set to "any", the pattern
1275 ph10 134 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
1276     This doesn't seem right; it now treats the CRLF combination as the line
1277     ending, and so does not match in that case. It's only a pattern such as ^$
1278     that would hit this one: something like ^ABC$ would have failed after \r
1279     and then tried again after \r\n.
1280    
1281 ph10 131 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
1282 ph10 134 in an attempt to make files that differ only in their line terminators
1283     compare equal. This works on Linux.
1284 ph10 142
1285 ph10 141 17. Under certain error circumstances pcregrep might try to free random memory
1286     as it exited. This is now fixed, thanks to valgrind.
1287 ph10 142
1288 ph10 141 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
1289 ph10 142 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
1290     was because its rules for how to advance for /g after matching an empty
1291 ph10 143 string at the end of a line did not allow for this case. They now check for
1292     it specially.
1293 ph10 150
1294     20. pcretest is supposed to handle patterns and data of any length, by
1295     extending its buffers when necessary. It was getting this wrong when the
1296 ph10 147 buffer for a data line had to be extended.
1297 ph10 150
1298 ph10 149 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
1299 ph10 150 CRLF as a newline sequence.
1300 ph10 152
1301 ph10 151 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
1302 ph10 152 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
1303     I have nevertheless tidied it up.
1304 ph10 111
1305 ph10 152 23. Added some casts to kill warnings from HP-UX ia64 compiler.
1306 ph10 123
1307 ph10 153 24. Added a man page for pcre-config.
1308 ph10 152
1309 ph10 153
1310 nigel 93 Version 7.0 19-Dec-06
1311     ---------------------
1312    
1313     1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
1314     moving to gcc 4.1.1.
1315    
1316     2. The -S option for pcretest uses setrlimit(); I had omitted to #include
1317     sys/time.h, which is documented as needed for this function. It doesn't
1318     seem to matter on Linux, but it showed up on some releases of OS X.
1319    
1320     3. It seems that there are systems where bytes whose values are greater than
1321     127 match isprint() in the "C" locale. The "C" locale should be the
1322     default when a C program starts up. In most systems, only ASCII printing
1323     characters match isprint(). This difference caused the output from pcretest
1324     to vary, making some of the tests fail. I have changed pcretest so that:
1325    
1326     (a) When it is outputting text in the compiled version of a pattern, bytes
1327     other than 32-126 are always shown as hex escapes.
1328    
1329     (b) When it is outputting text that is a matched part of a subject string,
1330     it does the same, unless a different locale has been set for the match
1331     (using the /L modifier). In this case, it uses isprint() to decide.
1332    
1333     4. Fixed a major bug that caused incorrect computation of the amount of memory
1334     required for a compiled pattern when options that changed within the
1335     pattern affected the logic of the preliminary scan that determines the
1336     length. The relevant options are -x, and -i in UTF-8 mode. The result was
1337     that the computed length was too small. The symptoms of this bug were
1338     either the PCRE error "internal error: code overflow" from pcre_compile(),
1339     or a glibc crash with a message such as "pcretest: free(): invalid next
1340     size (fast)". Examples of patterns that provoked this bug (shown in
1341     pcretest format) are:
1342    
1343     /(?-x: )/x
1344     /(?x)(?-x: \s*#\s*)/
1345     /((?i)[\x{c0}])/8
1346     /(?i:[\x{c0}])/8
1347    
1348     HOWEVER: Change 17 below makes this fix obsolete as the memory computation
1349     is now done differently.
1350    
1351     5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
1352     wrapper classes; (b) implement a new function in the C++ scanner that is
1353     more efficient than the old way of doing things because it avoids levels of
1354     recursion in the regex matching; (c) add a paragraph to the documentation
1355     for the FullMatch() function.
1356    
1357     6. The escape sequence \n was being treated as whatever was defined as
1358     "newline". Not only was this contrary to the documentation, which states
1359     that \n is character 10 (hex 0A), but it also went horribly wrong when
1360     "newline" was defined as CRLF. This has been fixed.
1361    
1362     7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
1363     was being set to -1 for the "end of line" case (supposedly a value that no
1364     character can have). Though this value is never used (the check for end of
1365     line is "zero bytes in current character"), it caused compiler complaints.
1366     I've changed it to 0xffffffff.
1367    
1368     8. In pcre_version.c, the version string was being built by a sequence of
1369     C macros that, in the event of PCRE_PRERELEASE being defined as an empty
1370     string (as it is for production releases) called a macro with an empty
1371     argument. The C standard says the result of this is undefined. The gcc
1372     compiler treats it as an empty string (which was what was wanted) but it is
1373     reported that Visual C gives an error. The source has been hacked around to
1374     avoid this problem.
1375    
1376     9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
1377     builds of pcretest, and changed the call to _setmode() to use _O_BINARY
1378     instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
1379     of them did).
1380    
1381     10. Originally, pcretest opened its input and output without "b"; then I was
1382     told that "b" was needed in some environments, so it was added for release
1383     5.0 to both the input and output. (It makes no difference on Unix-like
1384     systems.) Later I was told that it is wrong for the input on Windows. I've
1385     now abstracted the modes into two macros, to make it easier to fiddle with
1386     them, and removed "b" from the input mode under Windows.
1387    
1388     11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
1389    
1390     12. Added -help and --help to pcretest as an official way of being reminded
1391     of the options.
1392    
1393     13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
1394     and pcrecpp.cc because they annoy compilers at high warning levels.
1395    
1396     14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
1397    
1398     15. Fixed an occurrence of == in configure.ac that should have been = (shell
1399     scripts are not C programs :-) and which was not noticed because it works
1400     on Linux.
1401    
1402     16. pcretest is supposed to handle any length of pattern and data line (as one
1403     line or as a continued sequence of lines) by extending its input buffer if
1404     necessary. This feature was broken for very long pattern lines, leading to
1405     a string of junk being passed to pcre_compile() if the pattern was longer
1406     than about 50K.
1407    
1408     17. I have done a major re-factoring of the way pcre_compile() computes the
1409     amount of memory needed for a compiled pattern. Previously, there was code
1410     that made a preliminary scan of the pattern in order to do this. That was
1411     OK when PCRE was new, but as the facilities have expanded, it has become
1412     harder and harder to keep it in step with the real compile phase, and there
1413     have been a number of bugs (see for example, 4 above). I have now found a
1414     cunning way of running the real compile function in a "fake" mode that
1415     enables it to compute how much memory it would need, while actually only
1416     ever using a few hundred bytes of working memory and without too many
1417     tests of the mode. This should make future maintenance and development
1418     easier. A side effect of this work is that the limit of 200 on the nesting
1419     depth of parentheses has been removed (though this was never a serious
1420     limitation, I suspect). However, there is a downside: pcre_compile() now
1421     runs more slowly than before (30% or more, depending on the pattern). I
1422     hope this isn't a big issue. There is no effect on runtime performance.
1423    
1424     18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
1425     newline (only possible for the last line of a file) and it was a
1426     pattern that set a locale (followed by /Lsomething), pcretest crashed.
1427    
1428     19. Added additional timing features to pcretest. (1) The -tm option now times
1429     matching only, not compiling. (2) Both -t and -tm can be followed, as a
1430     separate command line item, by a number that specifies the number of
1431     repeats to use when timing. The default is 50000; this gives better
1432     precision, but takes uncomfortably long for very large patterns.
1433    
1434     20. Extended pcre_study() to be more clever in cases where a branch of a
1435     subpattern has no definite first character. For example, (a*|b*)[cd] would
1436     previously give no result from pcre_study(). Now it recognizes that the
1437     first character must be a, b, c, or d.
1438    
1439     21. There was an incorrect error "recursive call could loop indefinitely" if
1440     a subpattern (or the entire pattern) that was being tested for matching an
1441     empty string contained only one non-empty item after a nested subpattern.
1442     For example, the pattern (?>\x{100}*)\d(?R) provoked this error
1443     incorrectly, because the \d was being skipped in the check.
1444    
1445     22. The pcretest program now has a new pattern option /B and a command line
1446     option -b, which is equivalent to adding /B to every pattern. This causes
1447     it to show the compiled bytecode, without the additional information that
1448     -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
1449     is the same as /B/I).
1450    
1451     23. A new optimization is now able automatically to treat some sequences such
1452     as a*b as a*+b. More specifically, if something simple (such as a character
1453     or a simple class like \d) has an unlimited quantifier, and is followed by
1454     something that cannot possibly match the quantified thing, the quantifier
1455     is automatically "possessified".
1456    
1457     24. A recursive reference to a subpattern whose number was greater than 39
1458     went wrong under certain circumstances in UTF-8 mode. This bug could also
1459     have affected the operation of pcre_study().
1460    
1461     25. Realized that a little bit of performance could be had by replacing
1462     (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
1463    
1464     26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
1465    
1466     27. Possessive quantifiers such as a++ were previously implemented by turning
1467     them into atomic groups such as ($>a+). Now they have their own opcodes,
1468     which improves performance. This includes the automatically created ones
1469     from 23 above.
1470    
1471     28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
1472     lookahead was broken if it was not anchored. PCRE was mistakenly expecting
1473     the first matched character to be a colon. This applied both to named and
1474     numbered groups.
1475    
1476     29. The ucpinternal.h header file was missing its idempotency #ifdef.
1477    
1478     30. I was sent a "project" file called libpcre.a.dev which I understand makes
1479     building PCRE on Windows easier, so I have included it in the distribution.
1480    
1481     31. There is now a check in pcretest against a ridiculously large number being
1482     returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
1483     loop, the loop is abandoned.
1484    
1485     32. Forward references to subpatterns in conditions such as (?(2)...) where
1486     subpattern 2 is defined later cause pcre_compile() to search forwards in
1487     the pattern for the relevant set of parentheses. This search went wrong
1488     when there were unescaped parentheses in a character class, parentheses
1489     escaped with \Q...\E, or parentheses in a #-comment in /x mode.
1490    
1491     33. "Subroutine" calls and backreferences were previously restricted to
1492     referencing subpatterns earlier in the regex. This restriction has now
1493     been removed.
1494    
1495     34. Added a number of extra features that are going to be in Perl 5.10. On the
1496     whole, these are just syntactic alternatives for features that PCRE had
1497     previously implemented using the Python syntax or my own invention. The
1498     other formats are all retained for compatibility.
1499    
1500     (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
1501     as (?P<name>...). The new forms, as well as being in Perl 5.10, are
1502     also .NET compatible.
1503    
1504     (b) A recursion or subroutine call to a named group can now be defined as
1505     (?&name) as well as (?P>name).
1506    
1507     (c) A backreference to a named group can now be defined as \k<name> or
1508     \k'name' as well as (?P=name). The new forms, as well as being in Perl
1509     5.10, are also .NET compatible.
1510    
1511     (d) A conditional reference to a named group can now use the syntax
1512     (?(<name>) or (?('name') as well as (?(name).
1513    
1514     (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
1515     groups (named and numbered) that are never evaluated inline, but can be
1516     called as "subroutines" from elsewhere. In effect, the DEFINE condition
1517     is always false. There may be only one alternative in such a group.
1518    
1519     (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
1520     as the simple (?(R). The condition is true only if the most recent
1521     recursion is that of the given number or name. It does not search out
1522     through the entire recursion stack.
1523    
1524     (g) The escape \gN or \g{N} has been added, where N is a positive or
1525     negative number, specifying an absolute or relative reference.
1526    
1527     35. Tidied to get rid of some further signed/unsigned compiler warnings and
1528     some "unreachable code" warnings.
1529    
1530     36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
1531     things, this adds five new scripts.
1532    
1533     37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
1534     There were also incompatibilities regarding the handling of \Q..\E inside
1535     character classes, for example with patterns like [\Qa\E-\Qz\E] where the
1536     hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
1537    
1538     38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
1539     matches an empty string, and forcibly breaks the loop. There were bugs in
1540     this code in non-simple cases. For a pattern such as ^(a()*)* matched
1541     against aaaa the result was just "a" rather than "aaaa", for example. Two
1542     separate and independent bugs (that affected different cases) have been
1543     fixed.
1544    
1545     39. Refactored the code to abolish the use of different opcodes for small
1546     capturing bracket numbers. This is a tidy that I avoided doing when I
1547     removed the limit on the number of capturing brackets for 3.5 back in 2001.
1548     The new approach is not only tidier, it makes it possible to reduce the
1549     memory needed to fix the previous bug (38).
1550    
1551     40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
1552     sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
1553     processing dot, circumflex, or dollar metacharacters, or #-comments in /x
1554     mode.
1555    
1556     41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
1557     report.
1558    
1559     42. Applied patch, originally from Ari Pollak, modified by Google, to allow
1560     copy construction and assignment in the C++ wrapper.
1561    
1562     43. Updated pcregrep to support "--newline=any". In the process, I fixed a
1563     couple of bugs that could have given wrong results in the "--newline=crlf"
1564     case.
1565    
1566     44. Added a number of casts and did some reorganization of signed/unsigned int
1567     variables following suggestions from Dair Grant. Also renamed the variable
1568     "this" as "item" because it is a C++ keyword.
1569    
1570     45. Arranged for dftables to add
1571    
1572     #include "pcre_internal.h"
1573    
1574     to pcre_chartables.c because without it, gcc 4.x may remove the array
1575     definition from the final binary if PCRE is built into a static library and
1576     dead code stripping is activated.
1577    
1578     46. For an unanchored pattern, if a match attempt fails at the start of a
1579     newline sequence, and the newline setting is CRLF or ANY, and the next two
1580     characters are CRLF, advance by two characters instead of one.
1581    
1582    
1583 nigel 91 Version 6.7 04-Jul-06
1584     ---------------------
1585    
1586     1. In order to handle tests when input lines are enormously long, pcretest has
1587     been re-factored so that it automatically extends its buffers when
1588     necessary. The code is crude, but this _is_ just a test program. The
1589     default size has been increased from 32K to 50K.
1590    
1591     2. The code in pcre_study() was using the value of the re argument before
1592     testing it for NULL. (Of course, in any sensible call of the function, it
1593     won't be NULL.)
1594    
1595     3. The memmove() emulation function in pcre_internal.h, which is used on
1596     systems that lack both memmove() and bcopy() - that is, hardly ever -
1597     was missing a "static" storage class specifier.
1598    
1599     4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
1600     containing an extended class (one that cannot be represented by a bitmap
1601     because it contains high-valued characters or Unicode property items, e.g.
1602     [\pZ]). Almost always one would set UTF-8 mode when processing such a
1603     pattern, but PCRE should not loop if you do not (it no longer does).
1604     [Detail: two cases were found: (a) a repeated subpattern containing an
1605     extended class; (b) a recursive reference to a subpattern that followed a
1606     previous extended class. It wasn't skipping over the extended class
1607     correctly when UTF-8 mode was not set.]
1608    
1609     5. A negated single-character class was not being recognized as fixed-length
1610     in lookbehind assertions such as (?<=[^f]), leading to an incorrect
1611     compile error "lookbehind assertion is not fixed length".
1612    
1613     6. The RunPerlTest auxiliary script was showing an unexpected difference
1614     between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
1615     write a Perl script that can interpret lines of an input file either as
1616     byte characters or as UTF-8, which is what "perltest" was being required to
1617     do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
1618     can't do is switch easily at run time between having the "use utf8;" pragma
1619     or not. In the end, I fudged it by using the RunPerlTest script to insert
1620     "use utf8;" explicitly for the UTF-8 tests.
1621    
1622     7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
1623     the end of the subject string, contrary to the documentation and to what
1624     Perl does. This was true of both matching functions. Now it matches only at
1625     the start of the subject and immediately after *internal* newlines.
1626    
1627     8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
1628     a pointer to an int instead of a pointer to an unsigned long int. This
1629     caused problems on 64-bit systems.
1630    
1631     9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
1632     instance of the 'standard' template library not being so standard".
1633    
1634     10. There was no check on the number of named subpatterns nor the maximum
1635     length of a subpattern name. The product of these values is used to compute
1636     the size of the memory block for a compiled pattern. By supplying a very
1637     long subpattern name and a large number of named subpatterns, the size
1638     computation could be caused to overflow. This is now prevented by limiting
1639     the length of names to 32 characters, and the number of named subpatterns
1640     to 10,000.
1641    
1642     11. Subpatterns that are repeated with specific counts have to be replicated in
1643     the compiled pattern. The size of memory for this was computed from the
1644     length of the subpattern and the repeat count. The latter is limited to
1645     65535, but there was no limit on the former, meaning that integer overflow
1646     could in principle occur. The compiled length of a repeated subpattern is
1647     now limited to 30,000 bytes in order to prevent this.
1648    
1649     12. Added the optional facility to have named substrings with the same name.
1650    
1651     13. Added the ability to use a named substring as a condition, using the
1652     Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
1653     are numbers (not recommended). Forward references are permitted.
1654    
1655     14. Added forward references in named backreferences (if you see what I mean).
1656    
1657     15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
1658     pattern could run off the end of the subject. For example, the pattern
1659     "(?s)(.{1,5})"8 did this with the subject "ab".
1660    
1661     16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
1662     PCRE_CASELESS was set when matching characters that were quantified with ?
1663     or *.
1664    
1665     17. A character class other than a single negated character that had a minimum
1666     but no maximum quantifier - for example [ab]{6,} - was not handled
1667     correctly by pce_dfa_exec(). It would match only one character.
1668    
1669     18. A valid (though odd) pattern that looked like a POSIX character
1670     class but used an invalid character after [ (for example [[,abc,]]) caused
1671     pcre_compile() to give the error "Failed: internal error: code overflow" or
1672     in some cases to crash with a glibc free() error. This could even happen if
1673     the pattern terminated after [[ but there just happened to be a sequence of
1674     letters, a binary zero, and a closing ] in the memory that followed.
1675    
1676     19. Perl's treatment of octal escapes in the range \400 to \777 has changed
1677     over the years. Originally (before any Unicode support), just the bottom 8
1678     bits were taken. Thus, for example, \500 really meant \100. Nowadays the
1679     output from "man perlunicode" includes this:
1680    
1681     The regular expression compiler produces polymorphic opcodes. That
1682     is, the pattern adapts to the data and automatically switches to
1683     the Unicode character scheme when presented with Unicode data--or
1684     instead uses a traditional byte scheme when presented with byte
1685     data.
1686    
1687     Sadly, a wide octal escape does not cause a switch, and in a string with
1688     no other multibyte characters, these octal escapes are treated as before.
1689     Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
1690     /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
1691     Unicode string.
1692    
1693     I have not perpetrated such confusion in PCRE. Up till now, it took just
1694     the bottom 8 bits, as in old Perl. I have now made octal escapes with
1695     values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
1696     translate to the appropriate multibyte character.
1697    
1698     29. Applied some refactoring to reduce the number of warnings from Microsoft
1699     and Borland compilers. This has included removing the fudge introduced
1700     seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
1701     a warning about an unused variable.
1702    
1703     21. PCRE has not included VT (character 0x0b) in the set of whitespace
1704     characters since release 4.0, because Perl (from release 5.004) does not.
1705     [Or at least, is documented not to: some releases seem to be in conflict
1706     with the documentation.] However, when a pattern was studied with
1707     pcre_study() and all its branches started with \s, PCRE still included VT
1708     as a possible starting character. Of course, this did no harm; it just
1709     caused an unnecessary match attempt.
1710    
1711     22. Removed a now-redundant internal flag bit that recorded the fact that case
1712     dependency changed within the pattern. This was once needed for "required
1713     byte" processing, but is no longer used. This recovers a now-scarce options
1714     bit. Also moved the least significant internal flag bit to the most-
1715     significant bit of the word, which was not previously used (hangover from
1716     the days when it was an int rather than a uint) to free up another bit for
1717     the future.
1718    
1719     23. Added support for CRLF line endings as well as CR and LF. As well as the
1720     default being selectable at build time, it can now be changed at runtime
1721     via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
1722     specify that it is scanning data with non-default line endings.
1723    
1724     24. Changed the definition of CXXLINK to make it agree with the definition of
1725     LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
1726    
1727     25. Applied Ian Taylor's patches to avoid using another stack frame for tail
1728     recursions. This makes a big different to stack usage for some patterns.
1729    
1730     26. If a subpattern containing a named recursion or subroutine reference such
1731     as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
1732     the space required for the compiled pattern went wrong and gave too small a
1733     value. Depending on the environment, this could lead to "Failed: internal
1734     error: code overflow at offset 49" or "glibc detected double free or
1735     corruption" errors.
1736    
1737     27. Applied patches from Google (a) to support the new newline modes and (b) to
1738     advance over multibyte UTF-8 characters in GlobalReplace.
1739    
1740     28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
1741     difference for some implementation of PCRE in some Windows version.
1742    
1743     29. Added some extra testing facilities to pcretest:
1744    
1745     \q<number> in a data line sets the "match limit" value
1746     \Q<number> in a data line sets the "match recursion limt" value
1747     -S <number> sets the stack size, where <number> is in megabytes
1748    
1749     The -S option isn't available for Windows.
1750    
1751    
1752 nigel 89 Version 6.6 06-Feb-06
1753     ---------------------
1754    
1755     1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
1756     in pcreposix.h. I have copied the definition from pcre.h.
1757    
1758     2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
1759     because pcre.h is no longer a built file.
1760    
1761     3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
1762     not normally included in the compiled code.
1763    
1764    
1765 nigel 87 Version 6.5 01-Feb-06
1766     ---------------------
1767    
1768     1. When using the partial match feature with pcre_dfa_exec(), it was not
1769     anchoring the second and subsequent partial matches at the new starting
1770     point. This could lead to incorrect results. For example, with the pattern
1771     /1234/, partially matching against "123" and then "a4" gave a match.
1772    
1773     2. Changes to pcregrep:
1774    
1775     (a) All non-match returns from pcre_exec() were being treated as failures
1776     to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
1777     error message is output. Some extra information is given for the
1778     PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
1779     probably the only errors that are likely to be caused by users (by
1780     specifying a regex that has nested indefinite repeats, for instance).
1781     If there are more than 20 of these errors, pcregrep is abandoned.
1782    
1783     (b) A binary zero was treated as data while matching, but terminated the
1784     output line if it was written out. This has been fixed: binary zeroes
1785     are now no different to any other data bytes.
1786    
1787     (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
1788     used to set a locale for matching. The --locale=xxxx long option has
1789     been added (no short equivalent) to specify a locale explicitly on the
1790     pcregrep command, overriding the environment variables.
1791    
1792     (d) When -B was used with -n, some line numbers in the output were one less
1793     than they should have been.
1794    
1795     (e) Added the -o (--only-matching) option.
1796    
1797     (f) If -A or -C was used with -c (count only), some lines of context were
1798     accidentally printed for the final match.
1799    
1800     (g) Added the -H (--with-filename) option.
1801    
1802     (h) The combination of options -rh failed to suppress file names for files
1803     that were found from directory arguments.
1804    
1805     (i) Added the -D (--devices) and -d (--directories) options.
1806    
1807     (j) Added the -F (--fixed-strings) option.
1808    
1809     (k) Allow "-" to be used as a file name for -f as well as for a data file.
1810    
1811     (l) Added the --colo(u)r option.
1812    
1813     (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
1814     is not present by default.
1815    
1816     3. A nasty bug was discovered in the handling of recursive patterns, that is,
1817     items such as (?R) or (?1), when the recursion could match a number of
1818     alternatives. If it matched one of the alternatives, but subsequently,
1819     outside the recursion, there was a failure, the code tried to back up into
1820     the recursion. However, because of the way PCRE is implemented, this is not
1821     possible, and the result was an incorrect result from the match.
1822    
1823     In order to prevent this happening, the specification of recursion has
1824     been changed so that all such subpatterns are automatically treated as
1825     atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
1826    
1827     4. I had overlooked the fact that, in some locales, there are characters for
1828     which isalpha() is true but neither isupper() nor islower() are true. In
1829     the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
1830     and ordfeminine) are like this. This affected the treatment of \w and \W
1831     when they appeared in character classes, but not when they appeared outside
1832     a character class. The bit map for "word" characters is now created
1833     separately from the results of isalnum() instead of just taking it from the
1834     upper, lower, and digit maps. (Plus the underscore character, of course.)
1835    
1836     5. The above bug also affected the handling of POSIX character classes such as
1837     [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
1838     permanent tables. Instead, the bit maps for such a class were previously
1839     created as the appropriate unions of the upper, lower, and digit bitmaps.
1840     Now they are created by subtraction from the [[:word:]] class, which has
1841     its own bitmap.
1842    
1843     6. The [[:blank:]] character class matches horizontal, but not vertical space.
1844     It is created by subtracting the vertical space characters (\x09, \x0a,
1845     \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
1846     subtraction was done in the overall bitmap for a character class, meaning
1847     that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
1848     be recognized. This bug has been fixed.
1849    
1850     7. Patches from the folks at Google:
1851    
1852     (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
1853     real life, but is still worth protecting against".
1854    
1855     (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
1856     regular expressions".
1857    
1858     (c) pcre_scanner.cc: avoid use of std::count() because not all systems
1859     have it.
1860    
1861     (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
1862     "configure" and the latter not, in order to fix a problem somebody had
1863     with compiling the Arg class on HP-UX.
1864    
1865     (e) Improve the error-handling of the C++ wrapper a little bit.
1866    
1867     (f) New tests for checking recursion limiting.
1868    
1869     8. The pcre_memmove() function, which is used only if the environment does not
1870     have a standard memmove() function (and is therefore rarely compiled),
1871     contained two bugs: (a) use of int instead of size_t, and (b) it was not
1872     returning a result (though PCRE never actually uses the result).
1873    
1874     9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
1875     large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
1876     returned instead of calling malloc() with an overflowing number that would
1877     most likely cause subsequent chaos.
1878    
1879     10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
1880    
1881     11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
1882     with this option is matched, the nmatch and pmatch options of regexec() are
1883     ignored.
1884    
1885     12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
1886     provided in case anyone wants to the the POSIX interface with UTF-8
1887     strings.
1888    
1889     13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
1890     C++ linking (needed for some HP-UX environments).
1891    
1892     14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
1893     (unused parameter) and in the pcre_printint() function (omitted "default"
1894     switch label when the default is to do nothing).
1895    
1896     15. Added some code to make it possible, when PCRE is compiled as a C++
1897     library, to replace subject pointers for pcre_exec() with a smart pointer
1898     class, thus making it possible to process discontinuous strings.
1899    
1900     16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
1901     much the same function. They were added by different people who were trying
1902     to make PCRE easy to compile on non-Unix systems. It has been suggested
1903     that PCRE_EXPORT be abolished now that there is more automatic apparatus
1904     for compiling on Windows systems. I have therefore replaced it with
1905     PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
1906     defaults to "extern" for C or "extern C" for C++, which works fine on
1907     Unix-like systems. It is now possible to override the value of PCRE_DATA_
1908     SCOPE with something explicit in config.h. In addition:
1909    
1910     (a) pcreposix.h still had just "extern" instead of either of these macros;
1911     I have replaced it with PCRE_DATA_SCOPE.
1912    
1913     (b) Functions such as _pcre_xclass(), which are internal to the library,
1914     but external in the C sense, all had PCRE_EXPORT in their definitions.
1915     This is apparently wrong for the Windows case, so I have removed it.
1916     (It makes no difference on Unix-like systems.)
1917    
1918     17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
1919     of recursive calls to match(). This is different to MATCH_LIMIT because
1920     that limits the total number of calls to match(), not all of which increase
1921     the depth of recursion. Limiting the recursion depth limits the amount of
1922     stack (or heap if NO_RECURSE is set) that is used. The default can be set
1923     when PCRE is compiled, and changed at run time. A patch from Google adds
1924     this functionality to the C++ interface.
1925    
1926     18. Changes to the handling of Unicode character properties:
1927    
1928     (a) Updated the table to Unicode 4.1.0.
1929    
1930     (b) Recognize characters that are not in the table as "Cn" (undefined).
1931    
1932     (c) I revised the way the table is implemented to a much improved format
1933     which includes recognition of ranges. It now supports the ranges that
1934     are defined in UnicodeData.txt, and it also amalgamates other
1935     characters into ranges. This has reduced the number of entries in the
1936     table from around 16,000 to around 3,000, thus reducing its size
1937     considerably. I realized I did not need to use a tree structure after
1938     all - a binary chop search is just as efficient. Having reduced the
1939     number of entries, I extended their size from 6 bytes to 8 bytes to
1940     allow for more data.
1941    
1942     (d) Added support for Unicode script names via properties such as \p{Han}.
1943    
1944     19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
1945     matching that character.
1946    
1947     20. When matching a repeated Unicode property with a minimum greater than zero,
1948     (for example \pL{2,}), PCRE could look past the end of the subject if it
1949     reached it while seeking the minimum number of characters. This could
1950     happen only if some of the characters were more than one byte long, because
1951     there is a check for at least the minimum number of bytes.
1952    
1953     21. Refactored the implementation of \p and \P so as to be more general, to
1954     allow for more different types of property in future. This has changed the
1955     compiled form incompatibly. Anybody with saved compiled patterns that use
1956     \p or \P will have to recompile them.
1957    
1958     22. Added "Any" and "L&" to the supported property types.
1959    
1960     23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
1961     but give a compile time error if the value is greater than 0xff.
1962    
1963     24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
1964     accidentally not being installed or uninstalled.
1965    
1966     25. The pcre.h file was built from pcre.h.in, but the only changes that were
1967     made were to insert the current release number. This seemed silly, because
1968     it made things harder for people building PCRE on systems that don't run
1969     "configure". I have turned pcre.h into a distributed file, no longer built
1970     by "configure", with the version identification directly included. There is
1971     no longer a pcre.h.in file.
1972    
1973     However, this change necessitated a change to the pcre-config script as
1974     well. It is built from pcre-config.in, and one of the substitutions was the
1975     release number. I have updated configure.ac so that ./configure now finds
1976     the release number by grepping pcre.h.
1977    
1978     26. Added the ability to run the tests under valgrind.
1979    
1980    
1981 nigel 85 Version 6.4 05-Sep-05
1982     ---------------------
1983    
1984     1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
1985     "--" to be printed when multiple files were scanned, even when none of the
1986     -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
1987     consider it to be a bug, and have restored the previous behaviour.
1988    
1989     2. A couple of code tidies to get rid of compiler warnings.
1990    
1991     3. The pcretest program used to cheat by referring to symbols in the library
1992     whose names begin with _pcre_. These are internal symbols that are not
1993     really supposed to be visible externally, and in some environments it is
1994     possible to suppress them. The cheating is now confined to including
1995     certain files from the library's source, which is a bit cleaner.
1996    
1997     4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
1998     file's purpose clearer.
1999    
2000     5. Reorganized pcre_ucp_findchar().
2001    
2002    
2003 nigel 83 Version 6.3 15-Aug-05
2004     ---------------------
2005    
2006     1. The file libpcre.pc.in did not have general read permission in the tarball.
2007    
2008     2. There were some problems when building without C++ support:
2009    
2010     (a) If C++ support was not built, "make install" and "make test" still
2011     tried to test it.
2012    
2013     (b) There were problems when the value of CXX was explicitly set. Some
2014     changes have been made to try to fix these, and ...
2015    
2016     (c) --disable-cpp can now be used to explicitly disable C++ support.
2017    
2018     (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
2019     backslash in a target when C++ was disabled. This confuses some
2020     versions of "make", apparently. Using an intermediate variable solves
2021     this. (Same for CPP_LOBJ.)
2022    
2023     3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
2024     (non-Windows) now includes $(CFLAGS) because these flags are sometimes
2025     necessary on certain architectures.
2026    
2027     4. Added a setting of -export-symbols-regex to the link command to remove
2028     those symbols that are exported in the C sense, but actually are local
2029     within the library, and not documented. Their names all begin with
2030     "_pcre_". This is not a perfect job, because (a) we have to except some
2031     symbols that pcretest ("illegally") uses, and (b) the facility isn't always
2032     available (and never for static libraries). I have made a note to try to
2033     find a way round (a) in the future.
2034    
2035    
2036 nigel 81 Version 6.2 01-Aug-05
2037     ---------------------
2038    
2039     1. There was no test for integer overflow of quantifier values. A construction
2040     such as {1111111111111111} would give undefined results. What is worse, if
2041     a minimum quantifier for a parenthesized subpattern overflowed and became
2042     negative, the calculation of the memory size went wrong. This could have
2043     led to memory overwriting.
2044    
2045     2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
2046    
2047     3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
2048     operating environments where this matters.
2049    
2050     4. Applied Giuseppe Maxia's patch to add additional features for controlling
2051     PCRE options from within the C++ wrapper.
2052    
2053     5. Named capturing subpatterns were not being correctly counted when a pattern
2054     was compiled. This caused two problems: (a) If there were more than 100
2055     such subpatterns, the calculation of the memory needed for the whole
2056     compiled pattern went wrong, leading to an overflow error. (b) Numerical
2057     back references of the form \12, where the number was greater than 9, were
2058     not recognized as back references, even though there were sufficient
2059     previous subpatterns.
2060    
2061     6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
2062     versions of gcc, e.g. 2.95.4.
2063    
2064    
2065 nigel 79 Version 6.1 21-Jun-05
2066     ---------------------
2067    
2068     1. There was one reference to the variable "posix" in pcretest.c that was not
2069     surrounded by "#if !defined NOPOSIX".
2070    
2071     2. Make it possible to compile pcretest without DFA support, UTF8 support, or
2072     the cross-check on the old pcre_info() function, for the benefit of the
2073     cut-down version of PCRE that is currently imported into Exim.
2074    
2075     3. A (silly) pattern starting with (?i)(?-i) caused an internal space
2076     allocation error. I've done the easy fix, which wastes 2 bytes for sensible
2077     patterns that start (?i) but I don't think that matters. The use of (?i) is
2078     just an example; this all applies to the other options as well.
2079    
2080     4. Since libtool seems to echo the compile commands it is issuing, the output
2081     from "make" can be reduced a bit by putting "@" in front of each libtool
2082     compile command.
2083    
2084     5. Patch from the folks at Google for configure.in to be a bit more thorough
2085     in checking for a suitable C++ installation before trying to compile the
2086     C++ stuff. This should fix a reported problem when a compiler was present,
2087     but no suitable headers.
2088    
2089     6. The man pages all had just "PCRE" as their title. I have changed them to
2090     be the relevant file name. I have also arranged that these names are
2091     retained in the file doc/pcre.txt, which is a concatenation in text format
2092     of all the man pages except the little individual ones for each function.
2093    
2094     7. The NON-UNIX-USE file had not been updated for the different set of source
2095     files that come with release 6. I also added a few comments about the C++
2096     wrapper.
2097    
2098    
2099 nigel 77 Version 6.0 07-Jun-05
2100     ---------------------
2101    
2102     1. Some minor internal re-organization to help with my DFA experiments.
2103    
2104     2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
2105     didn't matter for the library itself when fully configured, but did matter
2106     when compiling without UCP support, or within Exim, where the ucp files are
2107     not imported.
2108    
2109     3. Refactoring of the library code to split up the various functions into
2110     different source modules. The addition of the new DFA matching code (see
2111     below) to a single monolithic source would have made it really too
2112     unwieldy, quite apart from causing all the code to be include in a
2113     statically linked application, when only some functions are used. This is
2114     relevant even without the DFA addition now that patterns can be compiled in
2115     one application and matched in another.
2116    
2117     The downside of splitting up is that there have to be some external
2118     functions and data tables that are used internally in different modules of
2119     the library but which are not part of the API. These have all had their
2120     names changed to start with "_pcre_" so that they are unlikely to clash
2121     with other external names.
2122    
2123     4. Added an alternate matching function, pcre_dfa_exec(), which matches using
2124     a different (DFA) algorithm. Although it is slower than the original
2125     function, it does have some advantages for certain types of matching
2126     problem.
2127    
2128     5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
2129     including restarting after a partial match.
2130    
2131     6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
2132     defined when compiling for Windows was sent to me. I have put it into the
2133     code, though I have no means of testing or verifying it.
2134    
2135     7. Added the pcre_refcount() auxiliary function.
2136    
2137     8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
2138     match before or at the first newline in the subject string. In pcretest,
2139     the /f option on a pattern can be used to set this.
2140    
2141     9. A repeated \w when used in UTF-8 mode with characters greater than 256
2142     would behave wrongly. This has been present in PCRE since release 4.0.
2143    
2144     10. A number of changes to the pcregrep command:
2145    
2146     (a) Refactored how -x works; insert ^(...)$ instead of setting
2147     PCRE_ANCHORED and checking the length, in preparation for adding
2148     something similar for -w.
2149    
2150     (b) Added the -w (match as a word) option.
2151    
2152     (c) Refactored the way lines are read and buffered so as to have more
2153     than one at a time available.
2154    
2155     (d) Implemented a pcregrep test script.
2156    
2157     (e) Added the -M (multiline match) option. This allows patterns to match
2158     over several lines of the subject. The buffering ensures that at least
2159     8K, or the rest of the document (whichever is the shorter) is available
2160     for matching (and similarly the previous 8K for lookbehind assertions).
2161    
2162     (f) Changed the --help output so that it now says
2163    
2164     -w, --word-regex(p)
2165    
2166     instead of two lines, one with "regex" and the other with "regexp"
2167     because that confused at least one person since the short forms are the
2168     same. (This required a bit of code, as the output is generated
2169     automatically from a table. It wasn't just a text change.)
2170    
2171     (g) -- can be used to terminate pcregrep options if the next thing isn't an
2172     option but starts with a hyphen. Could be a pattern or a path name
2173     starting with a hyphen, for instance.
2174    
2175     (h) "-" can be given as a file name to represent stdin.
2176    
2177     (i) When file names are being printed, "(standard input)" is used for
2178     the standard input, for compatibility with GNU grep. Previously
2179     "<stdin>" was used.
2180    
2181     (j) The option --label=xxx can be used to supply a name to be used for
2182     stdin when file names are being printed. There is no short form.
2183    
2184     (k) Re-factored the options decoding logic because we are going to add
2185     two more options that take data. Such options can now be given in four
2186     different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
2187    
2188     (l) Added the -A, -B, and -C options for requesting that lines of context
2189     around matches be printed.
2190    
2191     (m) Added the -L option to print the names of files that do not contain
2192     any matching lines, that is, the complement of -l.
2193    
2194     (n) The return code is 2 if any file cannot be opened, but pcregrep does
2195     continue to scan other files.
2196    
2197     (o) The -s option was incorrectly implemented. For compatibility with other
2198     greps, it now suppresses the error message for a non-existent or non-
2199     accessible file (but not the return code). There is a new option called
2200     -q that suppresses the output of matching lines, which was what -s was
2201     previously doing.
2202    
2203     (p) Added --include and --exclude options to specify files for inclusion
2204     and exclusion when recursing.
2205    
2206     11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
2207     Hopefully, it now does.
2208    
2209     12. Missing cast in pcre_study().
2210    
2211     13. Added an "uninstall" target to the makefile.
2212    
2213     14. Replaced "extern" in the function prototypes in Makefile.in with
2214     "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
2215     world, but is set differently for Windows.
2216    
2217     15. Added a second compiling function called pcre_compile2(). The only
2218     difference is that it has an extra argument, which is a pointer to an
2219     integer error code. When there is a compile-time failure, this is set
2220     non-zero, in addition to the error test pointer being set to point to an
2221     error message. The new argument may be NULL if no error number is required
2222     (but then you may as well call pcre_compile(), which is now just a
2223     wrapper). This facility is provided because some applications need a
2224     numeric error indication, but it has also enabled me to tidy up the way
2225     compile-time errors are handled in the POSIX wrapper.
2226    
2227     16. Added VPATH=.libs to the makefile; this should help when building with one
2228     prefix path and installing with another. (Or so I'm told by someone who
2229     knows more about this stuff than I do.)
2230    
2231     17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
2232     passes PCRE_DOTALL to the pcre_compile() function, making the "." character
2233     match everything, including newlines. This is not POSIX-compatible, but
2234     somebody wanted the feature. From pcretest it can be activated by using
2235     both the P and the s flags.
2236    
2237     18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
2238    
2239     19. libpcre.pc was being incorrectly installed as executable.
2240    
2241     20. A couple of places in pcretest check for end-of-line by looking for '\n';
2242     it now also looks for '\r' so that it will work unmodified on Windows.
2243    
2244     21. Added Google's contributed C++ wrapper to the distribution.
2245    
2246     22. Added some untidy missing memory free() calls in pcretest, to keep
2247     Electric Fence happy when testing.
2248    
2249    
2250    
2251 nigel 75 Version 5.0 13-Sep-04
2252     ---------------------
2253    
2254     1. Internal change: literal characters are no longer packed up into items
2255     containing multiple characters in a single byte-string. Each character
2256     is now matched using a separate opcode. However, there may be more than one
2257     byte in the character in UTF-8 mode.
2258    
2259     2. The pcre_callout_block structure has two new fields: pattern_position and
2260     next_item_length. These contain the offset in the pattern to the next match
2261     item, and its length, respectively.
2262    
2263     3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
2264     insertion of callouts before each pattern item. Added the /C option to
2265     pcretest to make use of this.
2266    
2267     4. On the advice of a Windows user, the lines
2268    
2269     #if defined(_WIN32) || defined(WIN32)
2270     _setmode( _fileno( stdout ), 0x8000 );
2271     #endif /* defined(_WIN32) || defined(WIN32) */
2272    
2273     have been added to the source of pcretest. This apparently does useful
2274     magic in relation to line terminators.
2275    
2276     5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
2277     for the benefit of those environments where the "b" makes a difference.
2278    
2279     6. The icc compiler has the same options as gcc, but "configure" doesn't seem
2280     to know about it. I have put a hack into configure.in that adds in code
2281     to set GCC=yes if CC=icc. This seems to end up at a point in the
2282     generated configure script that is early enough to affect the setting of
2283     compiler options, which is what is needed, but I have no means of testing
2284     whether it really works. (The user who reported this had patched the
2285     generated configure script, which of course I cannot do.)
2286    
2287     LATER: After change 22 below (new libtool files), the configure script
2288     seems to know about icc (and also ecc). Therefore, I have commented out
2289     this hack in configure.in.
2290    
2291     7. Added support for pkg-config (2 patches were sent in).
2292    
2293     8. Negated POSIX character classes that used a combination of internal tables
2294     were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
2295     [[:^ascii]]. Typically, they would match almost any characters. The other
2296     POSIX classes were not broken in this way.
2297    
2298     9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
2299     to find the match, as PCRE was deluded into thinking that the match had to
2300     start at the start point or following a newline. The same bug applied to
2301     patterns with negative forward assertions or any backward assertions
2302     preceding ".*" at the start, unless the pattern required a fixed first
2303     character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
2304    
2305     10. In UTF-8 mode, when moving forwards in the subject after a failed match
2306     starting at the last subject character, bytes beyond the end of the subject
2307     string were read.
2308    
2309     11. Renamed the variable "class" as "classbits" to make life easier for C++
2310     users. (Previously there was a macro definition, but it apparently wasn't
2311     enough.)
2312    
2313     12. Added the new field "tables" to the extra data so that tables can be passed
2314     in at exec time, or the internal tables can be re-selected. This allows
2315     a compiled regex to be saved and re-used at a later time by a different
2316     program that might have everything at different addresses.
2317    
2318     13. Modified the pcre-config script so that, when run on Solaris, it shows a
2319     -R library as well as a -L library.
2320    
2321     14. The debugging options of pcretest (-d on the command line or D on a
2322     pattern) showed incorrect output for anything following an extended class
2323     that contained multibyte characters and which was followed by a quantifier.
2324    
2325     15. Added optional support for general category Unicode character properties
2326     via the \p, \P, and \X escapes. Unicode property support implies UTF-8
2327     support. It adds about 90K to the size of the library. The meanings of the
2328     inbuilt class escapes such as \d and \s have NOT been changed.
2329    
2330     16. Updated pcredemo.c to include calls to free() to release the memory for the
2331     compiled pattern.
2332    
2333     17. The generated file chartables.c was being created in the source directory
2334     instead of in the building directory. This caused the build to fail if the
2335     source directory was different from the building directory, and was
2336     read-only.
2337    
2338     18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
2339     file. No doubt somebody will tell me if they don't make sense... Also added
2340     Dan Mooney's comments about building on OpenVMS.
2341    
2342     19. Added support for partial matching via the PCRE_PARTIAL option for
2343     pcre_exec() and the \P data escape in pcretest.
2344    
2345     20. Extended pcretest with 3 new pattern features:
2346    
2347     (i) A pattern option of the form ">rest-of-line" causes pcretest to
2348     write the compiled pattern to the file whose name is "rest-of-line".
2349     This is a straight binary dump of the data, with the saved pointer to
2350     the character tables forced to be NULL. The study data, if any, is
2351     written too. After writing, pcretest reads a new pattern.
2352    
2353     (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
2354     compiled pattern from the given file. There must not be any
2355     occurrences of "<" in the file name (pretty unlikely); if there are,
2356     pcretest will instead treat the initial "<" as a pattern delimiter.
2357     After reading in the pattern, pcretest goes on to read data lines as
2358     usual.
2359    
2360     (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
2361     and 16-bit fields in a compiled pattern, to simulate a pattern that
2362     was compiled on a host of opposite endianness.
2363    
2364     21. The pcre-exec() function can now cope with patterns that were compiled on
2365     hosts of opposite endianness, with this restriction:
2366    
2367     As for any compiled expression that is saved and used later, the tables
2368     pointer field cannot be preserved; the extra_data field in the arguments
2369     to pcre_exec() should be used to pass in a tables address if a value
2370     other than the default internal tables were used at compile time.
2371    
2372     22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
2373     now diagnosed as an error. Previously, most of the time, a negative number
2374     would have been treated as zero, but if in addition "ovector" was passed as
2375     NULL, a crash could occur.
2376    
2377     23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
2378     new versions from the libtool 1.5 distribution (the last one is a copy of
2379     a file called libtool.m4). This seems to have fixed the need to patch
2380     "configure" to support Darwin 1.3 (which I used to do). However, I still
2381     had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
2382     workstation).
2383    
2384     24. Changed the PCRE licence to be the more standard "BSD" licence.
2385    
2386    
2387 nigel 73 Version 4.5 01-Dec-03
2388     ---------------------
2389    
2390     1. There has been some re-arrangement of the code for the match() function so
2391     that it can be compiled in a version that does not call itself recursively.
2392     Instead, it keeps those local variables that need separate instances for
2393     each "recursion" in a frame on the heap, and gets/frees frames whenever it
2394     needs to "recurse". Keeping track of where control must go is done by means
2395     of setjmp/longjmp. The whole thing is implemented by a set of macros that
2396     hide most of the details from the main code, and operates only if
2397     NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
2398     "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
2399     operating.
2400    
2401     To make it easier for callers to provide specially tailored get/free
2402     functions for this usage, two new functions, pcre_stack_malloc, and
2403     pcre_stack_free, are used. They are always called in strict stacking order,
2404     and the size of block requested is always the same.
2405    
2406     The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
2407     PCRE has been compiled to use the stack or the heap for recursion. The
2408     -C option of pcretest uses this to show which version is compiled.
2409    
2410     A new data escape \S, is added to pcretest; it causes the amounts of store
2411     obtained and freed by both kinds of malloc/free at match time to be added
2412     to the output.
2413    
2414     2. Changed the locale test to use "fr_FR" instead of "fr" because that's
2415     what's available on my current Linux desktop machine.
2416    
2417     3. When matching a UTF-8 string, the test for a valid string at the start has
2418     been extended. If start_offset is not zero, PCRE now checks that it points
2419     to a byte that is the start of a UTF-8 character. If not, it returns
2420     PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
2421     this is necessary because there may be backward assertions in the pattern.
2422     When matching the same subject several times, it may save resources to use
2423     PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
2424    
2425     4. The code for checking the validity of UTF-8 strings has been tightened so
2426     that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
2427     containing "overlong sequences".
2428    
2429     5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
2430     I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
2431     should have been "&", but it just so happened that all the cases this let
2432     through by mistake were picked up later in the function.
2433    
2434     6. I had used a variable called "isblank" - this is a C99 function, causing
2435     some compilers to warn. To avoid this, I renamed it (as "blankclass").
2436    
2437     7. Cosmetic: (a) only output another newline at the end of pcretest if it is
2438     prompting; (b) run "./pcretest /dev/null" at the start of the test script
2439     so the version is shown; (c) stop "make test" echoing "./RunTest".
2440    
2441     8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
2442    
2443     9. The prototype for memmove() for systems that don't have it was using
2444     size_t, but the inclusion of the header that defines size_t was later. I've
2445     moved the #includes for the C headers earlier to avoid this.
2446    
2447     10. Added some adjustments to the code to make it easier to compiler on certain
2448     special systems:
2449    
2450     (a) Some "const" qualifiers were missing.
2451     (b) Added the macro EXPORT before all exported functions; by default this
2452     is defined to be empty.
2453     (c) Changed the dftables auxiliary program (that builds chartables.c) so
2454     that it reads its output file name as an argument instead of writing
2455     to the standard output and assuming this can be redirected.
2456    
2457     11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
2458     class containing characters with values greater than 255, PCRE compilation
2459     went into a loop.
2460    
2461     12. A recursive reference to a subpattern that was within another subpattern
2462     that had a minimum quantifier of zero caused PCRE to crash. For example,
2463     (x(y(?2))z)? provoked this bug with a subject that got as far as the
2464     recursion. If the recursively-called subpattern itself had a zero repeat,
2465     that was OK.
2466    
2467     13. In pcretest, the buffer for reading a data line was set at 30K, but the
2468     buffer into which it was copied (for escape processing) was still set at
2469     1024, so long lines caused crashes.
2470    
2471     14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
2472     "internal error: code overflow...". This applied to any character class
2473     that was followed by a possessive quantifier.
2474    
2475     15. Modified the Makefile to add libpcre.la as a prerequisite for
2476     libpcreposix.la because I was told this is needed for a parallel build to
2477     work.
2478    
2479     16. If a pattern that contained .* following optional items at the start was
2480     studied, the wrong optimizing data was generated, leading to matching
2481     errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
2482     matching string must start with a or b or c. The correct conclusion for
2483     this pattern is that a match can start with any character.
2484    
2485    
2486 nigel 71 Version 4.4 13-Aug-03
2487     ---------------------
2488    
2489     1. In UTF-8 mode, a character class containing characters with values between
2490     127 and 255 was not handled correctly if the compiled pattern was studied.
2491     In fixing this, I have also improved the studying algorithm for such
2492     classes (slightly).
2493    
2494     2. Three internal functions had redundant arguments passed to them. Removal
2495     might give a very teeny performance improvement.
2496    
2497     3. Documentation bug: the value of the capture_top field in a callout is *one
2498     more than* the number of the hightest numbered captured substring.
2499    
2500     4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
2501     in incorrectly linking with a previously installed version. They now link
2502     explicitly with libpcre.la.
2503    
2504     5. configure.in no longer needs to recognize Cygwin specially.
2505    
2506     6. A problem in pcre.in for Windows platforms is fixed.
2507    
2508     7. If a pattern was successfully studied, and the -d (or /D) flag was given to
2509     pcretest, it used to include the size of the study block as part of its
2510     output. Unfortunately, the structure contains a field that has a different
2511     size on different hardware architectures. This meant that the tests that
2512     showed this size failed. As the block is currently always of a fixed size,
2513     this information isn't actually particularly useful in pcretest output, so
2514     I have just removed it.
2515    
2516     8. Three pre-processor statements accidentally did not start in column 1.
2517     Sadly, there are *still* compilers around that complain, even though
2518     standard C has not required this for well over a decade. Sigh.
2519    
2520     9. In pcretest, the code for checking callouts passed small integers in the
2521     callout_data field, which is a void * field. However, some picky compilers
2522     complained about the casts involved for this on 64-bit systems. Now
2523     pcretest passes the address of the small integer instead, which should get
2524     rid of the warnings.
2525    
2526     10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
2527     both compile and run time, and gives an error if an invalid UTF-8 sequence
2528     is found. There is a option for disabling this check in cases where the
2529     string is known to be correct and/or the maximum performance is wanted.
2530    
2531     11. In response to a bug report, I changed one line in Makefile.in from
2532    
2533     -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
2534     to
2535     -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
2536    
2537     to look similar to other lines, but I have no way of telling whether this
2538     is the right thing to do, as I do not use Windows. No doubt I'll get told
2539     if it's wrong...
2540    
2541    
2542 nigel 69 Version 4.3 21-May-03
2543     ---------------------
2544    
2545     1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
2546     Makefile.
2547    
2548     2. Some refactoring to improve the quality of the code:
2549    
2550     (i) The utf8_table... variables are now declared "const".
2551    
2552     (ii) The code for \cx, which used the "case flipping" table to upper case
2553     lower case letters, now just substracts 32. This is ASCII-specific,
2554     but the whole concept of \cx is ASCII-specific, so it seems
2555     reasonable.
2556    
2557     (iii) PCRE was using its character types table to recognize decimal and
2558     hexadecimal digits in the pattern. This is silly, because it handles
2559     only 0-9, a-f, and A-F, but the character types table is locale-
2560     specific, which means strange things might happen. A private
2561     table is now used for this - though it costs 256 bytes, a table is
2562     much faster than multiple explicit tests. Of course, the standard
2563     character types table is still used for matching digits in subject
2564     strings against \d.
2565    
2566     (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
2567     ending in _t are). So I've renamed it as ESC_tee.
2568    
2569     3. The first argument for regexec() in the POSIX wrapper should have been
2570     defined as "const".
2571    
2572     4. Changed pcretest to use malloc() for its buffers so that they can be
2573     Electric Fenced for debugging.
2574    
2575     5. There were several places in the code where, in UTF-8 mode, PCRE would try
2576     to read one or more bytes before the start of the subject string. Often this
2577     had no effect on PCRE's behaviour, but in some circumstances it could
2578     provoke a segmentation fault.
2579    
2580     6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
2581     to try to read one or more bytes before the start of the subject string.
2582    
2583     7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
2584     UTF-8 support could misbehave in various ways if the subject string
2585     contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
2586     area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
2587     back over UTF-8 characters.)
2588    
2589    
2590 nigel 67 Version 4.2 14-Apr-03
2591     ---------------------
2592    
2593     1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
2594    
2595     2. Changes to the building process, supplied by Ronald Landheer-Cieslak
2596     [ON_WINDOWS]: new variable, "#" on non-Windows platforms
2597     [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
2598     [WIN_PREFIX]: new variable, "cyg" for Cygwin
2599     * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
2600     and BUILD_EXEEXT
2601     Note: automatic setting of the BUILD variables is not yet working
2602     set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
2603     compile-time but not at link-time
2604     [LINK]: use for linking executables only
2605     make different versions for Windows and non-Windows
2606     [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
2607     libraries
2608     [LINK_FOR_BUILD]: new variable
2609     [OBJEXT]: use throughout
2610     [EXEEXT]: use throughout
2611     <winshared>: new target
2612     <wininstall>: new target
2613     <dftables.o>: use native compiler
2614     <dftables>: use native linker
2615     <install>: handle Windows platform correctly
2616     <clean>: ditto
2617     <check>: ditto
2618     copy DLL to top builddir before testing
2619    
2620     As part of these changes, -no-undefined was removed again. This was reported
2621     to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
2622     in any case.
2623    
2624     3. Some tidies to get rid of compiler warnings:
2625    
2626     . In the match_data structure, match_limit was an unsigned long int, whereas
2627     match_call_count was an int. I've made them both unsigned long ints.
2628    
2629     . In pcretest the fact that a const uschar * doesn't automatically cast to
2630     a void * provoked a warning.
2631    
2632     . Turning on some more compiler warnings threw up some "shadow" variables
2633     and a few more missing casts.
2634    
2635     4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
2636     option, a class that contained a single character with a value between 128
2637     and 255 (e.g. /[\xFF]/) caused PCRE to crash.
2638    
2639     5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
2640     option, a class that contained several characters, but with at least one
2641     whose value was between 128 and 255 caused PCRE to crash.
2642    
2643    
2644 nigel 65 Version 4.1 12-Mar-03
2645     ---------------------
2646 nigel 63
2647 nigel 65 1. Compiling with gcc -pedantic found a couple of places where casts were
2648     needed, and a string in dftables.c that was longer than standard compilers are
2649     required to support.
2650    
2651     2. Compiling with Sun's compiler found a few more places where the code could
2652     be tidied up in order to avoid warnings.
2653    
2654     3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
2655     first of these names is deprecated in the latest Autoconf in favour of the name
2656     CC_FOR_BUILD, because "host" is typically used to mean the system on which the
2657     compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
2658     analogy I have changed it to CFLAGS_FOR_BUILD.
2659    
2660     4. Added -no-undefined to the linking command in the Makefile, because this is
2661     apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
2662     linking step for the pcreposix library.
2663    
2664     5. PCRE was failing to diagnose the case of two named groups with the same
2665     name.
2666    
2667     6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
2668     literal character that is needed in the subject for a match, and scans along to
2669     ensure that it is present before embarking on the full matching process. This
2670     saves time in cases of nested unlimited repeats that are never going to match.
2671     Problem: the scan can take a lot of time if the subject is very long (e.g.
2672     megabytes), thus penalizing straightforward matches. It is now done only if the
2673     amount of subject to be scanned is less than 1000 bytes.
2674    
2675     7. A lesser problem with the same optimization is that it was recording the
2676     first character of an anchored pattern as "needed", thus provoking a search
2677     right along the subject, even when the first match of the pattern was going to
2678     fail. The "needed" character is now not set for anchored patterns, unless it
2679     follows something in the pattern that is of non-fixed length. Thus, it still
2680     fulfils its original purpose of finding quick non-matches in cases of nested
2681     unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
2682    
2683    
2684     Version 4.0 17-Feb-03
2685     ---------------------
2686    
2687 nigel 63 1. If a comment in an extended regex that started immediately after a meta-item
2688     extended to the end of string, PCRE compiled incorrect data. This could lead to
2689     all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
2690    
2691     2. Moved to autoconf 2.53 and libtool 1.4.2.
2692    
2693     3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
2694     the special perltest8 script is no longer needed - all the tests can be run
2695     from a single perltest script.
2696    
2697     4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
2698     by \s. It has now been removed in PCRE. This means it isn't recognized as
2699     whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
2700     class [:space:] *does* include VT, thereby creating a mess.
2701    
2702     5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
2703     space and tab.
2704    
2705     6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
2706     its new features into the main test script, reducing the number of scripts.
2707    
2708     7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
2709     were backward compatible, and made the (?i) apply to the whole pattern, as if
2710     /i were given. Now it behaves more logically, and applies the option setting
2711     only to what follows. PCRE has been changed to follow suit. However, if it
2712     finds options settings right at the start of the pattern, it extracts them into
2713     the global options, as before. Thus, they show up in the info data.
2714    
2715     8. Added support for the \Q...\E escape sequence. Characters in between are
2716     treated as literals. This is slightly different from Perl in that $ and @ are
2717     also handled as literals inside the quotes. In Perl, they will cause variable
2718     interpolation. Note the following examples:
2719    
2720     Pattern PCRE matches Perl matches
2721    
2722     \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
2723     \Qabc\$xyz\E abc\$xyz abc\$xyz
2724     \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2725    
2726     For compatibility with Perl, \Q...\E sequences are recognized inside character
2727     classes as well as outside them.
2728    
2729     9. Re-organized 3 code statements in pcretest to avoid "overflow in
2730     floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
2731     (size_t) cast to one statement in pcretest and one in pcreposix to avoid
2732     signed/unsigned warnings.
2733    
2734     10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
2735     option for pcretest, so I've replaced it by a simple function that does just
2736     that job.
2737    
2738     11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
2739     "pcregrep -".
2740    
2741     12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
2742     Java package. This provides some syntactic sugar for simple cases of what my
2743     documentation calls "once-only subpatterns". A pattern such as x*+ is the same
2744     as (?>x*). In other words, if what is inside (?>...) is just a single repeated
2745     item, you can use this simplified notation. Note that only makes sense with
2746     greedy quantifiers. Consequently, the use of the possessive quantifier forces
2747     greediness, whatever the setting of the PCRE_UNGREEDY option.
2748    
2749     13. A change of greediness default within a pattern was not taking effect at
2750     the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
2751     subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
2752     was abstracted outside.
2753    
2754     14. PCRE now supports the \G assertion. It is true when the current matching
2755     position is at the start point of the match. This differs from \A when the
2756     starting offset is non-zero. Used with the /g option of pcretest (or similar
2757     code), it works in the same way as it does for Perl's /g option. If all
2758     alternatives of a regex begin with \G, the expression is anchored to the start
2759     match position, and the "anchored" flag is set in the compiled expression.
2760    
2761     15. Some bugs concerning the handling of certain option changes within patterns
2762     have been fixed. These applied to options other than (?ims). For example,
2763     "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
2764     been the other way round. Some of this was related to change 7 above.
2765    
2766     16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
2767     features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
2768     and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
2769     POSIX classes only within a class (e.g. /[[:alpha:]]/).
2770    
2771     17. Added support for Perl's \C escape. This matches one byte, even in UTF8
2772     mode. Unlike ".", it always matches newline, whatever the setting of
2773     PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
2774     assertions. Perl allows it, but it doesn't (in general) work because it can't
2775     calculate the length of the lookbehind. At least, that's the case for Perl
2776     5.8.0 - I've been told they are going to document that it doesn't work in
2777     future.
2778    
2779     18. Added an error diagnosis for escapes that PCRE does not support: these are
2780     \L, \l, \N, \P, \p, \U, \u, and \X.
2781    
2782     19. Although correctly diagnosing a missing ']' in a character class, PCRE was
2783     reading past the end of the pattern in cases such as /[abcd/.
2784    
2785     20. PCRE was getting more memory than necessary for patterns with classes that
2786     contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
2787    
2788     21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
2789     compiling PCRE for use with Virtual Pascal.
2790    
2791     22. Small fix to the Makefile to make it work properly if the build is done
2792     outside the source tree.
2793    
2794     23. Added a new extension: a condition to go with recursion. If a conditional
2795     subpattern starts with (?(R) the "true" branch is used if recursion has
2796     happened, whereas the "false" branch is used only at the top level.
2797    
2798     24. When there was a very long string of literal characters (over 255 bytes
2799     without UTF support, over 250 bytes with UTF support), the computation of how
2800     much memory was required could be incorrect, leading to segfaults or other
2801     strange effects.
2802    
2803     25. PCRE was incorrectly assuming anchoring (either to start of subject or to
2804     start of line for a non-DOTALL pattern) when a pattern started with (.*) and
2805     there was a subsequent back reference to those brackets. This meant that, for
2806     example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
2807     possible to check for precisely this case. All we can do is abandon the
2808     optimization if .* occurs inside capturing brackets when there are any back
2809     references whatsoever. (See below for a better fix that came later.)
2810    
2811     26. The handling of the optimization for finding the first character of a
2812     non-anchored pattern, and for finding a character that is required later in the
2813     match were failing in some cases. This didn't break the matching; it just
2814     failed to optimize when it could. The way this is done has been re-implemented.
2815    
2816     27. Fixed typo in error message for invalid (?R item (it said "(?p").
2817    
2818     28. Added a new feature that provides some of the functionality that Perl
2819     provides with (?{...}). The facility is termed a "callout". The way it is done
2820     in PCRE is for the caller to provide an optional function, by setting
2821     pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
2822     global variable. By default it is unset, which disables all calling out. To get
2823     the function called, the regex must include (?C) at appropriate points. This
2824     is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
2825     This provides a means of identifying different callout points. When PCRE
2826     reaches such a point in the regex, if pcre_callout has been set, the external
2827     function is called. It is provided with data in a structure called
2828     pcre_callout_block, which is defined in pcre.h. If the function returns 0,
2829     matching continues; if it returns a non-zero value, the match at the current
2830     point fails. However, backtracking will occur if possible. [This was changed
2831     later and other features added - see item 49 below.]
2832    
2833     29. pcretest is upgraded to test the callout functionality. It provides a
2834     callout function that displays information. By default, it shows the start of
2835     the match and the current position in the text. There are some new data escapes
2836     to vary what happens:
2837    
2838     \C+ in addition, show current contents of captured substrings
2839     \C- do not supply a callout function
2840     \C!n return 1 when callout number n is reached
2841     \C!n!m return 1 when callout number n is reached for the mth time
2842    
2843     30. If pcregrep was called with the -l option and just a single file name, it
2844     output "<stdin>" if a match was found, instead of the file name.
2845    
2846     31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
2847     slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
2848     pcre_exec(). This saves a malloc/free per call. The default value of
2849     POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
2850     when configuring.
2851    
2852     32. The default maximum size of a compiled pattern is 64K. There have been a
2853     few cases of people hitting this limit. The code now uses macros to handle the
2854     storing of links as offsets within the compiled pattern. It defaults to 2-byte
2855     links, but this can be changed to 3 or 4 bytes by --with-link-size when
2856     configuring. Tests 2 and 5 work only with 2-byte links because they output
2857     debugging information about compiled patterns.
2858    
2859     33. Internal code re-arrangements:
2860    
2861     (a) Moved the debugging function for printing out a compiled regex into
2862     its own source file (printint.c) and used #include to pull it into
2863     pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
2864     separate copies.
2865    
2866     (b) Defined the list of op-code names for debugging as a macro in
2867     internal.h so that it is next to the definition of the opcodes.
2868    
2869     (c) Defined a table of op-code lengths for simpler skipping along compiled
2870     code. This is again a macro in internal.h so that it is next to the
2871     definition of the opcodes.
2872    
2873     34. Added support for recursive calls to individual subpatterns, along the
2874     lines of Robin Houston's patch (but implemented somewhat differently).
2875    
2876     35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
2877     allow it to read and process whole directories in Win32. This code was
2878     contributed by Lionel Fourquaux; it has not been tested by me.
2879    
2880     36. Added support for named subpatterns. The Python syntax (?P<name>...) is
2881     used to name a group. Names consist of alphanumerics and underscores, and must
2882     be unique. Back references use the syntax (?P=name) and recursive calls use
2883     (?P>name) which is a PCRE extension to the Python extension. Groups still have
2884     numbers. The function pcre_fullinfo() can be used after compilation to extract
2885     a name/number map. There are three relevant calls:
2886    
2887     PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
2888     PCRE_INFO_NAMECOUNT yields the number of entries
2889     PCRE_INFO_NAMETABLE yields a pointer to the map.
2890    
2891     The map is a vector of fixed-size entries. The size of each entry depends on
2892     the length of the longest name used. The first two bytes of each entry are the
2893     group number, most significant byte first. There follows the corresponding
2894     name, zero terminated. The names are in alphabetical order.
2895    
2896     37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
2897     case instead of 255. Making it the same both with and without UTF-8 support
2898     means that the same test output works with both.
2899    
2900     38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
2901     calling malloc() with a zero argument.
2902    
2903     39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
2904     optimization. I've improved things by keeping a bitmap of backreferences with
2905     numbers 1-31 so that if .* occurs inside capturing brackets that are not in
2906     fact referenced, the optimization can be applied. It is unlikely that a
2907     relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
2908     the match to follow \n) will appear inside brackets with a number greater than
2909     31, but if it does, any back reference > 31 suppresses the optimization.
2910    
2911     40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
2912     of disabling numbered capturing parentheses. Any opening parenthesis that is
2913     not followed by ? behaves as if it were followed by ?: but named parentheses
2914     can still be used for capturing (and they will acquire numbers in the usual
2915     way).
2916    
2917     41. Redesigned the return codes from the match() function into yes/no/error so
2918     that errors can be passed back from deep inside the nested calls. A malloc
2919     failure while inside a recursive subpattern call now causes the
2920     PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
2921    
2922     42. It is now possible to set a limit on the number of times the match()
2923     function is called in a call to pcre_exec(). This facility makes it possible to
2924     limit the amount of recursion and backtracking, though not in a directly
2925     obvious way, because the match() function is used in a number of different
2926     circumstances. The count starts from zero for each position in the subject
2927     string (for non-anchored patterns). The default limit is, for compatibility, a
2928     large number, namely 10 000 000. You can change this in two ways:
2929    
2930     (a) When configuring PCRE before making, you can use --with-match-limit=n
2931     to set a default value for the compiled library.
2932    
2933     (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
2934     a different value is set. See 45 below.
2935    
2936     If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
2937    
2938     43. Added a new function pcre_config(int, void *) to enable run-time extraction
2939     of things that can be changed at compile time. The first argument specifies
2940     what is wanted and the second points to where the information is to be placed.
2941     The current list of available information is:
2942    
2943     PCRE_CONFIG_UTF8
2944    
2945     The output is an integer that is set to one if UTF-8 support is available;
2946     otherwise it is set to zero.
2947    
2948     PCRE_CONFIG_NEWLINE
2949    
2950     The output is an integer that it set to the value of the code that is used for
2951     newline. It is either LF (10) or CR (13).
2952    
2953     PCRE_CONFIG_LINK_SIZE
2954    
2955     The output is an integer that contains the number of bytes used for internal
2956     linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
2957    
2958     PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
2959    
2960     The output is an integer that contains the threshold above which the POSIX
2961     interface uses malloc() for output vectors. See item 31 above.
2962    
2963     PCRE_CONFIG_MATCH_LIMIT
2964    
2965     The output is an unsigned integer that contains the default limit of the number
2966     of match() calls in a pcre_exec() execution. See 42 above.
2967    
2968     44. pcretest has been upgraded by the addition of the -C option. This causes it
2969     to extract all the available output from the new pcre_config() function, and to
2970     output it. The program then exits immediately.
2971    
2972     45. A need has arisen to pass over additional data with calls to pcre_exec() in
2973     order to support additional features. One way would have been to define
2974     pcre_exec2() (for example) with extra arguments, but this would not have been
2975     extensible, and would also have required all calls to the original function to
2976     be mapped to the new one. Instead, I have chosen to extend the mechanism that
2977     is used for passing in "extra" data from pcre_study().
2978    
2979     The pcre_extra structure is now exposed and defined in pcre.h. It currently
2980     contains the following fields:
2981    
2982     flags a bitmap indicating which of the following fields are set
2983     study_data opaque data from pcre_study()
2984     match_limit a way of specifying a limit on match() calls for a specific
2985     call to pcre_exec()
2986     callout_data data for callouts (see 49 below)
2987    
2988     The flag bits are also defined in pcre.h, and are
2989    
2990     PCRE_EXTRA_STUDY_DATA
2991     PCRE_EXTRA_MATCH_LIMIT
2992     PCRE_EXTRA_CALLOUT_DATA
2993    
2994     The pcre_study() function now returns one of these new pcre_extra blocks, with
2995     the actual study data pointed to by the study_data field, and the
2996     PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
2997     before. That is, this change is entirely upwards-compatible and requires no
2998     change to existing code.
2999    
3000     If you want to pass in additional data to pcre_exec(), you can either place it
3001     in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
3002     block.
3003    
3004     46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
3005     data string contains the escape sequence \M, pcretest calls pcre_exec() several
3006     times with different match limits, until it finds the minimum value needed for
3007     pcre_exec() to complete. The value is then output. This can be instructive; for
3008     most simple matches the number is quite small, but for pathological cases it
3009     gets very large very quickly.
3010    
3011     47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
3012     returns the size of the data block pointed to by the study_data field in a
3013     pcre_extra block, that is, the value that was passed as the argument to
3014     pcre_malloc() when PCRE was getting memory in which to place the information
3015     created by pcre_study(). The fourth argument should point to a size_t variable.
3016     pcretest has been extended so that this information is shown after a successful
3017     pcre_study() call when information about the compiled regex is being displayed.
3018    
3019     48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
3020     because what follows is always an absolute path. (Later: it turns out that this
3021     is more than cosmetic for MinGW, because it doesn't like empty path
3022     components.)
3023    
3024     49. Some changes have been made to the callout feature (see 28 above):
3025    
3026     (i) A callout function now has three choices for what it returns:
3027    
3028     0 => success, carry on matching
3029     > 0 => failure at this point, but backtrack if possible
3030     < 0 => serious error, return this value from pcre_exec()
3031    
3032     Negative values should normally be chosen from the set of PCRE_ERROR_xxx
3033     values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
3034     "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
3035     use by callout functions. It will never be used by PCRE itself.
3036    
3037     (ii) The pcre_extra structure (see 45 above) has a void * field called
3038     callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
3039     pcre_callout_block structure has a field of the same name. The contents of
3040     the field passed in the pcre_extra structure are passed to the callout
3041     function in the corresponding field in the callout block. This makes it
3042     easier to use the same callout-containing regex from multiple threads. For
3043     testing, the pcretest program has a new data escape
3044    
3045     \C*n pass the number n (may be negative) as callout_data
3046    
3047     If the callout function in pcretest receives a non-zero value as
3048     callout_data, it returns that value.
3049    
3050     50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
3051     there were some redundant $(CFLAGS) in commands that are now specified as
3052     $(LINK), which already includes $(CFLAGS).
3053    
3054     51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
3055     has been compiled with UTF-8 support *and* pcre_compile() has been compiled
3056     with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
3057     one-byte characters throughout. Note that case-insensitive matching applies
3058     only to characters whose values are less than 256. PCRE doesn't support the
3059     notion of cases for higher-valued characters.
3060    
3061     (i) A character class whose characters are all within 0-255 is handled as
3062     a bit map, and the map is inverted for negative classes. Previously, a
3063     character > 255 always failed to match such a class; however it should
3064     match if the class was a negative one (e.g. [^ab]). This has been fixed.
3065    
3066     (ii) A negated character class with a single character < 255 is coded as
3067     "not this character" (OP_NOT). This wasn't working properly when the test
3068     character was multibyte, either singly or repeated.
3069    
3070     (iii) Repeats of multibyte characters are now handled correctly in UTF-8
3071     mode, for example: \x{100}{2,3}.
3072    
3073     (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
3074     singly or repeated) now correctly test multibyte characters. However,
3075     PCRE doesn't recognize any characters with values greater than 255 as
3076     digits, spaces, or word characters. Such characters always match \D, \S,
3077     and \W, and never match \d, \s, or \w.
3078    
3079     (v) Classes may now contain characters and character ranges with values
3080     greater than 255. For example: [ab\x{100}-\x{400}].
3081    
3082     (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
3083     PCRE in UTF-8 mode.
3084    
3085     52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
3086     PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
3087     retained for backwards compatibility. (Note that LASTLITERAL is also a byte
3088     value.)
3089    
3090     53. The single man page has become too large. I have therefore split it up into
3091     a number of separate man pages. These also give rise to individual HTML pages;
3092     these are now put in a separate directory, and there is an index.html page that
3093     lists them all. Some hyperlinking between the pages has been installed.
3094    
3095     54. Added convenience functions for handling named capturing parentheses.
3096    
3097     55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
3098     aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
3099     true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
3100     are faulted.
3101    
3102     56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
3103     calling configure. These values are used when compiling the dftables.c program
3104     which is run to generate the source of the default character tables. They
3105     default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
3106     you will need to set these values.
3107    
3108     57. Updated the building process for Windows DLL, as provided by Fred Cox.
3109    
3110    
3111     Version 3.9 02-Jan-02
3112 nigel 61 ---------------------
3113    
3114     1. A bit of extraneous text had somehow crept into the pcregrep documentation.
3115    
3116     2. If --disable-static was given, the building process failed when trying to
3117     build pcretest and pcregrep. (For some reason it was using libtool to compile
3118     them, which is not right, as they aren't part of the library.)
3119    
3120    
3121 nigel 59 Version 3.8 18-Dec-01
3122     ---------------------
3123    
3124     1. The experimental UTF-8 code was completely screwed up. It was packing the
3125     bytes in the wrong order. How dumb can you get?
3126    
3127    
3128 nigel 57 Version 3.7 29-Oct-01
3129     ---------------------
3130    
3131     1. In updating pcretest to check change 1 of version 3.6, I screwed up.
3132     This caused pcretest, when used on the test data, to segfault. Unfortunately,
3133     this didn't happen under Solaris 8, where I normally test things.
3134    
3135 nigel 59 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
3136     doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
3137     isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
3138     this fix an hour or so after the initial 3.7 release.)
3139 nigel 57
3140 nigel 59
3141 nigel 55 Version 3.6 23-Oct-01
3142     ---------------------
3143    
3144     1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
3145     offsets passed as NULL with zero offset count.
3146    
3147     2. The config.guess and config.sub files had not been updated when I moved to
3148     the latest autoconf.
3149    
3150    
3151 nigel 53 Version 3.5 15-Aug-01
3152     ---------------------
3153 nigel 5
3154 nigel 53 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
3155     had been forgotten.
3156    
3157     2. By using declared but undefined structures, we can avoid using "void"
3158     definitions in pcre.h while keeping the internal definitions of the structures
3159     private.
3160    
3161     3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
3162     user point of view, this means that both static and shared libraries are built
3163     by default, but this can be individually controlled. More of the work of
3164     handling this static/shared cases is now inside libtool instead of PCRE's make
3165     file.
3166    
3167     4. The pcretest utility is now installed along with pcregrep because it is
3168     useful for users (to test regexs) and by doing this, it automatically gets
3169     relinked by libtool. The documentation has been turned into a man page, so
3170     there are now .1, .txt, and .html versions in /doc.
3171    
3172     5. Upgrades to pcregrep:
3173     (i) Added long-form option names like gnu grep.
3174     (ii) Added --help to list all options with an explanatory phrase.
3175     (iii) Added -r, --recursive to recurse into sub-directories.
3176     (iv) Added -f, --file to read patterns from a file.
3177    
3178     6. pcre_exec() was referring to its "code" argument before testing that
3179     argument for NULL (and giving an error if it was NULL).
3180    
3181     7. Upgraded Makefile.in to allow for compiling in a different directory from
3182     the source directory.
3183    
3184     8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
3185     options bits, the pointer it was passed was to an int instead of to an unsigned
3186     long int. This mattered only on 64-bit systems.
3187    
3188     9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
3189     generated) instead of pcre.in, which it its source. Also made the same change
3190     in several of the .c files.
3191    
3192     10. A new release of gcc defines printf() as a macro, which broke pcretest
3193     because it had an ifdef in the middle of a string argument for printf(). Fixed
3194     by using separate calls to printf().
3195    
3196     11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
3197     script, to force use of CR or LF instead of \n in the source. On non-Unix
3198     systems, the value can be set in config.h.
3199    
3200     12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
3201     absolute limit. Changed the text of the error message to make this clear, and
3202     likewise updated the man page.
3203    
3204     13. The limit of 99 on the number of capturing subpatterns has been removed.
3205     The new limit is 65535, which I hope will not be a "real" limit.
3206    
3207    
3208 nigel 51 Version 3.4 22-Aug-00
3209     ---------------------
3210    
3211     1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
3212    
3213     2. Diagnose condition (?(0) as an error instead of crashing on matching.
3214    
3215    
3216 nigel 49 Version 3.3 01-Aug-00
3217     ---------------------
3218    
3219     1. If an octal character was given, but the value was greater than \377, it
3220     was not getting masked to the least significant bits, as documented. This could
3221     lead to crashes in some systems.
3222    
3223     2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
3224     the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
3225    
3226     3. Added the functions pcre_free_substring() and pcre_free_substring_list().
3227     These just pass their arguments on to (pcre_free)(), but they are provided
3228     because some uses of PCRE bind it to non-C systems that can call its functions,
3229     but cannot call free() or pcre_free() directly.
3230    
3231     4. Add "make test" as a synonym for "make check". Corrected some comments in
3232     the Makefile.
3233    
3234     5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
3235     Makefile.
3236    
3237     6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
3238     command called pgrep for grepping around the active processes.
3239    
3240     7. Added the beginnings of support for UTF-8 character strings.
3241    
3242     8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
3243     RANLIB to ./ltconfig so that they are used by libtool. I think these are all
3244     the relevant ones. (AR is not passed because ./ltconfig does its own figuring
3245     out for the ar command.)
3246    
3247    
3248 nigel 47 Version 3.2 12-May-00
3249     ---------------------
3250    
3251     This is purely a bug fixing release.
3252    
3253     1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
3254     of ZA. This was just one example of several cases that could provoke this bug,
3255     which was introduced by change 9 of version 2.00. The code for breaking
3256     infinite loops after an iteration that matches an empty string was't working
3257     correctly.
3258    
3259     2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
3260     when matched against abbab (for example). After matching an empty string, it
3261     wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
3262     caused it to match further down the string than it should.
3263    
3264     3. The code contained an inclusion of sys/types.h. It isn't clear why this
3265     was there because it doesn't seem to be needed, and it causes trouble on some
3266     systems, as it is not a Standard C header. It has been removed.
3267    
3268     4. Made 4 silly changes to the source to avoid stupid compiler warnings that
3269     were reported on the Macintosh. The changes were from
3270    
3271     while ((c = *(++ptr)) != 0 && c != '\n');
3272     to
3273     while ((c = *(++ptr)) != 0 && c != '\n') ;
3274    
3275     Totally extraordinary, but if that's what it takes...
3276    
3277     5. PCRE is being used in one environment where neither memmove() nor bcopy() is
3278     available. Added HAVE_BCOPY and an autoconf test for it; if neither
3279     HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
3280     assumes the way PCRE uses memmove() (always moving upwards).
3281    
3282     6. PCRE is being used in one environment where strchr() is not available. There
3283     was only one use in pcre.c, and writing it out to avoid strchr() probably gives
3284     faster code anyway.
3285    
3286    
3287 nigel 45 Version 3.1 09-Feb-00
3288     ---------------------
3289    
3290     The only change in this release is the fixing of some bugs in Makefile.in for
3291     the "install" target:
3292    
3293     (1) It was failing to install pcreposix.h.
3294    
3295     (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
3296    
3297    
3298 nigel 43 Version 3.0 01-Feb-00
3299     ---------------------
3300 nigel 41
3301     1. Add support for the /+ modifier to perltest (to output $` like it does in
3302     pcretest).
3303    
3304     2. Add support for the /g modifier to perltest.
3305    
3306     3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
3307     matches null strings.
3308    
3309     4. Fix perltest so that it doesn't do unwanted things when fed an empty
3310     pattern. Perl treats empty patterns specially - it reuses the most recent
3311     pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
3312     effect.
3313    
3314     5. The POSIX interface was broken in that it was just handing over the POSIX
3315     captured string vector to pcre_exec(), but (since release 2.00) PCRE has
3316     required a bigger vector, with some working space on the end. This means that
3317     the POSIX wrapper now has to get and free some memory, and copy the results.
3318    
3319 nigel 43 6. Added some simple autoconf support, placing the test data and the
3320     documentation in separate directories, re-organizing some of the
3321     information files, and making it build pcre-config (a GNU standard). Also added
3322     libtool support for building PCRE as a shared library, which is now the
3323     default.
3324 nigel 41
3325 nigel 43 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
3326     09 are not valid octal constants. Single digits will be used for minor values
3327     less than 10.
3328    
3329     8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
3330     existing programs that set these in the POSIX interface can use PCRE without
3331     modification.
3332    
3333     9. Added a new function, pcre_fullinfo() with an extensible interface. It can
3334     return all that pcre_info() returns, plus additional data. The pcre_info()
3335     function is retained for compatibility, but is considered to be obsolete.
3336    
3337     10. Added experimental recursion feature (?R) to handle one common case that
3338     Perl 5.6 will be able to do with (?p{...}).
3339    
3340     11. Added support for POSIX character classes like [:alpha:], which Perl is
3341     adopting.
3342    
3343    
3344 nigel 39 Version 2.08 31-Aug-99
3345     ----------------------
3346    
3347     1. When startoffset was not zero and the pattern began with ".*", PCRE was not
3348     trying to match at the startoffset position, but instead was moving forward to
3349     the next newline as if a previous match had failed.
3350    
3351     2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
3352     and could get into a loop if a null string was matched other than at the start
3353     of the subject.
3354    
3355     3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
3356     be distinguished at compile time, and for completeness also added PCRE_DATE.
3357    
3358     5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
3359     in GnuWin32 environments.
3360    
3361    
3362 nigel 37 Version 2.07 29-Jul-99
3363     ----------------------
3364    
3365     1. The documentation is now supplied in plain text form and HTML as well as in
3366     the form of man page sources.
3367    
3368     2. C++ compilers don't like assigning (void *) values to other pointer types.
3369     In particular this affects malloc(). Although there is no problem in Standard
3370     C, I've put in casts to keep C++ compilers happy.
3371    
3372     3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
3373     should be (const char *).
3374    
3375     4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
3376     be useful for non-Unix systems who don't want to bother with the POSIX stuff.
3377     However, I haven't made this a standard facility. The documentation doesn't
3378     mention it, and the Makefile doesn't support it.
3379    
3380     5. The Makefile now contains an "install" target, with editable destinations at
3381     the top of the file. The pcretest program is not installed.
3382    
3383     6. pgrep -V now gives the PCRE version number and date.
3384    
3385     7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
3386     causing the entire string to be ignored, instead of just the last character.
3387    
3388     8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
3389     non-matching string, it can take a very, very long time, even for strings of
3390     quite modest length, because of the nested recursion. PCRE now does better in
3391     some of these cases. It does this by remembering the last required literal
3392     character in the pattern, and pre-searching the subject to ensure it is present
3393     before running the real match. In other words, it applies a heuristic to detect
3394     some types of certain failure quickly, and in the above example, if presented
3395     with a string that has no trailing " it gives "no match" very quickly.
3396    
3397     9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
3398     other alternatives are tried instead.
3399    
3400    
3401 nigel 35 Version 2.06 09-Jun-99
3402     ----------------------
3403    
3404     1. Change pcretest's output for amount of store used to show just the code
3405     space, because the remainder (the data block) varies in size between 32-bit and
3406     64-bit systems.
3407    
3408     2. Added an extra argument to pcre_exec() to supply an offset in the subject to
3409     start matching at. This allows lookbehinds to work when searching for multiple
3410     occurrences in a string.
3411    
3412     3. Added additional options to pcretest for testing multiple occurrences:
3413    
3414     /+ outputs the rest of the string that follows a match
3415     /g loops for multiple occurrences, using the new startoffset argument
3416     /G loops for multiple occurrences by passing an incremented pointer
3417    
3418     4. PCRE wasn't doing the "first character" optimization for patterns starting
3419     with \b or \B, though it was doing it for other lookbehind assertions. That is,
3420     it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
3421     the letter 'x'. On long subject strings, this gives a significant speed-up.
3422    
3423    
3424 nigel 33 Version 2.05 21-Apr-99
3425     ----------------------
3426    
3427     1. Changed the type of magic_number from int to long int so that it works
3428     properly on 16-bit systems.
3429    
3430     2. Fixed a bug which caused patterns starting with .* not to work correctly
3431     when the subject string contained newline characters. PCRE was assuming
3432     anchoring for such patterns in all cases, which is not correct because .* will
3433     not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
3434     DOTALL is set at top level; otherwise it knows that patterns starting with .*
3435     must be retried after every newline in the subject.
3436    
3437    
3438 nigel 31 Version 2.04 18-Feb-99
3439     ----------------------
3440    
3441     1. For parenthesized subpatterns with repeats whose minimum was zero, the
3442     computation of the store needed to hold the pattern was incorrect (too large).
3443     If such patterns were nested a few deep, this could multiply and become a real
3444     problem.
3445    
3446     2. Added /M option to pcretest to show the memory requirement of a specific
3447     pattern. Made -m a synonym of -s (which does this globally) for compatibility.
3448    
3449     3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
3450     compiled in such a way that the backtracking after subsequent failure was
3451     pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
3452     ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
3453    
3454    
3455 nigel 29 Version 2.03 02-Feb-99
3456     ----------------------
3457    
3458     1. Fixed typo and small mistake in man page.
3459    
3460 nigel 31 2. Added 4th condition (GPL supersedes if conflict) and created separate
3461     LICENCE file containing the conditions.
3462 nigel 29
3463     3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
3464     Perl, that is the internal \ allows the delimiter to be included in the
3465     pattern. Locked out the use of \ as a delimiter. If \ immediately follows
3466     the final delimiter, add \ to the end of the pattern (to test the error).
3467    
3468     4. Added the convenience functions for extracting substrings after a successful
3469     match. Updated pcretest to make it able to test these functions.
3470    
3471    
3472 nigel 27 Version 2.02 14-Jan-99
3473     ----------------------
3474    
3475     1. Initialized the working variables associated with each extraction so that
3476     their saving and restoring doesn't refer to uninitialized store.
3477    
3478     2. Put dummy code into study.c in order to trick the optimizer of the IBM C
3479     compiler for OS/2 into generating correct code. Apparently IBM isn't going to
3480     fix the problem.
3481    
3482     3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
3483     calls, and wasn't printing the correct value for compiling calls. Increased the
3484     default value of LOOPREPEAT, and the number of significant figures in the
3485     times.
3486    
3487     4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
3488    
3489     5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
3490     a building problem on Windows NT with a FAT file system.
3491    
3492    
3493 nigel 25 Version 2.01 21-Oct-98
3494     ----------------------
3495    
3496     1. Changed the API for pcre_compile() to allow for the provision of a pointer
3497     to character tables built by pcre_maketables() in the current locale. If NULL
3498     is passed, the default tables are used.
3499    
3500    
3501 nigel 23 Version 2.00 24-Sep-98
3502 nigel 21 ----------------------
3503    
3504 nigel 23 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
3505     it any more.
3506    
3507     2. Allow quantification of (?>) groups, and make it work correctly.
3508    
3509     3. The first character computation wasn't working for (?>) groups.
3510    
3511     4. Correct the implementation of \Z (it is permitted to match on the \n at the
3512     end of the subject) and add 5.005's \z, which really does match only at the
3513     very end of the subject.
3514    
3515     5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
3516    
3517     6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
3518     DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
3519     localized options. All options to pcre_study() were also removed.
3520    
3521     7. Add other new features from 5.005:
3522    
3523     $(?<= positive lookbehind
3524     $(?<! negative lookbehind
3525     (?imsx-imsx) added the unsetting capability
3526     such a setting is global if at outer level; local otherwise
3527     (?imsx-imsx:) non-capturing groups with option setting
3528     (?(cond)re|re) conditional pattern matching
3529    
3530     A backreference to itself in a repeated group matches the previous
3531     captured string.
3532    
3533     8. General tidying up of studying (both automatic and via "study")
3534     consequential on the addition of new assertions.
3535    
3536     9. As in 5.005, unlimited repeated groups that could match an empty substring
3537     are no longer faulted at compile time. Instead, the loop is forcibly broken at
3538     runtime if any iteration does actually match an empty substring.
3539    
3540     10. Include the RunTest script in the distribution.
3541    
3542     11. Added tests from the Perl 5.005_02 distribution. This showed up a few
3543     discrepancies, some of which were old and were also with respect to 5.004. They
3544     have now been fixed.
3545    
3546    
3547     Version 1.09 28-Apr-98
3548     ----------------------
3549    
3550 nigel 21 1. A negated single character class followed by a quantifier with a minimum
3551     value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
3552     program crashes, or just wrong answers. This did not apply to negated classes
3553     containing more than one character, or to minima other than one.
3554    
3555    
3556 nigel 19 Version 1.08 27-Mar-98
3557     ----------------------
3558    
3559     1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
3560    
3561     2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
3562     latter must appear before anything that relies on it in the pattern.
3563    
3564    
3565 nigel 17 Version 1.07 16-Feb-98
3566     ----------------------
3567    
3568     1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
3569     repeat of a potentially empty string).
3570    
3571    
3572 nigel 15 Version 1.06 23-Jan-98
3573     ----------------------
3574    
3575     1. Added Markus Oberhumer's little patches for C++.
3576    
3577     2. Literal strings longer than 255 characters were broken.
3578    
3579    
3580 nigel 13 Version 1.05 23-Dec-97
3581     ----------------------
3582    
3583     1. Negated character classes containing more than one character were failing if
3584     PCRE_CASELESS was set at run time.
3585    
3586    
3587 nigel 11 Version 1.04 19-Dec-97
3588     ----------------------
3589    
3590     1. Corrected the man page, where some "const" qualifiers had been omitted.
3591    
3592     2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
3593     input syntax.
3594    
3595     3. Fixed memory leak which occurred when a regex with back references was
3596     matched with an offsets vector that wasn't big enough. The temporary memory
3597     that is used in this case wasn't being freed if the match failed.
3598    
3599     4. Tidied pcretest to ensure it frees memory that it gets.
3600    
3601     5. Temporary memory was being obtained in the case where the passed offsets
3602     vector was exactly big enough.
3603    
3604     6. Corrected definition of offsetof() from change 5 below.
3605    
3606     7. I had screwed up change 6 below and broken the rules for the use of
3607     setjmp(). Now fixed.
3608    
3609    
3610 nigel 9 Version 1.03 18-Dec-97
3611     ----------------------
3612    
3613     1. A erroneous regex with a missing opening parenthesis was correctly
3614     diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
3615     on some systems.
3616    
3617     2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
3618     it was reported that one broken compiler failed on the former because "code" is
3619     also an independent variable.
3620    
3621     3. The erroneous regex a[]b caused an array overrun reference.
3622    
3623     4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
3624     fail on data ending with that character. (It was going on too far, and checking
3625     the next character, typically a binary zero.) This was specific to the
3626     optimized code for single-character negative classes.
3627    
3628     5. Added a contributed patch from the TIN world which does the following:
3629    
3630     + Add an undef for memmove, in case the the system defines a macro for it.
3631    
3632     + Add a definition of offsetof(), in case there isn't one. (I don't know
3633     the reason behind this - offsetof() is part of the ANSI standard - but
3634     it does no harm).
3635    
3636     + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
3637     most of the places where whitespace preceded '#'. I have given up and
3638     allowed the remaining 2 cases to be at the margin.
3639    
3640     + Rename some variables in pcre to eliminate shadowing. This seems very
3641     pedantic, but does no harm, of course.
3642    
3643     6. Moved the call to setjmp() into its own function, to get rid of warnings
3644     from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
3645    
3646     7. Constructs such as \d{8,} were compiling into the equivalent of
3647 nigel 11 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
3648 nigel 9 outcome, but in this particular case used more store than had been allocated,
3649     which caused the bug to be discovered because it threw up an internal error.
3650    
3651     8. The debugging code in both pcre and pcretest for outputting the compiled
3652     form of a regex was going wrong in the case of back references followed by
3653     curly-bracketed repeats.
3654    
3655    
3656 nigel 7 Version 1.02 12-Dec-97
3657     ----------------------
3658    
3659     1. Typos in pcre.3 and comments in the source fixed.
3660    
3661     2. Applied a contributed patch to get rid of places where it used to remove
3662     'const' from variables, and fixed some signed/unsigned and uninitialized
3663     variable warnings.
3664    
3665     3. Added the "runtest" target to Makefile.
3666    
3667     4. Set default compiler flag to -O2 rather than just -O.
3668    
3669    
3670 nigel 5 Version 1.01 19-Nov-97
3671     ----------------------
3672    
3673     1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
3674     like /([ab]*)*/, that is, for classes with more than one character in them.
3675    
3676     2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
3677     as /((?>a*))*/ (a PCRE_EXTRA facility).
3678    
3679    
3680     Version 1.00 18-Nov-97
3681     ----------------------
3682    
3683     1. Added compile-time macros to support systems such as SunOS4 which don't have
3684     memmove() or strerror() but have other things that can be used instead.
3685    
3686     2. Arranged that "make clean" removes the executables.
3687    
3688    
3689 nigel 3 Version 0.99 27-Oct-97
3690     ----------------------
3691    
3692     1. Fixed bug in code for optimizing classes with only one character. It was
3693     initializing a 32-byte map regardless, which could cause it to run off the end
3694     of the memory it had got.
3695    
3696     2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
3697    
3698    
3699     Version 0.98 22-Oct-97
3700     ----------------------
3701    
3702     1. Fixed bug in code for handling temporary memory usage when there are more
3703     back references than supplied space in the ovector. This could cause segfaults.
3704    
3705    
3706     Version 0.97 21-Oct-97
3707     ----------------------
3708    
3709     1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
3710    
3711     2. Optimized negated single characters not to use a bit map.
3712    
3713     3. Brought error texts together as macro definitions; clarified some of them;
3714     fixed one that was wrong - it said "range out of order" when it meant "invalid
3715     escape sequence".
3716    
3717     4. Changed some char * arguments to const char *.
3718    
3719     5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
3720    
3721     6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
3722     pcretest.
3723    
3724    
3725     Version 0.96 16-Oct-97
3726     ----------------------
3727    
3728     1. Added a simple "pgrep" utility to the distribution.
3729    
3730     2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
3731     unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
3732     where "ddd" means "one or more decimal digits".
3733    
3734     3. Fixed serious bug. If a pattern had a back reference, but the call to
3735     pcre_exec() didn't supply a large enough ovector to record the related
3736     identifying subpattern, the match always failed. PCRE now remembers the number
3737     of the largest back reference, and gets some temporary memory in which to save
3738     the offsets during matching if necessary, in order to ensure that
3739     backreferences always work.
3740    
3741     4. Increased the compatibility with Perl in a number of ways:
3742    
3743     (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
3744     to request this handling. The option can be set at compile or exec time.
3745    
3746     (b) $ matches before a terminating newline by default; an option
3747     PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
3748     mode). The option can be set at compile or exec time.
3749    
3750     (c) The handling of \ followed by a digit other than 0 is now supposed to be
3751     the same as Perl's. If the decimal number it represents is less than 10
3752     or there aren't that many previous left capturing parentheses, an octal
3753     escape is read. Inside a character class, it's always an octal escape,
3754     even if it is a single digit.
3755    
3756     (d) An escaped but undefined alphabetic character is taken as a literal,