Return to ChangeLog CVS log

Up to [ELWIX - Embedded LightWeight unIX -] / embedaddon / pcre

pcre

    1: ChangeLog for PCRE
    2: ------------------
    3: 
    4: Version 8.21 12-Dec-2011
    5: ------------------------
    6: 
    7: 1.  Updating the JIT compiler.
    8: 
    9: 2.  JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
   10:     are added as well.
   11: 
   12: 3.  Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
   13:     PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
   14:     calling _pcre_jit_exec. Some extra comments are added.
   15: 
   16: 4.  (*MARK) settings inside atomic groups that do not contain any capturing
   17:     parentheses, for example, (?>a(*:m)), were not being passed out. This bug
   18:     was introduced by change 18 for 8.20.
   19: 
   20: 5.  Supporting of \x, \U and \u in JavaScript compatibility mode based on the
   21:     ECMA-262 standard.
   22: 
   23: 6.  Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
   24:     erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
   25:     This bug was probably introduced by change 9 of 8.13.
   26: 
   27: 7.  While fixing 6 above, I noticed that a number of other items were being
   28:     incorrectly rejected as "not fixed length". This arose partly because newer
   29:     opcodes had not been added to the fixed-length checking code. I have (a)
   30:     corrected the bug and added tests for these items, and (b) arranged for an
   31:     error to occur if an unknown opcode is encountered while checking for fixed
   32:     length instead of just assuming "not fixed length". The items that were
   33:     rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
   34:     (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
   35:     repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
   36: 
   37: 8.  A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
   38:     being incorrectly compiled and would have given unpredicatble results.
   39: 
   40: 9.  A possessively repeated subpattern with minimum repeat count greater than
   41:     one behaved incorrectly. For example, (A){2,}+ behaved as if it was
   42:     (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
   43:     the first (A) could occur when it should not.
   44: 
   45: 10. Add a cast and remove a redundant test from the code.
   46: 
   47: 11. JIT should use pcre_malloc/pcre_free for allocation.
   48: 
   49: 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
   50:     best practice nowadays, and helps with cross-compiling. (If the exec_prefix
   51:     is anything other than /usr, -L is still shown).
   52: 
   53: 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
   54: 
   55: 14. Perl does not support \N without a following name in a [] class; PCRE now
   56:     also gives an error.
   57: 
   58: 15. If a forward reference was repeated with an upper limit of around 2000,
   59:     it caused the error "internal error: overran compiling workspace". The
   60:     maximum number of forward references (including repeats) was limited by the
   61:     internal workspace, and dependent on the LINK_SIZE. The code has been
   62:     rewritten so that the workspace expands (via pcre_malloc) if necessary, and
   63:     the default depends on LINK_SIZE. There is a new upper limit (for safety)
   64:     of around 200,000 forward references. While doing this, I also speeded up
   65:     the filling in of repeated forward references.
   66: 
   67: 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
   68:     incorrectly expecting the subject to contain another "a" after the start.
   69: 
   70: 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
   71:     in the match, the SKIP should be ignored. This was not happening; instead
   72:     the SKIP was being treated as NOMATCH. For patterns such as
   73:     /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
   74:     tested.
   75: 
   76: 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
   77:     now much more compatible with Perl, in particular in cases where the result
   78:     is a non-match for a non-anchored pattern. For example, if
   79:     /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
   80:     "m", where previously it did not return a name. A side effect of this
   81:     change is that for partial matches, the last encountered mark name is
   82:     returned, as for non matches. A number of tests that were previously not
   83:     Perl-compatible have been moved into the Perl-compatible test files. The
   84:     refactoring has had the pleasing side effect of removing one argument from
   85:     the match() function, thus reducing its stack requirements.
   86: 
   87: 19. If the /S+ option was used in pcretest to study a pattern using JIT,
   88:     subsequent uses of /S (without +) incorrectly behaved like /S+.
   89: 
   90: 21. Retrieve executable code size support for the JIT compiler and fixing
   91:     some warnings.
   92: 
   93: 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
   94:     not work when the shorter character appeared right at the end of the
   95:     subject string.
   96: 
   97: 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
   98:     systems.
   99: 
  100: 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
  101:     output it when the /M option is used in pcretest.
  102: 
  103: 25. The CheckMan script was not being included in the distribution. Also, added
  104:     an explicit "perl" to run Perl scripts from the PrepareRelease script
  105:     because this is reportedly needed in Windows.
  106: 
  107: 26. If study data was being save in a file and studying had not found a set of
  108:     "starts with" bytes for the pattern, the data written to the file (though
  109:     never used) was taken from uninitialized memory and so caused valgrind to
  110:     complain.
  111: 
  112: 27. Updated RunTest.bat as provided by Sheri Pierce.
  113: 
  114: 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
  115: 
  116: 29. Computation of memory usage for the table of capturing group names was
  117:     giving an unnecessarily large value.
  118: 
  119: 
  120: Version 8.20 21-Oct-2011
  121: ------------------------
  122: 
  123: 1.  Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
  124:     a POSIX class. After further experiments with Perl, which convinced me that
  125:     Perl has bugs and confusions, a closing square bracket is no longer allowed
  126:     in a POSIX name. This bug also affected patterns with classes that started
  127:     with full stops.
  128: 
  129: 2.  If a pattern such as /(a)b|ac/ is matched against "ac", there is no
  130:     captured substring, but while checking the failing first alternative,
  131:     substring 1 is temporarily captured. If the output vector supplied to
  132:     pcre_exec() was not big enough for this capture, the yield of the function
  133:     was still zero ("insufficient space for captured substrings"). This cannot
  134:     be totally fixed without adding another stack variable, which seems a lot
  135:     of expense for a edge case. However, I have improved the situation in cases
  136:     such as /(a)(b)x|abc/ matched against "abc", where the return code
  137:     indicates that fewer than the maximum number of slots in the ovector have
  138:     been set.
  139: 
  140: 3.  Related to (2) above: when there are more back references in a pattern than
  141:     slots in the output vector, pcre_exec() uses temporary memory during
  142:     matching, and copies in the captures as far as possible afterwards. It was
  143:     using the entire output vector, but this conflicts with the specification
  144:     that only 2/3 is used for passing back captured substrings. Now it uses
  145:     only the first 2/3, for compatibility. This is, of course, another edge
  146:     case.
  147: 
  148: 4.  Zoltan Herczeg's just-in-time compiler support has been integrated into the
  149:     main code base, and can be used by building with --enable-jit. When this is
  150:     done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
  151:     runtime --no-jit option is given.
  152: 
  153: 5.  When the number of matches in a pcre_dfa_exec() run exactly filled the
  154:     ovector, the return from the function was zero, implying that there were
  155:     other matches that did not fit. The correct "exactly full" value is now
  156:     returned.
  157: 
  158: 6.  If a subpattern that was called recursively or as a subroutine contained
  159:     (*PRUNE) or any other control that caused it to give a non-standard return,
  160:     invalid errors such as "Error -26 (nested recursion at the same subject
  161:     position)" or even infinite loops could occur.
  162: 
  163: 7.  If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
  164:     computing the minimum length on reaching *ACCEPT, and so ended up with the
  165:     wrong value of 1 rather than 0. Further investigation indicates that
  166:     computing a minimum subject length in the presence of *ACCEPT is difficult
  167:     (think back references, subroutine calls), and so I have changed the code
  168:     so that no minimum is registered for a pattern that contains *ACCEPT.
  169: 
  170: 8.  If (*THEN) was present in the first (true) branch of a conditional group,
  171:     it was not handled as intended. [But see 16 below.]
  172: 
  173: 9.  Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
  174:     Sheri Pierce.
  175: 
  176: 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
  177:     the first byte in a match must be "a".
  178: 
  179: 11. Change 17 for 8.13 increased the recursion depth for patterns like
  180:     /a(?:.)*?a/ drastically. I've improved things by remembering whether a
  181:     pattern contains any instances of (*THEN). If it does not, the old
  182:     optimizations are restored. It would be nice to do this on a per-group
  183:     basis, but at the moment that is not feasible.
  184: 
  185: 12. In some environments, the output of pcretest -C is CRLF terminated. This
  186:     broke RunTest's code that checks for the link size. A single white space
  187:     character after the value is now allowed for.
  188: 
  189: 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
  190:     For "fr", it uses the Windows-specific input and output files.
  191: 
  192: 14. If (*THEN) appeared in a group that was called recursively or as a
  193:     subroutine, it did not work as intended. [But see next item.]
  194: 
  195: 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
  196:     pattern fragments (but not containing any | characters). If A and B are
  197:     matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
  198:     was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
  199:     D. In other words, Perl considers parentheses that do not contain any |
  200:     characters to be part of a surrounding alternative, whereas PCRE was
  201:     treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
  202:     differently. PCRE now behaves in the same way as Perl, except in the case
  203:     of subroutine/recursion calls such as (?1) which have in any case always
  204:     been different (but PCRE had them first :-).
  205: 
  206: 16. Related to 15 above: Perl does not treat the | in a conditional group as
  207:     creating alternatives. Such a group is treated in the same way as an
  208:     ordinary group without any | characters when processing (*THEN). PCRE has
  209:     been changed to match Perl's behaviour.
  210: 
  211: 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
  212:     RunGrepTest script failed.
  213: 
  214: 18. Change 22 for version 13 caused atomic groups to use more stack. This is
  215:     inevitable for groups that contain captures, but it can lead to a lot of
  216:     stack use in large patterns. The old behaviour has been restored for atomic
  217:     groups that do not contain any capturing parentheses.
  218: 
  219: 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
  220:     suppress the check for a minimum subject length at run time. (If it was
  221:     given to pcre_exec() or pcre_dfa_exec() it did work.)
  222: 
  223: 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
  224:     fail to work when decoding hex characters in data strings in EBCDIC
  225:     environments.
  226: 
  227: 21. It appears that in at least one Mac OS environment, the isxdigit() function
  228:     is implemented as a macro that evaluates to its argument more than once,
  229:     contravening the C 90 Standard (I haven't checked a later standard). There
  230:     was an instance in pcretest which caused it to go wrong when processing
  231:     \x{...} escapes in subject strings. The has been rewritten to avoid using
  232:     things like p++ in the argument of isxdigit().
  233: 
  234: 
  235: Version 8.13 16-Aug-2011
  236: ------------------------
  237: 
  238: 1.  The Unicode data tables have been updated to Unicode 6.0.0.
  239: 
  240: 2.  Two minor typos in pcre_internal.h have been fixed.
  241: 
  242: 3.  Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
  243:     pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
  244:     in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
  245: 
  246: 4.  There were a number of related bugs in the code for matching backrefences
  247:     caselessly in UTF-8 mode when codes for the characters concerned were
  248:     different numbers of bytes. For example, U+023A and U+2C65 are an upper
  249:     and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
  250:     (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
  251:     code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
  252:     2-byte code at the end of the subject (it thought there wasn't enough data
  253:     left).
  254: 
  255: 5.  Comprehensive information about what went wrong is now returned by
  256:     pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
  257:     as the output vector has at least 2 elements. The offset of the start of
  258:     the failing character and a reason code are placed in the vector.
  259: 
  260: 6.  When the UTF-8 string check fails for pcre_compile(), the offset that is
  261:     now returned is for the first byte of the failing character, instead of the
  262:     last byte inspected. This is an incompatible change, but I hope it is small
  263:     enough not to be a problem. It makes the returned offset consistent with
  264:     pcre_exec() and pcre_dfa_exec().
  265: 
  266: 7.  pcretest now gives a text phrase as well as the error number when
  267:     pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
  268:     failure, the offset and reason code are output.
  269: 
  270: 8.  When \R was used with a maximizing quantifier it failed to skip backwards
  271:     over a \r\n pair if the subsequent match failed. Instead, it just skipped
  272:     back over a single character (\n). This seems wrong (because it treated the
  273:     two characters as a single entity when going forwards), conflicts with the
  274:     documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
  275:     behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
  276:     has been changed.
  277: 
  278: 9.  Some internal refactoring has changed the processing so that the handling
  279:     of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
  280:     time (the PCRE_DOTALL option was changed this way some time ago: version
  281:     7.7 change 16). This has made it possible to abolish the OP_OPT op code,
  282:     which was always a bit of a fudge. It also means that there is one less
  283:     argument for the match() function, which reduces its stack requirements
  284:     slightly. This change also fixes an incompatibility with Perl: the pattern
  285:     (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
  286: 
  287: 10. More internal refactoring has drastically reduced the number of recursive
  288:     calls to match() for possessively repeated groups such as (abc)++ when
  289:     using pcre_exec().
  290: 
  291: 11. While implementing 10, a number of bugs in the handling of groups were
  292:     discovered and fixed:
  293: 
  294:     (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
  295:     (a|)*(?1) gave a compile-time internal error.
  296:     ((a|)+)+  did not notice that the outer group could match an empty string.
  297:     (^a|^)+   was not marked as anchored.
  298:     (.*a|.*)+ was not marked as matching at start or after a newline.
  299: 
  300: 12. Yet more internal refactoring has removed another argument from the match()
  301:     function. Special calls to this function are now indicated by setting a
  302:     value in a variable in the "match data" data block.
  303: 
  304: 13. Be more explicit in pcre_study() instead of relying on "default" for
  305:     opcodes that mean there is no starting character; this means that when new
  306:     ones are added and accidentally left out of pcre_study(), testing should
  307:     pick them up.
  308: 
  309: 14. The -s option of pcretest has been documented for ages as being an old
  310:     synonym of -m (show memory usage). I have changed it to mean "force study
  311:     for every regex", that is, assume /S for every regex. This is similar to -i
  312:     and -d etc. It's slightly incompatible, but I'm hoping nobody is still
  313:     using it. It makes it easier to run collections of tests with and without
  314:     study enabled, and thereby test pcre_study() more easily. All the standard
  315:     tests are now run with and without -s (but some patterns can be marked as
  316:     "never study" - see 20 below).
  317: 
  318: 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
  319:     restoration of the capturing data to the outer values was not happening
  320:     correctly.
  321: 
  322: 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
  323:     empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
  324:     pattern had matched an empty string, and so incorrectly returned a no
  325:     match.
  326: 
  327: 17. There was optimizing code for the last branch of non-capturing parentheses,
  328:     and also for the obeyed branch of a conditional subexpression, which used
  329:     tail recursion to cut down on stack usage. Unfortunately, now that there is
  330:     the possibility of (*THEN) occurring in these branches, tail recursion is
  331:     no longer possible because the return has to be checked for (*THEN). These
  332:     two optimizations have therefore been removed. [But see 8.20/11 above.]
  333: 
  334: 18. If a pattern containing \R was studied, it was assumed that \R always
  335:     matched two bytes, thus causing the minimum subject length to be
  336:     incorrectly computed because \R can also match just one byte.
  337: 
  338: 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
  339:     was incorrectly computed.
  340: 
  341: 20. If /S is present twice on a test pattern in pcretest input, it now
  342:     *disables* studying, thereby overriding the use of -s on the command line
  343:     (see 14 above). This is necessary for one or two tests to keep the output
  344:     identical in both cases.
  345: 
  346: 21. When (*ACCEPT) was used in an assertion that matched an empty string and
  347:     PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
  348: 
  349: 22. When an atomic group that contained a capturing parenthesis was
  350:     successfully matched, but the branch in which it appeared failed, the
  351:     capturing was not being forgotten if a higher numbered group was later
  352:     captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
  353:     group 1 to "a", when in fact it should be unset. This applied to multi-
  354:     branched capturing and non-capturing groups, repeated or not, and also to
  355:     positive assertions (capturing in negative assertions does not happen
  356:     in PCRE) and also to nested atomic groups.
  357: 
  358: 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
  359:     subject after a captured substring, to make it easier to tell which of a
  360:     number of identical substrings has been captured.
  361: 
  362: 24. The way atomic groups are processed by pcre_exec() has been changed so that
  363:     if they are repeated, backtracking one repetition now resets captured
  364:     values correctly. For example, if ((?>(a+)b)+aabab) is matched against
  365:     "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
  366:     "aaa". Previously, it would have been "a". As part of this code
  367:     refactoring, the way recursive calls are handled has also been changed.
  368: 
  369: 25. If an assertion condition captured any substrings, they were not passed
  370:     back unless some other capturing happened later. For example, if
  371:     (?(?=(a))a) was matched against "a", no capturing was returned.
  372: 
  373: 26. When studying a pattern that contained subroutine calls or assertions,
  374:     the code for finding the minimum length of a possible match was handling
  375:     direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
  376:     group 1 called group 2 while simultaneously a separate group 2 called group
  377:     1). A stack overflow occurred in this case. I have fixed this by limiting
  378:     the recursion depth to 10.
  379: 
  380: 27. Updated RunTest.bat in the distribution to the version supplied by Tom
  381:     Fortmann. This supports explicit test numbers on the command line, and has
  382:     argument validation and error reporting.
  383: 
  384: 28. An instance of \X with an unlimited repeat could fail if at any point the
  385:     first character it looked at was a mark character.
  386: 
  387: 29. Some minor code refactoring concerning Unicode properties and scripts
  388:     should reduce the stack requirement of match() slightly.
  389: 
  390: 30. Added the '=' option to pcretest to check the setting of unused capturing
  391:     slots at the end of the pattern, which are documented as being -1, but are
  392:     not included in the return count.
  393: 
  394: 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
  395:     compiled something random. Now it gives a compile-time error (as does
  396:     Perl).
  397: 
  398: 32. A *MARK encountered during the processing of a positive assertion is now
  399:     recorded and passed back (compatible with Perl).
  400: 
  401: 33. If --only-matching or --colour was set on a pcregrep call whose pattern
  402:     had alternative anchored branches, the search for a second match in a line
  403:     was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
  404:     matched the line "0102" twice. The same bug affected patterns that started
  405:     with a backwards assertion. For example /\b01|\b02/ also matched "0102"
  406:     twice.
  407: 
  408: 34. Previously, PCRE did not allow quantification of assertions. However, Perl
  409:     does, and because of capturing effects, quantifying parenthesized
  410:     assertions may at times be useful. Quantifiers are now allowed for
  411:     parenthesized assertions.
  412: 
  413: 35. A minor code tidy in pcre_compile() when checking options for \R usage.
  414: 
  415: 36. \g was being checked for fancy things in a character class, when it should
  416:     just be a literal "g".
  417: 
  418: 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
  419:     appearance of a nested POSIX class supersedes an apparent external class.
  420:     For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
  421:     unescaped square brackets may also appear as part of class names. For
  422:     example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
  423:     more like Perl. (But see 8.20/1 above.)
  424: 
  425: 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
  426:     was because it thought it was \N{name}, which is not supported).
  427: 
  428: 39. Add minix to OS list not supporting the -S option in pcretest.
  429: 
  430: 40. PCRE tries to detect cases of infinite recursion at compile time, but it
  431:     cannot analyze patterns in sufficient detail to catch mutual recursions
  432:     such as ((?1))((?2)). There is now a runtime test that gives an error if a
  433:     subgroup is called recursively as a subpattern for a second time at the
  434:     same position in the subject string. In previous releases this might have
  435:     been caught by the recursion limit, or it might have run out of stack.
  436: 
  437: 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
  438:     happen only once. PCRE was, however incorrectly giving a compile time error
  439:     "recursive call could loop indefinitely" because it cannot analyze the
  440:     pattern in sufficient detail. The compile time test no longer happens when
  441:     PCRE is compiling a conditional subpattern, but actual runaway loops are
  442:     now caught at runtime (see 40 above).
  443: 
  444: 42. It seems that Perl allows any characters other than a closing parenthesis
  445:     to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
  446:     has been changed to be the same.
  447: 
  448: 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
  449:     as not to get warnings when autogen.sh is called. Also changed
  450:     AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
  451: 
  452: 44. To help people who use pcregrep to scan files containing exceedingly long
  453:     lines, the following changes have been made:
  454: 
  455:     (a) The default value of the buffer size parameter has been increased from
  456:         8K to 20K. (The actual buffer used is three times this size.)
  457: 
  458:     (b) The default can be changed by ./configure --with-pcregrep-bufsize when
  459:         PCRE is built.
  460: 
  461:     (c) A --buffer-size=n option has been added to pcregrep, to allow the size
  462:         to be set at run time.
  463: 
  464:     (d) Numerical values in pcregrep options can be followed by K or M, for
  465:         example --buffer-size=50K.
  466: 
  467:     (e) If a line being scanned overflows pcregrep's buffer, an error is now
  468:         given and the return code is set to 2.
  469: 
  470: 45. Add a pointer to the latest mark to the callout data block.
  471: 
  472: 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
  473:     partial match of an empty string instead of no match. This was specific to
  474:     the use of ".".
  475: 
  476: 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
  477:     complete match instead of a partial match. This bug was dependent on both
  478:     the PCRE_UTF8 and PCRE_DOTALL options being set.
  479: 
  480: 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
  481:     starting byte set, because \b was not being ignored.
  482: 
  483: 
  484: Version 8.12 15-Jan-2011
  485: ------------------------
  486: 
  487: 1.  Fixed some typos in the markup of the man pages, and wrote a script that
  488:     checks for such things as part of the documentation building process.
  489: 
  490: 2.  On a big-endian 64-bit system, pcregrep did not correctly process the
  491:     --match-limit and --recursion-limit options (added for 8.11). In
  492:     particular, this made one of the standard tests fail. (The integer value
  493:     went into the wrong half of a long int.)
  494: 
  495: 3.  If the --colour option was given to pcregrep with -v (invert match), it
  496:     did strange things, either producing crazy output, or crashing. It should,
  497:     of course, ignore a request for colour when reporting lines that do not
  498:     match.
  499: 
  500: 4.  Another pcregrep bug caused similar problems if --colour was specified with
  501:     -M (multiline) and the pattern match finished with a line ending.
  502: 
  503: 5.  In pcregrep, when a pattern that ended with a literal newline sequence was
  504:     matched in multiline mode, the following line was shown as part of the
  505:     match. This seems wrong, so I have changed it.
  506: 
  507: 6.  Another pcregrep bug in multiline mode, when --colour was specified, caused
  508:     the check for further matches in the same line (so they could be coloured)
  509:     to overrun the end of the current line. If another match was found, it was
  510:     incorrectly shown (and then shown again when found in the next line).
  511: 
  512: 7.  If pcregrep was compiled under Windows, there was a reference to the
  513:     function pcregrep_exit() before it was defined. I am assuming this was
  514:     the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
  515:     reported by a user. I've moved the definition above the reference.
  516: 
  517: 
  518: Version 8.11 10-Dec-2010
  519: ------------------------
  520: 
  521: 1.  (*THEN) was not working properly if there were untried alternatives prior
  522:     to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
  523:     backtracked to try for "b" instead of moving to the next alternative branch
  524:     at the same level (in this case, to look for "c"). The Perl documentation
  525:     is clear that when (*THEN) is backtracked onto, it goes to the "next
  526:     alternative in the innermost enclosing group".
  527: 
  528: 2.  (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
  529:     such as   (A(*COMMIT)B(*THEN)C|D)  any failure after matching A should
  530:     result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
  531:     (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
  532:     (*THEN).
  533: 
  534: 3.  If \s appeared in a character class, it removed the VT character from
  535:     the class, even if it had been included by some previous item, for example
  536:     in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
  537:     of \s, but is part of the POSIX "space" class.)
  538: 
  539: 4.  A partial match never returns an empty string (because you can always
  540:     match an empty string at the end of the subject); however the checking for
  541:     an empty string was starting at the "start of match" point. This has been
  542:     changed to the "earliest inspected character" point, because the returned
  543:     data for a partial match starts at this character. This means that, for
  544:     example, /(?<=abc)def/ gives a partial match for the subject "abc"
  545:     (previously it gave "no match").
  546: 
  547: 5.  Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
  548:     of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
  549:     previously a full match would be given. However, setting PCRE_PARTIAL_HARD
  550:     has an implication that the given string is incomplete (because a partial
  551:     match is preferred over a full match). For this reason, these items now
  552:     give a partial match in this situation. [Aside: previously, the one case
  553:     /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
  554:     match rather than a full match, which was wrong by the old rules, but is
  555:     now correct.]
  556: 
  557: 6.  There was a bug in the handling of #-introduced comments, recognized when
  558:     PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
  559:     If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
  560:     UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
  561:     scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
  562:     but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
  563:     places in pcre_compile().
  564: 
  565: 7.  Related to (6) above, when pcre_compile() was skipping #-introduced
  566:     comments when looking ahead for named forward references to subpatterns,
  567:     the only newline sequence it recognized was NL. It now handles newlines
  568:     according to the set newline convention.
  569: 
  570: 8.  SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
  571:     former, but used strtoul(), whereas pcretest avoided strtoul() but did not
  572:     cater for a lack of strerror(). These oversights have been fixed.
  573: 
  574: 9.  Added --match-limit and --recursion-limit to pcregrep.
  575: 
  576: 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
  577: 
  578: 11. When the -o option was used, pcregrep was setting a return code of 1, even
  579:     when matches were found, and --line-buffered was not being honoured.
  580: 
  581: 12. Added an optional parentheses number to the -o and --only-matching options
  582:     of pcregrep.
  583: 
  584: 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
  585:     can match an empty string. The code to do it in pcretest and pcredemo
  586:     needed fixing:
  587: 
  588:     (a) When the newline convention was "crlf", pcretest got it wrong, skipping
  589:         only one byte after an empty string match just before CRLF (this case
  590:         just got forgotten; "any" and "anycrlf" were OK).
  591: 
  592:     (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
  593:         mode when an empty string match preceded an ASCII character followed by
  594:         a non-ASCII character. (The code for advancing by one character rather
  595:         than one byte was nonsense.)
  596: 
  597:     (c) The pcredemo.c sample program did not have any code at all to handle
  598:         the cases when CRLF is a valid newline sequence.
  599: 
  600: 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
  601:     as a starting offset was within the subject string. There is now a new
  602:     error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
  603:     negative or greater than the length of the string. In order to test this,
  604:     pcretest is extended to allow the setting of negative starting offsets.
  605: 
  606: 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
  607:     starting offset points to the beginning of a UTF-8 character was
  608:     unnecessarily clumsy. I tidied it up.
  609: 
  610: 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
  611:     bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
  612: 
  613: 17. Nobody had reported that the --include_dir option, which was added in
  614:     release 7.7 should have been called --include-dir (hyphen, not underscore)
  615:     for compatibility with GNU grep. I have changed it to --include-dir, but
  616:     left --include_dir as an undocumented synonym, and the same for
  617:     --exclude-dir, though that is not available in GNU grep, at least as of
  618:     release 2.5.4.
  619: 
  620: 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
  621:     characters from a string of bytes) have been redefined so as not to use
  622:     loops, in order to improve performance in some environments. At the same
  623:     time, I abstracted some of the common code into auxiliary macros to save
  624:     repetition (this should not affect the compiled code).
  625: 
  626: 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
  627:     compile-time error is now given if \c is not followed by an ASCII
  628:     character, that is, a byte less than 128. (In EBCDIC mode, the code is
  629:     different, and any byte value is allowed.)
  630: 
  631: 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
  632:     START_OPTIMIZE option, which is now allowed at compile time - but just
  633:     passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
  634:     to pcregrep and other applications that have no direct access to PCRE
  635:     options. The new /Y option in pcretest sets this option when calling
  636:     pcre_compile().
  637: 
  638: 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
  639:     back references. Groups containing recursive back references were forced to
  640:     be atomic by that change, but in the case of named groups, the amount of
  641:     memory required was incorrectly computed, leading to "Failed: internal
  642:     error: code overflow". This has been fixed.
  643: 
  644: 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
  645:     pcretest.c, to avoid build problems in some Borland environments.
  646: 
  647: 
  648: Version 8.10 25-Jun-2010
  649: ------------------------
  650: 
  651: 1.  Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
  652:     THEN.
  653: 
  654: 2.  (*ACCEPT) was not working when inside an atomic group.
  655: 
  656: 3.  Inside a character class, \B is treated as a literal by default, but
  657:     faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
  658:     causes the error). The code is unchanged, but I tidied the documentation.
  659: 
  660: 4.  Inside a character class, PCRE always treated \R and \X as literals,
  661:     whereas Perl faults them if its -w option is set. I have changed PCRE so
  662:     that it faults them when PCRE_EXTRA is set.
  663: 
  664: 5.  Added support for \N, which always matches any character other than
  665:     newline. (It is the same as "." when PCRE_DOTALL is not set.)
  666: 
  667: 6.  When compiling pcregrep with newer versions of gcc which may have
  668:     FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
  669:     declared with attribute warn_unused_result" were given. Just casting the
  670:     result to (void) does not stop the warnings; a more elaborate fudge is
  671:     needed. I've used a macro to implement this.
  672: 
  673: 7.  Minor change to pcretest.c to avoid a compiler warning.
  674: 
  675: 8.  Added four artifical Unicode properties to help with an option to make
  676:     \s etc use properties (see next item). The new properties are: Xan
  677:     (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
  678: 
  679: 9.  Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
  680:     use Unicode properties. (*UCP) at the start of a pattern can be used to set
  681:     this option. Modified pcretest to add /W to test this facility. Added
  682:     REG_UCP to make it available via the POSIX interface.
  683: 
  684: 10. Added --line-buffered to pcregrep.
  685: 
  686: 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
  687:     studied, and the match started with a letter with a code point greater than
  688:     127 whose first byte was different to the first byte of the other case of
  689:     the letter, the other case of this starting letter was not recognized
  690:     (#976).
  691: 
  692: 12. If a pattern that was studied started with a repeated Unicode property
  693:     test, for example, \p{Nd}+, there was the theoretical possibility of
  694:     setting up an incorrect bitmap of starting bytes, but fortunately it could
  695:     not have actually happened in practice until change 8 above was made (it
  696:     added property types that matched character-matching opcodes).
  697: 
  698: 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
  699:     possible starting bytes for non-anchored patterns.
  700: 
  701: 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
  702:     \R, and also a number of cases that involve Unicode properties, both
  703:     explicit and implicit when PCRE_UCP is set.
  704: 
  705: 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
  706:     input, it could crash or give wrong results if characters with values
  707:     greater than 0xc0 were present in the subject string. (Detail: it assumed
  708:     UTF-8 input when processing these items.)
  709: 
  710: 16. Added a lot of (int) casts to avoid compiler warnings in systems where
  711:     size_t is 64-bit (#991).
  712: 
  713: 17. Added a check for running out of memory when PCRE is compiled with
  714:     --disable-stack-for-recursion (#990).
  715: 
  716: 18. If the last data line in a file for pcretest does not have a newline on
  717:     the end, a newline was missing in the output.
  718: 
  719: 19. The default pcre_chartables.c file recognizes only ASCII characters (values
  720:     less than 128) in its various bitmaps. However, there is a facility for
  721:     generating tables according to the current locale when PCRE is compiled. It
  722:     turns out that in some environments, 0x85 and 0xa0, which are Unicode space
  723:     characters, are recognized by isspace() and therefore were getting set in
  724:     these tables, and indeed these tables seem to approximate to ISO 8859. This
  725:     caused a problem in UTF-8 mode when pcre_study() was used to create a list
  726:     of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
  727:     which of course cannot start UTF-8 characters. I have changed the code so
  728:     that only real ASCII characters (less than 128) and the correct starting
  729:     bytes for UTF-8 encodings are set for characters greater than 127 when in
  730:     UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
  731:     altogether.)
  732: 
  733: 20. Added the /T option to pcretest so as to be able to run tests with non-
  734:     standard character tables, thus making it possible to include the tests
  735:     used for 19 above in the standard set of tests.
  736: 
  737: 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
  738:     reference to a subpattern the other side of a comment that contains an
  739:     opening parenthesis caused either an internal compiling error, or a
  740:     reference to the wrong subpattern.
  741: 
  742: 
  743: Version 8.02 19-Mar-2010
  744: ------------------------
  745: 
  746: 1.  The Unicode data tables have been updated to Unicode 5.2.0.
  747: 
  748: 2.  Added the option --libs-cpp to pcre-config, but only when C++ support is
  749:     configured.
  750: 
  751: 3.  Updated the licensing terms in the pcregexp.pas file, as agreed with the
  752:     original author of that file, following a query about its status.
  753: 
  754: 4.  On systems that do not have stdint.h (e.g. Solaris), check for and include
  755:     inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
  756: 
  757: 5.  A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
  758:     quantifier applied to a forward-referencing subroutine call, could compile
  759:     incorrect code or give the error "internal error: previously-checked
  760:     referenced subpattern not found".
  761: 
  762: 6.  Both MS Visual Studio and Symbian OS have problems with initializing
  763:     variables to point to external functions. For these systems, therefore,
  764:     pcre_malloc etc. are now initialized to local functions that call the
  765:     relevant global functions.
  766: 
  767: 7.  There were two entries missing in the vectors called coptable and poptable
  768:     in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
  769:     I've fixed the data, and added a kludgy way of testing at compile time that
  770:     the lengths are correct (equal to the number of opcodes).
  771: 
  772: 8.  Following on from 7, I added a similar kludge to check the length of the
  773:     eint vector in pcreposix.c.
  774: 
  775: 9.  Error texts for pcre_compile() are held as one long string to avoid too
  776:     much relocation at load time. To find a text, the string is searched,
  777:     counting zeros. There was no check for running off the end of the string,
  778:     which could happen if a new error number was added without updating the
  779:     string.
  780: 
  781: 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
  782: 
  783: 11. \K was not working if it appeared in an atomic group or in a group that
  784:     was called as a "subroutine", or in an assertion. Perl 5.11 documents that
  785:     \K is "not well defined" if used in an assertion. PCRE now accepts it if
  786:     the assertion is positive, but not if it is negative.
  787: 
  788: 12. Change 11 fortuitously reduced the size of the stack frame used in the
  789:     "match()" function of pcre_exec.c by one pointer. Forthcoming
  790:     implementation of support for (*MARK) will need an extra pointer on the
  791:     stack; I have reserved it now, so that the stack frame size does not
  792:     decrease.
  793: 
  794: 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
  795:     item in branch that calls a recursion is a subroutine call - as in the
  796:     second branch in the above example - was incorrectly given the compile-
  797:     time error "recursive call could loop indefinitely" because pcre_compile()
  798:     was not correctly checking the subroutine for matching a non-empty string.
  799: 
  800: 14. The checks for overrunning compiling workspace could trigger after an
  801:     overrun had occurred. This is a "should never occur" error, but it can be
  802:     triggered by pathological patterns such as hundreds of nested parentheses.
  803:     The checks now trigger 100 bytes before the end of the workspace.
  804: 
  805: 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
  806: 
  807: 
  808: Version 8.01 19-Jan-2010
  809: ------------------------
  810: 
  811: 1.  If a pattern contained a conditional subpattern with only one branch (in
  812:     particular, this includes all (*DEFINE) patterns), a call to pcre_study()
  813:     computed the wrong minimum data length (which is of course zero for such
  814:     subpatterns). This could cause incorrect "no match" results.
  815: 
  816: 2.  For patterns such as (?i)a(?-i)b|c where an option setting at the start of
  817:     the pattern is reset in the first branch, pcre_compile() failed with
  818:     "internal error: code overflow at offset...". This happened only when
  819:     the reset was to the original external option setting. (An optimization
  820:     abstracts leading options settings into an external setting, which was the
  821:     cause of this.)
  822: 
  823: 3.  A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
  824:     of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
  825:     assertion pattern did not match (meaning that the assertion was true), it
  826:     was incorrectly treated as false if the SKIP had been reached during the
  827:     matching. This also applied to assertions used as conditions.
  828: 
  829: 4.  If an item that is not supported by pcre_dfa_exec() was encountered in an
  830:     assertion subpattern, including such a pattern used as a condition,
  831:     unpredictable results occurred, instead of the error return
  832:     PCRE_ERROR_DFA_UITEM.
  833: 
  834: 5.  The C++ GlobalReplace function was not working like Perl for the special
  835:     situation when an empty string is matched. It now does the fancy magic
  836:     stuff that is necessary.
  837: 
  838: 6.  In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
  839:     removed. (These were left over from very, very early versions of PCRE.)
  840: 
  841: 7.  Some cosmetic changes to the code to make life easier when compiling it
  842:     as part of something else:
  843: 
  844:     (a) Change DEBUG to PCRE_DEBUG.
  845: 
  846:     (b) In pcre_compile(), rename the member of the "branch_chain" structure
  847:         called "current" as "current_branch", to prevent a collision with the
  848:         Linux macro when compiled as a kernel module.
  849: 
  850:     (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
  851:         prevent a collision with the Linux macro when compiled as a kernel
  852:         module.
  853: 
  854: 8.  In pcre_compile() there are some checks for integer overflows that used to
  855:     cast potentially large values to (double). This has been changed to that
  856:     when building, a check for int64_t is made, and if it is found, it is used
  857:     instead, thus avoiding the use of floating point arithmetic. (There is no
  858:     other use of FP in PCRE.) If int64_t is not found, the fallback is to
  859:     double.
  860: 
  861: 9.  Added two casts to avoid signed/unsigned warnings from VS Studio Express
  862:     2005 (difference between two addresses compared to an unsigned value).
  863: 
  864: 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
  865:     custom one, because of the following reported problem in Windows:
  866: 
  867:       - libbz2 uses the Pascal calling convention (WINAPI) for the functions
  868:           under Win32.
  869:       - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
  870:           therefore missing the function definition.
  871:       - The compiler thus generates a "C" signature for the test function.
  872:       - The linker fails to find the "C" function.
  873:       - PCRE fails to configure if asked to do so against libbz2.
  874: 
  875: 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
  876:     messages were output:
  877: 
  878:       Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
  879:       rerunning libtoolize, to keep the correct libtool macros in-tree.
  880:       Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
  881: 
  882:     I have done both of these things.
  883: 
  884: 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
  885:     most of the time, it *can* run out if it is given a pattern that contains a
  886:     runaway infinite recursion. I updated the discussion in the pcrestack man
  887:     page.
  888: 
  889: 13. Now that we have gone to the x.xx style of version numbers, the minor
  890:     version may start with zero. Using 08 or 09 is a bad idea because users
  891:     might check the value of PCRE_MINOR in their code, and 08 or 09 may be
  892:     interpreted as invalid octal numbers. I've updated the previous comment in
  893:     configure.ac, and also added a check that gives an error if 08 or 09 are
  894:     used.
  895: 
  896: 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
  897:     causing partial matching to fail when the end of the subject matched \W
  898:     in a UTF-8 pattern where \W was quantified with a minimum of 3.
  899: 
  900: 15. There were some discrepancies between the declarations in pcre_internal.h
  901:     of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
  902:     their definitions. The declarations used "const uschar *" and the
  903:     definitions used USPTR. Even though USPTR is normally defined as "const
  904:     unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
  905:     reported that: "This difference in casting confuses some C++ compilers, for
  906:     example, SunCC recognizes above declarations as different functions and
  907:     generates broken code for hbpcre." I have changed the declarations to use
  908:     USPTR.
  909: 
  910: 16. GNU libtool is named differently on some systems. The autogen.sh script now
  911:     tries several variants such as glibtoolize (MacOSX) and libtoolize1x
  912:     (FreeBSD).
  913: 
  914: 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
  915:     (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
  916:     comment: "Figure out how to create a longlong from a string: strtoll and
  917:     equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
  918:     instance, but it only takes 2 args instead of 3!"
  919: 
  920: 18. A subtle bug concerned with back references has been fixed by a change of
  921:     specification, with a corresponding code fix. A pattern such as
  922:     ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
  923:     refers, was giving matches when it shouldn't. For example, xa=xaaa would
  924:     match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
  925:     same bug. Such groups have to be quantified to be useful, or contained
  926:     inside another quantified group. (If there's no repetition, the reference
  927:     can never match.) The problem arises because, having left the group and
  928:     moved on to the rest of the pattern, a later failure that backtracks into
  929:     the group uses the captured value from the final iteration of the group
  930:     rather than the correct earlier one. I have fixed this in PCRE by forcing
  931:     any group that contains a reference to itself to be an atomic group; that
  932:     is, there cannot be any backtracking into it once it has completed. This is
  933:     similar to recursive and subroutine calls.
  934: 
  935: 
  936: Version 8.00 19-Oct-09
  937: ----------------------
  938: 
  939: 1.  The table for translating pcre_compile() error codes into POSIX error codes
  940:     was out-of-date, and there was no check on the pcre_compile() error code
  941:     being within the table. This could lead to an OK return being given in
  942:     error.
  943: 
  944: 2.  Changed the call to open a subject file in pcregrep from fopen(pathname,
  945:     "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
  946:     in a Windows environment.
  947: 
  948: 3.  The pcregrep --count option prints the count for each file even when it is
  949:     zero, as does GNU grep. However, pcregrep was also printing all files when
  950:     --files-with-matches was added. Now, when both options are given, it prints
  951:     counts only for those files that have at least one match. (GNU grep just
  952:     prints the file name in this circumstance, but including the count seems
  953:     more useful - otherwise, why use --count?) Also ensured that the
  954:     combination -clh just lists non-zero counts, with no names.
  955: 
  956: 4.  The long form of the pcregrep -F option was incorrectly implemented as
  957:     --fixed_strings instead of --fixed-strings. This is an incompatible change,
  958:     but it seems right to fix it, and I didn't think it was worth preserving
  959:     the old behaviour.
  960: 
  961: 5.  The command line items --regex=pattern and --regexp=pattern were not
  962:     recognized by pcregrep, which required --regex pattern or --regexp pattern
  963:     (with a space rather than an '='). The man page documented the '=' forms,
  964:     which are compatible with GNU grep; these now work.
  965: 
  966: 6.  No libpcreposix.pc file was created for pkg-config; there was just
  967:     libpcre.pc and libpcrecpp.pc. The omission has been rectified.
  968: 
  969: 7.  Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
  970:     when UCP support is not needed, by modifying the Python script that
  971:     generates it from Unicode data files. This should not matter if the module
  972:     is correctly used as a library, but I received one complaint about 50K of
  973:     unwanted data. My guess is that the person linked everything into his
  974:     program rather than using a library. Anyway, it does no harm.
  975: 
  976: 8.  A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
  977:     was a minimum greater than 1 for a wide character in a possessive
  978:     repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
  979:     which had an unlimited repeat of a nested, fixed maximum repeat of a wide
  980:     character. Chaos in the form of incorrect output or a compiling loop could
  981:     result.
  982: 
  983: 9.  The restrictions on what a pattern can contain when partial matching is
  984:     requested for pcre_exec() have been removed. All patterns can now be
  985:     partially matched by this function. In addition, if there are at least two
  986:     slots in the offset vector, the offset of the earliest inspected character
  987:     for the match and the offset of the end of the subject are set in them when
  988:     PCRE_ERROR_PARTIAL is returned.
  989: 
  990: 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
  991:     synonymous with PCRE_PARTIAL, for backwards compatibility, and
  992:     PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
  993:     and may be more useful for multi-segment matching.
  994: 
  995: 11. Partial matching with pcre_exec() is now more intuitive. A partial match
  996:     used to be given if ever the end of the subject was reached; now it is
  997:     given only if matching could not proceed because another character was
  998:     needed. This makes a difference in some odd cases such as Z(*FAIL) with the
  999:     string "Z", which now yields "no match" instead of "partial match". In the
 1000:     case of pcre_dfa_exec(), "no match" is given if every matching path for the
 1001:     final character ended with (*FAIL).
 1002: 
 1003: 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
 1004:     if the pattern had a "must contain" character that was already found in the
 1005:     earlier partial match, unless partial matching was again requested. For
 1006:     example, with the pattern /dog.(body)?/, the "must contain" character is
 1007:     "g". If the first part-match was for the string "dog", restarting with
 1008:     "sbody" failed. This bug has been fixed.
 1009: 
 1010: 13. The string returned by pcre_dfa_exec() after a partial match has been
 1011:     changed so that it starts at the first inspected character rather than the
 1012:     first character of the match. This makes a difference only if the pattern
 1013:     starts with a lookbehind assertion or \b or \B (\K is not supported by
 1014:     pcre_dfa_exec()). It's an incompatible change, but it makes the two
 1015:     matching functions compatible, and I think it's the right thing to do.
 1016: 
 1017: 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
 1018:     so that the demonstration program is easily available in environments where
 1019:     PCRE has not been installed from source.
 1020: 
 1021: 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
 1022:     libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
 1023:     library.
 1024: 
 1025: 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
 1026:     It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
 1027:     is not the first non-POSIX option to be added. Clearly some people find
 1028:     these options useful.
 1029: 
 1030: 17. If a caller to the POSIX matching function regexec() passes a non-zero
 1031:     value for nmatch with a NULL value for pmatch, the value of
 1032:     nmatch is forced to zero.
 1033: 
 1034: 18. RunGrepTest did not have a test for the availability of the -u option of
 1035:     the diff command, as RunTest does. It now checks in the same way as
 1036:     RunTest, and also checks for the -b option.
 1037: 
 1038: 19. If an odd number of negated classes containing just a single character
 1039:     interposed, within parentheses, between a forward reference to a named
 1040:     subpattern and the definition of the subpattern, compilation crashed with
 1041:     an internal error, complaining that it could not find the referenced
 1042:     subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
 1043:     [The bug was that it was starting one character too far in when skipping
 1044:     over the character class, thus treating the ] as data rather than
 1045:     terminating the class. This meant it could skip too much.]
 1046: 
 1047: 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
 1048:     /g option in pcretest when the pattern contains \K, which makes it possible
 1049:     to have an empty string match not at the start, even when the pattern is
 1050:     anchored. Updated pcretest and pcredemo to use this option.
 1051: 
 1052: 21. If the maximum number of capturing subpatterns in a recursion was greater
 1053:     than the maximum at the outer level, the higher number was returned, but
 1054:     with unset values at the outer level. The correct (outer level) value is
 1055:     now given.
 1056: 
 1057: 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
 1058:     PCRE did not set those parentheses (unlike Perl). I have now found a way to
 1059:     make it do so. The string so far is captured, making this feature
 1060:     compatible with Perl.
 1061: 
 1062: 23. The tests have been re-organized, adding tests 11 and 12, to make it
 1063:     possible to check the Perl 5.10 features against Perl 5.10.
 1064: 
 1065: 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
 1066:     pattern matches a fixed length string. PCRE did not allow this; now it
 1067:     does. Neither allows recursion.
 1068: 
 1069: 25. I finally figured out how to implement a request to provide the minimum
 1070:     length of subject string that was needed in order to match a given pattern.
 1071:     (It was back references and recursion that I had previously got hung up
 1072:     on.) This code has now been added to pcre_study(); it finds a lower bound
 1073:     to the length of subject needed. It is not necessarily the greatest lower
 1074:     bound, but using it to avoid searching strings that are too short does give
 1075:     some useful speed-ups. The value is available to calling programs via
 1076:     pcre_fullinfo().
 1077: 
 1078: 26. While implementing 25, I discovered to my embarrassment that pcretest had
 1079:     not been passing the result of pcre_study() to pcre_dfa_exec(), so the
 1080:     study optimizations had never been tested with that matching function.
 1081:     Oops. What is worse, even when it was passed study data, there was a bug in
 1082:     pcre_dfa_exec() that meant it never actually used it. Double oops. There
 1083:     were also very few tests of studied patterns with pcre_dfa_exec().
 1084: 
 1085: 27. If (?| is used to create subpatterns with duplicate numbers, they are now
 1086:     allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
 1087:     on the other side of the coin, they are no longer allowed to have different
 1088:     names, because these cannot be distinguished in PCRE, and this has caused
 1089:     confusion. (This is a difference from Perl.)
 1090: 
 1091: 28. When duplicate subpattern names are present (necessarily with different
 1092:     numbers, as required by 27 above), and a test is made by name in a
 1093:     conditional pattern, either for a subpattern having been matched, or for
 1094:     recursion in such a pattern, all the associated numbered subpatterns are
 1095:     tested, and the overall condition is true if the condition is true for any
 1096:     one of them. This is the way Perl works, and is also more like the way
 1097:     testing by number works.
 1098: 
 1099: 
 1100: Version 7.9 11-Apr-09
 1101: ---------------------
 1102: 
 1103: 1.  When building with support for bzlib/zlib (pcregrep) and/or readline
 1104:     (pcretest), all targets were linked against these libraries. This included
 1105:     libpcre, libpcreposix, and libpcrecpp, even though they do not use these
 1106:     libraries. This caused unwanted dependencies to be created. This problem
 1107:     has been fixed, and now only pcregrep is linked with bzlib/zlib and only
 1108:     pcretest is linked with readline.
 1109: 
 1110: 2.  The "typedef int BOOL" in pcre_internal.h that was included inside the
 1111:     "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
 1112:     moved outside it again, because FALSE and TRUE are already defined in AIX,
 1113:     but BOOL is not.
 1114: 
 1115: 3.  The pcre_config() function was treating the PCRE_MATCH_LIMIT and
 1116:     PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
 1117: 
 1118: 4.  The pcregrep documentation said spaces were inserted as well as colons (or
 1119:     hyphens) following file names and line numbers when outputting matching
 1120:     lines. This is not true; no spaces are inserted. I have also clarified the
 1121:     wording for the --colour (or --color) option.
 1122: 
 1123: 5.  In pcregrep, when --colour was used with -o, the list of matching strings
 1124:     was not coloured; this is different to GNU grep, so I have changed it to be
 1125:     the same.
 1126: 
 1127: 6.  When --colo(u)r was used in pcregrep, only the first matching substring in
 1128:     each matching line was coloured. Now it goes on to look for further matches
 1129:     of any of the test patterns, which is the same behaviour as GNU grep.
 1130: 
 1131: 7.  A pattern that could match an empty string could cause pcregrep to loop; it
 1132:     doesn't make sense to accept an empty string match in pcregrep, so I have
 1133:     locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
 1134:     seems to be how GNU grep behaves.
 1135: 
 1136: 8.  The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
 1137:     start or after a newline", because the conditional assertion was not being
 1138:     correctly handled. The rule now is that both the assertion and what follows
 1139:     in the first alternative must satisfy the test.
 1140: 
 1141: 9.  If auto-callout was enabled in a pattern with a conditional group whose
 1142:     condition was an assertion, PCRE could crash during matching, both with
 1143:     pcre_exec() and pcre_dfa_exec().
 1144: 
 1145: 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
 1146:     used for matching.
 1147: 
 1148: 11. Unicode property support in character classes was not working for
 1149:     characters (bytes) greater than 127 when not in UTF-8 mode.
 1150: 
 1151: 12. Added the -M command line option to pcretest.
 1152: 
 1153: 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
 1154: 
 1155: 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
 1156: 
 1157: 16. Added comments and documentation about mis-use of no_arg in the C++
 1158:     wrapper.
 1159: 
 1160: 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
 1161:     from Martin Jerabek that uses macro names for all relevant character and
 1162:     string constants.
 1163: 
 1164: 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
 1165:     SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
 1166:     SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
 1167:     these, but not everybody uses configure.
 1168: 
 1169: 19. A conditional group that had only one branch was not being correctly
 1170:     recognized as an item that could match an empty string. This meant that an
 1171:     enclosing group might also not be so recognized, causing infinite looping
 1172:     (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
 1173:     with the subject "ab", where knowledge that the repeated group can match
 1174:     nothing is needed in order to break the loop.
 1175: 
 1176: 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
 1177:     exec(), but without supplying a callout function, matching went wrong.
 1178: 
 1179: 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
 1180:     leak if the size of the offset vector was greater than 30. When the vector
 1181:     is smaller, the saved offsets during recursion go onto a local stack
 1182:     vector, but for larger vectors malloc() is used. It was failing to free
 1183:     when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
 1184:     error, in fact).
 1185: 
 1186: 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
 1187:     heapframe that is used only when UTF-8 support is enabled. This caused no
 1188:     problem, but was untidy.
 1189: 
 1190: 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
 1191:     CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
 1192:     included within another project.
 1193: 
 1194: 24. Steven Van Ingelgem's patches to add more options to the CMake support,
 1195:     slightly modified by me:
 1196: 
 1197:       (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
 1198:           not building pcregrep.
 1199: 
 1200:       (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
 1201:           if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
 1202: 
 1203: 25. Forward references, both numeric and by name, in patterns that made use of
 1204:     duplicate group numbers, could behave incorrectly or give incorrect errors,
 1205:     because when scanning forward to find the reference group, PCRE was not
 1206:     taking into account the duplicate group numbers. A pattern such as
 1207:     ^X(?3)(a)(?|(b)|(q))(Y) is an example.
 1208: 
 1209: 26. Changed a few more instances of "const unsigned char *" to USPTR, making
 1210:     the feature of a custom pointer more persuasive (as requested by a user).
 1211: 
 1212: 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
 1213:     pcretest.c, inside #ifndefs, because it seems they are sometimes already
 1214:     pre-defined.
 1215: 
 1216: 28. Added support for (*UTF8) at the start of a pattern.
 1217: 
 1218: 29. Arrange for flags added by the "release type" setting in CMake to be shown
 1219:     in the configuration summary.
 1220: 
 1221: 
 1222: Version 7.8 05-Sep-08
 1223: ---------------------
 1224: 
 1225: 1.  Replaced UCP searching code with optimized version as implemented for Ad
 1226:     Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
 1227:     stage table and inline lookup instead of a function, giving speed ups of 2
 1228:     to 5 times on some simple patterns that I tested. Permission was given to
 1229:     distribute the MultiStage2.py script that generates the tables (it's not in
 1230:     the tarball, but is in the Subversion repository).
 1231: 
 1232: 2.  Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
 1233:     scripts.
 1234: 
 1235: 3.  Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
 1236:     a group with a zero qualifier. The result of the study could be incorrect,
 1237:     or the function might crash, depending on the pattern.
 1238: 
 1239: 4.  Caseless matching was not working for non-ASCII characters in back
 1240:     references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
 1241:     It now works when Unicode Property Support is available.
 1242: 
 1243: 5.  In pcretest, an escape such as \x{de} in the data was always generating
 1244:     a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
 1245:     non-UTF-8 mode. If the value is greater than 255, it gives a warning about
 1246:     truncation.
 1247: 
 1248: 6.  Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
 1249: 
 1250: 7.  Added two (int) casts to pcregrep when printing the difference of two
 1251:     pointers, in case they are 64-bit values.
 1252: 
 1253: 8.  Added comments about Mac OS X stack usage to the pcrestack man page and to
 1254:     test 2 if it fails.
 1255: 
 1256: 9.  Added PCRE_CALL_CONVENTION just before the names of all exported functions,
 1257:     and a #define of that name to empty if it is not externally set. This is to
 1258:     allow users of MSVC to set it if necessary.
 1259: 
 1260: 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
 1261:     the convenience functions in the pcre_get.c source file.
 1262: 
 1263: 11. An option change at the start of a pattern that had top-level alternatives
 1264:     could cause overwriting and/or a crash. This command provoked a crash in
 1265:     some environments:
 1266: 
 1267:       printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
 1268: 
 1269:     This potential security problem was recorded as CVE-2008-2371.
 1270: 
 1271: 12. For a pattern where the match had to start at the beginning or immediately
 1272:     after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
 1273:     pcre_dfa_exec() could read past the end of the passed subject if there was
 1274:     no match. To help with detecting such bugs (e.g. with valgrind), I modified
 1275:     pcretest so that it places the subject at the end of its malloc-ed buffer.
 1276: 
 1277: 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
 1278:     exec() might read past the end of the data buffer in UTF-8 mode.
 1279: 
 1280: 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
 1281:     the data contained the byte 0x85 as part of a UTF-8 character within its
 1282:     first line. This applied both to normal and DFA matching.
 1283: 
 1284: 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
 1285:     /^[^d]*?$/8 failed to match "abc".
 1286: 
 1287: 16. Added a missing copyright notice to pcrecpp_internal.h.
 1288: 
 1289: 17. Make it more clear in the documentation that values returned from
 1290:     pcre_exec() in ovector are byte offsets, not character counts.
 1291: 
 1292: 18. Tidied a few places to stop certain compilers from issuing warnings.
 1293: 
 1294: 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
 1295:     supplied by Stefan Weber. I made a further small update for 7.8 because
 1296:     there is a change of source arrangements: the pcre_searchfuncs.c module is
 1297:     replaced by pcre_ucd.c.
 1298: 
 1299: 
 1300: Version 7.7 07-May-08
 1301: ---------------------
 1302: 
 1303: 1.  Applied Craig's patch to sort out a long long problem: "If we can't convert
 1304:     a string to a long long, pretend we don't even have a long long." This is
 1305:     done by checking for the strtoq, strtoll, and _strtoi64 functions.
 1306: 
 1307: 2.  Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
 1308:     pre-7.6 versions, which defined a global no_arg variable instead of putting
 1309:     it in the RE class. (See also #8 below.)
 1310: 
 1311: 3.  Remove a line of dead code, identified by coverity and reported by Nuno
 1312:     Lopes.
 1313: 
 1314: 4.  Fixed two related pcregrep bugs involving -r with --include or --exclude:
 1315: 
 1316:     (1) The include/exclude patterns were being applied to the whole pathnames
 1317:         of files, instead of just to the final components.
 1318: 
 1319:     (2) If there was more than one level of directory, the subdirectories were
 1320:         skipped unless they satisfied the include/exclude conditions. This is
 1321:         inconsistent with GNU grep (and could even be seen as contrary to the
 1322:         pcregrep specification - which I improved to make it absolutely clear).
 1323:         The action now is always to scan all levels of directory, and just
 1324:         apply the include/exclude patterns to regular files.
 1325: 
 1326: 5.  Added the --include_dir and --exclude_dir patterns to pcregrep, and used
 1327:     --exclude_dir in the tests to avoid scanning .svn directories.
 1328: 
 1329: 6.  Applied Craig's patch to the QuoteMeta function so that it escapes the
 1330:     NUL character as backslash + 0 rather than backslash + NUL, because PCRE
 1331:     doesn't support NULs in patterns.
 1332: 
 1333: 7.  Added some missing "const"s to declarations of static tables in
 1334:     pcre_compile.c and pcre_dfa_exec.c.
 1335: 
 1336: 8.  Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
 1337:     caused by fix #2  above. (Subsequently also a second patch to fix the
 1338:     first patch. And a third patch - this was a messy problem.)
 1339: 
 1340: 9.  Applied Craig's patch to remove the use of push_back().
 1341: 
 1342: 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
 1343:     matching function regexec().
 1344: 
 1345: 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
 1346:     which, however, unlike Perl's \g{...}, are subroutine calls, not back
 1347:     references. PCRE supports relative numbers with this syntax (I don't think
 1348:     Oniguruma does).
 1349: 
 1350: 12. Previously, a group with a zero repeat such as (...){0} was completely
 1351:     omitted from the compiled regex. However, this means that if the group
 1352:     was called as a subroutine from elsewhere in the pattern, things went wrong
 1353:     (an internal error was given). Such groups are now left in the compiled
 1354:     pattern, with a new opcode that causes them to be skipped at execution
 1355:     time.
 1356: 
 1357: 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
 1358:     to the way PCRE behaves:
 1359: 
 1360:     (a) A lone ] character is dis-allowed (Perl treats it as data).
 1361: 
 1362:     (b) A back reference to an unmatched subpattern matches an empty string
 1363:         (Perl fails the current match path).
 1364: 
 1365:     (c) A data ] in a character class must be notated as \] because if the
 1366:         first data character in a class is ], it defines an empty class. (In
 1367:         Perl it is not possible to have an empty class.) The empty class []
 1368:         never matches; it forces failure and is equivalent to (*FAIL) or (?!).
 1369:         The negative empty class [^] matches any one character, independently
 1370:         of the DOTALL setting.
 1371: 
 1372: 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
 1373:     non-existent subpattern following a character class starting with ']' and
 1374:     containing () gave an internal compiling error instead of "reference to
 1375:     non-existent subpattern". Fortunately, when the pattern did exist, the
 1376:     compiled code was correct. (When scanning forwards to check for the
 1377:     existencd of the subpattern, it was treating the data ']' as terminating
 1378:     the class, so got the count wrong. When actually compiling, the reference
 1379:     was subsequently set up correctly.)
 1380: 
 1381: 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
 1382:     it was being rejected as not supported by pcre_dfa_exec(), even though
 1383:     other assertions are supported. I have made pcre_dfa_exec() support
 1384:     (*FAIL).
 1385: 
 1386: 16. The implementation of 13c above involved the invention of a new opcode,
 1387:     OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
 1388:     cannot be changed at match time, I realized I could make a small
 1389:     improvement to matching performance by compiling OP_ALLANY instead of
 1390:     OP_ANY for "." when DOTALL was set, and then removing the runtime tests
 1391:     on the OP_ANY path.
 1392: 
 1393: 17. Compiling pcretest on Windows with readline support failed without the
 1394:     following two fixes: (1) Make the unistd.h include conditional on
 1395:     HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
 1396: 
 1397: 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
 1398:     ncurses library to be included for pcretest when ReadLine support is
 1399:     requested, but also to allow for it to be overridden. This patch came from
 1400:     Daniel Bergström.
 1401: 
 1402: 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
 1403:     as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
 1404:     any errors with the current Unicode tables. Thanks to Peter Kankowski for
 1405:     spotting this.
 1406: 
 1407: 
 1408: Version 7.6 28-Jan-08
 1409: ---------------------
 1410: 
 1411: 1.  A character class containing a very large number of characters with
 1412:     codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
 1413:     overflow.
 1414: 
 1415: 2.  Patch to cut out the "long long" test in pcrecpp_unittest when
 1416:     HAVE_LONG_LONG is not defined.
 1417: 
 1418: 3.  Applied Christian Ehrlicher's patch to update the CMake build files to
 1419:     bring them up to date and include new features. This patch includes:
 1420: 
 1421:     - Fixed PH's badly added libz and libbz2 support.
 1422:     - Fixed a problem with static linking.
 1423:     - Added pcredemo. [But later removed - see 7 below.]
 1424:     - Fixed dftables problem and added an option.
 1425:     - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
 1426:         HAVE_LONG_LONG.
 1427:     - Added readline support for pcretest.
 1428:     - Added an listing of the option settings after cmake has run.
 1429: 
 1430: 4.  A user submitted a patch to Makefile that makes it easy to create
 1431:     "pcre.dll" under mingw when using Configure/Make. I added stuff to
 1432:     Makefile.am that cause it to include this special target, without
 1433:     affecting anything else. Note that the same mingw target plus all
 1434:     the other distribution libraries and programs are now supported
 1435:     when configuring with CMake (see 6 below) instead of with
 1436:     Configure/Make.
 1437: 
 1438: 5.  Applied Craig's patch that moves no_arg into the RE class in the C++ code.
 1439:     This is an attempt to solve the reported problem "pcrecpp::no_arg is not
 1440:     exported in the Windows port". It has not yet been confirmed that the patch
 1441:     solves the problem, but it does no harm.
 1442: 
 1443: 6.  Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
 1444:     NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
 1445:     with CMake, and also correct the comment about stack recursion.
 1446: 
 1447: 7.  Remove the automatic building of pcredemo from the ./configure system and
 1448:     from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
 1449:     of a program that users should build themselves after PCRE is installed, so
 1450:     building it automatically is not really right. What is more, it gave
 1451:     trouble in some build environments.
 1452: 
 1453: 8.  Further tidies to CMakeLists.txt from Sheri and Christian.
 1454: 
 1455: 
 1456: Version 7.5 10-Jan-08
 1457: ---------------------
 1458: 
 1459: 1.  Applied a patch from Craig: "This patch makes it possible to 'ignore'
 1460:     values in parens when parsing an RE using the C++ wrapper."
 1461: 
 1462: 2.  Negative specials like \S did not work in character classes in UTF-8 mode.
 1463:     Characters greater than 255 were excluded from the class instead of being
 1464:     included.
 1465: 
 1466: 3.  The same bug as (2) above applied to negated POSIX classes such as
 1467:     [:^space:].
 1468: 
 1469: 4.  PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
 1470:     defined or documented. It seems to have been a typo for PCRE_STATIC, so
 1471:     I have changed it.
 1472: 
 1473: 5.  The construct (?&) was not diagnosed as a syntax error (it referenced the
 1474:     first named subpattern) and a construct such as (?&a) would reference the
 1475:     first named subpattern whose name started with "a" (in other words, the
 1476:     length check was missing). Both these problems are fixed. "Subpattern name
 1477:     expected" is now given for (?&) (a zero-length name), and this patch also
 1478:     makes it give the same error for \k'' (previously it complained that that
 1479:     was a reference to a non-existent subpattern).
 1480: 
 1481: 6.  The erroneous patterns (?+-a) and (?-+a) give different error messages;
 1482:     this is right because (?- can be followed by option settings as well as by
 1483:     digits. I have, however, made the messages clearer.
 1484: 
 1485: 7.  Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
 1486:     than the number used in the conditional) now cause a compile-time error.
 1487:     This is actually not compatible with Perl, which accepts such patterns, but
 1488:     treats the conditional as always being FALSE (as PCRE used to), but it
 1489:     seems to me that giving a diagnostic is better.
 1490: 
 1491: 8.  Change "alphameric" to the more common word "alphanumeric" in comments
 1492:     and messages.
 1493: 
 1494: 9.  Fix two occurrences of "backslash" in comments that should have been
 1495:     "backspace".
 1496: 
 1497: 10. Remove two redundant lines of code that can never be obeyed (their function
 1498:     was moved elsewhere).
 1499: 
 1500: 11. The program that makes PCRE's Unicode character property table had a bug
 1501:     which caused it to generate incorrect table entries for sequences of
 1502:     characters that have the same character type, but are in different scripts.
 1503:     It amalgamated them into a single range, with the script of the first of
 1504:     them. In other words, some characters were in the wrong script. There were
 1505:     thirteen such cases, affecting characters in the following ranges:
 1506: 
 1507:       U+002b0 - U+002c1
 1508:       U+0060c - U+0060d
 1509:       U+0061e - U+00612
 1510:       U+0064b - U+0065e
 1511:       U+0074d - U+0076d
 1512:       U+01800 - U+01805
 1513:       U+01d00 - U+01d77
 1514:       U+01d9b - U+01dbf
 1515:       U+0200b - U+0200f
 1516:       U+030fc - U+030fe
 1517:       U+03260 - U+0327f
 1518:       U+0fb46 - U+0fbb1
 1519:       U+10450 - U+1049d
 1520: 
 1521: 12. The -o option (show only the matching part of a line) for pcregrep was not
 1522:     compatible with GNU grep in that, if there was more than one match in a
 1523:     line, it showed only the first of them. It now behaves in the same way as
 1524:     GNU grep.
 1525: 
 1526: 13. If the -o and -v options were combined for pcregrep, it printed a blank
 1527:     line for every non-matching line. GNU grep prints nothing, and pcregrep now
 1528:     does the same. The return code can be used to tell if there were any
 1529:     non-matching lines.
 1530: 
 1531: 14. Added --file-offsets and --line-offsets to pcregrep.
 1532: 
 1533: 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
 1534:     infinitely looping recursion. The bug was that positive lookaheads were not
 1535:     being skipped when checking for a possible empty match (negative lookaheads
 1536:     and both kinds of lookbehind were skipped).
 1537: 
 1538: 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
 1539:     inclusion of <windows.h> to before rather than after the definition of
 1540:     INVALID_FILE_ATTRIBUTES (patch from David Byron).
 1541: 
 1542: 17. Specifying a possessive quantifier with a specific limit for a Unicode
 1543:     character property caused pcre_compile() to compile bad code, which led at
 1544:     runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
 1545:     are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
 1546:     caused the error; without that there was no problem.
 1547: 
 1548: 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
 1549: 
 1550: 19. Added --enable-pcretest-libreadline.
 1551: 
 1552: 20. In pcrecpp.cc, the variable 'count' was incremented twice in
 1553:     RE::GlobalReplace(). As a result, the number of replacements returned was
 1554:     double what it should be. I removed one of the increments, but Craig sent a
 1555:     later patch that removed the other one (the right fix) and added unit tests
 1556:     that check the return values (which was not done before).
 1557: 
 1558: 21. Several CMake things:
 1559: 
 1560:     (1) Arranged that, when cmake is used on Unix, the libraries end up with
 1561:         the names libpcre and libpcreposix, not just pcre and pcreposix.
 1562: 
 1563:     (2) The above change means that pcretest and pcregrep are now correctly
 1564:         linked with the newly-built libraries, not previously installed ones.
 1565: 
 1566:     (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
 1567: 
 1568: 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
 1569:     crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
 1570:     UTF-8 newline character). The key issue is that the pattern starts .*;
 1571:     this means that the match must be either at the beginning, or after a
 1572:     newline. The bug was in the code for advancing after a failed match and
 1573:     checking that the new position followed a newline. It was not taking
 1574:     account of UTF-8 characters correctly.
 1575: 
 1576: 23. PCRE was behaving differently from Perl in the way it recognized POSIX
 1577:     character classes. PCRE was not treating the sequence [:...:] as a
 1578:     character class unless the ... were all letters. Perl, however, seems to
 1579:     allow any characters between [: and :], though of course it rejects as
 1580:     unknown any "names" that contain non-letters, because all the known class
 1581:     names consist only of letters. Thus, Perl gives an error for [[:1234:]],
 1582:     for example, whereas PCRE did not - it did not recognize a POSIX character
 1583:     class. This seemed a bit dangerous, so the code has been changed to be
 1584:     closer to Perl. The behaviour is not identical to Perl, because PCRE will
 1585:     diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
 1586:     treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
 1587:     Perl does, and where it didn't before.
 1588: 
 1589: 24. Rewrite so as to remove the single use of %n from pcregrep because in some
 1590:     Windows environments %n is disabled by default.
 1591: 
 1592: 
 1593: Version 7.4 21-Sep-07
 1594: ---------------------
 1595: 
 1596: 1.  Change 7.3/28 was implemented for classes by looking at the bitmap. This
 1597:     means that a class such as [\s] counted as "explicit reference to CR or
 1598:     LF". That isn't really right - the whole point of the change was to try to
 1599:     help when there was an actual mention of one of the two characters. So now
 1600:     the change happens only if \r or \n (or a literal CR or LF) character is
 1601:     encountered.
 1602: 
 1603: 2.  The 32-bit options word was also used for 6 internal flags, but the numbers
 1604:     of both had grown to the point where there were only 3 bits left.
 1605:     Fortunately, there was spare space in the data structure, and so I have
 1606:     moved the internal flags into a new 16-bit field to free up more option
 1607:     bits.
 1608: 
 1609: 3.  The appearance of (?J) at the start of a pattern set the DUPNAMES option,
 1610:     but did not set the internal JCHANGED flag - either of these is enough to
 1611:     control the way the "get" function works - but the PCRE_INFO_JCHANGED
 1612:     facility is supposed to tell if (?J) was ever used, so now (?J) at the
 1613:     start sets both bits.
 1614: 
 1615: 4.  Added options (at build time, compile time, exec time) to change \R from
 1616:     matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
 1617: 
 1618: 5.  doc/pcresyntax.html was missing from the distribution.
 1619: 
 1620: 6.  Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
 1621:     compatibility, even though it is no longer used.
 1622: 
 1623: 7.  Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
 1624:     strtoull to pcrecpp.cc to select the available functions in WIN32 when the
 1625:     windows.h file is present (where different names are used). [This was
 1626:     reversed later after testing - see 16 below.]
 1627: 
 1628: 8.  Changed all #include <config.h> to #include "config.h". There were also
 1629:     some further <pcre.h> cases that I changed to "pcre.h".
 1630: 
 1631: 9.  When pcregrep was used with the --colour option, it missed the line ending
 1632:     sequence off the lines that it output.
 1633: 
 1634: 10. It was pointed out to me that arrays of string pointers cause lots of
 1635:     relocations when a shared library is dynamically loaded. A technique of
 1636:     using a single long string with a table of offsets can drastically reduce
 1637:     these. I have refactored PCRE in four places to do this. The result is
 1638:     dramatic:
 1639: 
 1640:       Originally:                          290
 1641:       After changing UCP table:            187
 1642:       After changing error message table:   43
 1643:       After changing table of "verbs"       36
 1644:       After changing table of Posix names   22
 1645: 
 1646:     Thanks to the folks working on Gregex for glib for this insight.
 1647: 
 1648: 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
 1649:     unicode-properties was also set.
 1650: 
 1651: 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
 1652: 
 1653: 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
 1654:     checked only for CRLF.
 1655: 
 1656: 14. Added casts to pcretest.c to avoid compiler warnings.
 1657: 
 1658: 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
 1659: 
 1660: 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
 1661:     and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
 1662:     entirely. This removes changes made in 7 above.
 1663: 
 1664: 17. The CMake files have been updated, and there is now more information about
 1665:     building with CMake in the NON-UNIX-USE document.
 1666: 
 1667: 
 1668: Version 7.3 28-Aug-07
 1669: ---------------------
 1670: 
 1671:  1. In the rejigging of the build system that eventually resulted in 7.1, the
 1672:     line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
 1673:     brackets there is not right, since it causes compilers to look for an
 1674:     installed pcre.h, not the version that is in the source that is being
 1675:     compiled (which of course may be different). I have changed it back to:
 1676: 
 1677:       #include "pcre.h"
 1678: 
 1679:     I have a vague recollection that the change was concerned with compiling in
 1680:     different directories, but in the new build system, that is taken care of
 1681:     by the VPATH setting the Makefile.
 1682: 
 1683:  2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
 1684:     when the subject happened to end in the byte 0x85 (e.g. if the last
 1685:     character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
 1686:     characters but of course it shouldn't be taken as a newline when it is part
 1687:     of another character. The bug was that, for an unlimited repeat of . in
 1688:     not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
 1689:     characters when looking for a newline.
 1690: 
 1691:  3. A small performance improvement in the DOTALL UTF-8 mode .* case.
 1692: 
 1693:  4. Debugging: adjusted the names of opcodes for different kinds of parentheses
 1694:     in debug output.
 1695: 
 1696:  5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
 1697:     long printing in the pcrecpp unittest when running under MinGW.
 1698: 
 1699:  6. ESC_K was left out of the EBCDIC table.
 1700: 
 1701:  7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
 1702:     parentheses; I made it 1000, which seemed large enough. Unfortunately, the
 1703:     limit also applies to "virtual nesting" when a pattern is recursive, and in
 1704:     this case 1000 isn't so big. I have been able to remove this limit at the
 1705:     expense of backing off one optimization in certain circumstances. Normally,
 1706:     when pcre_exec() would call its internal match() function recursively and
 1707:     immediately return the result unconditionally, it uses a "tail recursion"
 1708:     feature to save stack. However, when a subpattern that can match an empty
 1709:     string has an unlimited repetition quantifier, it no longer makes this
 1710:     optimization. That gives it a stack frame in which to save the data for
 1711:     checking that an empty string has been matched. Previously this was taken
 1712:     from the 1000-entry workspace that had been reserved. So now there is no
 1713:     explicit limit, but more stack is used.
 1714: 
 1715:  8. Applied Daniel's patches to solve problems with the import/export magic
 1716:     syntax that is required for Windows, and which was going wrong for the
 1717:     pcreposix and pcrecpp parts of the library. These were overlooked when this
 1718:     problem was solved for the main library.
 1719: 
 1720:  9. There were some crude static tests to avoid integer overflow when computing
 1721:     the size of patterns that contain repeated groups with explicit upper
 1722:     limits. As the maximum quantifier is 65535, the maximum group length was
 1723:     set at 30,000 so that the product of these two numbers did not overflow a
 1724:     32-bit integer. However, it turns out that people want to use groups that
 1725:     are longer than 30,000 bytes (though not repeat them that many times).
 1726:     Change 7.0/17 (the refactoring of the way the pattern size is computed) has
 1727:     made it possible to implement the integer overflow checks in a much more
 1728:     dynamic way, which I have now done. The artificial limitation on group
 1729:     length has been removed - we now have only the limit on the total length of
 1730:     the compiled pattern, which depends on the LINK_SIZE setting.
 1731: 
 1732: 10. Fixed a bug in the documentation for get/copy named substring when
 1733:     duplicate names are permitted. If none of the named substrings are set, the
 1734:     functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
 1735:     empty string.
 1736: 
 1737: 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
 1738:     instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
 1739:     because the ] is interpreted as the first data character and the
 1740:     terminating ] is not found. PCRE has been made compatible with Perl in this
 1741:     regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
 1742:     cause memory overwriting.
 1743: 
 1744: 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
 1745:     string has been matched (to stop an infinite loop). It was not recognizing
 1746:     a conditional subpattern that could match an empty string if that
 1747:     subpattern was within another subpattern. For example, it looped when
 1748:     trying to match  (((?(1)X|))*)  but it was OK with  ((?(1)X|)*)  where the
 1749:     condition was not nested. This bug has been fixed.
 1750: 
 1751: 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
 1752:     past the start of the subject in the presence of bytes with the top bit
 1753:     set, for example "\x8aBCD".
 1754: 
 1755: 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
 1756:     (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
 1757: 
 1758: 14. Optimized (?!) to (*FAIL).
 1759: 
 1760: 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
 1761:     This restricts code points to be within the range 0 to 0x10FFFF, excluding
 1762:     the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
 1763:     full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
 1764:     does: it's just the validity check that is more restrictive.
 1765: 
 1766: 16. Inserted checks for integer overflows during escape sequence (backslash)
 1767:     processing, and also fixed erroneous offset values for syntax errors during
 1768:     backslash processing.
 1769: 
 1770: 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
 1771:     for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
 1772: 
 1773: 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
 1774:     caused an overrun.
 1775: 
 1776: 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
 1777:     something other than just ASCII characters) inside a group that had an
 1778:     unlimited repeat caused a loop at compile time (while checking to see
 1779:     whether the group could match an empty string).
 1780: 
 1781: 20. Debugging a pattern containing \p or \P could cause a crash. For example,
 1782:     [\P{Any}] did so. (Error in the code for printing property names.)
 1783: 
 1784: 21. An orphan \E inside a character class could cause a crash.
 1785: 
 1786: 22. A repeated capturing bracket such as (A)? could cause a wild memory
 1787:     reference during compilation.
 1788: 
 1789: 23. There are several functions in pcre_compile() that scan along a compiled
 1790:     expression for various reasons (e.g. to see if it's fixed length for look
 1791:     behind). There were bugs in these functions when a repeated \p or \P was
 1792:     present in the pattern. These operators have additional parameters compared
 1793:     with \d, etc, and these were not being taken into account when moving along
 1794:     the compiled data. Specifically:
 1795: 
 1796:     (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
 1797:         length.
 1798: 
 1799:     (b) An item such as \pL+ within a repeated group could cause crashes or
 1800:         loops.
 1801: 
 1802:     (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
 1803:         "reference to non-existent subpattern" error.
 1804: 
 1805:     (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
 1806: 
 1807: 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
 1808:     characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
 1809: 
 1810: 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
 1811: 
 1812: 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
 1813:     character were causing crashes (broken optimization).
 1814: 
 1815: 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
 1816:     \p or \P) caused a compile-time loop.
 1817: 
 1818: 28. More problems have arisen in unanchored patterns when CRLF is a valid line
 1819:     break. For example, the unstudied pattern [\r\n]A does not match the string
 1820:     "\r\nA" because change 7.0/46 below moves the current point on by two
 1821:     characters after failing to match at the start. However, the pattern \nA
 1822:     *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
 1823:     the same is true. There doesn't seem any very clean way out of this, but
 1824:     what I have chosen to do makes the common cases work: PCRE now takes note
 1825:     of whether there can be an explicit match for \r or \n anywhere in the
 1826:     pattern, and if so, 7.0/46 no longer applies. As part of this change,
 1827:     there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
 1828:     pattern has explicit CR or LF references.
 1829: 
 1830: 29. Added (*CR) etc for changing newline setting at start of pattern.
 1831: 
 1832: 
 1833: Version 7.2 19-Jun-07
 1834: ---------------------
 1835: 
 1836:  1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
 1837:     which is apparently normally available under Windows.
 1838: 
 1839:  2. Re-jig the pcregrep tests with different newline settings in an attempt
 1840:     to make them independent of the local environment's newline setting.
 1841: 
 1842:  3. Add code to configure.ac to remove -g from the CFLAGS default settings.
 1843: 
 1844:  4. Some of the "internals" tests were previously cut out when the link size
 1845:     was not 2, because the output contained actual offsets. The recent new
 1846:     "Z" feature of pcretest means that these can be cut out, making the tests
 1847:     usable with all link sizes.
 1848: 
 1849:  5. Implemented Stan Switzer's goto replacement for longjmp() when not using
 1850:     stack recursion. This gives a massive performance boost under BSD, but just
 1851:     a small improvement under Linux. However, it saves one field in the frame
 1852:     in all cases.
 1853: 
 1854:  6. Added more features from the forthcoming Perl 5.10:
 1855: 
 1856:     (a) (?-n) (where n is a string of digits) is a relative subroutine or
 1857:         recursion call. It refers to the nth most recently opened parentheses.
 1858: 
 1859:     (b) (?+n) is also a relative subroutine call; it refers to the nth next
 1860:         to be opened parentheses.
 1861: 
 1862:     (c) Conditions that refer to capturing parentheses can be specified
 1863:         relatively, for example, (?(-2)... or (?(+3)...
 1864: 
 1865:     (d) \K resets the start of the current match so that everything before
 1866:         is not part of it.
 1867: 
 1868:     (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
 1869: 
 1870:     (f) \g{name} is another synonym - part of Perl 5.10's unification of
 1871:         reference syntax.
 1872: 
 1873:     (g) (?| introduces a group in which the numbering of parentheses in each
 1874:         alternative starts with the same number.
 1875: 
 1876:     (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
 1877: 
 1878:  7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
 1879:     PCRE_INFO_JCHANGED.
 1880: 
 1881:  8. A pattern such as  (.*(.)?)*  caused pcre_exec() to fail by either not
 1882:     terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
 1883:     for detecting groups that can match an empty string.
 1884: 
 1885:  9. A pattern with a very large number of alternatives (more than several
 1886:     hundred) was running out of internal workspace during the pre-compile
 1887:     phase, where pcre_compile() figures out how much memory will be needed. A
 1888:     bit of new cunning has reduced the workspace needed for groups with
 1889:     alternatives. The 1000-alternative test pattern now uses 12 bytes of
 1890:     workspace instead of running out of the 4096 that are available.
 1891: 
 1892: 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
 1893: 
 1894: 11. Applied patch from Google to remove an optimization that didn't quite work.
 1895:     The report of the bug said:
 1896: 
 1897:       pcrecpp::RE("a*").FullMatch("aaa") matches, while
 1898:       pcrecpp::RE("a*?").FullMatch("aaa") does not, and
 1899:       pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
 1900: 
 1901: 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
 1902:     it matched the wrong number of bytes.
 1903: 
 1904: 
 1905: Version 7.1 24-Apr-07
 1906: ---------------------
 1907: 
 1908:  1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
 1909:     that is more "standard", making use of automake and other Autotools. There
 1910:     is some re-arrangement of the files and adjustment of comments consequent
 1911:     on this.
 1912: 
 1913:  2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
 1914:     for recursive directory scanning broke on some systems because the files
 1915:     are not scanned in any specific order and on different systems the order
 1916:     was different. A call to "sort" has been inserted into RunGrepTest for the
 1917:     approprate test as a short-term fix. In the longer term there may be an
 1918:     alternative.
 1919: 
 1920:  3. I had an email from Eric Raymond about problems translating some of PCRE's
 1921:     man pages to HTML (despite the fact that I distribute HTML pages, some
 1922:     people do their own conversions for various reasons). The problems
 1923:     concerned the use of low-level troff macros .br and .in. I have therefore
 1924:     removed all such uses from the man pages (some were redundant, some could
 1925:     be replaced by .nf/.fi pairs). The 132html script that I use to generate
 1926:     HTML has been updated to handle .nf/.fi and to complain if it encounters
 1927:     .br or .in.
 1928: 
 1929:  4. Updated comments in configure.ac that get placed in config.h.in and also
 1930:     arranged for config.h to be included in the distribution, with the name
 1931:     config.h.generic, for the benefit of those who have to compile without
 1932:     Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
 1933: 
 1934:  5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
 1935:     Weber: (1) pcre_internal.h was missing some function renames; (2) updated
 1936:     makevp.bat for the current PCRE, using the additional files
 1937:     makevp_c.txt, makevp_l.txt, and pcregexp.pas.
 1938: 
 1939:  6. A Windows user reported a minor discrepancy with test 2, which turned out
 1940:     to be caused by a trailing space on an input line that had got lost in his
 1941:     copy. The trailing space was an accident, so I've just removed it.
 1942: 
 1943:  7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
 1944:     that is needed.
 1945: 
 1946:  8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
 1947:     as "const" (a) because they are and (b) because it helps the PHP
 1948:     maintainers who have recently made a script to detect big data structures
 1949:     in the php code that should be moved to the .rodata section. I remembered
 1950:     to update Builducptable as well, so it won't revert if ucptable.h is ever
 1951:     re-created.
 1952: 
 1953:  9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
 1954:     pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
 1955:     order to be able to cut out the UTF-8 tables in the latter when UTF-8
 1956:     support is not required. This saves 1.5-2K of code, which is important in
 1957:     some applications.
 1958: 
 1959:     Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
 1960:     so as not to refer to the tables, even though these functions will never be
 1961:     called when UTF-8 support is disabled. Otherwise there are problems with a
 1962:     shared library.
 1963: 
 1964: 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
 1965: 
 1966:     (a) It was defining its arguments as char * instead of void *.
 1967: 
 1968:     (b) It was assuming that all moves were upwards in memory; this was true
 1969:         a long time ago when I wrote it, but is no longer the case.
 1970: 
 1971:     The emulated memove() is provided for those environments that have neither
 1972:     memmove() nor bcopy(). I didn't think anyone used it these days, but that
 1973:     is clearly not the case, as these two bugs were recently reported.
 1974: 
 1975: 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
 1976:     and Detrail to create the HTML documentation, the .txt form of the man
 1977:     pages, and it removes trailing spaces from listed files. It also creates
 1978:     pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
 1979:     case, it wraps all the #defines with #ifndefs. This script should be run
 1980:     before "make dist".
 1981: 
 1982: 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
 1983:     with Unicode property support.
 1984: 
 1985:     (a) For a maximizing quantifier, if the two different cases of the
 1986:         character were of different lengths in their UTF-8 codings (there are
 1987:         some cases like this - I found 11), and the matching function had to
 1988:         back up over a mixture of the two cases, it incorrectly assumed they
 1989:         were both the same length.
 1990: 
 1991:     (b) When PCRE was configured to use the heap rather than the stack for
 1992:         recursion during matching, it was not correctly preserving the data for
 1993:         the other case of a UTF-8 character when checking ahead for a match
 1994:         while processing a minimizing repeat. If the check also involved
 1995:         matching a wide character, but failed, corruption could cause an
 1996:         erroneous result when trying to check for a repeat of the original
 1997:         character.
 1998: 
 1999: 13. Some tidying changes to the testing mechanism:
 2000: 
 2001:     (a) The RunTest script now detects the internal link size and whether there
 2002:         is UTF-8 and UCP support by running ./pcretest -C instead of relying on
 2003:         values substituted by "configure". (The RunGrepTest script already did
 2004:         this for UTF-8.) The configure.ac script no longer substitutes the
 2005:         relevant variables.
 2006: 
 2007:     (b) The debugging options /B and /D in pcretest show the compiled bytecode
 2008:         with length and offset values. This means that the output is different
 2009:         for different internal link sizes. Test 2 is skipped for link sizes
 2010:         other than 2 because of this, bypassing the problem. Unfortunately,
 2011:         there was also a test in test 3 (the locale tests) that used /B and
 2012:         failed for link sizes other than 2. Rather than cut the whole test out,
 2013:         I have added a new /Z option to pcretest that replaces the length and
 2014:         offset values with spaces. This is now used to make test 3 independent
 2015:         of link size. (Test 2 will be tidied up later.)
 2016: 
 2017: 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
 2018:     segmentation fault instead of returning the appropriate error message.
 2019: 
 2020: 15. In multiline mode when the newline sequence was set to "any", the pattern
 2021:     ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
 2022:     This doesn't seem right; it now treats the CRLF combination as the line
 2023:     ending, and so does not match in that case. It's only a pattern such as ^$
 2024:     that would hit this one: something like ^ABC$ would have failed after \r
 2025:     and then tried again after \r\n.
 2026: 
 2027: 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
 2028:     in an attempt to make files that differ only in their line terminators
 2029:     compare equal. This works on Linux.
 2030: 
 2031: 17. Under certain error circumstances pcregrep might try to free random memory
 2032:     as it exited. This is now fixed, thanks to valgrind.
 2033: 
 2034: 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
 2035:     "abc\r\n\r\n", it found an unwanted second match after the second \r. This
 2036:     was because its rules for how to advance for /g after matching an empty
 2037:     string at the end of a line did not allow for this case. They now check for
 2038:     it specially.
 2039: 
 2040: 20. pcretest is supposed to handle patterns and data of any length, by
 2041:     extending its buffers when necessary. It was getting this wrong when the
 2042:     buffer for a data line had to be extended.
 2043: 
 2044: 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
 2045:     CRLF as a newline sequence.
 2046: 
 2047: 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
 2048:     out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
 2049:     I have nevertheless tidied it up.
 2050: 
 2051: 23. Added some casts to kill warnings from HP-UX ia64 compiler.
 2052: 
 2053: 24. Added a man page for pcre-config.
 2054: 
 2055: 
 2056: Version 7.0 19-Dec-06
 2057: ---------------------
 2058: 
 2059:  1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
 2060:     moving to gcc 4.1.1.
 2061: 
 2062:  2. The -S option for pcretest uses setrlimit(); I had omitted to #include
 2063:     sys/time.h, which is documented as needed for this function. It doesn't
 2064:     seem to matter on Linux, but it showed up on some releases of OS X.
 2065: 
 2066:  3. It seems that there are systems where bytes whose values are greater than
 2067:     127 match isprint() in the "C" locale. The "C" locale should be the
 2068:     default when a C program starts up. In most systems, only ASCII printing
 2069:     characters match isprint(). This difference caused the output from pcretest
 2070:     to vary, making some of the tests fail. I have changed pcretest so that:
 2071: 
 2072:     (a) When it is outputting text in the compiled version of a pattern, bytes
 2073:         other than 32-126 are always shown as hex escapes.
 2074: 
 2075:     (b) When it is outputting text that is a matched part of a subject string,
 2076:         it does the same, unless a different locale has been set for the match
 2077:         (using the /L modifier). In this case, it uses isprint() to decide.
 2078: 
 2079:  4. Fixed a major bug that caused incorrect computation of the amount of memory
 2080:     required for a compiled pattern when options that changed within the
 2081:     pattern affected the logic of the preliminary scan that determines the
 2082:     length. The relevant options are -x, and -i in UTF-8 mode. The result was
 2083:     that the computed length was too small. The symptoms of this bug were
 2084:     either the PCRE error "internal error: code overflow" from pcre_compile(),
 2085:     or a glibc crash with a message such as "pcretest: free(): invalid next
 2086:     size (fast)". Examples of patterns that provoked this bug (shown in
 2087:     pcretest format) are:
 2088: 
 2089:       /(?-x: )/x
 2090:       /(?x)(?-x: \s*#\s*)/
 2091:       /((?i)[\x{c0}])/8
 2092:       /(?i:[\x{c0}])/8
 2093: 
 2094:     HOWEVER: Change 17 below makes this fix obsolete as the memory computation
 2095:     is now done differently.
 2096: 
 2097:  5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
 2098:     wrapper classes; (b) implement a new function in the C++ scanner that is
 2099:     more efficient than the old way of doing things because it avoids levels of
 2100:     recursion in the regex matching; (c) add a paragraph to the documentation
 2101:     for the FullMatch() function.
 2102: 
 2103:  6. The escape sequence \n was being treated as whatever was defined as
 2104:     "newline". Not only was this contrary to the documentation, which states
 2105:     that \n is character 10 (hex 0A), but it also went horribly wrong when
 2106:     "newline" was defined as CRLF. This has been fixed.
 2107: 
 2108:  7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
 2109:     was being set to -1 for the "end of line" case (supposedly a value that no
 2110:     character can have). Though this value is never used (the check for end of
 2111:     line is "zero bytes in current character"), it caused compiler complaints.
 2112:     I've changed it to 0xffffffff.
 2113: 
 2114:  8. In pcre_version.c, the version string was being built by a sequence of
 2115:     C macros that, in the event of PCRE_PRERELEASE being defined as an empty
 2116:     string (as it is for production releases) called a macro with an empty
 2117:     argument. The C standard says the result of this is undefined. The gcc
 2118:     compiler treats it as an empty string (which was what was wanted) but it is
 2119:     reported that Visual C gives an error. The source has been hacked around to
 2120:     avoid this problem.
 2121: 
 2122:  9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
 2123:     builds of pcretest, and changed the call to _setmode() to use _O_BINARY
 2124:     instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
 2125:     of them did).
 2126: 
 2127: 10. Originally, pcretest opened its input and output without "b"; then I was
 2128:     told that "b" was needed in some environments, so it was added for release
 2129:     5.0 to both the input and output. (It makes no difference on Unix-like
 2130:     systems.) Later I was told that it is wrong for the input on Windows. I've
 2131:     now abstracted the modes into two macros, to make it easier to fiddle with
 2132:     them, and removed "b" from the input mode under Windows.
 2133: 
 2134: 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
 2135: 
 2136: 12. Added -help and --help to pcretest as an official way of being reminded
 2137:     of the options.
 2138: 
 2139: 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
 2140:     and pcrecpp.cc because they annoy compilers at high warning levels.
 2141: 
 2142: 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
 2143: 
 2144: 15. Fixed an occurrence of == in configure.ac that should have been = (shell
 2145:     scripts are not C programs :-) and which was not noticed because it works
 2146:     on Linux.
 2147: 
 2148: 16. pcretest is supposed to handle any length of pattern and data line (as one
 2149:     line or as a continued sequence of lines) by extending its input buffer if
 2150:     necessary. This feature was broken for very long pattern lines, leading to
 2151:     a string of junk being passed to pcre_compile() if the pattern was longer
 2152:     than about 50K.
 2153: 
 2154: 17. I have done a major re-factoring of the way pcre_compile() computes the
 2155:     amount of memory needed for a compiled pattern. Previously, there was code
 2156:     that made a preliminary scan of the pattern in order to do this. That was
 2157:     OK when PCRE was new, but as the facilities have expanded, it has become
 2158:     harder and harder to keep it in step with the real compile phase, and there
 2159:     have been a number of bugs (see for example, 4 above). I have now found a
 2160:     cunning way of running the real compile function in a "fake" mode that
 2161:     enables it to compute how much memory it would need, while actually only
 2162:     ever using a few hundred bytes of working memory and without too many
 2163:     tests of the mode. This should make future maintenance and development
 2164:     easier. A side effect of this work is that the limit of 200 on the nesting
 2165:     depth of parentheses has been removed (though this was never a serious
 2166:     limitation, I suspect). However, there is a downside: pcre_compile() now
 2167:     runs more slowly than before (30% or more, depending on the pattern). I
 2168:     hope this isn't a big issue. There is no effect on runtime performance.
 2169: 
 2170: 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
 2171:     newline (only possible for the last line of a file) and it was a
 2172:     pattern that set a locale (followed by /Lsomething), pcretest crashed.
 2173: 
 2174: 19. Added additional timing features to pcretest. (1) The -tm option now times
 2175:     matching only, not compiling. (2) Both -t and -tm can be followed, as a
 2176:     separate command line item, by a number that specifies the number of
 2177:     repeats to use when timing. The default is 50000; this gives better
 2178:     precision, but takes uncomfortably long for very large patterns.
 2179: 
 2180: 20. Extended pcre_study() to be more clever in cases where a branch of a
 2181:     subpattern has no definite first character. For example, (a*|b*)[cd] would
 2182:     previously give no result from pcre_study(). Now it recognizes that the
 2183:     first character must be a, b, c, or d.
 2184: 
 2185: 21. There was an incorrect error "recursive call could loop indefinitely" if
 2186:     a subpattern (or the entire pattern) that was being tested for matching an
 2187:     empty string contained only one non-empty item after a nested subpattern.
 2188:     For example, the pattern (?>\x{100}*)\d(?R) provoked this error
 2189:     incorrectly, because the \d was being skipped in the check.
 2190: 
 2191: 22. The pcretest program now has a new pattern option /B and a command line
 2192:     option -b, which is equivalent to adding /B to every pattern. This causes
 2193:     it to show the compiled bytecode, without the additional information that
 2194:     -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
 2195:     is the same as /B/I).
 2196: 
 2197: 23. A new optimization is now able automatically to treat some sequences such
 2198:     as a*b as a*+b. More specifically, if something simple (such as a character
 2199:     or a simple class like \d) has an unlimited quantifier, and is followed by
 2200:     something that cannot possibly match the quantified thing, the quantifier
 2201:     is automatically "possessified".
 2202: 
 2203: 24. A recursive reference to a subpattern whose number was greater than 39
 2204:     went wrong under certain circumstances in UTF-8 mode. This bug could also
 2205:     have affected the operation of pcre_study().
 2206: 
 2207: 25. Realized that a little bit of performance could be had by replacing
 2208:     (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
 2209: 
 2210: 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
 2211: 
 2212: 27. Possessive quantifiers such as a++ were previously implemented by turning
 2213:     them into atomic groups such as ($>a+). Now they have their own opcodes,
 2214:     which improves performance. This includes the automatically created ones
 2215:     from 23 above.
 2216: 
 2217: 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
 2218:     lookahead was broken if it was not anchored. PCRE was mistakenly expecting
 2219:     the first matched character to be a colon. This applied both to named and
 2220:     numbered groups.
 2221: 
 2222: 29. The ucpinternal.h header file was missing its idempotency #ifdef.
 2223: 
 2224: 30. I was sent a "project" file called libpcre.a.dev which I understand makes
 2225:     building PCRE on Windows easier, so I have included it in the distribution.
 2226: 
 2227: 31. There is now a check in pcretest against a ridiculously large number being
 2228:     returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
 2229:     loop, the loop is abandoned.
 2230: 
 2231: 32. Forward references to subpatterns in conditions such as (?(2)...) where
 2232:     subpattern 2 is defined later cause pcre_compile() to search forwards in
 2233:     the pattern for the relevant set of parentheses. This search went wrong
 2234:     when there were unescaped parentheses in a character class, parentheses
 2235:     escaped with \Q...\E, or parentheses in a #-comment in /x mode.
 2236: 
 2237: 33. "Subroutine" calls and backreferences were previously restricted to
 2238:     referencing subpatterns earlier in the regex. This restriction has now
 2239:     been removed.
 2240: 
 2241: 34. Added a number of extra features that are going to be in Perl 5.10. On the
 2242:     whole, these are just syntactic alternatives for features that PCRE had
 2243:     previously implemented using the Python syntax or my own invention. The
 2244:     other formats are all retained for compatibility.
 2245: 
 2246:     (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
 2247:         as (?P<name>...). The new forms, as well as being in Perl 5.10, are
 2248:         also .NET compatible.
 2249: 
 2250:     (b) A recursion or subroutine call to a named group can now be defined as
 2251:         (?&name) as well as (?P>name).
 2252: 
 2253:     (c) A backreference to a named group can now be defined as \k<name> or
 2254:         \k'name' as well as (?P=name). The new forms, as well as being in Perl
 2255:         5.10, are also .NET compatible.
 2256: 
 2257:     (d) A conditional reference to a named group can now use the syntax
 2258:         (?(<name>) or (?('name') as well as (?(name).
 2259: 
 2260:     (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
 2261:         groups (named and numbered) that are never evaluated inline, but can be
 2262:         called as "subroutines" from elsewhere. In effect, the DEFINE condition
 2263:         is always false. There may be only one alternative in such a group.
 2264: 
 2265:     (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
 2266:         as the simple (?(R). The condition is true only if the most recent
 2267:         recursion is that of the given number or name. It does not search out
 2268:         through the entire recursion stack.
 2269: 
 2270:     (g) The escape \gN or \g{N} has been added, where N is a positive or
 2271:         negative number, specifying an absolute or relative reference.
 2272: 
 2273: 35. Tidied to get rid of some further signed/unsigned compiler warnings and
 2274:     some "unreachable code" warnings.
 2275: 
 2276: 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
 2277:     things, this adds five new scripts.
 2278: 
 2279: 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
 2280:     There were also incompatibilities regarding the handling of \Q..\E inside
 2281:     character classes, for example with patterns like [\Qa\E-\Qz\E] where the
 2282:     hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
 2283: 
 2284: 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
 2285:     matches an empty string, and forcibly breaks the loop. There were bugs in
 2286:     this code in non-simple cases. For a pattern such as  ^(a()*)*  matched
 2287:     against  aaaa  the result was just "a" rather than "aaaa", for example. Two
 2288:     separate and independent bugs (that affected different cases) have been
 2289:     fixed.
 2290: 
 2291: 39. Refactored the code to abolish the use of different opcodes for small
 2292:     capturing bracket numbers. This is a tidy that I avoided doing when I
 2293:     removed the limit on the number of capturing brackets for 3.5 back in 2001.
 2294:     The new approach is not only tidier, it makes it possible to reduce the
 2295:     memory needed to fix the previous bug (38).
 2296: 
 2297: 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
 2298:     sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
 2299:     processing dot, circumflex, or dollar metacharacters, or #-comments in /x
 2300:     mode.
 2301: 
 2302: 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
 2303:     report.
 2304: 
 2305: 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
 2306:     copy construction and assignment in the C++ wrapper.
 2307: 
 2308: 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
 2309:     couple of bugs that could have given wrong results in the "--newline=crlf"
 2310:     case.
 2311: 
 2312: 44. Added a number of casts and did some reorganization of signed/unsigned int
 2313:     variables following suggestions from Dair Grant. Also renamed the variable
 2314:     "this" as "item" because it is a C++ keyword.
 2315: 
 2316: 45. Arranged for dftables to add
 2317: 
 2318:       #include "pcre_internal.h"
 2319: 
 2320:     to pcre_chartables.c because without it, gcc 4.x may remove the array
 2321:     definition from the final binary if PCRE is built into a static library and
 2322:     dead code stripping is activated.
 2323: 
 2324: 46. For an unanchored pattern, if a match attempt fails at the start of a
 2325:     newline sequence, and the newline setting is CRLF or ANY, and the next two
 2326:     characters are CRLF, advance by two characters instead of one.
 2327: 
 2328: 
 2329: Version 6.7 04-Jul-06
 2330: ---------------------
 2331: 
 2332:  1. In order to handle tests when input lines are enormously long, pcretest has
 2333:     been re-factored so that it automatically extends its buffers when
 2334:     necessary. The code is crude, but this _is_ just a test program. The
 2335:     default size has been increased from 32K to 50K.
 2336: 
 2337:  2. The code in pcre_study() was using the value of the re argument before
 2338:     testing it for NULL. (Of course, in any sensible call of the function, it
 2339:     won't be NULL.)
 2340: 
 2341:  3. The memmove() emulation function in pcre_internal.h, which is used on
 2342:     systems that lack both memmove() and bcopy() - that is, hardly ever -
 2343:     was missing a "static" storage class specifier.
 2344: 
 2345:  4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
 2346:     containing an extended class (one that cannot be represented by a bitmap
 2347:     because it contains high-valued characters or Unicode property items, e.g.
 2348:     [\pZ]). Almost always one would set UTF-8 mode when processing such a
 2349:     pattern, but PCRE should not loop if you do not (it no longer does).
 2350:     [Detail: two cases were found: (a) a repeated subpattern containing an
 2351:     extended class; (b) a recursive reference to a subpattern that followed a
 2352:     previous extended class. It wasn't skipping over the extended class
 2353:     correctly when UTF-8 mode was not set.]
 2354: 
 2355:  5. A negated single-character class was not being recognized as fixed-length
 2356:     in lookbehind assertions such as (?<=[^f]), leading to an incorrect
 2357:     compile error "lookbehind assertion is not fixed length".
 2358: 
 2359:  6. The RunPerlTest auxiliary script was showing an unexpected difference
 2360:     between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
 2361:     write a Perl script that can interpret lines of an input file either as
 2362:     byte characters or as UTF-8, which is what "perltest" was being required to
 2363:     do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
 2364:     can't do is switch easily at run time between having the "use utf8;" pragma
 2365:     or not. In the end, I fudged it by using the RunPerlTest script to insert
 2366:     "use utf8;" explicitly for the UTF-8 tests.
 2367: 
 2368:  7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
 2369:     the end of the subject string, contrary to the documentation and to what
 2370:     Perl does. This was true of both matching functions. Now it matches only at
 2371:     the start of the subject and immediately after *internal* newlines.
 2372: 
 2373:  8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
 2374:     a pointer to an int instead of a pointer to an unsigned long int. This
 2375:     caused problems on 64-bit systems.
 2376: 
 2377:  9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
 2378:     instance of the 'standard' template library not being so standard".
 2379: 
 2380: 10. There was no check on the number of named subpatterns nor the maximum
 2381:     length of a subpattern name. The product of these values is used to compute
 2382:     the size of the memory block for a compiled pattern. By supplying a very
 2383:     long subpattern name and a large number of named subpatterns, the size
 2384:     computation could be caused to overflow. This is now prevented by limiting
 2385:     the length of names to 32 characters, and the number of named subpatterns
 2386:     to 10,000.
 2387: 
 2388: 11. Subpatterns that are repeated with specific counts have to be replicated in
 2389:     the compiled pattern. The size of memory for this was computed from the
 2390:     length of the subpattern and the repeat count. The latter is limited to
 2391:     65535, but there was no limit on the former, meaning that integer overflow
 2392:     could in principle occur. The compiled length of a repeated subpattern is
 2393:     now limited to 30,000 bytes in order to prevent this.
 2394: 
 2395: 12. Added the optional facility to have named substrings with the same name.
 2396: 
 2397: 13. Added the ability to use a named substring as a condition, using the
 2398:     Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
 2399:     are numbers (not recommended). Forward references are permitted.
 2400: 
 2401: 14. Added forward references in named backreferences (if you see what I mean).
 2402: 
 2403: 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
 2404:     pattern could run off the end of the subject. For example, the pattern
 2405:     "(?s)(.{1,5})"8 did this with the subject "ab".
 2406: 
 2407: 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
 2408:     PCRE_CASELESS was set when matching characters that were quantified with ?
 2409:     or *.
 2410: 
 2411: 17. A character class other than a single negated character that had a minimum
 2412:     but no maximum quantifier - for example [ab]{6,} - was not handled
 2413:     correctly by pce_dfa_exec(). It would match only one character.
 2414: 
 2415: 18. A valid (though odd) pattern that looked like a POSIX character
 2416:     class but used an invalid character after [ (for example [[,abc,]]) caused
 2417:     pcre_compile() to give the error "Failed: internal error: code overflow" or
 2418:     in some cases to crash with a glibc free() error. This could even happen if
 2419:     the pattern terminated after [[ but there just happened to be a sequence of
 2420:     letters, a binary zero, and a closing ] in the memory that followed.
 2421: 
 2422: 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
 2423:     over the years. Originally (before any Unicode support), just the bottom 8
 2424:     bits were taken. Thus, for example, \500 really meant \100. Nowadays the
 2425:     output from "man perlunicode" includes this:
 2426: 
 2427:       The regular expression compiler produces polymorphic opcodes.  That
 2428:       is, the pattern adapts to the data and automatically switches to
 2429:       the Unicode character scheme when presented with Unicode data--or
 2430:       instead uses a traditional byte scheme when presented with byte
 2431:       data.
 2432: 
 2433:     Sadly, a wide octal escape does not cause a switch, and in a string with
 2434:     no other multibyte characters, these octal escapes are treated as before.
 2435:     Thus, in Perl, the pattern  /\500/ actually matches \100 but the pattern
 2436:     /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
 2437:     Unicode string.
 2438: 
 2439:     I have not perpetrated such confusion in PCRE. Up till now, it took just
 2440:     the bottom 8 bits, as in old Perl. I have now made octal escapes with
 2441:     values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
 2442:     translate to the appropriate multibyte character.
 2443: 
 2444: 29. Applied some refactoring to reduce the number of warnings from Microsoft
 2445:     and Borland compilers. This has included removing the fudge introduced
 2446:     seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
 2447:     a warning about an unused variable.
 2448: 
 2449: 21. PCRE has not included VT (character 0x0b) in the set of whitespace
 2450:     characters since release 4.0, because Perl (from release 5.004) does not.
 2451:     [Or at least, is documented not to: some releases seem to be in conflict
 2452:     with the documentation.] However, when a pattern was studied with
 2453:     pcre_study() and all its branches started with \s, PCRE still included VT
 2454:     as a possible starting character. Of course, this did no harm; it just
 2455:     caused an unnecessary match attempt.
 2456: 
 2457: 22. Removed a now-redundant internal flag bit that recorded the fact that case
 2458:     dependency changed within the pattern. This was once needed for "required
 2459:     byte" processing, but is no longer used. This recovers a now-scarce options
 2460:     bit. Also moved the least significant internal flag bit to the most-
 2461:     significant bit of the word, which was not previously used (hangover from
 2462:     the days when it was an int rather than a uint) to free up another bit for
 2463:     the future.
 2464: 
 2465: 23. Added support for CRLF line endings as well as CR and LF. As well as the
 2466:     default being selectable at build time, it can now be changed at runtime
 2467:     via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
 2468:     specify that it is scanning data with non-default line endings.
 2469: 
 2470: 24. Changed the definition of CXXLINK to make it agree with the definition of
 2471:     LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
 2472: 
 2473: 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
 2474:     recursions. This makes a big different to stack usage for some patterns.
 2475: 
 2476: 26. If a subpattern containing a named recursion or subroutine reference such
 2477:     as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
 2478:     the space required for the compiled pattern went wrong and gave too small a
 2479:     value. Depending on the environment, this could lead to "Failed: internal
 2480:     error: code overflow at offset 49" or "glibc detected double free or
 2481:     corruption" errors.
 2482: 
 2483: 27. Applied patches from Google (a) to support the new newline modes and (b) to
 2484:     advance over multibyte UTF-8 characters in GlobalReplace.
 2485: 
 2486: 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
 2487:     difference for some implementation of PCRE in some Windows version.
 2488: 
 2489: 29. Added some extra testing facilities to pcretest:
 2490: 
 2491:     \q<number>   in a data line sets the "match limit" value
 2492:     \Q<number>   in a data line sets the "match recursion limt" value
 2493:     -S <number>  sets the stack size, where <number> is in megabytes
 2494: 
 2495:     The -S option isn't available for Windows.
 2496: 
 2497: 
 2498: Version 6.6 06-Feb-06
 2499: ---------------------
 2500: 
 2501:  1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
 2502:     in pcreposix.h. I have copied the definition from pcre.h.
 2503: 
 2504:  2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
 2505:     because pcre.h is no longer a built file.
 2506: 
 2507:  3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
 2508:     not normally included in the compiled code.
 2509: 
 2510: 
 2511: Version 6.5 01-Feb-06
 2512: ---------------------
 2513: 
 2514:  1. When using the partial match feature with pcre_dfa_exec(), it was not
 2515:     anchoring the second and subsequent partial matches at the new starting
 2516:     point. This could lead to incorrect results. For example, with the pattern
 2517:     /1234/, partially matching against "123" and then "a4" gave a match.
 2518: 
 2519:  2. Changes to pcregrep:
 2520: 
 2521:     (a) All non-match returns from pcre_exec() were being treated as failures
 2522:         to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
 2523:         error message is output. Some extra information is given for the
 2524:         PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
 2525:         probably the only errors that are likely to be caused by users (by
 2526:         specifying a regex that has nested indefinite repeats, for instance).
 2527:         If there are more than 20 of these errors, pcregrep is abandoned.
 2528: 
 2529:     (b) A binary zero was treated as data while matching, but terminated the
 2530:         output line if it was written out. This has been fixed: binary zeroes
 2531:         are now no different to any other data bytes.
 2532: 
 2533:     (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
 2534:         used to set a locale for matching. The --locale=xxxx long option has
 2535:         been added (no short equivalent) to specify a locale explicitly on the
 2536:         pcregrep command, overriding the environment variables.
 2537: 
 2538:     (d) When -B was used with -n, some line numbers in the output were one less
 2539:         than they should have been.
 2540: 
 2541:     (e) Added the -o (--only-matching) option.
 2542: 
 2543:     (f) If -A or -C was used with -c (count only), some lines of context were
 2544:         accidentally printed for the final match.
 2545: 
 2546:     (g) Added the -H (--with-filename) option.
 2547: 
 2548:     (h) The combination of options -rh failed to suppress file names for files
 2549:         that were found from directory arguments.
 2550: 
 2551:     (i) Added the -D (--devices) and -d (--directories) options.
 2552: 
 2553:     (j) Added the -F (--fixed-strings) option.
 2554: 
 2555:     (k) Allow "-" to be used as a file name for -f as well as for a data file.
 2556: 
 2557:     (l) Added the --colo(u)r option.
 2558: 
 2559:     (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
 2560:         is not present by default.
 2561: 
 2562:  3. A nasty bug was discovered in the handling of recursive patterns, that is,
 2563:     items such as (?R) or (?1), when the recursion could match a number of
 2564:     alternatives. If it matched one of the alternatives, but subsequently,
 2565:     outside the recursion, there was a failure, the code tried to back up into
 2566:     the recursion. However, because of the way PCRE is implemented, this is not
 2567:     possible, and the result was an incorrect result from the match.
 2568: 
 2569:     In order to prevent this happening, the specification of recursion has
 2570:     been changed so that all such subpatterns are automatically treated as
 2571:     atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
 2572: 
 2573:  4. I had overlooked the fact that, in some locales, there are characters for
 2574:     which isalpha() is true but neither isupper() nor islower() are true. In
 2575:     the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
 2576:     and ordfeminine) are like this. This affected the treatment of \w and \W
 2577:     when they appeared in character classes, but not when they appeared outside
 2578:     a character class. The bit map for "word" characters is now created
 2579:     separately from the results of isalnum() instead of just taking it from the
 2580:     upper, lower, and digit maps. (Plus the underscore character, of course.)
 2581: 
 2582:  5. The above bug also affected the handling of POSIX character classes such as
 2583:     [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
 2584:     permanent tables. Instead, the bit maps for such a class were previously
 2585:     created as the appropriate unions of the upper, lower, and digit bitmaps.
 2586:     Now they are created by subtraction from the [[:word:]] class, which has
 2587:     its own bitmap.
 2588: 
 2589:  6. The [[:blank:]] character class matches horizontal, but not vertical space.
 2590:     It is created by subtracting the vertical space characters (\x09, \x0a,
 2591:     \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
 2592:     subtraction was done in the overall bitmap for a character class, meaning
 2593:     that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
 2594:     be recognized. This bug has been fixed.
 2595: 
 2596:  7. Patches from the folks at Google:
 2597: 
 2598:       (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
 2599:       real life, but is still worth protecting against".
 2600: 
 2601:       (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
 2602:       regular expressions".
 2603: 
 2604:       (c) pcre_scanner.cc: avoid use of std::count() because not all systems
 2605:       have it.
 2606: 
 2607:       (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
 2608:       "configure" and the latter not, in order to fix a problem somebody had
 2609:       with compiling the Arg class on HP-UX.
 2610: 
 2611:       (e) Improve the error-handling of the C++ wrapper a little bit.
 2612: 
 2613:       (f) New tests for checking recursion limiting.
 2614: 
 2615:  8. The pcre_memmove() function, which is used only if the environment does not
 2616:     have a standard memmove() function (and is therefore rarely compiled),
 2617:     contained two bugs: (a) use of int instead of size_t, and (b) it was not
 2618:     returning a result (though PCRE never actually uses the result).
 2619: 
 2620:  9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
 2621:     large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
 2622:     returned instead of calling malloc() with an overflowing number that would
 2623:     most likely cause subsequent chaos.
 2624: 
 2625: 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
 2626: 
 2627: 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
 2628:     with this option is matched, the nmatch and pmatch options of regexec() are
 2629:     ignored.
 2630: 
 2631: 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
 2632:     provided in case anyone wants to the the POSIX interface with UTF-8
 2633:     strings.
 2634: 
 2635: 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
 2636:     C++ linking (needed for some HP-UX environments).
 2637: 
 2638: 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
 2639:     (unused parameter) and in the pcre_printint() function (omitted "default"
 2640:     switch label when the default is to do nothing).
 2641: 
 2642: 15. Added some code to make it possible, when PCRE is compiled as a C++
 2643:     library, to replace subject pointers for pcre_exec() with a smart pointer
 2644:     class, thus making it possible to process discontinuous strings.
 2645: 
 2646: 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
 2647:     much the same function. They were added by different people who were trying
 2648:     to make PCRE easy to compile on non-Unix systems. It has been suggested
 2649:     that PCRE_EXPORT be abolished now that there is more automatic apparatus
 2650:     for compiling on Windows systems. I have therefore replaced it with
 2651:     PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
 2652:     defaults to "extern" for C or "extern C" for C++, which works fine on
 2653:     Unix-like systems. It is now possible to override the value of PCRE_DATA_
 2654:     SCOPE with something explicit in config.h. In addition:
 2655: 
 2656:     (a) pcreposix.h still had just "extern" instead of either of these macros;
 2657:         I have replaced it with PCRE_DATA_SCOPE.
 2658: 
 2659:     (b) Functions such as _pcre_xclass(), which are internal to the library,
 2660:         but external in the C sense, all had PCRE_EXPORT in their definitions.
 2661:         This is apparently wrong for the Windows case, so I have removed it.
 2662:         (It makes no difference on Unix-like systems.)
 2663: 
 2664: 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
 2665:     of recursive calls to match(). This is different to MATCH_LIMIT because
 2666:     that limits the total number of calls to match(), not all of which increase
 2667:     the depth of recursion. Limiting the recursion depth limits the amount of
 2668:     stack (or heap if NO_RECURSE is set) that is used. The default can be set
 2669:     when PCRE is compiled, and changed at run time. A patch from Google adds
 2670:     this functionality to the C++ interface.
 2671: 
 2672: 18. Changes to the handling of Unicode character properties:
 2673: 
 2674:     (a) Updated the table to Unicode 4.1.0.
 2675: 
 2676:     (b) Recognize characters that are not in the table as "Cn" (undefined).
 2677: 
 2678:     (c) I revised the way the table is implemented to a much improved format
 2679:         which includes recognition of ranges. It now supports the ranges that
 2680:         are defined in UnicodeData.txt, and it also amalgamates other
 2681:         characters into ranges. This has reduced the number of entries in the
 2682:         table from around 16,000 to around 3,000, thus reducing its size
 2683:         considerably. I realized I did not need to use a tree structure after
 2684:         all - a binary chop search is just as efficient. Having reduced the
 2685:         number of entries, I extended their size from 6 bytes to 8 bytes to
 2686:         allow for more data.
 2687: 
 2688:     (d) Added support for Unicode script names via properties such as \p{Han}.
 2689: 
 2690: 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
 2691:     matching that character.
 2692: 
 2693: 20. When matching a repeated Unicode property with a minimum greater than zero,
 2694:     (for example \pL{2,}), PCRE could look past the end of the subject if it
 2695:     reached it while seeking the minimum number of characters. This could
 2696:     happen only if some of the characters were more than one byte long, because
 2697:     there is a check for at least the minimum number of bytes.
 2698: 
 2699: 21. Refactored the implementation of \p and \P so as to be more general, to
 2700:     allow for more different types of property in future. This has changed the
 2701:     compiled form incompatibly. Anybody with saved compiled patterns that use
 2702:     \p or \P will have to recompile them.
 2703: 
 2704: 22. Added "Any" and "L&" to the supported property types.
 2705: 
 2706: 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
 2707:     but give a compile time error if the value is greater than 0xff.
 2708: 
 2709: 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
 2710:     accidentally not being installed or uninstalled.
 2711: 
 2712: 25. The pcre.h file was built from pcre.h.in, but the only changes that were
 2713:     made were to insert the current release number. This seemed silly, because
 2714:     it made things harder for people building PCRE on systems that don't run
 2715:     "configure". I have turned pcre.h into a distributed file, no longer built
 2716:     by "configure", with the version identification directly included. There is
 2717:     no longer a pcre.h.in file.
 2718: 
 2719:     However, this change necessitated a change to the pcre-config script as
 2720:     well. It is built from pcre-config.in, and one of the substitutions was the
 2721:     release number. I have updated configure.ac so that ./configure now finds
 2722:     the release number by grepping pcre.h.
 2723: 
 2724: 26. Added the ability to run the tests under valgrind.
 2725: 
 2726: 
 2727: Version 6.4 05-Sep-05
 2728: ---------------------
 2729: 
 2730:  1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
 2731:     "--" to be printed when multiple files were scanned, even when none of the
 2732:     -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
 2733:     consider it to be a bug, and have restored the previous behaviour.
 2734: 
 2735:  2. A couple of code tidies to get rid of compiler warnings.
 2736: 
 2737:  3. The pcretest program used to cheat by referring to symbols in the library
 2738:     whose names begin with _pcre_. These are internal symbols that are not
 2739:     really supposed to be visible externally, and in some environments it is
 2740:     possible to suppress them. The cheating is now confined to including
 2741:     certain files from the library's source, which is a bit cleaner.
 2742: 
 2743:  4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
 2744:     file's purpose clearer.
 2745: 
 2746:  5. Reorganized pcre_ucp_findchar().
 2747: 
 2748: 
 2749: Version 6.3 15-Aug-05
 2750: ---------------------
 2751: 
 2752:  1. The file libpcre.pc.in did not have general read permission in the tarball.
 2753: 
 2754:  2. There were some problems when building without C++ support:
 2755: 
 2756:     (a) If C++ support was not built, "make install" and "make test" still
 2757:         tried to test it.
 2758: 
 2759:     (b) There were problems when the value of CXX was explicitly set. Some
 2760:         changes have been made to try to fix these, and ...
 2761: 
 2762:     (c) --disable-cpp can now be used to explicitly disable C++ support.
 2763: 
 2764:     (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
 2765:         backslash in a target when C++ was disabled. This confuses some
 2766:         versions of "make", apparently. Using an intermediate variable solves
 2767:         this. (Same for CPP_LOBJ.)
 2768: 
 2769:  3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
 2770:     (non-Windows) now includes $(CFLAGS) because these flags are sometimes
 2771:     necessary on certain architectures.
 2772: 
 2773:  4. Added a setting of -export-symbols-regex to the link command to remove
 2774:     those symbols that are exported in the C sense, but actually are local
 2775:     within the library, and not documented. Their names all begin with
 2776:     "_pcre_". This is not a perfect job, because (a) we have to except some
 2777:     symbols that pcretest ("illegally") uses, and (b) the facility isn't always
 2778:     available (and never for static libraries). I have made a note to try to
 2779:     find a way round (a) in the future.
 2780: 
 2781: 
 2782: Version 6.2 01-Aug-05
 2783: ---------------------
 2784: 
 2785:  1. There was no test for integer overflow of quantifier values. A construction
 2786:     such as {1111111111111111} would give undefined results. What is worse, if
 2787:     a minimum quantifier for a parenthesized subpattern overflowed and became
 2788:     negative, the calculation of the memory size went wrong. This could have
 2789:     led to memory overwriting.
 2790: 
 2791:  2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
 2792: 
 2793:  3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
 2794:     operating environments where this matters.
 2795: 
 2796:  4. Applied Giuseppe Maxia's patch to add additional features for controlling
 2797:     PCRE options from within the C++ wrapper.
 2798: 
 2799:  5. Named capturing subpatterns were not being correctly counted when a pattern
 2800:     was compiled. This caused two problems: (a) If there were more than 100
 2801:     such subpatterns, the calculation of the memory needed for the whole
 2802:     compiled pattern went wrong, leading to an overflow error. (b) Numerical
 2803:     back references of the form \12, where the number was greater than 9, were
 2804:     not recognized as back references, even though there were sufficient
 2805:     previous subpatterns.
 2806: 
 2807:  6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
 2808:     versions of gcc, e.g. 2.95.4.
 2809: 
 2810: 
 2811: Version 6.1 21-Jun-05
 2812: ---------------------
 2813: 
 2814:  1. There was one reference to the variable "posix" in pcretest.c that was not
 2815:     surrounded by "#if !defined NOPOSIX".
 2816: 
 2817:  2. Make it possible to compile pcretest without DFA support, UTF8 support, or
 2818:     the cross-check on the old pcre_info() function, for the benefit of the
 2819:     cut-down version of PCRE that is currently imported into Exim.
 2820: 
 2821:  3. A (silly) pattern starting with (?i)(?-i) caused an internal space
 2822:     allocation error. I've done the easy fix, which wastes 2 bytes for sensible
 2823:     patterns that start (?i) but I don't think that matters. The use of (?i) is
 2824:     just an example; this all applies to the other options as well.
 2825: 
 2826:  4. Since libtool seems to echo the compile commands it is issuing, the output
 2827:     from "make" can be reduced a bit by putting "@" in front of each libtool
 2828:     compile command.
 2829: 
 2830:  5. Patch from the folks at Google for configure.in to be a bit more thorough
 2831:     in checking for a suitable C++ installation before trying to compile the
 2832:     C++ stuff. This should fix a reported problem when a compiler was present,
 2833:     but no suitable headers.
 2834: 
 2835:  6. The man pages all had just "PCRE" as their title. I have changed them to
 2836:     be the relevant file name. I have also arranged that these names are
 2837:     retained in the file doc/pcre.txt, which is a concatenation in text format
 2838:     of all the man pages except the little individual ones for each function.
 2839: 
 2840:  7. The NON-UNIX-USE file had not been updated for the different set of source
 2841:     files that come with release 6. I also added a few comments about the C++
 2842:     wrapper.
 2843: 
 2844: 
 2845: Version 6.0 07-Jun-05
 2846: ---------------------
 2847: 
 2848:  1. Some minor internal re-organization to help with my DFA experiments.
 2849: 
 2850:  2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
 2851:     didn't matter for the library itself when fully configured, but did matter
 2852:     when compiling without UCP support, or within Exim, where the ucp files are
 2853:     not imported.
 2854: 
 2855:  3. Refactoring of the library code to split up the various functions into
 2856:     different source modules. The addition of the new DFA matching code (see
 2857:     below) to a single monolithic source would have made it really too
 2858:     unwieldy, quite apart from causing all the code to be include in a
 2859:     statically linked application, when only some functions are used. This is
 2860:     relevant even without the DFA addition now that patterns can be compiled in
 2861:     one application and matched in another.
 2862: 
 2863:     The downside of splitting up is that there have to be some external
 2864:     functions and data tables that are used internally in different modules of
 2865:     the library but which are not part of the API. These have all had their
 2866:     names changed to start with "_pcre_" so that they are unlikely to clash
 2867:     with other external names.
 2868: 
 2869:  4. Added an alternate matching function, pcre_dfa_exec(), which matches using
 2870:     a different (DFA) algorithm. Although it is slower than the original
 2871:     function, it does have some advantages for certain types of matching
 2872:     problem.
 2873: 
 2874:  5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
 2875:     including restarting after a partial match.
 2876: 
 2877:  6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
 2878:     defined when compiling for Windows was sent to me. I have put it into the
 2879:     code, though I have no means of testing or verifying it.
 2880: 
 2881:  7. Added the pcre_refcount() auxiliary function.
 2882: 
 2883:  8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
 2884:     match before or at the first newline in the subject string. In pcretest,
 2885:     the /f option on a pattern can be used to set this.
 2886: 
 2887:  9. A repeated \w when used in UTF-8 mode with characters greater than 256
 2888:     would behave wrongly. This has been present in PCRE since release 4.0.
 2889: 
 2890: 10. A number of changes to the pcregrep command:
 2891: 
 2892:     (a) Refactored how -x works; insert ^(...)$ instead of setting
 2893:         PCRE_ANCHORED and checking the length, in preparation for adding
 2894:         something similar for -w.
 2895: 
 2896:     (b) Added the -w (match as a word) option.
 2897: 
 2898:     (c) Refactored the way lines are read and buffered so as to have more
 2899:         than one at a time available.
 2900: 
 2901:     (d) Implemented a pcregrep test script.
 2902: 
 2903:     (e) Added the -M (multiline match) option. This allows patterns to match
 2904:         over several lines of the subject. The buffering ensures that at least
 2905:         8K, or the rest of the document (whichever is the shorter) is available
 2906:         for matching (and similarly the previous 8K for lookbehind assertions).
 2907: 
 2908:     (f) Changed the --help output so that it now says
 2909: 
 2910:           -w, --word-regex(p)
 2911: 
 2912:         instead of two lines, one with "regex" and the other with "regexp"
 2913:         because that confused at least one person since the short forms are the
 2914:         same. (This required a bit of code, as the output is generated
 2915:         automatically from a table. It wasn't just a text change.)
 2916: 
 2917:     (g) -- can be used to terminate pcregrep options if the next thing isn't an
 2918:         option but starts with a hyphen. Could be a pattern or a path name
 2919:         starting with a hyphen, for instance.
 2920: 
 2921:     (h) "-" can be given as a file name to represent stdin.
 2922: 
 2923:     (i) When file names are being printed, "(standard input)" is used for
 2924:         the standard input, for compatibility with GNU grep. Previously
 2925:         "<stdin>" was used.
 2926: 
 2927:     (j) The option --label=xxx can be used to supply a name to be used for
 2928:         stdin when file names are being printed. There is no short form.
 2929: 
 2930:     (k) Re-factored the options decoding logic because we are going to add
 2931:         two more options that take data. Such options can now be given in four
 2932:         different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
 2933: 
 2934:     (l) Added the -A, -B, and -C options for requesting that lines of context
 2935:         around matches be printed.
 2936: 
 2937:     (m) Added the -L option to print the names of files that do not contain
 2938:         any matching lines, that is, the complement of -l.
 2939: 
 2940:     (n) The return code is 2 if any file cannot be opened, but pcregrep does
 2941:         continue to scan other files.
 2942: 
 2943:     (o) The -s option was incorrectly implemented. For compatibility with other
 2944:         greps, it now suppresses the error message for a non-existent or non-
 2945:         accessible file (but not the return code). There is a new option called
 2946:         -q that suppresses the output of matching lines, which was what -s was
 2947:         previously doing.
 2948: 
 2949:     (p) Added --include and --exclude options to specify files for inclusion
 2950:         and exclusion when recursing.
 2951: 
 2952: 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
 2953:     Hopefully, it now does.
 2954: 
 2955: 12. Missing cast in pcre_study().
 2956: 
 2957: 13. Added an "uninstall" target to the makefile.
 2958: 
 2959: 14. Replaced "extern" in the function prototypes in Makefile.in with
 2960:     "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
 2961:     world, but is set differently for Windows.
 2962: 
 2963: 15. Added a second compiling function called pcre_compile2(). The only
 2964:     difference is that it has an extra argument, which is a pointer to an
 2965:     integer error code. When there is a compile-time failure, this is set
 2966:     non-zero, in addition to the error test pointer being set to point to an
 2967:     error message. The new argument may be NULL if no error number is required
 2968:     (but then you may as well call pcre_compile(), which is now just a
 2969:     wrapper). This facility is provided because some applications need a
 2970:     numeric error indication, but it has also enabled me to tidy up the way
 2971:     compile-time errors are handled in the POSIX wrapper.
 2972: 
 2973: 16. Added VPATH=.libs to the makefile; this should help when building with one
 2974:     prefix path and installing with another. (Or so I'm told by someone who
 2975:     knows more about this stuff than I do.)
 2976: 
 2977: 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
 2978:     passes PCRE_DOTALL to the pcre_compile() function, making the "." character
 2979:     match everything, including newlines. This is not POSIX-compatible, but
 2980:     somebody wanted the feature. From pcretest it can be activated by using
 2981:     both the P and the s flags.
 2982: 
 2983: 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
 2984: 
 2985: 19. libpcre.pc was being incorrectly installed as executable.
 2986: 
 2987: 20. A couple of places in pcretest check for end-of-line by looking for '\n';
 2988:     it now also looks for '\r' so that it will work unmodified on Windows.
 2989: 
 2990: 21. Added Google's contributed C++ wrapper to the distribution.
 2991: 
 2992: 22. Added some untidy missing memory free() calls in pcretest, to keep
 2993:     Electric Fence happy when testing.
 2994: 
 2995: 
 2996: 
 2997: Version 5.0 13-Sep-04
 2998: ---------------------
 2999: 
 3000:  1. Internal change: literal characters are no longer packed up into items
 3001:     containing multiple characters in a single byte-string. Each character
 3002:     is now matched using a separate opcode. However, there may be more than one
 3003:     byte in the character in UTF-8 mode.
 3004: 
 3005:  2. The pcre_callout_block structure has two new fields: pattern_position and
 3006:     next_item_length. These contain the offset in the pattern to the next match
 3007:     item, and its length, respectively.
 3008: 
 3009:  3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
 3010:     insertion of callouts before each pattern item. Added the /C option to
 3011:     pcretest to make use of this.
 3012: 
 3013:  4. On the advice of a Windows user, the lines
 3014: 
 3015:       #if defined(_WIN32) || defined(WIN32)
 3016:       _setmode( _fileno( stdout ), 0x8000 );
 3017:       #endif  /* defined(_WIN32) || defined(WIN32) */
 3018: 
 3019:     have been added to the source of pcretest. This apparently does useful
 3020:     magic in relation to line terminators.
 3021: 
 3022:  5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
 3023:     for the benefit of those environments where the "b" makes a difference.
 3024: 
 3025:  6. The icc compiler has the same options as gcc, but "configure" doesn't seem
 3026:     to know about it. I have put a hack into configure.in that adds in code
 3027:     to set GCC=yes if CC=icc. This seems to end up at a point in the
 3028:     generated configure script that is early enough to affect the setting of
 3029:     compiler options, which is what is needed, but I have no means of testing
 3030:     whether it really works. (The user who reported this had patched the
 3031:     generated configure script, which of course I cannot do.)
 3032: 
 3033:     LATER: After change 22 below (new libtool files), the configure script
 3034:     seems to know about icc (and also ecc). Therefore, I have commented out
 3035:     this hack in configure.in.
 3036: 
 3037:  7. Added support for pkg-config (2 patches were sent in).
 3038: 
 3039:  8. Negated POSIX character classes that used a combination of internal tables
 3040:     were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
 3041:     [[:^ascii]]. Typically, they would match almost any characters. The other
 3042:     POSIX classes were not broken in this way.
 3043: 
 3044:  9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
 3045:     to find the match, as PCRE was deluded into thinking that the match had to
 3046:     start at the start point or following a newline. The same bug applied to
 3047:     patterns with negative forward assertions or any backward assertions
 3048:     preceding ".*" at the start, unless the pattern required a fixed first
 3049:     character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
 3050: 
 3051: 10. In UTF-8 mode, when moving forwards in the subject after a failed match
 3052:     starting at the last subject character, bytes beyond the end of the subject
 3053:     string were read.
 3054: 
 3055: 11. Renamed the variable "class" as "classbits" to make life easier for C++
 3056:     users. (Previously there was a macro definition, but it apparently wasn't
 3057:     enough.)
 3058: 
 3059: 12. Added the new field "tables" to the extra data so that tables can be passed
 3060:     in at exec time, or the internal tables can be re-selected. This allows
 3061:     a compiled regex to be saved and re-used at a later time by a different
 3062:     program that might have everything at different addresses.
 3063: 
 3064: 13. Modified the pcre-config script so that, when run on Solaris, it shows a
 3065:     -R library as well as a -L library.
 3066: 
 3067: 14. The debugging options of pcretest (-d on the command line or D on a
 3068:     pattern) showed incorrect output for anything following an extended class
 3069:     that contained multibyte characters and which was followed by a quantifier.
 3070: 
 3071: 15. Added optional support for general category Unicode character properties
 3072:     via the \p, \P, and \X escapes. Unicode property support implies UTF-8
 3073:     support. It adds about 90K to the size of the library. The meanings of the
 3074:     inbuilt class escapes such as \d and \s have NOT been changed.
 3075: 
 3076: 16. Updated pcredemo.c to include calls to free() to release the memory for the
 3077:     compiled pattern.
 3078: 
 3079: 17. The generated file chartables.c was being created in the source directory
 3080:     instead of in the building directory. This caused the build to fail if the
 3081:     source directory was different from the building directory, and was
 3082:     read-only.
 3083: 
 3084: 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
 3085:     file. No doubt somebody will tell me if they don't make sense... Also added
 3086:     Dan Mooney's comments about building on OpenVMS.
 3087: 
 3088: 19. Added support for partial matching via the PCRE_PARTIAL option for
 3089:     pcre_exec() and the \P data escape in pcretest.
 3090: 
 3091: 20. Extended pcretest with 3 new pattern features:
 3092: 
 3093:     (i)   A pattern option of the form ">rest-of-line" causes pcretest to
 3094:           write the compiled pattern to the file whose name is "rest-of-line".
 3095:           This is a straight binary dump of the data, with the saved pointer to
 3096:           the character tables forced to be NULL. The study data, if any, is
 3097:           written too. After writing, pcretest reads a new pattern.
 3098: 
 3099:     (ii)  If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
 3100:           compiled pattern from the given file. There must not be any
 3101:           occurrences of "<" in the file name (pretty unlikely); if there are,
 3102:           pcretest will instead treat the initial "<" as a pattern delimiter.
 3103:           After reading in the pattern, pcretest goes on to read data lines as
 3104:           usual.
 3105: 
 3106:     (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
 3107:           and 16-bit fields in a compiled pattern, to simulate a pattern that
 3108:           was compiled on a host of opposite endianness.
 3109: 
 3110: 21. The pcre-exec() function can now cope with patterns that were compiled on
 3111:     hosts of opposite endianness, with this restriction:
 3112: 
 3113:       As for any compiled expression that is saved and used later, the tables
 3114:       pointer field cannot be preserved; the extra_data field in the arguments
 3115:       to pcre_exec() should be used to pass in a tables address if a value
 3116:       other than the default internal tables were used at compile time.
 3117: 
 3118: 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
 3119:     now diagnosed as an error. Previously, most of the time, a negative number
 3120:     would have been treated as zero, but if in addition "ovector" was passed as
 3121:     NULL, a crash could occur.
 3122: 
 3123: 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
 3124:     new versions from the libtool 1.5 distribution (the last one is a copy of
 3125:     a file called libtool.m4). This seems to have fixed the need to patch
 3126:     "configure" to support Darwin 1.3 (which I used to do). However, I still
 3127:     had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
 3128:     workstation).
 3129: 
 3130: 24. Changed the PCRE licence to be the more standard "BSD" licence.
 3131: 
 3132: 
 3133: Version 4.5 01-Dec-03
 3134: ---------------------
 3135: 
 3136:  1. There has been some re-arrangement of the code for the match() function so
 3137:     that it can be compiled in a version that does not call itself recursively.
 3138:     Instead, it keeps those local variables that need separate instances for
 3139:     each "recursion" in a frame on the heap, and gets/frees frames whenever it
 3140:     needs to "recurse". Keeping track of where control must go is done by means
 3141:     of setjmp/longjmp. The whole thing is implemented by a set of macros that
 3142:     hide most of the details from the main code, and operates only if
 3143:     NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
 3144:     "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
 3145:     operating.
 3146: 
 3147:     To make it easier for callers to provide specially tailored get/free
 3148:     functions for this usage, two new functions, pcre_stack_malloc, and
 3149:     pcre_stack_free, are used. They are always called in strict stacking order,
 3150:     and the size of block requested is always the same.
 3151: 
 3152:     The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
 3153:     PCRE has been compiled to use the stack or the heap for recursion. The
 3154:     -C option of pcretest uses this to show which version is compiled.
 3155: 
 3156:     A new data escape \S, is added to pcretest; it causes the amounts of store
 3157:     obtained and freed by both kinds of malloc/free at match time to be added
 3158:     to the output.
 3159: 
 3160:  2. Changed the locale test to use "fr_FR" instead of "fr" because that's
 3161:     what's available on my current Linux desktop machine.
 3162: 
 3163:  3. When matching a UTF-8 string, the test for a valid string at the start has
 3164:     been extended. If start_offset is not zero, PCRE now checks that it points
 3165:     to a byte that is the start of a UTF-8 character. If not, it returns
 3166:     PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
 3167:     this is necessary because there may be backward assertions in the pattern.
 3168:     When matching the same subject several times, it may save resources to use
 3169:     PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
 3170: 
 3171:  4. The code for checking the validity of UTF-8 strings has been tightened so
 3172:     that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
 3173:     containing "overlong sequences".
 3174: 
 3175:  5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
 3176:     I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
 3177:     should have been "&", but it just so happened that all the cases this let
 3178:     through by mistake were picked up later in the function.
 3179: 
 3180:  6. I had used a variable called "isblank" - this is a C99 function, causing
 3181:     some compilers to warn. To avoid this, I renamed it (as "blankclass").
 3182: 
 3183:  7. Cosmetic: (a) only output another newline at the end of pcretest if it is
 3184:     prompting; (b) run "./pcretest /dev/null" at the start of the test script
 3185:     so the version is shown; (c) stop "make test" echoing "./RunTest".
 3186: 
 3187:  8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
 3188: 
 3189:  9. The prototype for memmove() for systems that don't have it was using
 3190:     size_t, but the inclusion of the header that defines size_t was later. I've
 3191:     moved the #includes for the C headers earlier to avoid this.
 3192: 
 3193: 10. Added some adjustments to the code to make it easier to compiler on certain
 3194:     special systems:
 3195: 
 3196:       (a) Some "const" qualifiers were missing.
 3197:       (b) Added the macro EXPORT before all exported functions; by default this
 3198:           is defined to be empty.
 3199:       (c) Changed the dftables auxiliary program (that builds chartables.c) so
 3200:           that it reads its output file name as an argument instead of writing
 3201:           to the standard output and assuming this can be redirected.
 3202: 
 3203: 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
 3204:     class containing characters with values greater than 255, PCRE compilation
 3205:     went into a loop.
 3206: 
 3207: 12. A recursive reference to a subpattern that was within another subpattern
 3208:     that had a minimum quantifier of zero caused PCRE to crash. For example,
 3209:     (x(y(?2))z)? provoked this bug with a subject that got as far as the
 3210:     recursion. If the recursively-called subpattern itself had a zero repeat,
 3211:     that was OK.
 3212: 
 3213: 13. In pcretest, the buffer for reading a data line was set at 30K, but the
 3214:     buffer into which it was copied (for escape processing) was still set at
 3215:     1024, so long lines caused crashes.
 3216: 
 3217: 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
 3218:     "internal error: code overflow...". This applied to any character class
 3219:     that was followed by a possessive quantifier.
 3220: 
 3221: 15. Modified the Makefile to add libpcre.la as a prerequisite for
 3222:     libpcreposix.la because I was told this is needed for a parallel build to
 3223:     work.
 3224: 
 3225: 16. If a pattern that contained .* following optional items at the start was
 3226:     studied, the wrong optimizing data was generated, leading to matching
 3227:     errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
 3228:     matching string must start with a or b or c. The correct conclusion for
 3229:     this pattern is that a match can start with any character.
 3230: 
 3231: 
 3232: Version 4.4 13-Aug-03
 3233: ---------------------
 3234: 
 3235:  1. In UTF-8 mode, a character class containing characters with values between
 3236:     127 and 255 was not handled correctly if the compiled pattern was studied.
 3237:     In fixing this, I have also improved the studying algorithm for such
 3238:     classes (slightly).
 3239: 
 3240:  2. Three internal functions had redundant arguments passed to them. Removal
 3241:     might give a very teeny performance improvement.
 3242: 
 3243:  3. Documentation bug: the value of the capture_top field in a callout is *one
 3244:     more than* the number of the hightest numbered captured substring.
 3245: 
 3246:  4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
 3247:     in incorrectly linking with a previously installed version. They now link
 3248:     explicitly with libpcre.la.
 3249: 
 3250:  5. configure.in no longer needs to recognize Cygwin specially.
 3251: 
 3252:  6. A problem in pcre.in for Windows platforms is fixed.
 3253: 
 3254:  7. If a pattern was successfully studied, and the -d (or /D) flag was given to
 3255:     pcretest, it used to include the size of the study block as part of its
 3256:     output. Unfortunately, the structure contains a field that has a different
 3257:     size on different hardware architectures. This meant that the tests that
 3258:     showed this size failed. As the block is currently always of a fixed size,
 3259:     this information isn't actually particularly useful in pcretest output, so
 3260:     I have just removed it.
 3261: 
 3262:  8. Three pre-processor statements accidentally did not start in column 1.
 3263:     Sadly, there are *still* compilers around that complain, even though
 3264:     standard C has not required this for well over a decade. Sigh.
 3265: 
 3266:  9. In pcretest, the code for checking callouts passed small integers in the
 3267:     callout_data field, which is a void * field. However, some picky compilers
 3268:     complained about the casts involved for this on 64-bit systems. Now
 3269:     pcretest passes the address of the small integer instead, which should get
 3270:     rid of the warnings.
 3271: 
 3272: 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
 3273:     both compile and run time, and gives an error if an invalid UTF-8 sequence
 3274:     is found. There is a option for disabling this check in cases where the
 3275:     string is known to be correct and/or the maximum performance is wanted.
 3276: 
 3277: 11. In response to a bug report, I changed one line in Makefile.in from
 3278: 
 3279:         -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
 3280:     to
 3281:         -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
 3282: 
 3283:     to look similar to other lines, but I have no way of telling whether this
 3284:     is the right thing to do, as I do not use Windows. No doubt I'll get told
 3285:     if it's wrong...
 3286: 
 3287: 
 3288: Version 4.3 21-May-03
 3289: ---------------------
 3290: 
 3291: 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
 3292:    Makefile.
 3293: 
 3294: 2. Some refactoring to improve the quality of the code:
 3295: 
 3296:    (i)   The utf8_table... variables are now declared "const".
 3297: 
 3298:    (ii)  The code for \cx, which used the "case flipping" table to upper case
 3299:          lower case letters, now just substracts 32. This is ASCII-specific,
 3300:          but the whole concept of \cx is ASCII-specific, so it seems
 3301:          reasonable.
 3302: 
 3303:    (iii) PCRE was using its character types table to recognize decimal and
 3304:          hexadecimal digits in the pattern. This is silly, because it handles
 3305:          only 0-9, a-f, and A-F, but the character types table is locale-
 3306:          specific, which means strange things might happen. A private
 3307:          table is now used for this - though it costs 256 bytes, a table is
 3308:          much faster than multiple explicit tests. Of course, the standard
 3309:          character types table is still used for matching digits in subject
 3310:          strings against \d.
 3311: 
 3312:    (iv)  Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
 3313:          ending in _t are). So I've renamed it as ESC_tee.
 3314: 
 3315: 3. The first argument for regexec() in the POSIX wrapper should have been
 3316:    defined as "const".
 3317: 
 3318: 4. Changed pcretest to use malloc() for its buffers so that they can be
 3319:    Electric Fenced for debugging.
 3320: 
 3321: 5. There were several places in the code where, in UTF-8 mode, PCRE would try
 3322:    to read one or more bytes before the start of the subject string. Often this
 3323:    had no effect on PCRE's behaviour, but in some circumstances it could
 3324:    provoke a segmentation fault.
 3325: 
 3326: 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
 3327:    to try to read one or more bytes before the start of the subject string.
 3328: 
 3329: 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
 3330:    UTF-8 support could misbehave in various ways if the subject string
 3331:    contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
 3332:    area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
 3333:    back over UTF-8 characters.)
 3334: 
 3335: 
 3336: Version 4.2 14-Apr-03
 3337: ---------------------
 3338: 
 3339: 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
 3340: 
 3341: 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
 3342:      [ON_WINDOWS]: new variable, "#" on non-Windows platforms
 3343:      [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
 3344:      [WIN_PREFIX]: new variable, "cyg" for Cygwin
 3345:      * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
 3346:        and BUILD_EXEEXT
 3347:      Note: automatic setting of the BUILD variables is not yet working
 3348:      set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
 3349:        compile-time but not at link-time
 3350:      [LINK]: use for linking executables only
 3351:      make different versions for Windows and non-Windows
 3352:      [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
 3353:        libraries
 3354:      [LINK_FOR_BUILD]: new variable
 3355:      [OBJEXT]: use throughout
 3356:      [EXEEXT]: use throughout
 3357:      <winshared>: new target
 3358:      <wininstall>: new target
 3359:      <dftables.o>: use native compiler
 3360:      <dftables>: use native linker
 3361:      <install>: handle Windows platform correctly
 3362:      <clean>: ditto
 3363:      <check>: ditto
 3364:      copy DLL to top builddir before testing
 3365: 
 3366:    As part of these changes, -no-undefined was removed again. This was reported
 3367:    to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
 3368:    in any case.
 3369: 
 3370: 3. Some tidies to get rid of compiler warnings:
 3371: 
 3372:    . In the match_data structure, match_limit was an unsigned long int, whereas
 3373:      match_call_count was an int. I've made them both unsigned long ints.
 3374: 
 3375:    . In pcretest the fact that a const uschar * doesn't automatically cast to
 3376:      a void * provoked a warning.
 3377: 
 3378:    . Turning on some more compiler warnings threw up some "shadow" variables
 3379:      and a few more missing casts.
 3380: 
 3381: 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
 3382:    option, a class that contained a single character with a value between 128
 3383:    and 255 (e.g. /[\xFF]/) caused PCRE to crash.
 3384: 
 3385: 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
 3386:    option, a class that contained several characters, but with at least one
 3387:    whose value was between 128 and 255 caused PCRE to crash.
 3388: 
 3389: 
 3390: Version 4.1 12-Mar-03
 3391: ---------------------
 3392: 
 3393: 1. Compiling with gcc -pedantic found a couple of places where casts were
 3394: needed, and a string in dftables.c that was longer than standard compilers are
 3395: required to support.
 3396: 
 3397: 2. Compiling with Sun's compiler found a few more places where the code could
 3398: be tidied up in order to avoid warnings.
 3399: 
 3400: 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
 3401: first of these names is deprecated in the latest Autoconf in favour of the name
 3402: CC_FOR_BUILD, because "host" is typically used to mean the system on which the
 3403: compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
 3404: analogy I have changed it to CFLAGS_FOR_BUILD.
 3405: 
 3406: 4. Added -no-undefined to the linking command in the Makefile, because this is
 3407: apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
 3408: linking step for the pcreposix library.
 3409: 
 3410: 5. PCRE was failing to diagnose the case of two named groups with the same
 3411: name.
 3412: 
 3413: 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
 3414: literal character that is needed in the subject for a match, and scans along to
 3415: ensure that it is present before embarking on the full matching process. This
 3416: saves time in cases of nested unlimited repeats that are never going to match.
 3417: Problem: the scan can take a lot of time if the subject is very long (e.g.
 3418: megabytes), thus penalizing straightforward matches. It is now done only if the
 3419: amount of subject to be scanned is less than 1000 bytes.
 3420: 
 3421: 7. A lesser problem with the same optimization is that it was recording the
 3422: first character of an anchored pattern as "needed", thus provoking a search
 3423: right along the subject, even when the first match of the pattern was going to
 3424: fail. The "needed" character is now not set for anchored patterns, unless it
 3425: follows something in the pattern that is of non-fixed length. Thus, it still
 3426: fulfils its original purpose of finding quick non-matches in cases of nested
 3427: unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
 3428: 
 3429: 
 3430: Version 4.0 17-Feb-03
 3431: ---------------------
 3432: 
 3433: 1. If a comment in an extended regex that started immediately after a meta-item
 3434: extended to the end of string, PCRE compiled incorrect data. This could lead to
 3435: all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
 3436: 
 3437: 2. Moved to autoconf 2.53 and libtool 1.4.2.
 3438: 
 3439: 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
 3440: the special perltest8 script is no longer needed - all the tests can be run
 3441: from a single perltest script.
 3442: 
 3443: 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
 3444: by \s. It has now been removed in PCRE. This means it isn't recognized as
 3445: whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
 3446: class [:space:] *does* include VT, thereby creating a mess.
 3447: 
 3448: 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
 3449: space and tab.
 3450: 
 3451: 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
 3452: its new features into the main test script, reducing the number of scripts.
 3453: 
 3454: 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
 3455: were backward compatible, and made the (?i) apply to the whole pattern, as if
 3456: /i were given. Now it behaves more logically, and applies the option setting
 3457: only to what follows. PCRE has been changed to follow suit. However, if it
 3458: finds options settings right at the start of the pattern, it extracts them into
 3459: the global options, as before. Thus, they show up in the info data.
 3460: 
 3461: 8. Added support for the \Q...\E escape sequence. Characters in between are
 3462: treated as literals. This is slightly different from Perl in that $ and @ are
 3463: also handled as literals inside the quotes. In Perl, they will cause variable
 3464: interpolation. Note the following examples:
 3465: 
 3466:     Pattern            PCRE matches      Perl matches
 3467: 
 3468:     \Qabc$xyz\E        abc$xyz           abc followed by the contents of $xyz
 3469:     \Qabc\$xyz\E       abc\$xyz          abc\$xyz
 3470:     \Qabc\E\$\Qxyz\E   abc$xyz           abc$xyz
 3471: 
 3472: For compatibility with Perl, \Q...\E sequences are recognized inside character
 3473: classes as well as outside them.
 3474: 
 3475: 9. Re-organized 3 code statements in pcretest to avoid "overflow in
 3476: floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
 3477: (size_t) cast to one statement in pcretest and one in pcreposix to avoid
 3478: signed/unsigned warnings.
 3479: 
 3480: 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
 3481: option for pcretest, so I've replaced it by a simple function that does just
 3482: that job.
 3483: 
 3484: 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
 3485: "pcregrep -".
 3486: 
 3487: 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
 3488: Java package. This provides some syntactic sugar for simple cases of what my
 3489: documentation calls "once-only subpatterns". A pattern such as x*+ is the same
 3490: as (?>x*). In other words, if what is inside (?>...) is just a single repeated
 3491: item, you can use this simplified notation. Note that only makes sense with
 3492: greedy quantifiers. Consequently, the use of the possessive quantifier forces
 3493: greediness, whatever the setting of the PCRE_UNGREEDY option.
 3494: 
 3495: 13. A change of greediness default within a pattern was not taking effect at
 3496: the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
 3497: subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
 3498: was abstracted outside.
 3499: 
 3500: 14. PCRE now supports the \G assertion. It is true when the current matching
 3501: position is at the start point of the match. This differs from \A when the
 3502: starting offset is non-zero. Used with the /g option of pcretest (or similar
 3503: code), it works in the same way as it does for Perl's /g option. If all
 3504: alternatives of a regex begin with \G, the expression is anchored to the start
 3505: match position, and the "anchored" flag is set in the compiled expression.
 3506: 
 3507: 15. Some bugs concerning the handling of certain option changes within patterns
 3508: have been fixed. These applied to options other than (?ims). For example,
 3509: "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
 3510: been the other way round. Some of this was related to change 7 above.
 3511: 
 3512: 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
 3513: features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
 3514: and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
 3515: POSIX classes only within a class (e.g. /[[:alpha:]]/).
 3516: 
 3517: 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
 3518: mode. Unlike ".", it always matches newline, whatever the setting of
 3519: PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
 3520: assertions. Perl allows it, but it doesn't (in general) work because it can't
 3521: calculate the length of the lookbehind. At least, that's the case for Perl
 3522: 5.8.0 - I've been told they are going to document that it doesn't work in
 3523: future.
 3524: 
 3525: 18. Added an error diagnosis for escapes that PCRE does not support: these are
 3526: \L, \l, \N, \P, \p, \U, \u, and \X.
 3527: 
 3528: 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
 3529: reading past the end of the pattern in cases such as /[abcd/.
 3530: 
 3531: 20. PCRE was getting more memory than necessary for patterns with classes that
 3532: contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
 3533: 
 3534: 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
 3535: compiling PCRE for use with Virtual Pascal.
 3536: 
 3537: 22. Small fix to the Makefile to make it work properly if the build is done
 3538: outside the source tree.
 3539: 
 3540: 23. Added a new extension: a condition to go with recursion. If a conditional
 3541: subpattern starts with (?(R) the "true" branch is used if recursion has
 3542: happened, whereas the "false" branch is used only at the top level.
 3543: 
 3544: 24. When there was a very long string of literal characters (over 255 bytes
 3545: without UTF support, over 250 bytes with UTF support), the computation of how
 3546: much memory was required could be incorrect, leading to segfaults or other
 3547: strange effects.
 3548: 
 3549: 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
 3550: start of line for a non-DOTALL pattern) when a pattern started with (.*) and
 3551: there was a subsequent back reference to those brackets. This meant that, for
 3552: example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
 3553: possible to check for precisely this case. All we can do is abandon the
 3554: optimization if .* occurs inside capturing brackets when there are any back
 3555: references whatsoever. (See below for a better fix that came later.)
 3556: 
 3557: 26. The handling of the optimization for finding the first character of a
 3558: non-anchored pattern, and for finding a character that is required later in the
 3559: match were failing in some cases. This didn't break the matching; it just
 3560: failed to optimize when it could. The way this is done has been re-implemented.
 3561: 
 3562: 27. Fixed typo in error message for invalid (?R item (it said "(?p").
 3563: 
 3564: 28. Added a new feature that provides some of the functionality that Perl
 3565: provides with (?{...}). The facility is termed a "callout". The way it is done
 3566: in PCRE is for the caller to provide an optional function, by setting
 3567: pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
 3568: global variable. By default it is unset, which disables all calling out. To get
 3569: the function called, the regex must include (?C) at appropriate points. This
 3570: is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
 3571: This provides a means of identifying different callout points. When PCRE
 3572: reaches such a point in the regex, if pcre_callout has been set, the external
 3573: function is called. It is provided with data in a structure called
 3574: pcre_callout_block, which is defined in pcre.h. If the function returns 0,
 3575: matching continues; if it returns a non-zero value, the match at the current
 3576: point fails. However, backtracking will occur if possible. [This was changed
 3577: later and other features added - see item 49 below.]
 3578: 
 3579: 29. pcretest is upgraded to test the callout functionality. It provides a
 3580: callout function that displays information. By default, it shows the start of
 3581: the match and the current position in the text. There are some new data escapes
 3582: to vary what happens:
 3583: 
 3584:     \C+         in addition, show current contents of captured substrings
 3585:     \C-         do not supply a callout function
 3586:     \C!n        return 1 when callout number n is reached
 3587:     \C!n!m      return 1 when callout number n is reached for the mth time
 3588: 
 3589: 30. If pcregrep was called with the -l option and just a single file name, it
 3590: output "<stdin>" if a match was found, instead of the file name.
 3591: 
 3592: 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
 3593: slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
 3594: pcre_exec(). This saves a malloc/free per call. The default value of
 3595: POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
 3596: when configuring.
 3597: 
 3598: 32. The default maximum size of a compiled pattern is 64K. There have been a
 3599: few cases of people hitting this limit. The code now uses macros to handle the
 3600: storing of links as offsets within the compiled pattern. It defaults to 2-byte
 3601: links, but this can be changed to 3 or 4 bytes by --with-link-size when
 3602: configuring. Tests 2 and 5 work only with 2-byte links because they output
 3603: debugging information about compiled patterns.
 3604: 
 3605: 33. Internal code re-arrangements:
 3606: 
 3607: (a) Moved the debugging function for printing out a compiled regex into
 3608:     its own source file (printint.c) and used #include to pull it into
 3609:     pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
 3610:     separate copies.
 3611: 
 3612: (b) Defined the list of op-code names for debugging as a macro in
 3613:     internal.h so that it is next to the definition of the opcodes.
 3614: 
 3615: (c) Defined a table of op-code lengths for simpler skipping along compiled
 3616:     code. This is again a macro in internal.h so that it is next to the
 3617:     definition of the opcodes.
 3618: 
 3619: 34. Added support for recursive calls to individual subpatterns, along the
 3620: lines of Robin Houston's patch (but implemented somewhat differently).
 3621: 
 3622: 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
 3623: allow it to read and process whole directories in Win32. This code was
 3624: contributed by Lionel Fourquaux; it has not been tested by me.
 3625: 
 3626: 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
 3627: used to name a group. Names consist of alphanumerics and underscores, and must
 3628: be unique. Back references use the syntax (?P=name) and recursive calls use
 3629: (?P>name) which is a PCRE extension to the Python extension. Groups still have
 3630: numbers. The function pcre_fullinfo() can be used after compilation to extract
 3631: a name/number map. There are three relevant calls:
 3632: 
 3633:   PCRE_INFO_NAMEENTRYSIZE        yields the size of each entry in the map
 3634:   PCRE_INFO_NAMECOUNT            yields the number of entries
 3635:   PCRE_INFO_NAMETABLE            yields a pointer to the map.
 3636: 
 3637: The map is a vector of fixed-size entries. The size of each entry depends on
 3638: the length of the longest name used. The first two bytes of each entry are the
 3639: group number, most significant byte first. There follows the corresponding
 3640: name, zero terminated. The names are in alphabetical order.
 3641: 
 3642: 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
 3643: case instead of 255. Making it the same both with and without UTF-8 support
 3644: means that the same test output works with both.
 3645: 
 3646: 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
 3647: calling malloc() with a zero argument.
 3648: 
 3649: 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
 3650: optimization. I've improved things by keeping a bitmap of backreferences with
 3651: numbers 1-31 so that if .* occurs inside capturing brackets that are not in
 3652: fact referenced, the optimization can be applied. It is unlikely that a
 3653: relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
 3654: the match to follow \n) will appear inside brackets with a number greater than
 3655: 31, but if it does, any back reference > 31 suppresses the optimization.
 3656: 
 3657: 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
 3658: of disabling numbered capturing parentheses. Any opening parenthesis that is
 3659: not followed by ? behaves as if it were followed by ?: but named parentheses
 3660: can still be used for capturing (and they will acquire numbers in the usual
 3661: way).
 3662: 
 3663: 41. Redesigned the return codes from the match() function into yes/no/error so
 3664: that errors can be passed back from deep inside the nested calls. A malloc
 3665: failure while inside a recursive subpattern call now causes the
 3666: PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
 3667: 
 3668: 42. It is now possible to set a limit on the number of times the match()
 3669: function is called in a call to pcre_exec(). This facility makes it possible to
 3670: limit the amount of recursion and backtracking, though not in a directly
 3671: obvious way, because the match() function is used in a number of different
 3672: circumstances. The count starts from zero for each position in the subject
 3673: string (for non-anchored patterns). The default limit is, for compatibility, a
 3674: large number, namely 10 000 000. You can change this in two ways:
 3675: 
 3676: (a) When configuring PCRE before making, you can use --with-match-limit=n
 3677:     to set a default value for the compiled library.
 3678: 
 3679: (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
 3680:     a different value is set. See 45 below.
 3681: 
 3682: If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
 3683: 
 3684: 43. Added a new function pcre_config(int, void *) to enable run-time extraction
 3685: of things that can be changed at compile time. The first argument specifies
 3686: what is wanted and the second points to where the information is to be placed.
 3687: The current list of available information is:
 3688: 
 3689:   PCRE_CONFIG_UTF8
 3690: 
 3691: The output is an integer that is set to one if UTF-8 support is available;
 3692: otherwise it is set to zero.
 3693: 
 3694:   PCRE_CONFIG_NEWLINE
 3695: 
 3696: The output is an integer that it set to the value of the code that is used for
 3697: newline. It is either LF (10) or CR (13).
 3698: 
 3699:   PCRE_CONFIG_LINK_SIZE
 3700: 
 3701: The output is an integer that contains the number of bytes used for internal
 3702: linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
 3703: 
 3704:   PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
 3705: 
 3706: The output is an integer that contains the threshold above which the POSIX
 3707: interface uses malloc() for output vectors. See item 31 above.
 3708: 
 3709:   PCRE_CONFIG_MATCH_LIMIT
 3710: 
 3711: The output is an unsigned integer that contains the default limit of the number
 3712: of match() calls in a pcre_exec() execution. See 42 above.
 3713: 
 3714: 44. pcretest has been upgraded by the addition of the -C option. This causes it
 3715: to extract all the available output from the new pcre_config() function, and to
 3716: output it. The program then exits immediately.
 3717: 
 3718: 45. A need has arisen to pass over additional data with calls to pcre_exec() in
 3719: order to support additional features. One way would have been to define
 3720: pcre_exec2() (for example) with extra arguments, but this would not have been
 3721: extensible, and would also have required all calls to the original function to
 3722: be mapped to the new one. Instead, I have chosen to extend the mechanism that
 3723: is used for passing in "extra" data from pcre_study().
 3724: 
 3725: The pcre_extra structure is now exposed and defined in pcre.h. It currently
 3726: contains the following fields:
 3727: 
 3728:   flags         a bitmap indicating which of the following fields are set
 3729:   study_data    opaque data from pcre_study()
 3730:   match_limit   a way of specifying a limit on match() calls for a specific
 3731:                   call to pcre_exec()
 3732:   callout_data  data for callouts (see 49 below)
 3733: 
 3734: The flag bits are also defined in pcre.h, and are
 3735: 
 3736:   PCRE_EXTRA_STUDY_DATA
 3737:   PCRE_EXTRA_MATCH_LIMIT
 3738:   PCRE_EXTRA_CALLOUT_DATA
 3739: 
 3740: The pcre_study() function now returns one of these new pcre_extra blocks, with
 3741: the actual study data pointed to by the study_data field, and the
 3742: PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
 3743: before. That is, this change is entirely upwards-compatible and requires no
 3744: change to existing code.
 3745: 
 3746: If you want to pass in additional data to pcre_exec(), you can either place it
 3747: in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
 3748: block.
 3749: 
 3750: 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
 3751: data string contains the escape sequence \M, pcretest calls pcre_exec() several
 3752: times with different match limits, until it finds the minimum value needed for
 3753: pcre_exec() to complete. The value is then output. This can be instructive; for
 3754: most simple matches the number is quite small, but for pathological cases it
 3755: gets very large very quickly.
 3756: 
 3757: 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
 3758: returns the size of the data block pointed to by the study_data field in a
 3759: pcre_extra block, that is, the value that was passed as the argument to
 3760: pcre_malloc() when PCRE was getting memory in which to place the information
 3761: created by pcre_study(). The fourth argument should point to a size_t variable.
 3762: pcretest has been extended so that this information is shown after a successful
 3763: pcre_study() call when information about the compiled regex is being displayed.
 3764: 
 3765: 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
 3766: because what follows is always an absolute path. (Later: it turns out that this
 3767: is more than cosmetic for MinGW, because it doesn't like empty path
 3768: components.)
 3769: 
 3770: 49. Some changes have been made to the callout feature (see 28 above):
 3771: 
 3772: (i)  A callout function now has three choices for what it returns:
 3773: 
 3774:        0  =>  success, carry on matching
 3775:      > 0  =>  failure at this point, but backtrack if possible
 3776:      < 0  =>  serious error, return this value from pcre_exec()
 3777: 
 3778:      Negative values should normally be chosen from the set of PCRE_ERROR_xxx
 3779:      values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
 3780:      "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
 3781:      use by callout functions. It will never be used by PCRE itself.
 3782: 
 3783: (ii) The pcre_extra structure (see 45 above) has a void * field called
 3784:      callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
 3785:      pcre_callout_block structure has a field of the same name. The contents of
 3786:      the field passed in the pcre_extra structure are passed to the callout
 3787:      function in the corresponding field in the callout block. This makes it
 3788:      easier to use the same callout-containing regex from multiple threads. For
 3789:      testing, the pcretest program has a new data escape
 3790: 
 3791:        \C*n        pass the number n (may be negative) as callout_data
 3792: 
 3793:      If the callout function in pcretest receives a non-zero value as
 3794:      callout_data, it returns that value.
 3795: 
 3796: 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
 3797: there were some redundant $(CFLAGS) in commands that are now specified as
 3798: $(LINK), which already includes $(CFLAGS).
 3799: 
 3800: 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
 3801: has been compiled with UTF-8 support *and* pcre_compile() has been compiled
 3802: with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
 3803: one-byte characters throughout. Note that case-insensitive matching applies
 3804: only to characters whose values are less than 256. PCRE doesn't support the
 3805: notion of cases for higher-valued characters.
 3806: 
 3807: (i)   A character class whose characters are all within 0-255 is handled as
 3808:       a bit map, and the map is inverted for negative classes. Previously, a
 3809:       character > 255 always failed to match such a class; however it should
 3810:       match if the class was a negative one (e.g. [^ab]). This has been fixed.
 3811: 
 3812: (ii)  A negated character class with a single character < 255 is coded as
 3813:       "not this character" (OP_NOT). This wasn't working properly when the test
 3814:       character was multibyte, either singly or repeated.
 3815: 
 3816: (iii) Repeats of multibyte characters are now handled correctly in UTF-8
 3817:       mode, for example: \x{100}{2,3}.
 3818: 
 3819: (iv)  The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
 3820:       singly or repeated) now correctly test multibyte characters. However,
 3821:       PCRE doesn't recognize any characters with values greater than 255 as
 3822:       digits, spaces, or word characters. Such characters always match \D, \S,
 3823:       and \W, and never match \d, \s, or \w.
 3824: 
 3825: (v)   Classes may now contain characters and character ranges with values
 3826:       greater than 255. For example: [ab\x{100}-\x{400}].
 3827: 
 3828: (vi)  pcregrep now has a --utf-8 option (synonym -u) which makes it call
 3829:       PCRE in UTF-8 mode.
 3830: 
 3831: 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
 3832: PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
 3833: retained for backwards compatibility. (Note that LASTLITERAL is also a byte
 3834: value.)
 3835: 
 3836: 53. The single man page has become too large. I have therefore split it up into
 3837: a number of separate man pages. These also give rise to individual HTML pages;
 3838: these are now put in a separate directory, and there is an index.html page that
 3839: lists them all. Some hyperlinking between the pages has been installed.
 3840: 
 3841: 54. Added convenience functions for handling named capturing parentheses.
 3842: 
 3843: 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
 3844: aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
 3845: true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
 3846: are faulted.
 3847: 
 3848: 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
 3849: calling configure. These values are used when compiling the dftables.c program
 3850: which is run to generate the source of the default character tables. They
 3851: default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
 3852: you will need to set these values.
 3853: 
 3854: 57. Updated the building process for Windows DLL, as provided by Fred Cox.
 3855: 
 3856: 
 3857: Version 3.9 02-Jan-02
 3858: ---------------------
 3859: 
 3860: 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
 3861: 
 3862: 2. If --disable-static was given, the building process failed when trying to
 3863: build pcretest and pcregrep. (For some reason it was using libtool to compile
 3864: them, which is not right, as they aren't part of the library.)
 3865: 
 3866: 
 3867: Version 3.8 18-Dec-01
 3868: ---------------------
 3869: 
 3870: 1. The experimental UTF-8 code was completely screwed up. It was packing the
 3871: bytes in the wrong order. How dumb can you get?
 3872: 
 3873: 
 3874: Version 3.7 29-Oct-01
 3875: ---------------------
 3876: 
 3877: 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
 3878: This caused pcretest, when used on the test data, to segfault. Unfortunately,
 3879: this didn't happen under Solaris 8, where I normally test things.
 3880: 
 3881: 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
 3882: doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
 3883: isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
 3884: this fix an hour or so after the initial 3.7 release.)
 3885: 
 3886: 
 3887: Version 3.6 23-Oct-01
 3888: ---------------------
 3889: 
 3890: 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
 3891: offsets passed as NULL with zero offset count.
 3892: 
 3893: 2. The config.guess and config.sub files had not been updated when I moved to
 3894: the latest autoconf.
 3895: 
 3896: 
 3897: Version 3.5 15-Aug-01
 3898: ---------------------
 3899: 
 3900: 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
 3901: had been forgotten.
 3902: 
 3903: 2. By using declared but undefined structures, we can avoid using "void"
 3904: definitions in pcre.h while keeping the internal definitions of the structures
 3905: private.
 3906: 
 3907: 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
 3908: user point of view, this means that both static and shared libraries are built
 3909: by default, but this can be individually controlled. More of the work of
 3910: handling this static/shared cases is now inside libtool instead of PCRE's make
 3911: file.
 3912: 
 3913: 4. The pcretest utility is now installed along with pcregrep because it is
 3914: useful for users (to test regexs) and by doing this, it automatically gets
 3915: relinked by libtool. The documentation has been turned into a man page, so
 3916: there are now .1, .txt, and .html versions in /doc.
 3917: 
 3918: 5. Upgrades to pcregrep:
 3919:    (i)   Added long-form option names like gnu grep.
 3920:    (ii)  Added --help to list all options with an explanatory phrase.
 3921:    (iii) Added -r, --recursive to recurse into sub-directories.
 3922:    (iv)  Added -f, --file to read patterns from a file.
 3923: 
 3924: 6. pcre_exec() was referring to its "code" argument before testing that
 3925: argument for NULL (and giving an error if it was NULL).
 3926: 
 3927: 7. Upgraded Makefile.in to allow for compiling in a different directory from
 3928: the source directory.
 3929: 
 3930: 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
 3931: options bits, the pointer it was passed was to an int instead of to an unsigned
 3932: long int. This mattered only on 64-bit systems.
 3933: 
 3934: 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
 3935: generated) instead of pcre.in, which it its source. Also made the same change
 3936: in several of the .c files.
 3937: 
 3938: 10. A new release of gcc defines printf() as a macro, which broke pcretest
 3939: because it had an ifdef in the middle of a string argument for printf(). Fixed
 3940: by using separate calls to printf().
 3941: 
 3942: 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
 3943: script, to force use of CR or LF instead of \n in the source. On non-Unix
 3944: systems, the value can be set in config.h.
 3945: 
 3946: 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
 3947: absolute limit. Changed the text of the error message to make this clear, and
 3948: likewise updated the man page.
 3949: 
 3950: 13. The limit of 99 on the number of capturing subpatterns has been removed.
 3951: The new limit is 65535, which I hope will not be a "real" limit.
 3952: 
 3953: 
 3954: Version 3.4 22-Aug-00
 3955: ---------------------
 3956: 
 3957: 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
 3958: 
 3959: 2. Diagnose condition (?(0) as an error instead of crashing on matching.
 3960: 
 3961: 
 3962: Version 3.3 01-Aug-00
 3963: ---------------------
 3964: 
 3965: 1. If an octal character was given, but the value was greater than \377, it
 3966: was not getting masked to the least significant bits, as documented. This could
 3967: lead to crashes in some systems.
 3968: 
 3969: 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
 3970: the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
 3971: 
 3972: 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
 3973: These just pass their arguments on to (pcre_free)(), but they are provided
 3974: because some uses of PCRE bind it to non-C systems that can call its functions,
 3975: but cannot call free() or pcre_free() directly.
 3976: 
 3977: 4. Add "make test" as a synonym for "make check". Corrected some comments in
 3978: the Makefile.
 3979: 
 3980: 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
 3981: Makefile.
 3982: 
 3983: 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
 3984: command called pgrep for grepping around the active processes.
 3985: 
 3986: 7. Added the beginnings of support for UTF-8 character strings.
 3987: 
 3988: 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
 3989: RANLIB to ./ltconfig so that they are used by libtool. I think these are all
 3990: the relevant ones. (AR is not passed because ./ltconfig does its own figuring
 3991: out for the ar command.)
 3992: 
 3993: 
 3994: Version 3.2 12-May-00
 3995: ---------------------
 3996: 
 3997: This is purely a bug fixing release.
 3998: 
 3999: 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
 4000: of ZA. This was just one example of several cases that could provoke this bug,
 4001: which was introduced by change 9 of version 2.00. The code for breaking
 4002: infinite loops after an iteration that matches an empty string was't working
 4003: correctly.
 4004: 
 4005: 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
 4006: when matched against abbab (for example). After matching an empty string, it
 4007: wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
 4008: caused it to match further down the string than it should.
 4009: 
 4010: 3. The code contained an inclusion of sys/types.h. It isn't clear why this
 4011: was there because it doesn't seem to be needed, and it causes trouble on some
 4012: systems, as it is not a Standard C header. It has been removed.
 4013: 
 4014: 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
 4015: were reported on the Macintosh. The changes were from
 4016: 
 4017:   while ((c = *(++ptr)) != 0 && c != '\n');
 4018: to
 4019:   while ((c = *(++ptr)) != 0 && c != '\n') ;
 4020: 
 4021: Totally extraordinary, but if that's what it takes...
 4022: 
 4023: 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
 4024: available. Added HAVE_BCOPY and an autoconf test for it; if neither
 4025: HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
 4026: assumes the way PCRE uses memmove() (always moving upwards).
 4027: 
 4028: 6. PCRE is being used in one environment where strchr() is not available. There
 4029: was only one use in pcre.c, and writing it out to avoid strchr() probably gives
 4030: faster code anyway.
 4031: 
 4032: 
 4033: Version 3.1 09-Feb-00
 4034: ---------------------
 4035: 
 4036: The only change in this release is the fixing of some bugs in Makefile.in for
 4037: the "install" target:
 4038: 
 4039: (1) It was failing to install pcreposix.h.
 4040: 
 4041: (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
 4042: 
 4043: 
 4044: Version 3.0 01-Feb-00
 4045: ---------------------
 4046: 
 4047: 1. Add support for the /+ modifier to perltest (to output $` like it does in
 4048: pcretest).
 4049: 
 4050: 2. Add support for the /g modifier to perltest.
 4051: 
 4052: 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
 4053: matches null strings.
 4054: 
 4055: 4. Fix perltest so that it doesn't do unwanted things when fed an empty
 4056: pattern. Perl treats empty patterns specially - it reuses the most recent
 4057: pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
 4058: effect.
 4059: 
 4060: 5. The POSIX interface was broken in that it was just handing over the POSIX
 4061: captured string vector to pcre_exec(), but (since release 2.00) PCRE has
 4062: required a bigger vector, with some working space on the end. This means that
 4063: the POSIX wrapper now has to get and free some memory, and copy the results.
 4064: 
 4065: 6. Added some simple autoconf support, placing the test data and the
 4066: documentation in separate directories, re-organizing some of the
 4067: information files, and making it build pcre-config (a GNU standard). Also added
 4068: libtool support for building PCRE as a shared library, which is now the
 4069: default.
 4070: 
 4071: 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
 4072: 09 are not valid octal constants. Single digits will be used for minor values
 4073: less than 10.
 4074: 
 4075: 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
 4076: existing programs that set these in the POSIX interface can use PCRE without
 4077: modification.
 4078: 
 4079: 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
 4080: return all that pcre_info() returns, plus additional data. The pcre_info()
 4081: function is retained for compatibility, but is considered to be obsolete.
 4082: 
 4083: 10. Added experimental recursion feature (?R) to handle one common case that
 4084: Perl 5.6 will be able to do with (?p{...}).
 4085: 
 4086: 11. Added support for POSIX character classes like [:alpha:], which Perl is
 4087: adopting.
 4088: 
 4089: 
 4090: Version 2.08 31-Aug-99
 4091: ----------------------
 4092: 
 4093: 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
 4094: trying to match at the startoffset position, but instead was moving forward to
 4095: the next newline as if a previous match had failed.
 4096: 
 4097: 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
 4098: and could get into a loop if a null string was matched other than at the start
 4099: of the subject.
 4100: 
 4101: 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
 4102: be distinguished at compile time, and for completeness also added PCRE_DATE.
 4103: 
 4104: 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
 4105: in GnuWin32 environments.
 4106: 
 4107: 
 4108: Version 2.07 29-Jul-99
 4109: ----------------------
 4110: 
 4111: 1. The documentation is now supplied in plain text form and HTML as well as in
 4112: the form of man page sources.
 4113: 
 4114: 2. C++ compilers don't like assigning (void *) values to other pointer types.
 4115: In particular this affects malloc(). Although there is no problem in Standard
 4116: C, I've put in casts to keep C++ compilers happy.
 4117: 
 4118: 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
 4119: should be (const char *).
 4120: 
 4121: 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
 4122: be useful for non-Unix systems who don't want to bother with the POSIX stuff.
 4123: However, I haven't made this a standard facility. The documentation doesn't
 4124: mention it, and the Makefile doesn't support it.
 4125: 
 4126: 5. The Makefile now contains an "install" target, with editable destinations at
 4127: the top of the file. The pcretest program is not installed.
 4128: 
 4129: 6. pgrep -V now gives the PCRE version number and date.
 4130: 
 4131: 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
 4132: causing the entire string to be ignored, instead of just the last character.
 4133: 
 4134: 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
 4135: non-matching string, it can take a very, very long time, even for strings of
 4136: quite modest length, because of the nested recursion. PCRE now does better in
 4137: some of these cases. It does this by remembering the last required literal
 4138: character in the pattern, and pre-searching the subject to ensure it is present
 4139: before running the real match. In other words, it applies a heuristic to detect
 4140: some types of certain failure quickly, and in the above example, if presented
 4141: with a string that has no trailing " it gives "no match" very quickly.
 4142: 
 4143: 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
 4144: other alternatives are tried instead.
 4145: 
 4146: 
 4147: Version 2.06 09-Jun-99
 4148: ----------------------
 4149: 
 4150: 1. Change pcretest's output for amount of store used to show just the code
 4151: space, because the remainder (the data block) varies in size between 32-bit and
 4152: 64-bit systems.
 4153: 
 4154: 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
 4155: start matching at. This allows lookbehinds to work when searching for multiple
 4156: occurrences in a string.
 4157: 
 4158: 3. Added additional options to pcretest for testing multiple occurrences:
 4159: 
 4160:    /+   outputs the rest of the string that follows a match
 4161:    /g   loops for multiple occurrences, using the new startoffset argument
 4162:    /G   loops for multiple occurrences by passing an incremented pointer
 4163: 
 4164: 4. PCRE wasn't doing the "first character" optimization for patterns starting
 4165: with \b or \B, though it was doing it for other lookbehind assertions. That is,
 4166: it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
 4167: the letter 'x'. On long subject strings, this gives a significant speed-up.
 4168: 
 4169: 
 4170: Version 2.05 21-Apr-99
 4171: ----------------------
 4172: 
 4173: 1. Changed the type of magic_number from int to long int so that it works
 4174: properly on 16-bit systems.
 4175: 
 4176: 2. Fixed a bug which caused patterns starting with .* not to work correctly
 4177: when the subject string contained newline characters. PCRE was assuming
 4178: anchoring for such patterns in all cases, which is not correct because .* will
 4179: not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
 4180: DOTALL is set at top level; otherwise it knows that patterns starting with .*
 4181: must be retried after every newline in the subject.
 4182: 
 4183: 
 4184: Version 2.04 18-Feb-99
 4185: ----------------------
 4186: 
 4187: 1. For parenthesized subpatterns with repeats whose minimum was zero, the
 4188: computation of the store needed to hold the pattern was incorrect (too large).
 4189: If such patterns were nested a few deep, this could multiply and become a real
 4190: problem.
 4191: 
 4192: 2. Added /M option to pcretest to show the memory requirement of a specific
 4193: pattern. Made -m a synonym of -s (which does this globally) for compatibility.
 4194: 
 4195: 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
 4196: compiled in such a way that the backtracking after subsequent failure was
 4197: pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
 4198: ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
 4199: 
 4200: 
 4201: Version 2.03 02-Feb-99
 4202: ----------------------
 4203: 
 4204: 1. Fixed typo and small mistake in man page.
 4205: 
 4206: 2. Added 4th condition (GPL supersedes if conflict) and created separate
 4207: LICENCE file containing the conditions.
 4208: 
 4209: 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
 4210: Perl, that is the internal \ allows the delimiter to be included in the
 4211: pattern. Locked out the use of \ as a delimiter. If \ immediately follows
 4212: the final delimiter, add \ to the end of the pattern (to test the error).
 4213: 
 4214: 4. Added the convenience functions for extracting substrings after a successful
 4215: match. Updated pcretest to make it able to test these functions.
 4216: 
 4217: 
 4218: Version 2.02 14-Jan-99
 4219: ----------------------
 4220: 
 4221: 1. Initialized the working variables associated with each extraction so that
 4222: their saving and restoring doesn't refer to uninitialized store.
 4223: 
 4224: 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
 4225: compiler for OS/2 into generating correct code. Apparently IBM isn't going to
 4226: fix the problem.
 4227: 
 4228: 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
 4229: calls, and wasn't printing the correct value for compiling calls. Increased the
 4230: default value of LOOPREPEAT, and the number of significant figures in the
 4231: times.
 4232: 
 4233: 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
 4234: 
 4235: 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
 4236: a building problem on Windows NT with a FAT file system.
 4237: 
 4238: 
 4239: Version 2.01 21-Oct-98
 4240: ----------------------
 4241: 
 4242: 1. Changed the API for pcre_compile() to allow for the provision of a pointer
 4243: to character tables built by pcre_maketables() in the current locale. If NULL
 4244: is passed, the default tables are used.
 4245: 
 4246: 
 4247: Version 2.00 24-Sep-98
 4248: ----------------------
 4249: 
 4250: 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
 4251: it any more.
 4252: 
 4253: 2. Allow quantification of (?>) groups, and make it work correctly.
 4254: 
 4255: 3. The first character computation wasn't working for (?>) groups.
 4256: 
 4257: 4. Correct the implementation of \Z (it is permitted to match on the \n at the
 4258: end of the subject) and add 5.005's \z, which really does match only at the
 4259: very end of the subject.
 4260: 
 4261: 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
 4262: 
 4263: 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
 4264: DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
 4265: localized options. All options to pcre_study() were also removed.
 4266: 
 4267: 7. Add other new features from 5.005:
 4268: 
 4269:    $(?<=           positive lookbehind
 4270:    $(?<!           negative lookbehind
 4271:    (?imsx-imsx)    added the unsetting capability
 4272:                    such a setting is global if at outer level; local otherwise
 4273:    (?imsx-imsx:)   non-capturing groups with option setting
 4274:    (?(cond)re|re)  conditional pattern matching
 4275: 
 4276:    A backreference to itself in a repeated group matches the previous
 4277:    captured string.
 4278: 
 4279: 8. General tidying up of studying (both automatic and via "study")
 4280: consequential on the addition of new assertions.
 4281: 
 4282: 9. As in 5.005, unlimited repeated groups that could match an empty substring
 4283: are no longer faulted at compile time. Instead, the loop is forcibly broken at
 4284: runtime if any iteration does actually match an empty substring.
 4285: 
 4286: 10. Include the RunTest script in the distribution.
 4287: 
 4288: 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
 4289: discrepancies, some of which were old and were also with respect to 5.004. They
 4290: have now been fixed.
 4291: 
 4292: 
 4293: Version 1.09 28-Apr-98
 4294: ----------------------
 4295: 
 4296: 1. A negated single character class followed by a quantifier with a minimum
 4297: value of one (e.g.  [^x]{1,6}  ) was not compiled correctly. This could lead to
 4298: program crashes, or just wrong answers. This did not apply to negated classes
 4299: containing more than one character, or to minima other than one.
 4300: 
 4301: 
 4302: Version 1.08 27-Mar-98
 4303: ----------------------
 4304: 
 4305: 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
 4306: 
 4307: 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
 4308: latter must appear before anything that relies on it in the pattern.
 4309: 
 4310: 
 4311: Version 1.07 16-Feb-98
 4312: ----------------------
 4313: 
 4314: 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
 4315: repeat of a potentially empty string).
 4316: 
 4317: 
 4318: Version 1.06 23-Jan-98
 4319: ----------------------
 4320: 
 4321: 1. Added Markus Oberhumer's little patches for C++.
 4322: 
 4323: 2. Literal strings longer than 255 characters were broken.
 4324: 
 4325: 
 4326: Version 1.05 23-Dec-97
 4327: ----------------------
 4328: 
 4329: 1. Negated character classes containing more than one character were failing if
 4330: PCRE_CASELESS was set at run time.
 4331: 
 4332: 
 4333: Version 1.04 19-Dec-97
 4334: ----------------------
 4335: 
 4336: 1. Corrected the man page, where some "const" qualifiers had been omitted.
 4337: 
 4338: 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
 4339: input syntax.
 4340: 
 4341: 3. Fixed memory leak which occurred when a regex with back references was
 4342: matched with an offsets vector that wasn't big enough. The temporary memory
 4343: that is used in this case wasn't being freed if the match failed.
 4344: 
 4345: 4. Tidied pcretest to ensure it frees memory that it gets.
 4346: 
 4347: 5. Temporary memory was being obtained in the case where the passed offsets
 4348: vector was exactly big enough.
 4349: 
 4350: 6. Corrected definition of offsetof() from change 5 below.
 4351: 
 4352: 7. I had screwed up change 6 below and broken the rules for the use of
 4353: setjmp(). Now fixed.
 4354: 
 4355: 
 4356: Version 1.03 18-Dec-97
 4357: ----------------------
 4358: 
 4359: 1. A erroneous regex with a missing opening parenthesis was correctly
 4360: diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
 4361: on some systems.
 4362: 
 4363: 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
 4364: it was reported that one broken compiler failed on the former because "code" is
 4365: also an independent variable.
 4366: 
 4367: 3. The erroneous regex a[]b caused an array overrun reference.
 4368: 
 4369: 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
 4370: fail on data ending with that character. (It was going on too far, and checking
 4371: the next character, typically a binary zero.) This was specific to the
 4372: optimized code for single-character negative classes.
 4373: 
 4374: 5. Added a contributed patch from the TIN world which does the following:
 4375: 
 4376:   + Add an undef for memmove, in case the the system defines a macro for it.
 4377: 
 4378:   + Add a definition of offsetof(), in case there isn't one. (I don't know
 4379:     the reason behind this - offsetof() is part of the ANSI standard - but
 4380:     it does no harm).
 4381: 
 4382:   + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
 4383:     most of the places where whitespace preceded '#'. I have given up and
 4384:     allowed the remaining 2 cases to be at the margin.
 4385: 
 4386:   + Rename some variables in pcre to eliminate shadowing. This seems very
 4387:     pedantic, but does no harm, of course.
 4388: 
 4389: 6. Moved the call to setjmp() into its own function, to get rid of warnings
 4390: from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
 4391: 
 4392: 7. Constructs such as \d{8,} were compiling into the equivalent of
 4393: \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
 4394: outcome, but in this particular case used more store than had been allocated,
 4395: which caused the bug to be discovered because it threw up an internal error.
 4396: 
 4397: 8. The debugging code in both pcre and pcretest for outputting the compiled
 4398: form of a regex was going wrong in the case of back references followed by
 4399: curly-bracketed repeats.
 4400: 
 4401: 
 4402: Version 1.02 12-Dec-97
 4403: ----------------------
 4404: 
 4405: 1. Typos in pcre.3 and comments in the source fixed.
 4406: 
 4407: 2. Applied a contributed patch to get rid of places where it used to remove
 4408: 'const' from variables, and fixed some signed/unsigned and uninitialized
 4409: variable warnings.
 4410: 
 4411: 3. Added the "runtest" target to Makefile.
 4412: 
 4413: 4. Set default compiler flag to -O2 rather than just -O.
 4414: 
 4415: 
 4416: Version 1.01 19-Nov-97
 4417: ----------------------
 4418: 
 4419: 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
 4420: like /([ab]*)*/, that is, for classes with more than one character in them.
 4421: 
 4422: 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
 4423: as /((?>a*))*/ (a PCRE_EXTRA facility).
 4424: 
 4425: 
 4426: Version 1.00 18-Nov-97
 4427: ----------------------
 4428: 
 4429: 1. Added compile-time macros to support systems such as SunOS4 which don't have
 4430: memmove() or strerror() but have other things that can be used instead.
 4431: 
 4432: 2. Arranged that "make clean" removes the executables.
 4433: 
 4434: 
 4435: Version 0.99 27-Oct-97
 4436: ----------------------
 4437: 
 4438: 1. Fixed bug in code for optimizing classes with only one character. It was
 4439: initializing a 32-byte map regardless, which could cause it to run off the end
 4440: of the memory it had got.
 4441: 
 4442: 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
 4443: 
 4444: 
 4445: Version 0.98 22-Oct-97
 4446: ----------------------
 4447: 
 4448: 1. Fixed bug in code for handling temporary memory usage when there are more
 4449: back references than supplied space in the ovector. This could cause segfaults.
 4450: 
 4451: 
 4452: Version 0.97 21-Oct-97
 4453: ----------------------
 4454: 
 4455: 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
 4456: 
 4457: 2. Optimized negated single characters not to use a bit map.
 4458: 
 4459: 3. Brought error texts together as macro definitions; clarified some of them;
 4460: fixed one that was wrong - it said "range out of order" when it meant "invalid
 4461: escape sequence".
 4462: 
 4463: 4. Changed some char * arguments to const char *.
 4464: 
 4465: 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
 4466: 
 4467: 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
 4468: pcretest.
 4469: 
 4470: 
 4471: Version 0.96 16-Oct-97
 4472: ----------------------
 4473: 
 4474: 1. Added a simple "pgrep" utility to the distribution.
 4475: 
 4476: 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
 4477: unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
 4478: where "ddd" means "one or more decimal digits".
 4479: 
 4480: 3. Fixed serious bug. If a pattern had a back reference, but the call to
 4481: pcre_exec() didn't supply a large enough ovector to record the related
 4482: identifying subpattern, the match always failed. PCRE now remembers the number
 4483: of the largest back reference, and gets some temporary memory in which to save
 4484: the offsets during matching if necessary, in order to ensure that
 4485: backreferences always work.
 4486: 
 4487: 4. Increased the compatibility with Perl in a number of ways:
 4488: 
 4489:   (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
 4490:       to request this handling. The option can be set at compile or exec time.
 4491: 
 4492:   (b) $ matches before a terminating newline by default; an option
 4493:       PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
 4494:       mode). The option can be set at compile or exec time.
 4495: 
 4496:   (c) The handling of \ followed by a digit other than 0 is now supposed to be
 4497:       the same as Perl's. If the decimal number it represents is less than 10
 4498:       or there aren't that many previous left capturing parentheses, an octal
 4499:       escape is read. Inside a character class, it's always an octal escape,
 4500:       even if it is a single digit.
 4501: 
 4502:   (d) An escaped but undefined alphabetic character is taken as a literal,
 4503:       unless PCRE_EXTRA is set. Currently this just reserves the remaining
 4504:       escapes.
 4505: 
 4506:   (e) {0} is now permitted. (The previous item is removed from the compiled
 4507:       pattern).
 4508: 
 4509: 5. Changed all the names of code files so that the basic parts are no longer
 4510: than 10 characters, and abolished the teeny "globals.c" file.
 4511: 
 4512: 6. Changed the handling of character classes; they are now done with a 32-byte
 4513: bit map always.
 4514: 
 4515: 7. Added the -d and /D options to pcretest to make it possible to look at the
 4516: internals of compilation without having to recompile pcre.
 4517: 
 4518: 
 4519: Version 0.95 23-Sep-97
 4520: ----------------------
 4521: 
 4522: 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
 4523: \x20 at the start of a run of normal characters. These were being treated as
 4524: real characters, instead of the source characters being re-checked.
 4525: 
 4526: 
 4527: Version 0.94 18-Sep-97
 4528: ----------------------
 4529: 
 4530: 1. The functions are now thread-safe, with the caveat that the global variables
 4531: containing pointers to malloc() and free() or alternative functions are the
 4532: same for all threads.
 4533: 
 4534: 2. Get pcre_study() to generate a bitmap of initial characters for non-
 4535: anchored patterns when this is possible, and use it if passed to pcre_exec().
 4536: 
 4537: 
 4538: Version 0.93 15-Sep-97
 4539: ----------------------
 4540: 
 4541: 1. /(b)|(:+)/ was computing an incorrect first character.
 4542: 
 4543: 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
 4544: but not actually doing anything yet.
 4545: 
 4546: 3. Treat "-" characters in classes that cannot be part of ranges as literals,
 4547: as Perl does (e.g. [-az] or [az-]).
 4548: 
 4549: 4. Set the anchored flag if a branch starts with .* or .*? because that tests
 4550: all possible positions.
 4551: 
 4552: 5. Split up into different modules to avoid including unneeded functions in a
 4553: compiled binary. However, compile and exec are still in one module. The "study"
 4554: function is split off.
 4555: 
 4556: 6. The character tables are now in a separate module whose source is generated
 4557: by an auxiliary program - but can then be edited by hand if required. There are
 4558: now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
 4559: toupper() in the code.
 4560: 
 4561: 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
 4562: make them global. Abolish the function for setting them, as the caller can now
 4563: set them directly.
 4564: 
 4565: 
 4566: Version 0.92 11-Sep-97
 4567: ----------------------
 4568: 
 4569: 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
 4570: (e.g. /a{1,3}/) was broken (I mis-optimized it).
 4571: 
 4572: 2. Caseless matching was not working in character classes if the characters in
 4573: the pattern were in upper case.
 4574: 
 4575: 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
 4576: 
 4577: 4. Make PCRE_ANCHORED public and accept as a compile option.
 4578: 
 4579: 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
 4580: PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
 4581: pass them.
 4582: 
 4583: 6. Give an error if bad option bits passed at compile or run time.
 4584: 
 4585: 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
 4586: pcretest to cause it to pass that flag.
 4587: 
 4588: 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
 4589: options, and the first character, if set.
 4590: 
 4591: 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
 4592: 
 4593: 
 4594: Version 0.91 10-Sep-97
 4595: ----------------------
 4596: 
 4597: 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
 4598: match the empty string as in /(a*)*/. It was looping and ultimately crashing.
 4599: 
 4600: 2. PCRE was looping on encountering an indefinitely repeated back reference to
 4601: a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
 4602: Perl does - treats the match as successful.
 4603: 
 4604: ****