"Fossies" - the Fresh Open Source Software Archive

Member "pcre-8.43/ChangeLog" (23 Feb 2019, 298092 Bytes) of package /linux/misc/pcre-8.43.tar.bz2:

As a special service "Fossies" has tried to format the requested text file into HTML format (style: standard) with prefixed line numbers. Alternatively you can here view or download the uninterpreted source code file. See also the latest Fossies "Diffs" side-by-side code changes report for "ChangeLog": 8.42_vs_8.43.

    1 ChangeLog for PCRE
    2 ------------------
    4 Note that the PCRE 8.xx series (PCRE1) is now in a bugfix-only state. All
    5 development is happening in the PCRE2 10.xx series.
    8 Version 8.43 23-February-2019
    9 -----------------------------
   11 1. Some time ago the config macro SUPPORT_UTF8 was changed to SUPPORT_UTF
   12 because it also applies to UTF-16 and UTF-32. However, this change was not made
   13 in the pcre2cpp files; consequently the C++ wrapper has from then been compiled
   14 with a bug in it, which would have been picked up by the unit test except that
   15 it also had its UTF8 code cut out. The bug was in a global replace when moving
   16 forward after matching an empty string.
   18 2. The C++ wrapper got broken a long time ago (version 7.3, August 2007) when
   19 (*CR) was invented (assuming it was the first such start-of-pattern option).
   20 The wrapper could never handle such patterns because it wraps patterns in
   21 (?:...)\z in order to support end anchoring. I have hacked in some code to fix
   22 this, that is, move the wrapping till after any existing start-of-pattern
   23 special settings.
   25 3. "pcre2grep" (sic) was accidentally mentioned in an error message (fix was
   26 ported from PCRE2).
   28 4. Typo LCC_ALL for LC_ALL fixed in pcregrep.
   30 5. In a pattern such as /[^\x{100}-\x{ffff}]*[\x80-\xff]/ which has a repeated
   31 negative class with no characters less than 0x100 followed by a positive class
   32 with only characters less than 0x100, the first class was incorrectly being
   33 auto-possessified, causing incorrect match failures.
   35 6. If the only branch in a conditional subpattern was anchored, the whole
   36 subpattern was treated as anchored, when it should not have been, since the
   37 assumed empty second branch cannot be anchored. Demonstrated by test patterns
   38 such as /(?(1)^())b/ or /(?(?=^))b/.
   40 7. Fix subject buffer overread in JIT when UTF is disabled and \X or \R has
   41 a greater than 1 fixed quantifier. This issue was found by Yunho Kim.
   43 8. If a pattern started with a subroutine call that had a quantifier with a
   44 minimum of zero, an incorrect "match must start with this character" could be
   45 recorded. Example: /(?&xxx)*ABC(?<xxx>XYZ)/ would (incorrectly) expect 'A' to
   46 be the first character of a match.
   48 9. Improve MAP_JIT flag usage on MacOS. Patch by Rich Siegel.
   51 Version 8.42 20-March-2018
   52 --------------------------
   54 1.  Fixed a MIPS issue in the JIT compiler reported by Joshua Kinard.
   56 2.  Fixed outdated real_pcre definitions in pcre.h.in (patch by Evgeny Kotkov).
   58 3.  pcregrep was truncating components of file names to 128 characters when
   59 processing files with the -r option, and also (some very odd code) truncating
   60 path names to 512 characters. There is now a check on the absolute length of
   61 full path file names, which may be up to 2047 characters long.
   63 4.  Using pcre_dfa_exec(), in UTF mode when UCP support was not defined, there
   64 was the possibility of a false positive match when caselessly matching a "not
   65 this character" item such as [^\x{1234}] (with a code point greater than 127)
   66 because the "other case" variable was not being initialized.
   68 5. Although pcre_jit_exec checks whether the pattern is compiled
   69 in a given mode, it was also expected that at least one mode is available.
   70 This is fixed and pcre_jit_exec returns with PCRE_ERROR_JIT_BADOPTION
   71 when the pattern is not optimized by JIT at all.
   73 6. The line number and related variables such as match counts in pcregrep
   74 were all int variables, causing overflow when files with more than 2147483647
   75 lines were processed (assuming 32-bit ints). They have all been changed to
   76 unsigned long ints.
   78 7. If a backreference with a minimum repeat count of zero was first in a
   79 pattern, apart from assertions, an incorrect first matching character could be
   80 recorded. For example, for the pattern /(?=(a))\1?b/, "b" was incorrectly set
   81 as the first character of a match.
   83 8. Fix out-of-bounds read for partial matching of /./ against an empty string
   84 when the newline type is CRLF.
   86 9. When matching using the the REG_STARTEND feature of the POSIX API with a
   87 non-zero starting offset, unset capturing groups with lower numbers than a
   88 group that did capture something were not being correctly returned as "unset"
   89 (that is, with offset values of -1).
   91 10. Matching the pattern /(*UTF)\C[^\v]+\x80/ against an 8-bit string
   92 containing multi-code-unit characters caused bad behaviour and possibly a
   93 crash. This issue was fixed for other kinds of repeat in release 8.37 by change
   94 38, but repeating character classes were overlooked.
   96 11. A small fix to pcregrep to avoid compiler warnings for -Wformat-overflow=2.
   98 12. Added --enable-jit=auto support to configure.ac.
  100 13. Fix misleading error message in configure.ac.
  103 Version 8.41 05-July-2017
  104 -------------------------
  106 1.  Fixed typo in CMakeLists.txt (wrong number of arguments for
  107 PCRE_STATIC_RUNTIME (affects MSVC only).
  109 2.  Issue 1 for 8.40 below was not correctly fixed. If pcregrep in multiline
  110 mode with --only-matching matched several lines, it restarted scanning at the
  111 next line instead of moving on to the end of the matched string, which can be
  112 several lines after the start.
  114 3.  Fix a missing else in the JIT compiler reported by 'idaifish'.
  116 4.  A (?# style comment is now ignored between a basic quantifier and a
  117 following '+' or '?' (example: /X+(?#comment)?Y/.
  119 5.  Avoid use of a potentially overflowing buffer in pcregrep (patch by Petr
  120 Pisar).
  122 6.  Fuzzers have reported issues in pcretest. These are NOT serious (it is,
  123 after all, just a test program). However, to stop the reports, some easy ones
  124 are fixed:
  126     (a) Check for values < 256 when calling isprint() in pcretest.
  127     (b) Give an error for too big a number after \O.
  129 7.  In the 32-bit library in non-UTF mode, an attempt to find a Unicode
  130 property for a character with a code point greater than 0x10ffff (the Unicode
  131 maximum) caused a crash.
  133 8. The alternative matching function, pcre_dfa_exec() misbehaved if it
  134 encountered a character class with a possessive repeat, for example [a-f]{3}+.
  136 9. When pcretest called pcre_copy_substring() in 32-bit mode, it set the buffer
  137 length incorrectly, which could result in buffer overflow.
  139 10. Remove redundant line of code (accidentally left in ages ago).
  141 11. Applied C++ patch from Irfan Adilovic to guard 'using std::' directives
  142 with namespace pcrecpp (Bugzilla #2084).
  144 12. Remove a duplication typo in pcre_tables.c.
  146 13. Fix returned offsets from regexec() when REG_STARTEND is used with a
  147 starting offset greater than zero.
  150 Version 8.40 11-January-2017
  151 ----------------------------
  153 1.  Using -o with -M in pcregrep could cause unnecessary repeated output when
  154     the match extended over a line boundary.
  156 2.  Applied Chris Wilson's second patch (Bugzilla #1681) to CMakeLists.txt for
  157     MSVC static compilation, putting the first patch under a new option.
  159 3.  Fix register overwite in JIT when SSE2 acceleration is enabled.
  161 4.  Ignore "show all captures" (/=) for DFA matching.
  163 5.  Fix JIT unaligned accesses on x86. Patch by Marc Mutz.
  165 6.  In any wide-character mode (8-bit UTF or any 16-bit or 32-bit mode),
  166     without PCRE_UCP set, a negative character type such as \D in a positive
  167     class should cause all characters greater than 255 to match, whatever else
  168     is in the class. There was a bug that caused this not to happen if a
  169     Unicode property item was added to such a class, for example [\D\P{Nd}] or
  170     [\W\pL].
  172 7.  When pcretest was outputing information from a callout, the caret indicator
  173     for the current position in the subject line was incorrect if it was after
  174     an escape sequence for a character whose code point was greater than
  175     \x{ff}.
  177 8.  A pattern such as (?<RA>abc)(?(R)xyz) was incorrectly compiled such that
  178     the conditional was interpreted as a reference to capturing group 1 instead
  179     of a test for recursion. Any group whose name began with R was
  180     misinterpreted in this way. (The reference interpretation should only
  181     happen if the group's name is precisely "R".)
  183 9.  A number of bugs have been mended relating to match start-up optimizations
  184     when the first thing in a pattern is a positive lookahead. These all
  185     applied only when PCRE_NO_START_OPTIMIZE was *not* set:
  187     (a) A pattern such as (?=.*X)X$ was incorrectly optimized as if it needed
  188         both an initial 'X' and a following 'X'.
  189     (b) Some patterns starting with an assertion that started with .* were
  190         incorrectly optimized as having to match at the start of the subject or
  191         after a newline. There are cases where this is not true, for example,
  192         (?=.*[A-Z])(?=.{8,16})(?!.*[\s]) matches after the start in lines that
  193         start with spaces. Starting .* in an assertion is no longer taken as an
  194         indication of matching at the start (or after a newline).
  197 Version 8.39 14-June-2016
  198 -------------------------
  200 1.  If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
  201     an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
  202     misbehaved. This bug was found by the LLVM fuzzer.
  204 2.  Similar to the above, if an isolated \E was present between an item and its
  205     qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
  206     bug was found by the LLVM fuzzer.
  208 3.  Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
  209     working correctly in UCP mode.
  211 4.  The POSIX wrapper function regexec() crashed if the option REG_STARTEND
  212     was set when the pmatch argument was NULL. It now returns REG_INVARG.
  214 5.  Allow for up to 32-bit numbers in the ordin() function in pcregrep.
  216 6.  An empty \Q\E sequence between an item and its qualifier caused
  217     pcre_compile() to misbehave when auto callouts were enabled. This bug was
  218     found by the LLVM fuzzer.
  220 7.  If a pattern that was compiled with PCRE_EXTENDED started with white
  221     space or a #-type comment that was followed by (?-x), which turns off
  222     PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
  223     pcre_compile() assumed that (?-x) applied to the whole pattern and
  224     consequently mis-compiled it. This bug was found by the LLVM fuzzer.
  226 8.  A call of pcre_copy_named_substring() for a named substring whose number
  227     was greater than the space in the ovector could cause a crash.
  229 9.  Yet another buffer overflow bug involved duplicate named groups with a
  230     group that reset capture numbers (compare 8.38/7 below). Once again, I have
  231     just allowed for more memory, even if not needed. (A proper fix is
  232     implemented in PCRE2, but it involves a lot of refactoring.)
  234 10. pcre_get_substring_list() crashed if the use of \K in a match caused the
  235     start of the match to be earlier than the end.
  237 11. Migrating appropriate PCRE2 JIT improvements to PCRE.
  239 12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
  240     assertion, caused pcretest to generate incorrect output, and also to read
  241     uninitialized memory (detected by ASAN or valgrind).
  243 13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
  244     nested set of parentheses of sufficient size caused an overflow of the
  245     compiling workspace (which was diagnosed, but of course is not desirable).
  247 14. And yet another buffer overflow bug involving duplicate named groups, this
  248     time nested, with a nested back reference. Yet again, I have just allowed
  249     for more memory, because anything more needs all the refactoring that has
  250     been done for PCRE2. An example pattern that provoked this bug is:
  251     /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
  252     registered as CVE-2016-1283.
  254 15. pcretest went into a loop if global matching was requested with an ovector
  255     size less than 2. It now gives an error message. This bug was found by
  256     afl-fuzz.
  258 16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
  259     ("assertion expected") when (?(?C) was not followed by an opening
  260     parenthesis.
  262 17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
  263     actually affect anything, by sheer luck.
  265 18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
  266     static compilation.
  268 19. Modified the RunTest script to incorporate a valgrind suppressions file so
  269     that certain errors, provoked by the SSE2 instruction set when JIT is used,
  270     are ignored.
  272 20. A racing condition is fixed in JIT reported by Mozilla.
  274 21. Minor code refactor to avoid "array subscript is below array bounds"
  275     compiler warning.
  277 22. Minor code refactor to avoid "left shift of negative number" warning.
  279 23. Fix typo causing compile error when 16- or 32-bit JIT is compiled without
  280     UCP support.
  282 24. Refactor to avoid compiler warnings in pcrecpp.cc.
  284 25. Refactor to fix a typo in pcre_jit_test.c
  286 26. Patch to support compiling pcrecpp.cc with Intel compiler.
  289 Version 8.38 23-November-2015
  290 -----------------------------
  292 1.  If a group that contained a recursive back reference also contained a
  293     forward reference subroutine call followed by a non-forward-reference
  294     subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
  295     compile correct code, leading to undefined behaviour or an internally
  296     detected error. This bug was discovered by the LLVM fuzzer.
  298 2.  Quantification of certain items (e.g. atomic back references) could cause
  299     incorrect code to be compiled when recursive forward references were
  300     involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
  301     This bug was discovered by the LLVM fuzzer.
  303 3.  A repeated conditional group whose condition was a reference by name caused
  304     a buffer overflow if there was more than one group with the given name.
  305     This bug was discovered by the LLVM fuzzer.
  307 4.  A recursive back reference by name within a group that had the same name as
  308     another group caused a buffer overflow. For example:
  309     /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
  311 5.  A forward reference by name to a group whose number is the same as the
  312     current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
  313     a buffer overflow at compile time. This bug was discovered by the LLVM
  314     fuzzer.
  316 6.  A lookbehind assertion within a set of mutually recursive subpatterns could
  317     provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
  319 7.  Another buffer overflow bug involved duplicate named groups with a
  320     reference between their definition, with a group that reset capture
  321     numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
  322     fixed by always allowing for more memory, even if not needed. (A proper fix
  323     is implemented in PCRE2, but it involves more refactoring.)
  325 8.  There was no check for integer overflow in subroutine calls such as (?123).
  327 9.  The table entry for \l in EBCDIC environments was incorrect, leading to its
  328     being treated as a literal 'l' instead of causing an error.
  330 10. There was a buffer overflow if pcre_exec() was called with an ovector of
  331     size 1. This bug was found by american fuzzy lop.
  333 11. If a non-capturing group containing a conditional group that could match
  334     an empty string was repeated, it was not identified as matching an empty
  335     string itself. For example: /^(?:(?(1)x|)+)+$()/.
  337 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
  338     \a and \e in test subject lines.
  340 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
  341     instead of the EBCDIC value.
  343 14. The handling of \c in an EBCDIC environment has been revised so that it is
  344     now compatible with the specification in Perl's perlebcdic page.
  346 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
  347     ASCII/Unicode. This has now been added to the list of characters that are
  348     recognized as white space in EBCDIC.
  350 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
  351     error (correctly) when used outside a class, but did not give an error
  352     within a class.
  354 17. \h within a class was incorrectly compiled in EBCDIC environments.
  356 18. A pattern with an unmatched closing parenthesis that contained a backward
  357     assertion which itself contained a forward reference caused buffer
  358     overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
  360 19. JIT should return with error when the compiled pattern requires more stack
  361     space than the maximum.
  363 20. A possessively repeated conditional group that could match an empty string,
  364     for example, /(?(R))*+/, was incorrectly compiled.
  366 21. Fix infinite recursion in the JIT compiler when certain patterns such as
  367     /(?:|a|){100}x/ are analysed.
  369 22. Some patterns with character classes involving [: and \\ were incorrectly
  370     compiled and could cause reading from uninitialized memory or an incorrect
  371     error diagnosis.
  373 23. Pathological patterns containing many nested occurrences of [: caused
  374     pcre_compile() to run for a very long time.
  376 24. A conditional group with only one branch has an implicit empty alternative
  377     branch and must therefore be treated as potentially matching an empty
  378     string.
  380 25. If (?R was followed by - or + incorrect behaviour happened instead of a
  381     diagnostic.
  383 26. Arrange to give up on finding the minimum matching length for overly
  384     complex patterns.
  386 27. Similar to (4) above: in a pattern with duplicated named groups and an
  387     occurrence of (?| it is possible for an apparently non-recursive back
  388     reference to become recursive if a later named group with the relevant
  389     number is encountered. This could lead to a buffer overflow. Wen Guanxing
  390     from Venustech ADLAB discovered this bug.
  392 28. If pcregrep was given the -q option with -c or -l, or when handling a
  393     binary file, it incorrectly wrote output to stdout.
  395 29. The JIT compiler did not restore the control verb head in case of *THEN
  396     control verbs. This issue was found by Karl Skomski with a custom LLVM
  397     fuzzer.
  399 30. Error messages for syntax errors following \g and \k were giving inaccurate
  400     offsets in the pattern.
  402 31. Added a check for integer overflow in conditions (?(<digits>) and
  403     (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
  404     fuzzer.
  406 32. Handling recursive references such as (?2) when the reference is to a group
  407     later in the pattern uses code that is very hacked about and error-prone.
  408     It has been re-written for PCRE2. Here in PCRE1, a check has been added to
  409     give an internal error if it is obvious that compiling has gone wrong.
  411 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
  412     This issue was found by Karl Skomski with a custom LLVM fuzzer.
  414 34. The JIT compiler should restore the control chain for empty possessive
  415     repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
  417 35. Match limit check added to JIT recursion. This issue was found by Karl
  418     Skomski with a custom LLVM fuzzer.
  420 36. Yet another case similar to 27 above has been circumvented by an
  421     unconditional allocation of extra memory. This issue is fixed "properly" in
  422     PCRE2 by refactoring the way references are handled. Wen Guanxing
  423     from Venustech ADLAB discovered this bug.
  425 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
  426     with a custom LLVM fuzzer.
  428 38. Fixed a corner case of range optimization in JIT.
  430 39. An incorrect error "overran compiling workspace" was given if there were
  431     exactly enough group forward references such that the last one extended
  432     into the workspace safety margin. The next one would have expanded the
  433     workspace. The test for overflow was not including the safety margin.
  435 40. A match limit issue is fixed in JIT which was found by Karl Skomski
  436     with a custom LLVM fuzzer.
  438 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
  439     work under Windows. (Why has it taken so long for anyone to notice?)
  441 42. In a character class such as [\W\p{Any}] where both a negative-type escape
  442     ("not a word character") and a property escape were present, the property
  443     escape was being ignored.
  445 43. Fix crash caused by very long (*MARK) or (*THEN) names.
  447 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
  448     by a single ASCII character in a class item, was incorrectly compiled in
  449     UCP mode. The POSIX class got lost, but only if the single character
  450     followed it.
  452 45. [:punct:] in UCP mode was matching some characters in the range 128-255
  453     that should not have been matched.
  455 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
  456     class, all characters with code points greater than 255 are in the class.
  457     When a Unicode property was also in the class (if PCRE_UCP is set, escapes
  458     such as \w are turned into Unicode properties), wide characters were not
  459     correctly handled, and could fail to match.
  462 Version 8.37 28-April-2015
  463 --------------------------
  465 1.  When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
  466     for those parentheses to be closed with whatever has been captured so far.
  467     However, it was failing to mark any other groups between the hightest
  468     capture so far and the currrent group as "unset". Thus, the ovector for
  469     those groups contained whatever was previously there. An example is the
  470     pattern /(x)|((*ACCEPT))/ when matched against "abcd".
  472 2.  If an assertion condition was quantified with a minimum of zero (an odd
  473     thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
  475 3.  If a pattern in pcretest input had the P (POSIX) modifier followed by an
  476     unrecognized modifier, a crash could occur.
  478 4.  An attempt to do global matching in pcretest with a zero-length ovector
  479     caused a crash.
  481 5.  Fixed a memory leak during matching that could occur for a subpattern
  482     subroutine call (recursive or otherwise) if the number of captured groups
  483     that had to be saved was greater than ten.
  485 6.  Catch a bad opcode during auto-possessification after compiling a bad UTF
  486     string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
  487     UTF with NO_UTF_CHECK is documented as having an undefined outcome.
  489 7.  A UTF pattern containing a "not" match of a non-ASCII character and a
  490     subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
  492 8. When a pattern is compiled, it remembers the highest back reference so that
  493    when matching, if the ovector is too small, extra memory can be obtained to
  494    use instead. A conditional subpattern whose condition is a check on a
  495    capture having happened, such as, for example in the pattern
  496    /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
  497    setting the highest backreference number. This mattered only if pcre_exec()
  498    was called with an ovector that was too small to hold the capture, and there
  499    was no other kind of back reference (a situation which is probably quite
  500    rare). The effect of the bug was that the condition was always treated as
  501    FALSE when the capture could not be consulted, leading to a incorrect
  502    behaviour by pcre_exec(). This bug has been fixed.
  504 9. A reference to a duplicated named group (either a back reference or a test
  505    for being set in a conditional) that occurred in a part of the pattern where
  506    PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
  507    to be incorrectly calculated, leading to overwriting.
  509 10. A mutually recursive set of back references such as (\2)(\1) caused a
  510     segfault at study time (while trying to find the minimum matching length).
  511     The infinite loop is now broken (with the minimum length unset, that is,
  512     zero).
  514 11. If an assertion that was used as a condition was quantified with a minimum
  515     of zero, matching went wrong. In particular, if the whole group had
  516     unlimited repetition and could match an empty string, a segfault was
  517     likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
  518     assertions to be quantified, but not if they are being used as conditions,
  519     so the above pattern is faulted by Perl. PCRE has now been changed so that
  520     it also rejects such patterns.
  522 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
  523     failed to allow the zero-repeat case if pcre2_exec() was called with an
  524     ovector too small to capture the group.
  526 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
  527     Red Hat Product Security:
  529     (a) A crash if /K and /F were both set with the option to save the compiled
  530     pattern.
  532     (b) Another crash if the option to print captured substrings in a callout
  533     was combined with setting a null ovector, for example \O\C+ as a subject
  534     string.
  536 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
  537     forward reference repeated a large (but limited) number of times within a
  538     repeated outer group that has a zero minimum quantifier, caused incorrect
  539     code to be compiled, leading to the error "internal error:
  540     previously-checked referenced subpattern not found" when an incorrect
  541     memory address was read. This bug was reported as "heap overflow",
  542     discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
  543     CVE-2015-2325.
  545 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
  546     call within a group that also contained a recursive back reference caused
  547     incorrect code to be compiled. This bug was reported as "heap overflow",
  548     discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
  549     number CVE-2015-2326.
  551 24. Computing the size of the JIT read-only data in advance has been a source
  552     of various issues, and new ones are still appear unfortunately. To fix
  553     existing and future issues, size computation is eliminated from the code,
  554     and replaced by on-demand memory allocation.
  556 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
  557     adjacent to the end of the range, and the range contained characters with
  558     more than one other case, caused incorrect behaviour when compiled in UTF
  559     mode. In that example, the range a-j was left out of the class.
  561 26. Fix JIT compilation of conditional blocks, which assertion
  562     is converted to (*FAIL). E.g: /(?(?!))/.
  564 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
  565     discovered by the LLVM fuzzer.
  567 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
  568     when this assertion was used as a condition, for example (?(?!)a|b). In
  569     pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
  570     error about an unsupported item.
  572 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
  573     possessification code could take exponential time to complete. A recursion
  574     depth limit of 1000 has been imposed to limit the resources used by this
  575     optimization.
  577 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
  578     such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
  579     because \S ensures they are all in the class. The code for doing this was
  580     interacting badly with the code for computing the amount of space needed to
  581     compile the pattern, leading to a buffer overflow. This bug was discovered
  582     by the LLVM fuzzer.
  584 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
  585     other kinds of group caused stack overflow at compile time. This bug was
  586     discovered by the LLVM fuzzer.
  588 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
  589     between a subroutine call and its quantifier was incorrectly compiled,
  590     leading to buffer overflow or other errors. This bug was discovered by the
  591     LLVM fuzzer.
  593 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
  594     assertion after (?(. The code was failing to check the character after
  595     (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
  596     was discovered by the LLVM fuzzer.
  598 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
  599     a fixed maximum following a group that contains a subroutine reference was
  600     incorrectly compiled and could trigger buffer overflow. This bug was
  601     discovered by the LLVM fuzzer.
  603 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
  604     caused a stack overflow instead of the diagnosis of a non-fixed length
  605     lookbehind assertion. This bug was discovered by the LLVM fuzzer.
  607 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
  608     (e.g. /(?<=\Ka)/) could make pcregrep loop.
  610 37. There was a similar problem to 36 in pcretest for global matches.
  612 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
  613     and a subsequent item in the pattern caused a non-match, backtracking over
  614     the repeated \X did not stop, but carried on past the start of the subject,
  615     causing reference to random memory and/or a segfault. There were also some
  616     other cases where backtracking after \C could crash. This set of bugs was
  617     discovered by the LLVM fuzzer.
  619 39. The function for finding the minimum length of a matching string could take
  620     a very long time if mutual recursion was present many times in a pattern,
  621     for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
  622     method has been implemented. This infelicity was discovered by the LLVM
  623     fuzzer.
  625 40. Static linking against the PCRE library using the pkg-config module was
  626     failing on missing pthread symbols.
  629 Version 8.36 26-September-2014
  630 ------------------------------
  632 1.  Got rid of some compiler warnings in the C++ modules that were shown up by
  633     -Wmissing-field-initializers and -Wunused-parameter.
  635 2.  The tests for quantifiers being too big (greater than 65535) were being
  636     applied after reading the number, and stupidly assuming that integer
  637     overflow would give a negative number. The tests are now applied as the
  638     numbers are read.
  640 3.  Tidy code in pcre_exec.c where two branches that used to be different are
  641     now the same.
  643 4.  The JIT compiler did not generate match limit checks for certain
  644     bracketed expressions with quantifiers. This may lead to exponential
  645     backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
  646     issue should be resolved now.
  648 5.  Fixed an issue, which occures when nested alternatives are optimized
  649     with table jumps.
  651 6.  Inserted two casts and changed some ints to size_t in the light of some
  652     reported 64-bit compiler warnings (Bugzilla 1477).
  654 7.  Fixed a bug concerned with zero-minimum possessive groups that could match
  655     an empty string, which sometimes were behaving incorrectly in the
  656     interpreter (though correctly in the JIT matcher). This pcretest input is
  657     an example:
  659       '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
  662     the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
  663     JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
  664     for an empty string was breaking the inner loop and carrying on at a lower
  665     level, when possessive repeated groups should always return to a higher
  666     level as they have no backtrack points in them. The empty string test now
  667     occurs at the outer level.
  669 8.  Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
  670     ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
  672 9.  Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
  673     doesn't).
  675 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
  676     to hold a pattern to be incorrectly computed (too small) when there were
  677     named back references to duplicated names. This could cause "internal
  678     error: code overflow" or "double free or corruption" or other memory
  679     handling errors.
  681 11. When named subpatterns had the same prefixes, back references could be
  682     confused. For example, in this pattern:
  684       /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
  686     the reference to 'Name' was incorrectly treated as a reference to a
  687     duplicate name.
  689 12. A pattern such as /^s?c/mi8 where the optional character has more than
  690     one "other case" was incorrectly compiled such that it would only try to
  691     match starting at "c".
  693 13. When a pattern starting with \s was studied, VT was not included in the
  694     list of possible starting characters; this should have been part of the
  695     8.34/18 patch.
  697 14. If a character class started [\Qx]... where x is any character, the class
  698     was incorrectly terminated at the ].
  700 15. If a pattern that started with a caseless match for a character with more
  701     than one "other case" was studied, PCRE did not set up the starting code
  702     unit bit map for the list of possible characters. Now it does. This is an
  703     optimization improvement, not a bug fix.
  705 16. The Unicode data tables have been updated to Unicode 7.0.0.
  707 17. Fixed a number of memory leaks in pcregrep.
  709 18. Avoid a compiler warning (from some compilers) for a function call with
  710     a cast that removes "const" from an lvalue by using an intermediate
  711     variable (to which the compiler does not object).
  713 19. Incorrect code was compiled if a group that contained an internal recursive
  714     back reference was optional (had quantifier with a minimum of zero). This
  715     example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
  716     caused segmentation faults because of stack overflows at compile time.
  718 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
  719     group that is quantified with an indefinite repeat, caused a compile-time
  720     loop which used up all the system stack and provoked a segmentation fault.
  721     This was not the same bug as 19 above.
  723 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
  724     Patch by Mike Frysinger.
  727 Version 8.35 04-April-2014
  728 --------------------------
  730 1.  A new flag is set, when property checks are present in an XCLASS.
  731     When this flag is not set, PCRE can perform certain optimizations
  732     such as studying these XCLASS-es.
  734 2.  The auto-possessification of character sets were improved: a normal
  735     and an extended character set can be compared now. Furthermore
  736     the JIT compiler optimizes more character set checks.
  738 3.  Got rid of some compiler warnings for potentially uninitialized variables
  739     that show up only when compiled with -O2.
  741 4.  A pattern such as (?=ab\K) that uses \K in an assertion can set the start
  742     of a match later then the end of the match. The pcretest program was not
  743     handling the case sensibly - it was outputting from the start to the next
  744     binary zero. It now reports this situation in a message, and outputs the
  745     text from the end to the start.
  747 5.  Fast forward search is improved in JIT. Instead of the first three
  748     characters, any three characters with fixed position can be searched.
  749     Search order: first, last, middle.
  751 6.  Improve character range checks in JIT. Characters are read by an inprecise
  752     function now, which returns with an unknown value if the character code is
  753     above a certain threshold (e.g: 256). The only limitation is that the value
  754     must be bigger than the threshold as well. This function is useful when
  755     the characters above the threshold are handled in the same way.
  757 7.  The macros whose names start with RAWUCHAR are placeholders for a future
  758     mode in which only the bottom 21 bits of 32-bit data items are used. To
  759     make this more memorable for those maintaining the code, the names have
  760     been changed to start with UCHAR21, and an extensive comment has been added
  761     to their definition.
  763 8.  Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
  764     to the export list in Makefile.am (they were accidentally omitted from the
  765     8.34 tarball).
  767 9.  The informational output from pcretest used the phrase "starting byte set"
  768     which is inappropriate for the 16-bit and 32-bit libraries. As the output
  769     for "first char" and "need char" really means "non-UTF-char", I've changed
  770     "byte" to "char", and slightly reworded the output. The documentation about
  771     these values has also been (I hope) clarified.
  773 10. Another JIT related optimization: use table jumps for selecting the correct
  774     backtracking path, when more than four alternatives are present inside a
  775     bracket.
  777 11. Empty match is not possible, when the minimum length is greater than zero,
  778     and there is no \K in the pattern. JIT should avoid empty match checks in
  779     such cases.
  781 12. In a caseless character class with UCP support, when a character with more
  782     than one alternative case was not the first character of a range, not all
  783     the alternative cases were added to the class. For example, s and \x{17f}
  784     are both alternative cases for S: the class [RST] was handled correctly,
  785     but [R-T] was not.
  787 13. The configure.ac file always checked for pthread support when JIT was
  788     enabled. This is not used in Windows, so I have put this test inside a
  789     check for the presence of windows.h (which was already tested for).
  791 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
  792     The algorithm provides a way to skip certain starting offsets, and usually
  793     faster than linear prefix searches.
  795 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
  796     as for 'fr_FR' and 'french'. For some reason, however, it then used the
  797     Windows-specific input and output files, which have 'french' screwed in.
  798     So this could never have worked. One of the problems with locales is that
  799     they aren't always the same. I have now updated RunTest so that it checks
  800     the output of the locale test (test 3) against three different output
  801     files, and it allows the test to pass if any one of them matches. With luck
  802     this should make the test pass on some versions of Solaris where it was
  803     failing. Because of the uncertainty, the script did not used to stop if
  804     test 3 failed; it now does. If further versions of a French locale ever
  805     come to light, they can now easily be added.
  807 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
  808     there was a message during ./configure, but it did not stop. This now
  809     provokes an error. The invalid example in README has been corrected.
  810     If a value less than the minimum is given, the minimum value has always
  811     been used, but now a warning is given.
  813 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
  814     was a bug in the test system, which is now fixed. Also, the list of various
  815     configurations that are tested for each release did not have one with both
  816     16/32 bits and --enable-bar-anycrlf. It now does.
  818 18. pcretest was missing "-C bsr" for displaying the \R default setting.
  820 19. Little endian PowerPC systems are supported now by the JIT compiler.
  822 20. The fast forward newline mechanism could enter to an infinite loop on
  823     certain invalid UTF-8 input. Although we don't support these cases
  824     this issue can be fixed by a performance optimization.
  826 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
  827     not take account if existing stack usage. There is now a new global
  828     variable called pcre_stack_guard that can be set to point to an external
  829     function to check stack availability. It is called at the start of
  830     processing every parenthesized group.
  832 22. A typo in the code meant that in ungreedy mode the max/min qualifier
  833     behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
  834     not match "ab".
  836 23. When UTF was disabled, the JIT program reported some incorrect compile
  837     errors. These messages are silenced now.
  839 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
  840     compiler.
  842 25. Change all the temporary files used in RunGrepTest to be different to those
  843     used by RunTest so that the tests can be run simultaneously, for example by
  844     "make -j check".
  847 Version 8.34 15-December-2013
  848 -----------------------------
  850 1.  Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
  851     executable memory. Patch inspired by Carsten Klein.
  853 2.  ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
  854     this macro is never tested and has no effect, because the work to support
  855     coverage involves only compiling and linking options and special targets in
  856     the Makefile. The comment in config.h implied that defining the macro would
  857     enable coverage support, which is totally false. There was also support for
  858     setting this macro in the CMake files (my fault, I just copied it from
  859     configure). SUPPORT_GCOV has now been removed.
  861 3.  Make a small performance improvement in strlen16() and strlen32() in
  862     pcretest.
  864 4.  Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
  865     detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
  866     these cases). There was also one in pcretest.c.
  868 5.  Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
  870 6.  In UTF mode, the code for checking whether a group could match an empty
  871     string (which is used for indefinitely repeated groups to allow for
  872     breaking an infinite loop) was broken when the group contained a repeated
  873     negated single-character class with a character that occupied more than one
  874     data item and had a minimum repetition of zero (for example, [^\x{100}]* in
  875     UTF-8 mode). The effect was undefined: the group might or might not be
  876     deemed as matching an empty string, or the program might have crashed.
  878 7.  The code for checking whether a group could match an empty string was not
  879     recognizing that \h, \H, \v, \V, and \R must match a character.
  881 8.  Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
  882     an empty string. If it can, pcretest shows this in its information output.
  884 9.  Fixed two related bugs that applied to Unicode extended grapheme clusters
  885     that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
  886     matched by pcre_exec() without using JIT:
  888     (a) If the rest of the pattern did not match after a maximal run of
  889         grapheme clusters, the code for backing up to try with fewer of them
  890         did not always back up over a full grapheme when characters that do not
  891         have the modifier quality were involved, e.g. Hangul syllables.
  893     (b) If the match point in a subject started with modifier character, and
  894         there was no match, the code could incorrectly back up beyond the match
  895         point, and potentially beyond the first character in the subject,
  896         leading to a segfault or an incorrect match result.
  898 10. A conditional group with an assertion condition could lead to PCRE
  899     recording an incorrect first data item for a match if no other first data
  900     item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
  901     first data item, and therefore matched "ca" after "c" instead of at the
  902     start.
  904 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
  905     bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
  907 12. The source of pcregrep now includes z/OS-specific code so that it can be
  908     compiled for z/OS as part of the special z/OS distribution.
  910 13. Added the -T and -TM options to pcretest.
  912 14. The code in pcre_compile.c for creating the table of named capturing groups
  913     has been refactored. Instead of creating the table dynamically during the
  914     actual compiling pass, the information is remembered during the pre-compile
  915     pass (on the stack unless there are more than 20 named groups, in which
  916     case malloc() is used) and the whole table is created before the actual
  917     compile happens. This has simplified the code (it is now nearly 150 lines
  918     shorter) and prepared the way for better handling of references to groups
  919     with duplicate names.
  921 15. A back reference to a named subpattern when there is more than one of the
  922     same name now checks them in the order in which they appear in the pattern.
  923     The first one that is set is used for the reference. Previously only the
  924     first one was inspected. This change makes PCRE more compatible with Perl.
  926 16. Unicode character properties were updated from Unicode 6.3.0.
  928 17. The compile-time code for auto-possessification has been refactored, based
  929     on a patch by Zoltan Herczeg. It now happens after instead of during
  930     compilation. The code is cleaner, and more cases are handled. The option
  931     PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
  932     options in pcretest are provided to set it. It can also be set by
  933     (*NO_AUTO_POSSESS) at the start of a pattern.
  935 18. The character VT has been added to the default ("C" locale) set of
  936     characters that match \s and are generally treated as white space,
  937     following this same change in Perl 5.18. There is now no difference between
  938     "Perl space" and "POSIX space". Whether VT is treated as white space in
  939     other locales depends on the locale.
  941 19. The code for checking named groups as conditions, either for being set or
  942     for being recursed, has been refactored (this is related to 14 and 15
  943     above). Processing unduplicated named groups should now be as fast at
  944     numerical groups, and processing duplicated groups should be faster than
  945     before.
  947 20. Two patches to the CMake build system, by Alexander Barkov:
  949       (1) Replace the "source" command by "." in CMakeLists.txt because
  950           "source" is a bash-ism.
  952       (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
  953           without these the CMake build does not work on Solaris.
  955 21. Perl has changed its handling of \8 and \9. If there is no previously
  956     encountered capturing group of those numbers, they are treated as the
  957     literal characters 8 and 9 instead of a binary zero followed by the
  958     literals. PCRE now does the same.
  960 22. Following Perl, added \o{} to specify codepoints in octal, making it
  961     possible to specify values greater than 0777 and also making them
  962     unambiguous.
  964 23. Perl now gives an error for missing closing braces after \x{... instead of
  965     treating the string as literal. PCRE now does the same.
  967 24. RunTest used to grumble if an inappropriate test was selected explicitly,
  968     but just skip it when running all tests. This make it awkward to run ranges
  969     of tests when one of them was inappropriate. Now it just skips any
  970     inappropriate tests, as it always did when running all tests.
  972 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
  973     character types such as \d or \w, too many callouts were inserted, and the
  974     data that they returned was rubbish.
  976 26. In UCP mode, \s was not matching two of the characters that Perl matches,
  977     namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
  978     were matched by \h. The code has now been refactored so that the lists of
  979     the horizontal and vertical whitespace characters used for \h and \v (which
  980     are defined only in one place) are now also used for \s.
  982 27. Add JIT support for the 64 bit TileGX architecture.
  983     Patch by Jiong Wang (Tilera Corporation).
  985 28. Possessive quantifiers for classes (both explicit and automatically
  986     generated) now use special opcodes instead of wrapping in ONCE brackets.
  988 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
  989     (because it's meaningless), this was not happening when PCRE_CASELESS was
  990     set. Not wrong, but inefficient.
  992 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
  993     properties for \w, \d, etc) is present in a test regex. Otherwise if the
  994     test contains no characters greater than 255, Perl doesn't realise it
  995     should be using Unicode semantics.
  997 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
  998     [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
  999     does in Unicode mode.
 1001 32. Added the "forbid" facility to pcretest so that putting tests into the
 1002     wrong test files can sometimes be quickly detected.
 1004 33. There is now a limit (default 250) on the depth of nesting of parentheses.
 1005     This limit is imposed to control the amount of system stack used at compile
 1006     time. It can be changed at build time by --with-parens-nest-limit=xxx or
 1007     the equivalent in CMake.
 1009 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
 1010     errors. Perl warns for these when in warning mode, but PCRE has no facility
 1011     for giving warnings.
 1013 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
 1014     However, this was not working for (?!) because it is optimized to (*FAIL),
 1015     for which PCRE does not allow quantifiers. The optimization is now disabled
 1016     when a quantifier follows (?!). I can't see any use for this, but it makes
 1017     things uniform.
 1019 36. Perl no longer allows group names to start with digits, so I have made this
 1020     change also in PCRE. It simplifies the code a bit.
 1022 37. In extended mode, Perl ignores spaces before a + that indicates a
 1023     possessive quantifier. PCRE allowed a space before the quantifier, but not
 1024     before the possessive +. It now does.
 1026 38. The use of \K (reset reported match start) within a repeated possessive
 1027     group such as (a\Kb)*+ was not working.
 1029 40. Document that the same character tables must be used at compile time and
 1030     run time, and that the facility to pass tables to pcre_exec() and
 1031     pcre_dfa_exec() is for use only with saved/restored patterns.
 1033 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
 1034     features for Builds with MSVC:
 1036     1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
 1037        builds). The .rc files can be used to set FileDescription and many other
 1038        attributes.
 1040     2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
 1041        This allows higher-level build scripts which want .pdb files to avoid
 1042        hard-coding the exact files needed."
 1044 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
 1045     mean "start of word" and "end of word", respectively, as a transition aid.
 1047 43. A minimizing repeat of a class containing codepoints greater than 255 in
 1048     non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
 1049     compiled to use the heap for recursion.
 1051 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
 1052     is configured.
 1055 Version 8.33 28-May-2013
 1056 ------------------------
 1058 1.  Added 'U' to some constants that are compared to unsigned integers, to
 1059     avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
 1060     variables that are added to signed variables, to ensure the result is
 1061     signed and can be negated.
 1063 2.  Applied patch by Daniel Richard G for quashing MSVC warnings to the
 1064     CMake config files.
 1066 3.  Revise the creation of config.h.generic so that all boolean macros are
 1067     #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
 1068     overriding via -D on the command line possible.
 1070 4.  Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
 1071     to unsigned int is reported to make a quite noticeable speed difference in
 1072     a specific Windows environment. Testing on Linux did also appear to show
 1073     some benefit (and it is clearly not harmful). Also fixed the definition of
 1074     Xop which should be unsigned.
 1076 5.  Related to (4), changing the definition of the intermediate variable cc
 1077     in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
 1078     improvements.
 1080 6.  Fix forward search in JIT when link size is 3 or greater. Also removed some
 1081     unnecessary spaces.
 1083 7.  Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
 1084     and later.
 1086 8.  Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
 1088 9.  Optimizing fast_forward_start_bits in JIT.
 1090 10. Adding support for callouts in JIT, and fixing some issues revealed
 1091     during this work. Namely:
 1093     (a) Unoptimized capturing brackets incorrectly reset on backtrack.
 1095     (b) Minimum length was not checked before the matching is started.
 1097 11. The value of capture_last that is passed to callouts was incorrect in some
 1098     cases when there was a capture on one path that was subsequently abandoned
 1099     after a backtrack. Also, the capture_last value is now reset after a
 1100     recursion, since all captures are also reset in this case.
 1102 12. The interpreter no longer returns the "too many substrings" error in the
 1103     case when an overflowing capture is in a branch that is subsequently
 1104     abandoned after a backtrack.
 1106 13. In the pathological case when an offset vector of size 2 is used, pcretest
 1107     now prints out the matched string after a yield of 0 or 1.
 1109 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
 1110     Only supported by the JIT compiler at the moment.
 1112 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
 1114 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
 1115     offset of the starting point of the matching process, provided the offsets
 1116     vector is large enough.
 1118 17. The \A escape now records a lookbehind value of 1, though its execution
 1119     does not actually inspect the previous character. This is to ensure that,
 1120     in partial multi-segment matching, at least one character from the old
 1121     segment is retained when a new segment is processed. Otherwise, if there
 1122     are no lookbehinds in the pattern, \A might match incorrectly at the start
 1123     of a new segment.
 1125 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
 1127 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
 1128     gives some modest performance improvement in 8-bit mode.
 1130 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
 1131     be expressed in certain programming languages using Universal Character
 1132     Names.
 1134 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
 1135     which points out that "non characters" are not "characters that may not
 1136     appear in Unicode strings" but rather "characters that are reserved for
 1137     internal use and have only local meaning".
 1139 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
 1140     there was a conditional group that depended on an assertion, if the
 1141     assertion was false, the callout that immediately followed the alternation
 1142     in the condition was skipped when pcre_exec() was used for matching.
 1144 23. Allow an explicit callout to be inserted before an assertion that is the
 1145     condition for a conditional group, for compatibility with automatic
 1146     callouts, which always insert a callout at this point.
 1148 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
 1149     confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
 1151 25. (*PRUNE) is now supported by the JIT compiler.
 1153 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
 1155 27. Fix the case where there are two or more SKIPs with arguments that may be
 1156     ignored.
 1158 28. (*SKIP) is now supported by the JIT compiler.
 1160 29. (*THEN) is now supported by the JIT compiler.
 1162 30. Update RunTest with additional test selector options.
 1164 31. The way PCRE handles backtracking verbs has been changed in two ways.
 1166     (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
 1167     SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
 1168     backtracking. In some cases this makes it more Perl-compatible, but Perl's
 1169     rather obscure rules do not always do the same thing.
 1171     (2) Previously, backtracking verbs were confined within assertions. This is
 1172     no longer the case for positive assertions, except for (*ACCEPT). Again,
 1173     this sometimes improves Perl compatibility, and sometimes does not.
 1175 32. A number of tests that were in test 2 because Perl did things differently
 1176     have been moved to test 1, because either Perl or PCRE has changed, and
 1177     these tests are now compatible.
 1179 32. Backtracking control verbs are now handled in the same way in JIT and
 1180     interpreter.
 1182 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
 1183     contained a forward subroutine reference caused a compile error.
 1185 34. Auto-detect and optimize limited repetitions in JIT.
 1187 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
 1188     blocking (*UTF) etc.
 1190 36. In the interpreter, maximizing pattern repetitions for characters and
 1191     character types now use tail recursion, which reduces stack usage.
 1193 37. The value of the max lookbehind was not correctly preserved if a compiled
 1194     and saved regex was reloaded on a host of different endianness.
 1196 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
 1197     of the compiled pattern block, expand the flags field from 16 to 32 bits
 1198     because it was almost full.
 1200 39. Try madvise first before posix_madvise.
 1202 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
 1203     with a pattern such as ^$. It has taken 4 years for anybody to notice! The
 1204     original change locked out all matches of empty strings. This has been
 1205     changed so that one match of an empty string per line is recognized.
 1206     Subsequent searches on the same line (for colouring or for --only-matching,
 1207     for example) do not recognize empty strings.
 1209 41. Applied a user patch to fix a number of spelling mistakes in comments.
 1211 42. Data lines longer than 65536 caused pcretest to crash.
 1213 43. Clarified the data type for length and startoffset arguments for pcre_exec
 1214     and pcre_dfa_exec in the function-specific man pages, where they were
 1215     explicitly stated to be in bytes, never having been updated. I also added
 1216     some clarification to the pcreapi man page.
 1218 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
 1219     a segmentation fault.
 1222 Version 8.32 30-November-2012
 1223 -----------------------------
 1225 1.  Improved JIT compiler optimizations for first character search and single
 1226     character iterators.
 1228 2.  Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
 1229     Patch by Daniel Richard G.
 1231 3.  Single character iterator optimizations in the JIT compiler.
 1233 4.  Improved JIT compiler optimizations for character ranges.
 1235 5.  Rename the "leave" variable names to "quit" to improve WinCE compatibility.
 1236     Reported by Giuseppe D'Angelo.
 1238 6.  The PCRE_STARTLINE bit, indicating that a match can occur only at the start
 1239     of a line, was being set incorrectly in cases where .* appeared inside
 1240     atomic brackets at the start of a pattern, or where there was a subsequent
 1241     *PRUNE or *SKIP.
 1243 7.  Improved instruction cache flush for POWER/PowerPC.
 1244     Patch by Daniel Richard G.
 1246 8.  Fixed a number of issues in pcregrep, making it more compatible with GNU
 1247     grep:
 1249     (a) There is now no limit to the number of patterns to be matched.
 1251     (b) An error is given if a pattern is too long.
 1253     (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
 1254         are now supported.
 1256     (d) --exclude-from and --include-from (multiple use) have been added.
 1258     (e) Exclusions and inclusions now apply to all files and directories, not
 1259         just to those obtained from scanning a directory recursively.
 1261     (f) Multiple uses of -f and --file-list are now supported.
 1263     (g) In a Windows environment, the default for -d has been changed from
 1264         "read" (the GNU grep default) to "skip", because otherwise the presence
 1265         of a directory in the file list provokes an error.
 1267     (h) The documentation has been revised and clarified in places.
 1269 9.  Improve the matching speed of capturing brackets.
 1271 10. Changed the meaning of \X so that it now matches a Unicode extended
 1272     grapheme cluster.
 1274 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
 1275     out POSIX threads when JIT support is configured.
 1277 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
 1279 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
 1280     structure could go wrong in environments where size_t is not the same size
 1281     as int.
 1283 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
 1284     set.
 1286 15. The EBCDIC support had decayed; later updates to the code had included
 1287     explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
 1288     general tidy up of EBCDIC-related issues, and the documentation was also
 1289     not quite right. There is now a test that can be run on ASCII systems to
 1290     check some of the EBCDIC-related things (but is it not a full test).
 1292 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
 1293     in a small tidy to the code.
 1295 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
 1297 18. If the --only-matching (-o) option in pcregrep is specified multiple
 1298     times, each one causes appropriate output. For example, -o1 -o2 outputs the
 1299     substrings matched by the 1st and 2nd capturing parentheses. A separating
 1300     string can be specified by --om-separator (default empty).
 1302 19. Improving the first n character searches.
 1304 20. Turn case lists for horizontal and vertical white space into macros so that
 1305     they are defined only once.
 1307 21. This set of changes together give more compatible Unicode case-folding
 1308     behaviour for characters that have more than one other case when UCP
 1309     support is available.
 1311     (a) The Unicode property table now has offsets into a new table of sets of
 1312         three or more characters that are case-equivalent. The MultiStage2.py
 1313         script that generates these tables (the pcre_ucd.c file) now scans
 1314         CaseFolding.txt instead of UnicodeData.txt for character case
 1315         information.
 1317     (b) The code for adding characters or ranges of characters to a character
 1318         class has been abstracted into a generalized function that also handles
 1319         case-independence. In UTF-mode with UCP support, this uses the new data
 1320         to handle characters with more than one other case.
 1322     (c) A bug that is fixed as a result of (b) is that codepoints less than 256
 1323         whose other case is greater than 256 are now correctly matched
 1324         caselessly. Previously, the high codepoint matched the low one, but not
 1325         vice versa.
 1327     (d) The processing of \h, \H, \v, and \ in character classes now makes use
 1328         of the new class addition function, using character lists defined as
 1329         macros alongside the case definitions of 20 above.
 1331     (e) Caseless back references now work with characters that have more than
 1332         one other case.
 1334     (f) General caseless matching of characters with more than one other case
 1335         is supported.
 1337 22. Unicode character properties were updated from Unicode 6.2.0
 1339 23. Improved CMake support under Windows. Patch by Daniel Richard G.
 1341 24. Add support for 32-bit character strings, and UTF-32
 1343 25. Major JIT compiler update (code refactoring and bugfixing).
 1344     Experimental Sparc 32 support is added.
 1346 26. Applied a modified version of Daniel Richard G's patch to create
 1347     pcre.h.generic and config.h.generic by "make" instead of in the
 1348     PrepareRelease script.
 1350 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
 1351     pcre_compile.c when checking for a zero character.
 1353 28. Introducing a native interface for JIT. Through this interface, the compiled
 1354     machine code can be directly executed. The purpose of this interface is to
 1355     provide fast pattern matching, so several sanity checks are not performed.
 1356     However, feature tests are still performed. The new interface provides
 1357     1.4x speedup compared to the old one.
 1359 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
 1360     the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
 1361     was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
 1363 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
 1364     greater than 0x7fffffff (which cannot be represented in UTF-8, even under
 1365     the "old" RFC 2279). Instead, it ended up passing a negative length to
 1366     pcre_exec().
 1368 31. Add support for GCC's visibility feature to hide internal functions.
 1370 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
 1371     "unknown -C option" after outputting 0 or 1.
 1373 33. There is now support for generating a code coverage report for the test
 1374     suite in environments where gcc is the compiler and lcov is installed. This
 1375     is mainly for the benefit of the developers.
 1377 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
 1378     unaddressable using valgrind annotations, allowing valgrind to detect
 1379     invalid memory accesses. This is mainly for the benefit of the developers.
 1381 25. (*UTF) can now be used to start a pattern in any of the three libraries.
 1383 26. Give configure error if --enable-cpp but no C++ compiler found.
 1386 Version 8.31 06-July-2012
 1387 -------------------------
 1389 1.  Fixing a wrong JIT test case and some compiler warnings.
 1391 2.  Removed a bashism from the RunTest script.
 1393 3.  Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
 1394     to unsigned type, result still unsigned" that was given by an MS compiler
 1395     on encountering the code "-sizeof(xxx)".
 1397 4.  Partial matching support is added to the JIT compiler.
 1399 5.  Fixed several bugs concerned with partial matching of items that consist
 1400     of more than one character:
 1402     (a) /^(..)\1/ did not partially match "aba" because checking references was
 1403         done on an "all or nothing" basis. This also applied to repeated
 1404         references.
 1406     (b) \R did not give a hard partial match if \r was found at the end of the
 1407         subject.
 1409     (c) \X did not give a hard partial match after matching one or more
 1410         characters at the end of the subject.
 1412     (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
 1413         a partial match for the string "\r".
 1415     (e) When newline was set to CRLF, the metacharacter "." did not recognize
 1416         a partial match for a CR character at the end of the subject string.
 1418 6.  If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
 1419     running pcretest, the text "(JIT)" added to the output whenever JIT is
 1420     actually used to run the match.
 1422 7.  Individual JIT compile options can be set in pcretest by following -s+[+]
 1423     or /S+[+] with a digit between 1 and 7.
 1425 8.  OP_NOT now supports any UTF character not just single-byte ones.
 1427 9.  (*MARK) control verb is now supported by the JIT compiler.
 1429 10. The command "./RunTest list" lists the available tests without actually
 1430     running any of them. (Because I keep forgetting what they all are.)
 1434 12. Applied a (slightly modified) user-supplied patch that improves performance
 1435     when the heap is used for recursion (compiled with --disable-stack-for-
 1436     recursion). Instead of malloc and free for each heap frame each time a
 1437     logical recursion happens, frames are retained on a chain and re-used where
 1438     possible. This sometimes gives as much as 30% improvement.
 1440 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
 1441     call.
 1443 14. As documented, (*COMMIT) is now confined to within a positive assertion.
 1445 15. It is now possible to link pcretest with libedit as an alternative to
 1446     libreadline.
 1448 16. (*COMMIT) control verb is now supported by the JIT compiler.
 1450 17. The Unicode data tables have been updated to Unicode 6.1.0.
 1452 18. Added --file-list option to pcregrep.
 1454 19. Added binary file support to pcregrep, including the -a, --binary-files,
 1455     -I, and --text options.
 1457 20. The madvise function is renamed for posix_madvise for QNX compatibility
 1458     reasons. Fixed by Giuseppe D'Angelo.
 1460 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
 1462 22. Changed the option for creating symbolic links for 16-bit man pages from
 1463     -s to -sf so that re-installing does not cause issues.
 1465 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
 1467 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
 1468     matches in certain environments (the workspace was not being correctly
 1469     retained). Also added to pcre_dfa_exec() a simple plausibility check on
 1470     some of the workspace data at the beginning of a restart.
 1472 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
 1473     was not doing so when it should - probably a typo introduced by SVN 528
 1474     (change 8.10/14).
 1476 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
 1477     \w+ when the character tables indicated that \x{c4} was a word character.
 1478     There were several related cases, all because the tests for doing a table
 1479     lookup were testing for characters less than 127 instead of 255.
 1481 27. If a pattern contains capturing parentheses that are not used in a match,
 1482     their slots in the ovector are set to -1. For those that are higher than
 1483     any matched groups, this happens at the end of processing. In the case when
 1484     there were back references that the ovector was too small to contain
 1485     (causing temporary malloc'd memory to be used during matching), and the
 1486     highest capturing number was not used, memory off the end of the ovector
 1487     was incorrectly being set to -1. (It was using the size of the temporary
 1488     memory instead of the true size.)
 1490 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
 1491     ovector size, it uses memory at the end of the block that it has got.
 1493 29. Check for an overlong MARK name and give an error at compile time. The
 1494     limit is 255 for the 8-bit library and 65535 for the 16-bit library.
 1496 30. JIT compiler update.
 1498 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
 1499     Rill for the patch.
 1501 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
 1503 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
 1505 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
 1506     SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
 1507     was enabled.
 1509 35. Improve JIT code generation for greedy plus quantifier.
 1511 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
 1512     1 to "aa" instead of to an empty string. The bug affected repeated groups
 1513     that could potentially match an empty string.
 1515 37. Optimizing single character iterators in JIT.
 1517 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
 1518     the same checks as \x{...} characters in non-JavaScript mode. Specifically,
 1519     codepoints that are too big for the mode are faulted, and in a UTF mode,
 1520     disallowed codepoints are also faulted.
 1522 39. If PCRE was compiled with UTF support, in three places in the DFA
 1523     matcher there was code that should only have been obeyed in UTF mode, but
 1524     was being obeyed unconditionally. In 8-bit mode this could cause incorrect
 1525     processing when bytes with values greater than 127 were present. In 16-bit
 1526     mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
 1527     both cases the values are those that cannot be the first data item in a UTF
 1528     character. The three items that might have provoked this were recursions,
 1529     possessively repeated groups, and atomic groups.
 1531 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
 1532     and pcregrep, because some OS require shared objects to be explicitly
 1533     passed to ld, causing the link step to fail if they are not.
 1535 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
 1536     mode, patterns that started with \h* or \R* might be incorrectly matched.
 1539 Version 8.30 04-February-2012
 1540 -----------------------------
 1542 1.  Renamed "isnumber" as "is_a_number" because in some Mac environments this
 1543     name is defined in ctype.h.
 1545 2.  Fixed a bug in fixed-length calculation for lookbehinds that would show up
 1546     only in quite long subpatterns.
 1548 3.  Removed the function pcre_info(), which has been obsolete and deprecated
 1549     since it was replaced by pcre_fullinfo() in February 2000.
 1551 4.  For a non-anchored pattern, if (*SKIP) was given with a name that did not
 1552     match a (*MARK), and the match failed at the start of the subject, a
 1553     reference to memory before the start of the subject could occur. This bug
 1554     was introduced by fix 17 of release 8.21.
 1556 5.  A reference to an unset group with zero minimum repetition was giving
 1557     totally wrong answers (in non-JavaScript-compatibility mode). For example,
 1558     /(another)?(\1?)test/ matched against "hello world test". This bug was
 1559     introduced in release 8.13.
 1561 6.  Add support for 16-bit character strings (a large amount of work involving
 1562     many changes and refactorings).
 1564 7.  RunGrepTest failed on msys because \r\n was replaced by whitespace when the
 1565     command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
 1566     from a file.
 1568 8.  Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
 1569     rounding is not applied in this particular case).
 1571 9.  The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
 1572     if they appear, or are escaped, in patterns.
 1574 10. Get rid of a number of -Wunused-but-set-variable warnings.
 1576 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
 1577     "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
 1578     Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
 1579     also returns the mark "x". This bug applied to capturing parentheses,
 1580     non-capturing parentheses, and atomic parentheses. It also applied to some
 1581     assertions.
 1583 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
 1584     information out of configure.ac instead of relying on pcre.h.generic, which
 1585     is not stored in the repository.
 1587 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
 1588     -lreadline.
 1590 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
 1592 15. Applied Graycode's patch to put the top-level frame on the stack rather
 1593     than the heap when not using the stack for recursion. This gives a
 1594     performance improvement in many cases when recursion is not deep.
 1596 16. Experimental code added to "pcretest -C" to output the stack frame size.
 1599 Version 8.21 12-Dec-2011
 1600 ------------------------
 1602 1.  Updating the JIT compiler.
 1604 2.  JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
 1605     are added as well.
 1607 3.  Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
 1608     PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
 1609     calling _pcre_jit_exec. Some extra comments are added.
 1611 4.  (*MARK) settings inside atomic groups that do not contain any capturing
 1612     parentheses, for example, (?>a(*:m)), were not being passed out. This bug
 1613     was introduced by change 18 for 8.20.
 1615 5.  Supporting of \x, \U and \u in JavaScript compatibility mode based on the
 1616     ECMA-262 standard.
 1618 6.  Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
 1619     erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
 1620     This bug was probably introduced by change 9 of 8.13.
 1622 7.  While fixing 6 above, I noticed that a number of other items were being
 1623     incorrectly rejected as "not fixed length". This arose partly because newer
 1624     opcodes had not been added to the fixed-length checking code. I have (a)
 1625     corrected the bug and added tests for these items, and (b) arranged for an
 1626     error to occur if an unknown opcode is encountered while checking for fixed
 1627     length instead of just assuming "not fixed length". The items that were
 1628     rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
 1629     (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
 1630     repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
 1632 8.  A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
 1633     being incorrectly compiled and would have given unpredicatble results.
 1635 9.  A possessively repeated subpattern with minimum repeat count greater than
 1636     one behaved incorrectly. For example, (A){2,}+ behaved as if it was
 1637     (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
 1638     the first (A) could occur when it should not.
 1640 10. Add a cast and remove a redundant test from the code.
 1642 11. JIT should use pcre_malloc/pcre_free for allocation.
 1644 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
 1645     best practice nowadays, and helps with cross-compiling. (If the exec_prefix
 1646     is anything other than /usr, -L is still shown).
 1648 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
 1650 14. Perl does not support \N without a following name in a [] class; PCRE now
 1651     also gives an error.
 1653 15. If a forward reference was repeated with an upper limit of around 2000,
 1654     it caused the error "internal error: overran compiling workspace". The
 1655     maximum number of forward references (including repeats) was limited by the
 1656     internal workspace, and dependent on the LINK_SIZE. The code has been
 1657     rewritten so that the workspace expands (via pcre_malloc) if necessary, and
 1658     the default depends on LINK_SIZE. There is a new upper limit (for safety)
 1659     of around 200,000 forward references. While doing this, I also speeded up
 1660     the filling in of repeated forward references.
 1662 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
 1663     incorrectly expecting the subject to contain another "a" after the start.
 1665 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
 1666     in the match, the SKIP should be ignored. This was not happening; instead
 1667     the SKIP was being treated as NOMATCH. For patterns such as
 1668     /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
 1669     tested.
 1671 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
 1672     now much more compatible with Perl, in particular in cases where the result
 1673     is a non-match for a non-anchored pattern. For example, if
 1674     /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
 1675     "m", where previously it did not return a name. A side effect of this
 1676     change is that for partial matches, the last encountered mark name is
 1677     returned, as for non matches. A number of tests that were previously not
 1678     Perl-compatible have been moved into the Perl-compatible test files. The
 1679     refactoring has had the pleasing side effect of removing one argument from
 1680     the match() function, thus reducing its stack requirements.
 1682 19. If the /S+ option was used in pcretest to study a pattern using JIT,
 1683     subsequent uses of /S (without +) incorrectly behaved like /S+.
 1685 21. Retrieve executable code size support for the JIT compiler and fixing
 1686     some warnings.
 1688 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
 1689     not work when the shorter character appeared right at the end of the
 1690     subject string.
 1692 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
 1693     systems.
 1695 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
 1696     output it when the /M option is used in pcretest.
 1698 25. The CheckMan script was not being included in the distribution. Also, added
 1699     an explicit "perl" to run Perl scripts from the PrepareRelease script
 1700     because this is reportedly needed in Windows.
 1702 26. If study data was being save in a file and studying had not found a set of
 1703     "starts with" bytes for the pattern, the data written to the file (though
 1704     never used) was taken from uninitialized memory and so caused valgrind to
 1705     complain.
 1707 27. Updated RunTest.bat as provided by Sheri Pierce.
 1709 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
 1711 29. Computation of memory usage for the table of capturing group names was
 1712     giving an unnecessarily large value.
 1715 Version 8.20 21-Oct-2011
 1716 ------------------------
 1718 1.  Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
 1719     a POSIX class. After further experiments with Perl, which convinced me that
 1720     Perl has bugs and confusions, a closing square bracket is no longer allowed
 1721     in a POSIX name. This bug also affected patterns with classes that started
 1722     with full stops.
 1724 2.  If a pattern such as /(a)b|ac/ is matched against "ac", there is no
 1725     captured substring, but while checking the failing first alternative,
 1726     substring 1 is temporarily captured. If the output vector supplied to
 1727     pcre_exec() was not big enough for this capture, the yield of the function
 1728     was still zero ("insufficient space for captured substrings"). This cannot
 1729     be totally fixed without adding another stack variable, which seems a lot
 1730     of expense for a edge case. However, I have improved the situation in cases
 1731     such as /(a)(b)x|abc/ matched against "abc", where the return code
 1732     indicates that fewer than the maximum number of slots in the ovector have
 1733     been set.
 1735 3.  Related to (2) above: when there are more back references in a pattern than
 1736     slots in the output vector, pcre_exec() uses temporary memory during
 1737     matching, and copies in the captures as far as possible afterwards. It was
 1738     using the entire output vector, but this conflicts with the specification
 1739     that only 2/3 is used for passing back captured substrings. Now it uses
 1740     only the first 2/3, for compatibility. This is, of course, another edge
 1741     case.
 1743 4.  Zoltan Herczeg's just-in-time compiler support has been integrated into the
 1744     main code base, and can be used by building with --enable-jit. When this is
 1745     done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
 1746     runtime --no-jit option is given.
 1748 5.  When the number of matches in a pcre_dfa_exec() run exactly filled the
 1749     ovector, the return from the function was zero, implying that there were
 1750     other matches that did not fit. The correct "exactly full" value is now
 1751     returned.
 1753 6.  If a subpattern that was called recursively or as a subroutine contained
 1754     (*PRUNE) or any other control that caused it to give a non-standard return,
 1755     invalid errors such as "Error -26 (nested recursion at the same subject
 1756     position)" or even infinite loops could occur.
 1758 7.  If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
 1759     computing the minimum length on reaching *ACCEPT, and so ended up with the
 1760     wrong value of 1 rather than 0. Further investigation indicates that
 1761     computing a minimum subject length in the presence of *ACCEPT is difficult
 1762     (think back references, subroutine calls), and so I have changed the code
 1763     so that no minimum is registered for a pattern that contains *ACCEPT.
 1765 8.  If (*THEN) was present in the first (true) branch of a conditional group,
 1766     it was not handled as intended. [But see 16 below.]
 1768 9.  Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
 1769     Sheri Pierce.
 1771 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
 1772     the first byte in a match must be "a".
 1774 11. Change 17 for 8.13 increased the recursion depth for patterns like
 1775     /a(?:.)*?a/ drastically. I've improved things by remembering whether a
 1776     pattern contains any instances of (*THEN). If it does not, the old
 1777     optimizations are restored. It would be nice to do this on a per-group
 1778     basis, but at the moment that is not feasible.
 1780 12. In some environments, the output of pcretest -C is CRLF terminated. This
 1781     broke RunTest's code that checks for the link size. A single white space
 1782     character after the value is now allowed for.
 1784 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
 1785     For "fr", it uses the Windows-specific input and output files.
 1787 14. If (*THEN) appeared in a group that was called recursively or as a
 1788     subroutine, it did not work as intended. [But see next item.]
 1790 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
 1791     pattern fragments (but not containing any | characters). If A and B are
 1792     matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
 1793     was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
 1794     D. In other words, Perl considers parentheses that do not contain any |
 1795     characters to be part of a surrounding alternative, whereas PCRE was
 1796     treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
 1797     differently. PCRE now behaves in the same way as Perl, except in the case
 1798     of subroutine/recursion calls such as (?1) which have in any case always
 1799     been different (but PCRE had them first :-).
 1801 16. Related to 15 above: Perl does not treat the | in a conditional group as
 1802     creating alternatives. Such a group is treated in the same way as an
 1803     ordinary group without any | characters when processing (*THEN). PCRE has
 1804     been changed to match Perl's behaviour.
 1806 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
 1807     RunGrepTest script failed.
 1809 18. Change 22 for version 13 caused atomic groups to use more stack. This is
 1810     inevitable for groups that contain captures, but it can lead to a lot of
 1811     stack use in large patterns. The old behaviour has been restored for atomic
 1812     groups that do not contain any capturing parentheses.
 1814 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
 1815     suppress the check for a minimum subject length at run time. (If it was
 1816     given to pcre_exec() or pcre_dfa_exec() it did work.)
 1818 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
 1819     fail to work when decoding hex characters in data strings in EBCDIC
 1820     environments.
 1822 21. It appears that in at least one Mac OS environment, the isxdigit() function
 1823     is implemented as a macro that evaluates to its argument more than once,
 1824     contravening the C 90 Standard (I haven't checked a later standard). There
 1825     was an instance in pcretest which caused it to go wrong when processing
 1826     \x{...} escapes in subject strings. The has been rewritten to avoid using
 1827     things like p++ in the argument of isxdigit().
 1830 Version 8.13 16-Aug-2011
 1831 ------------------------
 1833 1.  The Unicode data tables have been updated to Unicode 6.0.0.
 1835 2.  Two minor typos in pcre_internal.h have been fixed.
 1837 3.  Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
 1838     pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
 1839     in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
 1841 4.  There were a number of related bugs in the code for matching backrefences
 1842     caselessly in UTF-8 mode when codes for the characters concerned were
 1843     different numbers of bytes. For example, U+023A and U+2C65 are an upper
 1844     and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
 1845     (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
 1846     code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
 1847     2-byte code at the end of the subject (it thought there wasn't enough data
 1848     left).
 1850 5.  Comprehensive information about what went wrong is now returned by
 1851     pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
 1852     as the output vector has at least 2 elements. The offset of the start of
 1853     the failing character and a reason code are placed in the vector.
 1855 6.  When the UTF-8 string check fails for pcre_compile(), the offset that is
 1856     now returned is for the first byte of the failing character, instead of the
 1857     last byte inspected. This is an incompatible change, but I hope it is small
 1858     enough not to be a problem. It makes the returned offset consistent with
 1859     pcre_exec() and pcre_dfa_exec().
 1861 7.  pcretest now gives a text phrase as well as the error number when
 1862     pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
 1863     failure, the offset and reason code are output.
 1865 8.  When \R was used with a maximizing quantifier it failed to skip backwards
 1866     over a \r\n pair if the subsequent match failed. Instead, it just skipped
 1867     back over a single character (\n). This seems wrong (because it treated the
 1868     two characters as a single entity when going forwards), conflicts with the
 1869     documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
 1870     behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
 1871     has been changed.
 1873 9.  Some internal refactoring has changed the processing so that the handling
 1874     of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
 1875     time (the PCRE_DOTALL option was changed this way some time ago: version
 1876     7.7 change 16). This has made it possible to abolish the OP_OPT op code,
 1877     which was always a bit of a fudge. It also means that there is one less
 1878     argument for the match() function, which reduces its stack requirements
 1879     slightly. This change also fixes an incompatibility with Perl: the pattern
 1880     (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
 1882 10. More internal refactoring has drastically reduced the number of recursive
 1883     calls to match() for possessively repeated groups such as (abc)++ when
 1884     using pcre_exec().
 1886 11. While implementing 10, a number of bugs in the handling of groups were
 1887     discovered and fixed:
 1889     (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
 1890     (a|)*(?1) gave a compile-time internal error.
 1891     ((a|)+)+  did not notice that the outer group could match an empty string.
 1892     (^a|^)+   was not marked as anchored.
 1893     (.*a|.*)+ was not marked as matching at start or after a newline.
 1895 12. Yet more internal refactoring has removed another argument from the match()
 1896     function. Special calls to this function are now indicated by setting a
 1897     value in a variable in the "match data" data block.
 1899 13. Be more explicit in pcre_study() instead of relying on "default" for
 1900     opcodes that mean there is no starting character; this means that when new
 1901     ones are added and accidentally left out of pcre_study(), testing should
 1902     pick them up.
 1904 14. The -s option of pcretest has been documented for ages as being an old
 1905     synonym of -m (show memory usage). I have changed it to mean "force study
 1906     for every regex", that is, assume /S for every regex. This is similar to -i
 1907     and -d etc. It's slightly incompatible, but I'm hoping nobody is still
 1908     using it. It makes it easier to run collections of tests with and without
 1909     study enabled, and thereby test pcre_study() more easily. All the standard
 1910     tests are now run with and without -s (but some patterns can be marked as
 1911     "never study" - see 20 below).
 1913 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
 1914     restoration of the capturing data to the outer values was not happening
 1915     correctly.
 1917 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
 1918     empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
 1919     pattern had matched an empty string, and so incorrectly returned a no
 1920     match.
 1922 17. There was optimizing code for the last branch of non-capturing parentheses,
 1923     and also for the obeyed branch of a conditional subexpression, which used
 1924     tail recursion to cut down on stack usage. Unfortunately, now that there is
 1925     the possibility of (*THEN) occurring in these branches, tail recursion is
 1926     no longer possible because the return has to be checked for (*THEN). These
 1927     two optimizations have therefore been removed. [But see 8.20/11 above.]
 1929 18. If a pattern containing \R was studied, it was assumed that \R always
 1930     matched two bytes, thus causing the minimum subject length to be
 1931     incorrectly computed because \R can also match just one byte.
 1933 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
 1934     was incorrectly computed.
 1936 20. If /S is present twice on a test pattern in pcretest input, it now
 1937     *disables* studying, thereby overriding the use of -s on the command line
 1938     (see 14 above). This is necessary for one or two tests to keep the output
 1939     identical in both cases.
 1941 21. When (*ACCEPT) was used in an assertion that matched an empty string and
 1942     PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
 1944 22. When an atomic group that contained a capturing parenthesis was
 1945     successfully matched, but the branch in which it appeared failed, the
 1946     capturing was not being forgotten if a higher numbered group was later
 1947     captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
 1948     group 1 to "a", when in fact it should be unset. This applied to multi-
 1949     branched capturing and non-capturing groups, repeated or not, and also to
 1950     positive assertions (capturing in negative assertions does not happen
 1951     in PCRE) and also to nested atomic groups.
 1953 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
 1954     subject after a captured substring, to make it easier to tell which of a
 1955     number of identical substrings has been captured.
 1957 24. The way atomic groups are processed by pcre_exec() has been changed so that
 1958     if they are repeated, backtracking one repetition now resets captured
 1959     values correctly. For example, if ((?>(a+)b)+aabab) is matched against
 1960     "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
 1961     "aaa". Previously, it would have been "a". As part of this code
 1962     refactoring, the way recursive calls are handled has also been changed.
 1964 25. If an assertion condition captured any substrings, they were not passed
 1965     back unless some other capturing happened later. For example, if
 1966     (?(?=(a))a) was matched against "a", no capturing was returned.
 1968 26. When studying a pattern that contained subroutine calls or assertions,
 1969     the code for finding the minimum length of a possible match was handling
 1970     direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
 1971     group 1 called group 2 while simultaneously a separate group 2 called group
 1972     1). A stack overflow occurred in this case. I have fixed this by limiting
 1973     the recursion depth to 10.
 1975 27. Updated RunTest.bat in the distribution to the version supplied by Tom
 1976     Fortmann. This supports explicit test numbers on the command line, and has
 1977     argument validation and error reporting.
 1979 28. An instance of \X with an unlimited repeat could fail if at any point the
 1980     first character it looked at was a mark character.
 1982 29. Some minor code refactoring concerning Unicode properties and scripts
 1983     should reduce the stack requirement of match() slightly.
 1985 30. Added the '=' option to pcretest to check the setting of unused capturing
 1986     slots at the end of the pattern, which are documented as being -1, but are
 1987     not included in the return count.
 1989 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
 1990     compiled something random. Now it gives a compile-time error (as does
 1991     Perl).
 1993 32. A *MARK encountered during the processing of a positive assertion is now
 1994     recorded and passed back (compatible with Perl).
 1996 33. If --only-matching or --colour was set on a pcregrep call whose pattern
 1997     had alternative anchored branches, the search for a second match in a line
 1998     was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
 1999     matched the line "0102" twice. The same bug affected patterns that started
 2000     with a backwards assertion. For example /\b01|\b02/ also matched "0102"
 2001     twice.
 2003 34. Previously, PCRE did not allow quantification of assertions. However, Perl
 2004     does, and because of capturing effects, quantifying parenthesized
 2005     assertions may at times be useful. Quantifiers are now allowed for
 2006     parenthesized assertions.
 2008 35. A minor code tidy in pcre_compile() when checking options for \R usage.
 2010 36. \g was being checked for fancy things in a character class, when it should
 2011     just be a literal "g".
 2013 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
 2014     appearance of a nested POSIX class supersedes an apparent external class.
 2015     For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
 2016     unescaped square brackets may also appear as part of class names. For
 2017     example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
 2018     more like Perl. (But see 8.20/1 above.)
 2020 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
 2021     was because it thought it was \N{name}, which is not supported).
 2023 39. Add minix to OS list not supporting the -S option in pcretest.
 2025 40. PCRE tries to detect cases of infinite recursion at compile time, but it
 2026     cannot analyze patterns in sufficient detail to catch mutual recursions
 2027     such as ((?1))((?2)). There is now a runtime test that gives an error if a
 2028     subgroup is called recursively as a subpattern for a second time at the
 2029     same position in the subject string. In previous releases this might have
 2030     been caught by the recursion limit, or it might have run out of stack.
 2032 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
 2033     happen only once. PCRE was, however incorrectly giving a compile time error
 2034     "recursive call could loop indefinitely" because it cannot analyze the
 2035     pattern in sufficient detail. The compile time test no longer happens when
 2036     PCRE is compiling a conditional subpattern, but actual runaway loops are
 2037     now caught at runtime (see 40 above).
 2039 42. It seems that Perl allows any characters other than a closing parenthesis
 2040     to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
 2041     has been changed to be the same.
 2043 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
 2044     as not to get warnings when autogen.sh is called. Also changed
 2045     AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
 2047 44. To help people who use pcregrep to scan files containing exceedingly long
 2048     lines, the following changes have been made:
 2050     (a) The default value of the buffer size parameter has been increased from
 2051         8K to 20K. (The actual buffer used is three times this size.)
 2053     (b) The default can be changed by ./configure --with-pcregrep-bufsize when
 2054         PCRE is built.
 2056     (c) A --buffer-size=n option has been added to pcregrep, to allow the size
 2057         to be set at run time.
 2059     (d) Numerical values in pcregrep options can be followed by K or M, for
 2060         example --buffer-size=50K.
 2062     (e) If a line being scanned overflows pcregrep's buffer, an error is now
 2063         given and the return code is set to 2.
 2065 45. Add a pointer to the latest mark to the callout data block.
 2067 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
 2068     partial match of an empty string instead of no match. This was specific to
 2069     the use of ".".
 2071 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
 2072     complete match instead of a partial match. This bug was dependent on both
 2073     the PCRE_UTF8 and PCRE_DOTALL options being set.
 2075 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
 2076     starting byte set, because \b was not being ignored.
 2079 Version 8.12 15-Jan-2011
 2080 ------------------------
 2082 1.  Fixed some typos in the markup of the man pages, and wrote a script that
 2083     checks for such things as part of the documentation building process.
 2085 2.  On a big-endian 64-bit system, pcregrep did not correctly process the
 2086     --match-limit and --recursion-limit options (added for 8.11). In
 2087     particular, this made one of the standard tests fail. (The integer value
 2088     went into the wrong half of a long int.)
 2090 3.  If the --colour option was given to pcregrep with -v (invert match), it
 2091     did strange things, either producing crazy output, or crashing. It should,
 2092     of course, ignore a request for colour when reporting lines that do not
 2093     match.
 2095 4.  Another pcregrep bug caused similar problems if --colour was specified with
 2096     -M (multiline) and the pattern match finished with a line ending.
 2098 5.  In pcregrep, when a pattern that ended with a literal newline sequence was
 2099     matched in multiline mode, the following line was shown as part of the
 2100     match. This seems wrong, so I have changed it.
 2102 6.  Another pcregrep bug in multiline mode, when --colour was specified, caused
 2103     the check for further matches in the same line (so they could be coloured)
 2104     to overrun the end of the current line. If another match was found, it was
 2105     incorrectly shown (and then shown again when found in the next line).
 2107 7.  If pcregrep was compiled under Windows, there was a reference to the
 2108     function pcregrep_exit() before it was defined. I am assuming this was
 2109     the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
 2110     reported by a user. I've moved the definition above the reference.
 2113 Version 8.11 10-Dec-2010
 2114 ------------------------
 2116 1.  (*THEN) was not working properly if there were untried alternatives prior
 2117     to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
 2118     backtracked to try for "b" instead of moving to the next alternative branch
 2119     at the same level (in this case, to look for "c"). The Perl documentation
 2120     is clear that when (*THEN) is backtracked onto, it goes to the "next
 2121     alternative in the innermost enclosing group".
 2123 2.  (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
 2124     such as   (A(*COMMIT)B(*THEN)C|D)  any failure after matching A should
 2125     result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
 2126     (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
 2127     (*THEN).
 2129 3.  If \s appeared in a character class, it removed the VT character from
 2130     the class, even if it had been included by some previous item, for example
 2131     in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
 2132     of \s, but is part of the POSIX "space" class.)
 2134 4.  A partial match never returns an empty string (because you can always
 2135     match an empty string at the end of the subject); however the checking for
 2136     an empty string was starting at the "start of match" point. This has been
 2137     changed to the "earliest inspected character" point, because the returned
 2138     data for a partial match starts at this character. This means that, for
 2139     example, /(?<=abc)def/ gives a partial match for the subject "abc"
 2140     (previously it gave "no match").
 2142 5.  Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
 2143     of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
 2144     previously a full match would be given. However, setting PCRE_PARTIAL_HARD
 2145     has an implication that the given string is incomplete (because a partial
 2146     match is preferred over a full match). For this reason, these items now
 2147     give a partial match in this situation. [Aside: previously, the one case
 2148     /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
 2149     match rather than a full match, which was wrong by the old rules, but is
 2150     now correct.]
 2152 6.  There was a bug in the handling of #-introduced comments, recognized when
 2153     PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
 2154     If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
 2155     UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
 2156     scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
 2157     but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
 2158     places in pcre_compile().
 2160 7.  Related to (6) above, when pcre_compile() was skipping #-introduced
 2161     comments when looking ahead for named forward references to subpatterns,
 2162     the only newline sequence it recognized was NL. It now handles newlines
 2163     according to the set newline convention.
 2165 8.  SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
 2166     former, but used strtoul(), whereas pcretest avoided strtoul() but did not
 2167     cater for a lack of strerror(). These oversights have been fixed.
 2169 9.  Added --match-limit and --recursion-limit to pcregrep.
 2171 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
 2173 11. When the -o option was used, pcregrep was setting a return code of 1, even
 2174     when matches were found, and --line-buffered was not being honoured.
 2176 12. Added an optional parentheses number to the -o and --only-matching options
 2177     of pcregrep.
 2179 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
 2180     can match an empty string. The code to do it in pcretest and pcredemo
 2181     needed fixing:
 2183     (a) When the newline convention was "crlf", pcretest got it wrong, skipping
 2184         only one byte after an empty string match just before CRLF (this case
 2185         just got forgotten; "any" and "anycrlf" were OK).
 2187     (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
 2188         mode when an empty string match preceded an ASCII character followed by
 2189         a non-ASCII character. (The code for advancing by one character rather
 2190         than one byte was nonsense.)
 2192     (c) The pcredemo.c sample program did not have any code at all to handle
 2193         the cases when CRLF is a valid newline sequence.
 2195 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
 2196     as a starting offset was within the subject string. There is now a new
 2197     error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
 2198     negative or greater than the length of the string. In order to test this,
 2199     pcretest is extended to allow the setting of negative starting offsets.
 2201 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
 2202     starting offset points to the beginning of a UTF-8 character was
 2203     unnecessarily clumsy. I tidied it up.
 2205 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
 2206     bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
 2208 17. Nobody had reported that the --include_dir option, which was added in
 2209     release 7.7 should have been called --include-dir (hyphen, not underscore)
 2210     for compatibility with GNU grep. I have changed it to --include-dir, but
 2211     left --include_dir as an undocumented synonym, and the same for
 2212     --exclude-dir, though that is not available in GNU grep, at least as of
 2213     release 2.5.4.
 2215 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
 2216     characters from a string of bytes) have been redefined so as not to use
 2217     loops, in order to improve performance in some environments. At the same
 2218     time, I abstracted some of the common code into auxiliary macros to save
 2219     repetition (this should not affect the compiled code).
 2221 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
 2222     compile-time error is now given if \c is not followed by an ASCII
 2223     character, that is, a byte less than 128. (In EBCDIC mode, the code is
 2224     different, and any byte value is allowed.)
 2226 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
 2227     START_OPTIMIZE option, which is now allowed at compile time - but just
 2228     passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
 2229     to pcregrep and other applications that have no direct access to PCRE
 2230     options. The new /Y option in pcretest sets this option when calling
 2231     pcre_compile().
 2233 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
 2234     back references. Groups containing recursive back references were forced to
 2235     be atomic by that change, but in the case of named groups, the amount of
 2236     memory required was incorrectly computed, leading to "Failed: internal
 2237     error: code overflow". This has been fixed.
 2239 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
 2240     pcretest.c, to avoid build problems in some Borland environments.
 2243 Version 8.10 25-Jun-2010
 2244 ------------------------
 2246 1.  Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
 2247     THEN.
 2249 2.  (*ACCEPT) was not working when inside an atomic group.
 2251 3.  Inside a character class, \B is treated as a literal by default, but
 2252     faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
 2253     causes the error). The code is unchanged, but I tidied the documentation.
 2255 4.  Inside a character class, PCRE always treated \R and \X as literals,
 2256     whereas Perl faults them if its -w option is set. I have changed PCRE so
 2257     that it faults them when PCRE_EXTRA is set.
 2259 5.  Added support for \N, which always matches any character other than
 2260     newline. (It is the same as "." when PCRE_DOTALL is not set.)
 2262 6.  When compiling pcregrep with newer versions of gcc which may have
 2263     FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
 2264     declared with attribute warn_unused_result" were given. Just casting the
 2265     result to (void) does not stop the warnings; a more elaborate fudge is
 2266     needed. I've used a macro to implement this.
 2268 7.  Minor change to pcretest.c to avoid a compiler warning.
 2270 8.  Added four artifical Unicode properties to help with an option to make
 2271     \s etc use properties (see next item). The new properties are: Xan
 2272     (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
 2274 9.  Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
 2275     use Unicode properties. (*UCP) at the start of a pattern can be used to set
 2276     this option. Modified pcretest to add /W to test this facility. Added
 2277     REG_UCP to make it available via the POSIX interface.
 2279 10. Added --line-buffered to pcregrep.
 2281 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
 2282     studied, and the match started with a letter with a code point greater than
 2283     127 whose first byte was different to the first byte of the other case of
 2284     the letter, the other case of this starting letter was not recognized
 2285     (#976).
 2287 12. If a pattern that was studied started with a repeated Unicode property
 2288     test, for example, \p{Nd}+, there was the theoretical possibility of
 2289     setting up an incorrect bitmap of starting bytes, but fortunately it could
 2290     not have actually happened in practice until change 8 above was made (it
 2291     added property types that matched character-matching opcodes).
 2293 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
 2294     possible starting bytes for non-anchored patterns.
 2296 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
 2297     \R, and also a number of cases that involve Unicode properties, both
 2298     explicit and implicit when PCRE_UCP is set.
 2300 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
 2301     input, it could crash or give wrong results if characters with values
 2302     greater than 0xc0 were present in the subject string. (Detail: it assumed
 2303     UTF-8 input when processing these items.)
 2305 16. Added a lot of (int) casts to avoid compiler warnings in systems where
 2306     size_t is 64-bit (#991).
 2308 17. Added a check for running out of memory when PCRE is compiled with
 2309     --disable-stack-for-recursion (#990).
 2311 18. If the last data line in a file for pcretest does not have a newline on
 2312     the end, a newline was missing in the output.
 2314 19. The default pcre_chartables.c file recognizes only ASCII characters (values
 2315     less than 128) in its various bitmaps. However, there is a facility for
 2316     generating tables according to the current locale when PCRE is compiled. It
 2317     turns out that in some environments, 0x85 and 0xa0, which are Unicode space
 2318     characters, are recognized by isspace() and therefore were getting set in
 2319     these tables, and indeed these tables seem to approximate to ISO 8859. This
 2320     caused a problem in UTF-8 mode when pcre_study() was used to create a list
 2321     of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
 2322     which of course cannot start UTF-8 characters. I have changed the code so
 2323     that only real ASCII characters (less than 128) and the correct starting
 2324     bytes for UTF-8 encodings are set for characters greater than 127 when in
 2325     UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
 2326     altogether.)
 2328 20. Added the /T option to pcretest so as to be able to run tests with non-
 2329     standard character tables, thus making it possible to include the tests
 2330     used for 19 above in the standard set of tests.
 2332 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
 2333     reference to a subpattern the other side of a comment that contains an
 2334     opening parenthesis caused either an internal compiling error, or a
 2335     reference to the wrong subpattern.
 2338 Version 8.02 19-Mar-2010
 2339 ------------------------
 2341 1.  The Unicode data tables have been updated to Unicode 5.2.0.
 2343 2.  Added the option --libs-cpp to pcre-config, but only when C++ support is
 2344     configured.
 2346 3.  Updated the licensing terms in the pcregexp.pas file, as agreed with the
 2347     original author of that file, following a query about its status.
 2349 4.  On systems that do not have stdint.h (e.g. Solaris), check for and include
 2350     inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
 2352 5.  A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
 2353     quantifier applied to a forward-referencing subroutine call, could compile
 2354     incorrect code or give the error "internal error: previously-checked
 2355     referenced subpattern not found".
 2357 6.  Both MS Visual Studio and Symbian OS have problems with initializing
 2358     variables to point to external functions. For these systems, therefore,
 2359     pcre_malloc etc. are now initialized to local functions that call the
 2360     relevant global functions.
 2362 7.  There were two entries missing in the vectors called coptable and poptable
 2363     in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
 2364     I've fixed the data, and added a kludgy way of testing at compile time that
 2365     the lengths are correct (equal to the number of opcodes).
 2367 8.  Following on from 7, I added a similar kludge to check the length of the
 2368     eint vector in pcreposix.c.
 2370 9.  Error texts for pcre_compile() are held as one long string to avoid too
 2371     much relocation at load time. To find a text, the string is searched,
 2372     counting zeros. There was no check for running off the end of the string,
 2373     which could happen if a new error number was added without updating the
 2374     string.
 2376 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
 2378 11. \K was not working if it appeared in an atomic group or in a group that
 2379     was called as a "subroutine", or in an assertion. Perl 5.11 documents that
 2380     \K is "not well defined" if used in an assertion. PCRE now accepts it if
 2381     the assertion is positive, but not if it is negative.
 2383 12. Change 11 fortuitously reduced the size of the stack frame used in the
 2384     "match()" function of pcre_exec.c by one pointer. Forthcoming
 2385     implementation of support for (*MARK) will need an extra pointer on the
 2386     stack; I have reserved it now, so that the stack frame size does not
 2387     decrease.
 2389 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
 2390     item in branch that calls a recursion is a subroutine call - as in the
 2391     second branch in the above example - was incorrectly given the compile-
 2392     time error "recursive call could loop indefinitely" because pcre_compile()
 2393     was not correctly checking the subroutine for matching a non-empty string.
 2395 14. The checks for overrunning compiling workspace could trigger after an
 2396     overrun had occurred. This is a "should never occur" error, but it can be
 2397     triggered by pathological patterns such as hundreds of nested parentheses.
 2398     The checks now trigger 100 bytes before the end of the workspace.
 2400 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
 2403 Version 8.01 19-Jan-2010
 2404 ------------------------
 2406 1.  If a pattern contained a conditional subpattern with only one branch (in
 2407     particular, this includes all (*DEFINE) patterns), a call to pcre_study()
 2408     computed the wrong minimum data length (which is of course zero for such
 2409     subpatterns). This could cause incorrect "no match" results.
 2411 2.  For patterns such as (?i)a(?-i)b|c where an option setting at the start of
 2412     the pattern is reset in the first branch, pcre_compile() failed with
 2413     "internal error: code overflow at offset...". This happened only when
 2414     the reset was to the original external option setting. (An optimization
 2415     abstracts leading options settings into an external setting, which was the
 2416     cause of this.)
 2418 3.  A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
 2419     of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
 2420     assertion pattern did not match (meaning that the assertion was true), it
 2421     was incorrectly treated as false if the SKIP had been reached during the
 2422     matching. This also applied to assertions used as conditions.
 2424 4.  If an item that is not supported by pcre_dfa_exec() was encountered in an
 2425     assertion subpattern, including such a pattern used as a condition,
 2426     unpredictable results occurred, instead of the error return
 2429 5.  The C++ GlobalReplace function was not working like Perl for the special
 2430     situation when an empty string is matched. It now does the fancy magic
 2431     stuff that is necessary.
 2433 6.  In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
 2434     removed. (These were left over from very, very early versions of PCRE.)
 2436 7.  Some cosmetic changes to the code to make life easier when compiling it
 2437     as part of something else:
 2439     (a) Change DEBUG to PCRE_DEBUG.
 2441     (b) In pcre_compile(), rename the member of the "branch_chain" structure
 2442         called "current" as "current_branch", to prevent a collision with the
 2443         Linux macro when compiled as a kernel module.
 2445     (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
 2446         prevent a collision with the Linux macro when compiled as a kernel
 2447         module.
 2449 8.  In pcre_compile() there are some checks for integer overflows that used to
 2450     cast potentially large values to (double). This has been changed to that
 2451     when building, a check for int64_t is made, and if it is found, it is used
 2452     instead, thus avoiding the use of floating point arithmetic. (There is no
 2453     other use of FP in PCRE.) If int64_t is not found, the fallback is to
 2454     double.
 2456 9.  Added two casts to avoid signed/unsigned warnings from VS Studio Express
 2457     2005 (difference between two addresses compared to an unsigned value).
 2459 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
 2460     custom one, because of the following reported problem in Windows:
 2462       - libbz2 uses the Pascal calling convention (WINAPI) for the functions
 2463           under Win32.
 2464       - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
 2465           therefore missing the function definition.
 2466       - The compiler thus generates a "C" signature for the test function.
 2467       - The linker fails to find the "C" function.
 2468       - PCRE fails to configure if asked to do so against libbz2.
 2470 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
 2471     messages were output:
 2473       Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
 2474       rerunning libtoolize, to keep the correct libtool macros in-tree.
 2475       Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
 2477     I have done both of these things.
 2479 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
 2480     most of the time, it *can* run out if it is given a pattern that contains a
 2481     runaway infinite recursion. I updated the discussion in the pcrestack man
 2482     page.
 2484 13. Now that we have gone to the x.xx style of version numbers, the minor
 2485     version may start with zero. Using 08 or 09 is a bad idea because users
 2486     might check the value of PCRE_MINOR in their code, and 08 or 09 may be
 2487     interpreted as invalid octal numbers. I've updated the previous comment in
 2488     configure.ac, and also added a check that gives an error if 08 or 09 are
 2489     used.
 2491 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
 2492     causing partial matching to fail when the end of the subject matched \W
 2493     in a UTF-8 pattern where \W was quantified with a minimum of 3.
 2495 15. There were some discrepancies between the declarations in pcre_internal.h
 2496     of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
 2497     their definitions. The declarations used "const uschar *" and the
 2498     definitions used USPTR. Even though USPTR is normally defined as "const
 2499     unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
 2500     reported that: "This difference in casting confuses some C++ compilers, for
 2501     example, SunCC recognizes above declarations as different functions and
 2502     generates broken code for hbpcre." I have changed the declarations to use
 2503     USPTR.
 2505 16. GNU libtool is named differently on some systems. The autogen.sh script now
 2506     tries several variants such as glibtoolize (MacOSX) and libtoolize1x
 2507     (FreeBSD).
 2509 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
 2510     (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
 2511     comment: "Figure out how to create a longlong from a string: strtoll and
 2512     equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
 2513     instance, but it only takes 2 args instead of 3!"
 2515 18. A subtle bug concerned with back references has been fixed by a change of
 2516     specification, with a corresponding code fix. A pattern such as
 2517     ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
 2518     refers, was giving matches when it shouldn't. For example, xa=xaaa would
 2519     match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
 2520     same bug. Such groups have to be quantified to be useful, or contained
 2521     inside another quantified group. (If there's no repetition, the reference
 2522     can never match.) The problem arises because, having left the group and
 2523     moved on to the rest of the pattern, a later failure that backtracks into
 2524     the group uses the captured value from the final iteration of the group
 2525     rather than the correct earlier one. I have fixed this in PCRE by forcing
 2526     any group that contains a reference to itself to be an atomic group; that
 2527     is, there cannot be any backtracking into it once it has completed. This is
 2528     similar to recursive and subroutine calls.
 2531 Version 8.00 19-Oct-09
 2532 ----------------------
 2534 1.  The table for translating pcre_compile() error codes into POSIX error codes
 2535     was out-of-date, and there was no check on the pcre_compile() error code
 2536     being within the table. This could lead to an OK return being given in
 2537     error.
 2539 2.  Changed the call to open a subject file in pcregrep from fopen(pathname,
 2540     "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
 2541     in a Windows environment.
 2543 3.  The pcregrep --count option prints the count for each file even when it is
 2544     zero, as does GNU grep. However, pcregrep was also printing all files when
 2545     --files-with-matches was added. Now, when both options are given, it prints
 2546     counts only for those files that have at least one match. (GNU grep just
 2547     prints the file name in this circumstance, but including the count seems
 2548     more useful - otherwise, why use --count?) Also ensured that the
 2549     combination -clh just lists non-zero counts, with no names.
 2551 4.  The long form of the pcregrep -F option was incorrectly implemented as
 2552     --fixed_strings instead of --fixed-strings. This is an incompatible change,
 2553     but it seems right to fix it, and I didn't think it was worth preserving
 2554     the old behaviour.
 2556 5.  The command line items --regex=pattern and --regexp=pattern were not
 2557     recognized by pcregrep, which required --regex pattern or --regexp pattern
 2558     (with a space rather than an '='). The man page documented the '=' forms,
 2559     which are compatible with GNU grep; these now work.
 2561 6.  No libpcreposix.pc file was created for pkg-config; there was just
 2562     libpcre.pc and libpcrecpp.pc. The omission has been rectified.
 2564 7.  Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
 2565     when UCP support is not needed, by modifying the Python script that
 2566     generates it from Unicode data files. This should not matter if the module
 2567     is correctly used as a library, but I received one complaint about 50K of
 2568     unwanted data. My guess is that the person linked everything into his
 2569     program rather than using a library. Anyway, it does no harm.
 2571 8.  A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
 2572     was a minimum greater than 1 for a wide character in a possessive
 2573     repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
 2574     which had an unlimited repeat of a nested, fixed maximum repeat of a wide
 2575     character. Chaos in the form of incorrect output or a compiling loop could
 2576     result.
 2578 9.  The restrictions on what a pattern can contain when partial matching is
 2579     requested for pcre_exec() have been removed. All patterns can now be
 2580     partially matched by this function. In addition, if there are at least two
 2581     slots in the offset vector, the offset of the earliest inspected character
 2582     for the match and the offset of the end of the subject are set in them when
 2583     PCRE_ERROR_PARTIAL is returned.
 2585 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
 2586     synonymous with PCRE_PARTIAL, for backwards compatibility, and
 2587     PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
 2588     and may be more useful for multi-segment matching.
 2590 11. Partial matching with pcre_exec() is now more intuitive. A partial match
 2591     used to be given if ever the end of the subject was reached; now it is
 2592     given only if matching could not proceed because another character was
 2593     needed. This makes a difference in some odd cases such as Z(*FAIL) with the
 2594     string "Z", which now yields "no match" instead of "partial match". In the
 2595     case of pcre_dfa_exec(), "no match" is given if every matching path for the
 2596     final character ended with (*FAIL).
 2598 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
 2599     if the pattern had a "must contain" character that was already found in the
 2600     earlier partial match, unless partial matching was again requested. For
 2601     example, with the pattern /dog.(body)?/, the "must contain" character is
 2602     "g". If the first part-match was for the string "dog", restarting with
 2603     "sbody" failed. This bug has been fixed.
 2605 13. The string returned by pcre_dfa_exec() after a partial match has been
 2606     changed so that it starts at the first inspected character rather than the
 2607     first character of the match. This makes a difference only if the pattern
 2608     starts with a lookbehind assertion or \b or \B (\K is not supported by
 2609     pcre_dfa_exec()). It's an incompatible change, but it makes the two
 2610     matching functions compatible, and I think it's the right thing to do.
 2612 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
 2613     so that the demonstration program is easily available in environments where
 2614     PCRE has not been installed from source.
 2616 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
 2617     libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
 2618     library.
 2620 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
 2621     It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
 2622     is not the first non-POSIX option to be added. Clearly some people find
 2623     these options useful.
 2625 17. If a caller to the POSIX matching function regexec() passes a non-zero
 2626     value for nmatch with a NULL value for pmatch, the value of
 2627     nmatch is forced to zero.
 2629 18. RunGrepTest did not have a test for the availability of the -u option of
 2630     the diff command, as RunTest does. It now checks in the same way as
 2631     RunTest, and also checks for the -b option.
 2633 19. If an odd number of negated classes containing just a single character
 2634     interposed, within parentheses, between a forward reference to a named
 2635     subpattern and the definition of the subpattern, compilation crashed with
 2636     an internal error, complaining that it could not find the referenced
 2637     subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
 2638     [The bug was that it was starting one character too far in when skipping
 2639     over the character class, thus treating the ] as data rather than
 2640     terminating the class. This meant it could skip too much.]
 2642 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
 2643     /g option in pcretest when the pattern contains \K, which makes it possible
 2644     to have an empty string match not at the start, even when the pattern is
 2645     anchored. Updated pcretest and pcredemo to use this option.
 2647 21. If the maximum number of capturing subpatterns in a recursion was greater
 2648     than the maximum at the outer level, the higher number was returned, but
 2649     with unset values at the outer level. The correct (outer level) value is
 2650     now given.
 2652 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
 2653     PCRE did not set those parentheses (unlike Perl). I have now found a way to
 2654     make it do so. The string so far is captured, making this feature
 2655     compatible with Perl.
 2657 23. The tests have been re-organized, adding tests 11 and 12, to make it
 2658     possible to check the Perl 5.10 features against Perl 5.10.
 2660 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
 2661     pattern matches a fixed length string. PCRE did not allow this; now it
 2662     does. Neither allows recursion.
 2664 25. I finally figured out how to implement a request to provide the minimum
 2665     length of subject string that was needed in order to match a given pattern.
 2666     (It was back references and recursion that I had previously got hung up
 2667     on.) This code has now been added to pcre_study(); it finds a lower bound
 2668     to the length of subject needed. It is not necessarily the greatest lower
 2669     bound, but using it to avoid searching strings that are too short does give
 2670     some useful speed-ups. The value is available to calling programs via
 2671     pcre_fullinfo().
 2673 26. While implementing 25, I discovered to my embarrassment that pcretest had
 2674     not been passing the result of pcre_study() to pcre_dfa_exec(), so the
 2675     study optimizations had never been tested with that matching function.
 2676     Oops. What is worse, even when it was passed study data, there was a bug in
 2677     pcre_dfa_exec() that meant it never actually used it. Double oops. There
 2678     were also very few tests of studied patterns with pcre_dfa_exec().
 2680 27. If (?| is used to create subpatterns with duplicate numbers, they are now
 2681     allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
 2682     on the other side of the coin, they are no longer allowed to have different
 2683     names, because these cannot be distinguished in PCRE, and this has caused
 2684     confusion. (This is a difference from Perl.)
 2686 28. When duplicate subpattern names are present (necessarily with different
 2687     numbers, as required by 27 above), and a test is made by name in a
 2688     conditional pattern, either for a subpattern having been matched, or for
 2689     recursion in such a pattern, all the associated numbered subpatterns are
 2690     tested, and the overall condition is true if the condition is true for any
 2691     one of them. This is the way Perl works, and is also more like the way
 2692     testing by number works.
 2695 Version 7.9 11-Apr-09
 2696 ---------------------
 2698 1.  When building with support for bzlib/zlib (pcregrep) and/or readline
 2699     (pcretest), all targets were linked against these libraries. This included
 2700     libpcre, libpcreposix, and libpcrecpp, even though they do not use these
 2701     libraries. This caused unwanted dependencies to be created. This problem
 2702     has been fixed, and now only pcregrep is linked with bzlib/zlib and only
 2703     pcretest is linked with readline.
 2705 2.  The "typedef int BOOL" in pcre_internal.h that was included inside the
 2706     "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
 2707     moved outside it again, because FALSE and TRUE are already defined in AIX,
 2708     but BOOL is not.
 2710 3.  The pcre_config() function was treating the PCRE_MATCH_LIMIT and
 2711     PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
 2713 4.  The pcregrep documentation said spaces were inserted as well as colons (or
 2714     hyphens) following file names and line numbers when outputting matching
 2715     lines. This is not true; no spaces are inserted. I have also clarified the
 2716     wording for the --colour (or --color) option.
 2718 5.  In pcregrep, when --colour was used with -o, the list of matching strings
 2719     was not coloured; this is different to GNU grep, so I have changed it to be
 2720     the same.
 2722 6.  When --colo(u)r was used in pcregrep, only the first matching substring in
 2723     each matching line was coloured. Now it goes on to look for further matches
 2724     of any of the test patterns, which is the same behaviour as GNU grep.
 2726 7.  A pattern that could match an empty string could cause pcregrep to loop; it
 2727     doesn't make sense to accept an empty string match in pcregrep, so I have
 2728     locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
 2729     seems to be how GNU grep behaves. [But see later change 40 for release
 2730     8.33.]
 2732 8.  The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
 2733     start or after a newline", because the conditional assertion was not being
 2734     correctly handled. The rule now is that both the assertion and what follows
 2735     in the first alternative must satisfy the test.
 2737 9.  If auto-callout was enabled in a pattern with a conditional group whose
 2738     condition was an assertion, PCRE could crash during matching, both with
 2739     pcre_exec() and pcre_dfa_exec().
 2741 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
 2742     used for matching.
 2744 11. Unicode property support in character classes was not working for
 2745     characters (bytes) greater than 127 when not in UTF-8 mode.
 2747 12. Added the -M command line option to pcretest.
 2749 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
 2751 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
 2753 16. Added comments and documentation about mis-use of no_arg in the C++
 2754     wrapper.
 2756 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
 2757     from Martin Jerabek that uses macro names for all relevant character and
 2758     string constants.
 2760 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
 2761     SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
 2762     SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
 2763     these, but not everybody uses configure.
 2765 19. A conditional group that had only one branch was not being correctly
 2766     recognized as an item that could match an empty string. This meant that an
 2767     enclosing group might also not be so recognized, causing infinite looping
 2768     (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
 2769     with the subject "ab", where knowledge that the repeated group can match
 2770     nothing is needed in order to break the loop.
 2772 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
 2773     exec(), but without supplying a callout function, matching went wrong.
 2775 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
 2776     leak if the size of the offset vector was greater than 30. When the vector
 2777     is smaller, the saved offsets during recursion go onto a local stack
 2778     vector, but for larger vectors malloc() is used. It was failing to free
 2779     when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
 2780     error, in fact).
 2782 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
 2783     heapframe that is used only when UTF-8 support is enabled. This caused no
 2784     problem, but was untidy.
 2786 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
 2787     CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
 2788     included within another project.
 2790 24. Steven Van Ingelgem's patches to add more options to the CMake support,
 2791     slightly modified by me:
 2793       (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
 2794           not building pcregrep.
 2796       (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
 2797           if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
 2799 25. Forward references, both numeric and by name, in patterns that made use of
 2800     duplicate group numbers, could behave incorrectly or give incorrect errors,
 2801     because when scanning forward to find the reference group, PCRE was not
 2802     taking into account the duplicate group numbers. A pattern such as
 2803     ^X(?3)(a)(?|(b)|(q))(Y) is an example.
 2805 26. Changed a few more instances of "const unsigned char *" to USPTR, making
 2806     the feature of a custom pointer more persuasive (as requested by a user).
 2808 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
 2809     pcretest.c, inside #ifndefs, because it seems they are sometimes already
 2810     pre-defined.
 2812 28. Added support for (*UTF8) at the start of a pattern.
 2814 29. Arrange for flags added by the "release type" setting in CMake to be shown
 2815     in the configuration summary.
 2818 Version 7.8 05-Sep-08
 2819 ---------------------
 2821 1.  Replaced UCP searching code with optimized version as implemented for Ad
 2822     Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
 2823     stage table and inline lookup instead of a function, giving speed ups of 2
 2824     to 5 times on some simple patterns that I tested. Permission was given to
 2825     distribute the MultiStage2.py script that generates the tables (it's not in
 2826     the tarball, but is in the Subversion repository).
 2828 2.  Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
 2829     scripts.
 2831 3.  Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
 2832     a group with a zero qualifier. The result of the study could be incorrect,
 2833     or the function might crash, depending on the pattern.
 2835 4.  Caseless matching was not working for non-ASCII characters in back
 2836     references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
 2837     It now works when Unicode Property Support is available.
 2839 5.  In pcretest, an escape such as \x{de} in the data was always generating
 2840     a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
 2841     non-UTF-8 mode. If the value is greater than 255, it gives a warning about
 2842     truncation.
 2844 6.  Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
 2846 7.  Added two (int) casts to pcregrep when printing the difference of two
 2847     pointers, in case they are 64-bit values.
 2849 8.  Added comments about Mac OS X stack usage to the pcrestack man page and to
 2850     test 2 if it fails.
 2852 9.  Added PCRE_CALL_CONVENTION just before the names of all exported functions,
 2853     and a #define of that name to empty if it is not externally set. This is to
 2854     allow users of MSVC to set it if necessary.
 2856 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
 2857     the convenience functions in the pcre_get.c source file.
 2859 11. An option change at the start of a pattern that had top-level alternatives
 2860     could cause overwriting and/or a crash. This command provoked a crash in
 2861     some environments:
 2863       printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
 2865     This potential security problem was recorded as CVE-2008-2371.
 2867 12. For a pattern where the match had to start at the beginning or immediately
 2868     after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
 2869     pcre_dfa_exec() could read past the end of the passed subject if there was
 2870     no match. To help with detecting such bugs (e.g. with valgrind), I modified
 2871     pcretest so that it places the subject at the end of its malloc-ed buffer.
 2873 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
 2874     exec() might read past the end of the data buffer in UTF-8 mode.
 2876 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
 2877     the data contained the byte 0x85 as part of a UTF-8 character within its
 2878     first line. This applied both to normal and DFA matching.
 2880 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
 2881     /^[^d]*?$/8 failed to match "abc".
 2883 16. Added a missing copyright notice to pcrecpp_internal.h.
 2885 17. Make it more clear in the documentation that values returned from
 2886     pcre_exec() in ovector are byte offsets, not character counts.
 2888 18. Tidied a few places to stop certain compilers from issuing warnings.
 2890 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
 2891     supplied by Stefan Weber. I made a further small update for 7.8 because
 2892     there is a change of source arrangements: the pcre_searchfuncs.c module is
 2893     replaced by pcre_ucd.c.
 2896 Version 7.7 07-May-08
 2897 ---------------------
 2899 1.  Applied Craig's patch to sort out a long long problem: "If we can't convert
 2900     a string to a long long, pretend we don't even have a long long." This is
 2901     done by checking for the strtoq, strtoll, and _strtoi64 functions.
 2903 2.  Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
 2904     pre-7.6 versions, which defined a global no_arg variable instead of putting
 2905     it in the RE class. (See also #8 below.)
 2907 3.  Remove a line of dead code, identified by coverity and reported by Nuno
 2908     Lopes.
 2910 4.  Fixed two related pcregrep bugs involving -r with --include or --exclude:
 2912     (1) The include/exclude patterns were being applied to the whole pathnames
 2913         of files, instead of just to the final components.
 2915     (2) If there was more than one level of directory, the subdirectories were
 2916         skipped unless they satisfied the include/exclude conditions. This is
 2917         inconsistent with GNU grep (and could even be seen as contrary to the
 2918         pcregrep specification - which I improved to make it absolutely clear).
 2919         The action now is always to scan all levels of directory, and just
 2920         apply the include/exclude patterns to regular files.
 2922 5.  Added the --include_dir and --exclude_dir patterns to pcregrep, and used
 2923     --exclude_dir in the tests to avoid scanning .svn directories.
 2925 6.  Applied Craig's patch to the QuoteMeta function so that it escapes the
 2926     NUL character as backslash + 0 rather than backslash + NUL, because PCRE
 2927     doesn't support NULs in patterns.
 2929 7.  Added some missing "const"s to declarations of static tables in
 2930     pcre_compile.c and pcre_dfa_exec.c.
 2932 8.  Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
 2933     caused by fix #2  above. (Subsequently also a second patch to fix the
 2934     first patch. And a third patch - this was a messy problem.)
 2936 9.  Applied Craig's patch to remove the use of push_back().
 2938 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
 2939     matching function regexec().
 2941 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
 2942     which, however, unlike Perl's \g{...}, are subroutine calls, not back
 2943     references. PCRE supports relative numbers with this syntax (I don't think
 2944     Oniguruma does).
 2946 12. Previously, a group with a zero repeat such as (...){0} was completely
 2947     omitted from the compiled regex. However, this means that if the group
 2948     was called as a subroutine from elsewhere in the pattern, things went wrong
 2949     (an internal error was given). Such groups are now left in the compiled
 2950     pattern, with a new opcode that causes them to be skipped at execution
 2951     time.
 2953 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
 2954     to the way PCRE behaves:
 2956     (a) A lone ] character is dis-allowed (Perl treats it as data).
 2958     (b) A back reference to an unmatched subpattern matches an empty string
 2959         (Perl fails the current match path).
 2961     (c) A data ] in a character class must be notated as \] because if the
 2962         first data character in a class is ], it defines an empty class. (In
 2963         Perl it is not possible to have an empty class.) The empty class []
 2964         never matches; it forces failure and is equivalent to (*FAIL) or (?!).
 2965         The negative empty class [^] matches any one character, independently
 2966         of the DOTALL setting.
 2968 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
 2969     non-existent subpattern following a character class starting with ']' and
 2970     containing () gave an internal compiling error instead of "reference to
 2971     non-existent subpattern". Fortunately, when the pattern did exist, the
 2972     compiled code was correct. (When scanning forwards to check for the
 2973     existence of the subpattern, it was treating the data ']' as terminating
 2974     the class, so got the count wrong. When actually compiling, the reference
 2975     was subsequently set up correctly.)
 2977 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
 2978     it was being rejected as not supported by pcre_dfa_exec(), even though
 2979     other assertions are supported. I have made pcre_dfa_exec() support
 2980     (*FAIL).
 2982 16. The implementation of 13c above involved the invention of a new opcode,
 2983     OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
 2984     cannot be changed at match time, I realized I could make a small
 2985     improvement to matching performance by compiling OP_ALLANY instead of
 2986     OP_ANY for "." when DOTALL was set, and then removing the runtime tests
 2987     on the OP_ANY path.
 2989 17. Compiling pcretest on Windows with readline support failed without the
 2990     following two fixes: (1) Make the unistd.h include conditional on
 2991     HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
 2993 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
 2994     ncurses library to be included for pcretest when ReadLine support is
 2995     requested, but also to allow for it to be overridden. This patch came from
 2996     Daniel Bergström.
 2998 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
 2999     as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
 3000     any errors with the current Unicode tables. Thanks to Peter Kankowski for
 3001     spotting this.
 3004 Version 7.6 28-Jan-08
 3005 ---------------------
 3007 1.  A character class containing a very large number of characters with
 3008     codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
 3009     overflow.
 3011 2.  Patch to cut out the "long long" test in pcrecpp_unittest when
 3012     HAVE_LONG_LONG is not defined.
 3014 3.  Applied Christian Ehrlicher's patch to update the CMake build files to
 3015     bring them up to date and include new features. This patch includes:
 3017     - Fixed PH's badly added libz and libbz2 support.
 3018     - Fixed a problem with static linking.
 3019     - Added pcredemo. [But later removed - see 7 below.]
 3020     - Fixed dftables problem and added an option.
 3021     - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
 3022         HAVE_LONG_LONG.
 3023     - Added readline support for pcretest.
 3024     - Added an listing of the option settings after cmake has run.
 3026 4.  A user submitted a patch to Makefile that makes it easy to create
 3027     "pcre.dll" under mingw when using Configure/Make. I added stuff to
 3028     Makefile.am that cause it to include this special target, without
 3029     affecting anything else. Note that the same mingw target plus all
 3030     the other distribution libraries and programs are now supported
 3031     when configuring with CMake (see 6 below) instead of with
 3032     Configure/Make.
 3034 5.  Applied Craig's patch that moves no_arg into the RE class in the C++ code.
 3035     This is an attempt to solve the reported problem "pcrecpp::no_arg is not
 3036     exported in the Windows port". It has not yet been confirmed that the patch
 3037     solves the problem, but it does no harm.
 3039 6.  Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
 3040     NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
 3041     with CMake, and also correct the comment about stack recursion.
 3043 7.  Remove the automatic building of pcredemo from the ./configure system and
 3044     from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
 3045     of a program that users should build themselves after PCRE is installed, so
 3046     building it automatically is not really right. What is more, it gave
 3047     trouble in some build environments.
 3049 8.  Further tidies to CMakeLists.txt from Sheri and Christian.
 3052 Version 7.5 10-Jan-08
 3053 ---------------------
 3055 1.  Applied a patch from Craig: "This patch makes it possible to 'ignore'
 3056     values in parens when parsing an RE using the C++ wrapper."
 3058 2.  Negative specials like \S did not work in character classes in UTF-8 mode.
 3059     Characters greater than 255 were excluded from the class instead of being
 3060     included.
 3062 3.  The same bug as (2) above applied to negated POSIX classes such as
 3063     [:^space:].
 3065 4.  PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
 3066     defined or documented. It seems to have been a typo for PCRE_STATIC, so
 3067     I have changed it.
 3069 5.  The construct (?&) was not diagnosed as a syntax error (it referenced the
 3070     first named subpattern) and a construct such as (?&a) would reference the
 3071     first named subpattern whose name started with "a" (in other words, the
 3072     length check was missing). Both these problems are fixed. "Subpattern name
 3073     expected" is now given for (?&) (a zero-length name), and this patch also
 3074     makes it give the same error for \k'' (previously it complained that that
 3075     was a reference to a non-existent subpattern).
 3077 6.  The erroneous patterns (?+-a) and (?-+a) give different error messages;
 3078     this is right because (?- can be followed by option settings as well as by
 3079     digits. I have, however, made the messages clearer.
 3081 7.  Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
 3082     than the number used in the conditional) now cause a compile-time error.
 3083     This is actually not compatible with Perl, which accepts such patterns, but
 3084     treats the conditional as always being FALSE (as PCRE used to), but it
 3085     seems to me that giving a diagnostic is better.
 3087 8.  Change "alphameric" to the more common word "alphanumeric" in comments
 3088     and messages.
 3090 9.  Fix two occurrences of "backslash" in comments that should have been
 3091     "backspace".
 3093 10. Remove two redundant lines of code that can never be obeyed (their function
 3094     was moved elsewhere).
 3096 11. The program that makes PCRE's Unicode character property table had a bug
 3097     which caused it to generate incorrect table entries for sequences of
 3098     characters that have the same character type, but are in different scripts.
 3099     It amalgamated them into a single range, with the script of the first of
 3100     them. In other words, some characters were in the wrong script. There were
 3101     thirteen such cases, affecting characters in the following ranges:
 3103       U+002b0 - U+002c1
 3104       U+0060c - U+0060d
 3105       U+0061e - U+00612
 3106       U+0064b - U+0065e
 3107       U+0074d - U+0076d
 3108       U+01800 - U+01805
 3109       U+01d00 - U+01d77
 3110       U+01d9b - U+01dbf
 3111       U+0200b - U+0200f
 3112       U+030fc - U+030fe
 3113       U+03260 - U+0327f
 3114       U+0fb46 - U+0fbb1
 3115       U+10450 - U+1049d
 3117 12. The -o option (show only the matching part of a line) for pcregrep was not
 3118     compatible with GNU grep in that, if there was more than one match in a
 3119     line, it showed only the first of them. It now behaves in the same way as
 3120     GNU grep.
 3122 13. If the -o and -v options were combined for pcregrep, it printed a blank
 3123     line for every non-matching line. GNU grep prints nothing, and pcregrep now
 3124     does the same. The return code can be used to tell if there were any
 3125     non-matching lines.
 3127 14. Added --file-offsets and --line-offsets to pcregrep.
 3129 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
 3130     infinitely looping recursion. The bug was that positive lookaheads were not
 3131     being skipped when checking for a possible empty match (negative lookaheads
 3132     and both kinds of lookbehind were skipped).
 3134 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
 3135     inclusion of <windows.h> to before rather than after the definition of
 3136     INVALID_FILE_ATTRIBUTES (patch from David Byron).
 3138 17. Specifying a possessive quantifier with a specific limit for a Unicode
 3139     character property caused pcre_compile() to compile bad code, which led at
 3140     runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
 3141     are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
 3142     caused the error; without that there was no problem.
 3144 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
 3146 19. Added --enable-pcretest-libreadline.
 3148 20. In pcrecpp.cc, the variable 'count' was incremented twice in
 3149     RE::GlobalReplace(). As a result, the number of replacements returned was
 3150     double what it should be. I removed one of the increments, but Craig sent a
 3151     later patch that removed the other one (the right fix) and added unit tests
 3152     that check the return values (which was not done before).
 3154 21. Several CMake things:
 3156     (1) Arranged that, when cmake is used on Unix, the libraries end up with
 3157         the names libpcre and libpcreposix, not just pcre and pcreposix.
 3159     (2) The above change means that pcretest and pcregrep are now correctly
 3160         linked with the newly-built libraries, not previously installed ones.
 3164 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
 3165     crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
 3166     UTF-8 newline character). The key issue is that the pattern starts .*;
 3167     this means that the match must be either at the beginning, or after a
 3168     newline. The bug was in the code for advancing after a failed match and
 3169     checking that the new position followed a newline. It was not taking
 3170     account of UTF-8 characters correctly.
 3172 23. PCRE was behaving differently from Perl in the way it recognized POSIX
 3173     character classes. PCRE was not treating the sequence [:...:] as a
 3174     character class unless the ... were all letters. Perl, however, seems to
 3175     allow any characters between [: and :], though of course it rejects as
 3176     unknown any "names" that contain non-letters, because all the known class
 3177     names consist only of letters. Thus, Perl gives an error for [[:1234:]],
 3178     for example, whereas PCRE did not - it did not recognize a POSIX character
 3179     class. This seemed a bit dangerous, so the code has been changed to be
 3180     closer to Perl. The behaviour is not identical to Perl, because PCRE will
 3181     diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
 3182     treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
 3183     Perl does, and where it didn't before.
 3185 24. Rewrite so as to remove the single use of %n from pcregrep because in some
 3186     Windows environments %n is disabled by default.
 3189 Version 7.4 21-Sep-07
 3190 ---------------------
 3192 1.  Change 7.3/28 was implemented for classes by looking at the bitmap. This
 3193     means that a class such as [\s] counted as "explicit reference to CR or
 3194     LF". That isn't really right - the whole point of the change was to try to
 3195     help when there was an actual mention of one of the two characters. So now
 3196     the change happens only if \r or \n (or a literal CR or LF) character is
 3197     encountered.
 3199 2.  The 32-bit options word was also used for 6 internal flags, but the numbers
 3200     of both had grown to the point where there were only 3 bits left.
 3201     Fortunately, there was spare space in the data structure, and so I have
 3202     moved the internal flags into a new 16-bit field to free up more option
 3203     bits.
 3205 3.  The appearance of (?J) at the start of a pattern set the DUPNAMES option,
 3206     but did not set the internal JCHANGED flag - either of these is enough to
 3207     control the way the "get" function works - but the PCRE_INFO_JCHANGED
 3208     facility is supposed to tell if (?J) was ever used, so now (?J) at the
 3209     start sets both bits.
 3211 4.  Added options (at build time, compile time, exec time) to change \R from
 3212     matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
 3214 5.  doc/pcresyntax.html was missing from the distribution.
 3216 6.  Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
 3217     compatibility, even though it is no longer used.
 3219 7.  Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
 3220     strtoull to pcrecpp.cc to select the available functions in WIN32 when the
 3221     windows.h file is present (where different names are used). [This was
 3222     reversed later after testing - see 16 below.]
 3224 8.  Changed all #include <config.h> to #include "config.h". There were also
 3225     some further <pcre.h> cases that I changed to "pcre.h".
 3227 9.  When pcregrep was used with the --colour option, it missed the line ending
 3228     sequence off the lines that it output.
 3230 10. It was pointed out to me that arrays of string pointers cause lots of
 3231     relocations when a shared library is dynamically loaded. A technique of
 3232     using a single long string with a table of offsets can drastically reduce
 3233     these. I have refactored PCRE in four places to do this. The result is
 3234     dramatic:
 3236       Originally:                          290
 3237       After changing UCP table:            187
 3238       After changing error message table:   43
 3239       After changing table of "verbs"       36
 3240       After changing table of Posix names   22
 3242     Thanks to the folks working on Gregex for glib for this insight.
 3244 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
 3245     unicode-properties was also set.
 3247 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
 3249 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
 3250     checked only for CRLF.
 3252 14. Added casts to pcretest.c to avoid compiler warnings.
 3254 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
 3256 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
 3257     and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
 3258     entirely. This removes changes made in 7 above.
 3260 17. The CMake files have been updated, and there is now more information about
 3261     building with CMake in the NON-UNIX-USE document.
 3264 Version 7.3 28-Aug-07
 3265 ---------------------
 3267  1. In the rejigging of the build system that eventually resulted in 7.1, the
 3268     line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
 3269     brackets there is not right, since it causes compilers to look for an
 3270     installed pcre.h, not the version that is in the source that is being
 3271     compiled (which of course may be different). I have changed it back to:
 3273       #include "pcre.h"
 3275     I have a vague recollection that the change was concerned with compiling in
 3276     different directories, but in the new build system, that is taken care of
 3277     by the VPATH setting the Makefile.
 3279  2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
 3280     when the subject happened to end in the byte 0x85 (e.g. if the last
 3281     character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
 3282     characters but of course it shouldn't be taken as a newline when it is part
 3283     of another character. The bug was that, for an unlimited repeat of . in
 3284     not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
 3285     characters when looking for a newline.
 3287  3. A small performance improvement in the DOTALL UTF-8 mode .* case.
 3289  4. Debugging: adjusted the names of opcodes for different kinds of parentheses
 3290     in debug output.
 3292  5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
 3293     long printing in the pcrecpp unittest when running under MinGW.
 3295  6. ESC_K was left out of the EBCDIC table.
 3297  7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
 3298     parentheses; I made it 1000, which seemed large enough. Unfortunately, the
 3299     limit also applies to "virtual nesting" when a pattern is recursive, and in
 3300     this case 1000 isn't so big. I have been able to remove this limit at the
 3301     expense of backing off one optimization in certain circumstances. Normally,
 3302     when pcre_exec() would call its internal match() function recursively and
 3303     immediately return the result unconditionally, it uses a "tail recursion"
 3304     feature to save stack. However, when a subpattern that can match an empty
 3305     string has an unlimited repetition quantifier, it no longer makes this
 3306     optimization. That gives it a stack frame in which to save the data for
 3307     checking that an empty string has been matched. Previously this was taken
 3308     from the 1000-entry workspace that had been reserved. So now there is no
 3309     explicit limit, but more stack is used.
 3311  8. Applied Daniel's patches to solve problems with the import/export magic
 3312     syntax that is required for Windows, and which was going wrong for the
 3313     pcreposix and pcrecpp parts of the library. These were overlooked when this
 3314     problem was solved for the main library.
 3316  9. There were some crude static tests to avoid integer overflow when computing
 3317     the size of patterns that contain repeated groups with explicit upper
 3318     limits. As the maximum quantifier is 65535, the maximum group length was
 3319     set at 30,000 so that the product of these two numbers did not overflow a
 3320     32-bit integer. However, it turns out that people want to use groups that
 3321     are longer than 30,000 bytes (though not repeat them that many times).
 3322     Change 7.0/17 (the refactoring of the way the pattern size is computed) has
 3323     made it possible to implement the integer overflow checks in a much more
 3324     dynamic way, which I have now done. The artificial limitation on group
 3325     length has been removed - we now have only the limit on the total length of
 3326     the compiled pattern, which depends on the LINK_SIZE setting.
 3328 10. Fixed a bug in the documentation for get/copy named substring when
 3329     duplicate names are permitted. If none of the named substrings are set, the
 3330     functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
 3331     empty string.
 3333 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
 3334     instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
 3335     because the ] is interpreted as the first data character and the
 3336     terminating ] is not found. PCRE has been made compatible with Perl in this
 3337     regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
 3338     cause memory overwriting.
 3340 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
 3341     string has been matched (to stop an infinite loop). It was not recognizing
 3342     a conditional subpattern that could match an empty string if that
 3343     subpattern was within another subpattern. For example, it looped when
 3344     trying to match  (((?(1)X|))*)  but it was OK with  ((?(1)X|)*)  where the
 3345     condition was not nested. This bug has been fixed.
 3347 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
 3348     past the start of the subject in the presence of bytes with the top bit
 3349     set, for example "\x8aBCD".
 3351 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
 3352     (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
 3354 14. Optimized (?!) to (*FAIL).
 3356 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
 3357     This restricts code points to be within the range 0 to 0x10FFFF, excluding
 3358     the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
 3359     full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
 3360     does: it's just the validity check that is more restrictive.
 3362 16. Inserted checks for integer overflows during escape sequence (backslash)
 3363     processing, and also fixed erroneous offset values for syntax errors during
 3364     backslash processing.
 3366 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
 3367     for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
 3369 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
 3370     caused an overrun.
 3372 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
 3373     something other than just ASCII characters) inside a group that had an
 3374     unlimited repeat caused a loop at compile time (while checking to see
 3375     whether the group could match an empty string).
 3377 20. Debugging a pattern containing \p or \P could cause a crash. For example,
 3378     [\P{Any}] did so. (Error in the code for printing property names.)
 3380 21. An orphan \E inside a character class could cause a crash.
 3382 22. A repeated capturing bracket such as (A)? could cause a wild memory
 3383     reference during compilation.
 3385 23. There are several functions in pcre_compile() that scan along a compiled
 3386     expression for various reasons (e.g. to see if it's fixed length for look
 3387     behind). There were bugs in these functions when a repeated \p or \P was
 3388     present in the pattern. These operators have additional parameters compared
 3389     with \d, etc, and these were not being taken into account when moving along
 3390     the compiled data. Specifically:
 3392     (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
 3393         length.
 3395     (b) An item such as \pL+ within a repeated group could cause crashes or
 3396         loops.
 3398     (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
 3399         "reference to non-existent subpattern" error.
 3401     (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
 3403 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
 3404     characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
 3406 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
 3408 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
 3409     character were causing crashes (broken optimization).
 3411 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
 3412     \p or \P) caused a compile-time loop.
 3414 28. More problems have arisen in unanchored patterns when CRLF is a valid line
 3415     break. For example, the unstudied pattern [\r\n]A does not match the string
 3416     "\r\nA" because change 7.0/46 below moves the current point on by two
 3417     characters after failing to match at the start. However, the pattern \nA
 3418     *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
 3419     the same is true. There doesn't seem any very clean way out of this, but
 3420     what I have chosen to do makes the common cases work: PCRE now takes note
 3421     of whether there can be an explicit match for \r or \n anywhere in the
 3422     pattern, and if so, 7.0/46 no longer applies. As part of this change,
 3423     there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
 3424     pattern has explicit CR or LF references.
 3426 29. Added (*CR) etc for changing newline setting at start of pattern.
 3429 Version 7.2 19-Jun-07
 3430 ---------------------
 3432  1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
 3433     which is apparently normally available under Windows.
 3435  2. Re-jig the pcregrep tests with different newline settings in an attempt
 3436     to make them independent of the local environment's newline setting.
 3438  3. Add code to configure.ac to remove -g from the CFLAGS default settings.
 3440  4. Some of the "internals" tests were previously cut out when the link size
 3441     was not 2, because the output contained actual offsets. The recent new
 3442     "Z" feature of pcretest means that these can be cut out, making the tests
 3443     usable with all link sizes.
 3445  5. Implemented Stan Switzer's goto replacement for longjmp() when not using
 3446     stack recursion. This gives a massive performance boost under BSD, but just
 3447     a small improvement under Linux. However, it saves one field in the frame
 3448     in all cases.
 3450  6. Added more features from the forthcoming Perl 5.10:
 3452     (a) (?-n) (where n is a string of digits) is a relative subroutine or
 3453         recursion call. It refers to the nth most recently opened parentheses.
 3455     (b) (?+n) is also a relative subroutine call; it refers to the nth next
 3456         to be opened parentheses.
 3458     (c) Conditions that refer to capturing parentheses can be specified
 3459         relatively, for example, (?(-2)... or (?(+3)...
 3461     (d) \K resets the start of the current match so that everything before
 3462         is not part of it.
 3464     (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
 3466     (f) \g{name} is another synonym - part of Perl 5.10's unification of
 3467         reference syntax.
 3469     (g) (?| introduces a group in which the numbering of parentheses in each
 3470         alternative starts with the same number.
 3472     (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
 3474  7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
 3477  8. A pattern such as  (.*(.)?)*  caused pcre_exec() to fail by either not
 3478     terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
 3479     for detecting groups that can match an empty string.
 3481  9. A pattern with a very large number of alternatives (more than several
 3482     hundred) was running out of internal workspace during the pre-compile
 3483     phase, where pcre_compile() figures out how much memory will be needed. A
 3484     bit of new cunning has reduced the workspace needed for groups with
 3485     alternatives. The 1000-alternative test pattern now uses 12 bytes of
 3486     workspace instead of running out of the 4096 that are available.
 3488 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
 3490 11. Applied patch from Google to remove an optimization that didn't quite work.
 3491     The report of the bug said:
 3493       pcrecpp::RE("a*").FullMatch("aaa") matches, while
 3494       pcrecpp::RE("a*?").FullMatch("aaa") does not, and
 3495       pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
 3497 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
 3498     it matched the wrong number of bytes.
 3501 Version 7.1 24-Apr-07
 3502 ---------------------
 3504  1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
 3505     that is more "standard", making use of automake and other Autotools. There
 3506     is some re-arrangement of the files and adjustment of comments consequent
 3507     on this.
 3509  2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
 3510     for recursive directory scanning broke on some systems because the files
 3511     are not scanned in any specific order and on different systems the order
 3512     was different. A call to "sort" has been inserted into RunGrepTest for the
 3513     approprate test as a short-term fix. In the longer term there may be an
 3514     alternative.
 3516  3. I had an email from Eric Raymond about problems translating some of PCRE's
 3517     man pages to HTML (despite the fact that I distribute HTML pages, some
 3518     people do their own conversions for various reasons). The problems
 3519     concerned the use of low-level troff macros .br and .in. I have therefore
 3520     removed all such uses from the man pages (some were redundant, some could
 3521     be replaced by .nf/.fi pairs). The 132html script that I use to generate
 3522     HTML has been updated to handle .nf/.fi and to complain if it encounters
 3523     .br or .in.
 3525  4. Updated comments in configure.ac that get placed in config.h.in and also
 3526     arranged for config.h to be included in the distribution, with the name
 3527     config.h.generic, for the benefit of those who have to compile without
 3528     Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
 3530  5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
 3531     Weber: (1) pcre_internal.h was missing some function renames; (2) updated
 3532     makevp.bat for the current PCRE, using the additional files
 3533     makevp_c.txt, makevp_l.txt, and pcregexp.pas.
 3535  6. A Windows user reported a minor discrepancy with test 2, which turned out
 3536     to be caused by a trailing space on an input line that had got lost in his
 3537     copy. The trailing space was an accident, so I've just removed it.
 3539  7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
 3540     that is needed.
 3542  8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
 3543     as "const" (a) because they are and (b) because it helps the PHP
 3544     maintainers who have recently made a script to detect big data structures
 3545     in the php code that should be moved to the .rodata section. I remembered
 3546     to update Builducptable as well, so it won't revert if ucptable.h is ever
 3547     re-created.
 3549  9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
 3550     pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
 3551     order to be able to cut out the UTF-8 tables in the latter when UTF-8
 3552     support is not required. This saves 1.5-2K of code, which is important in
 3553     some applications.
 3555     Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
 3556     so as not to refer to the tables, even though these functions will never be
 3557     called when UTF-8 support is disabled. Otherwise there are problems with a
 3558     shared library.
 3560 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
 3562     (a) It was defining its arguments as char * instead of void *.
 3564     (b) It was assuming that all moves were upwards in memory; this was true
 3565         a long time ago when I wrote it, but is no longer the case.
 3567     The emulated memove() is provided for those environments that have neither
 3568     memmove() nor bcopy(). I didn't think anyone used it these days, but that
 3569     is clearly not the case, as these two bugs were recently reported.
 3571 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
 3572     and Detrail to create the HTML documentation, the .txt form of the man
 3573     pages, and it removes trailing spaces from listed files. It also creates
 3574     pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
 3575     case, it wraps all the #defines with #ifndefs. This script should be run
 3576     before "make dist".
 3578 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
 3579     with Unicode property support.
 3581     (a) For a maximizing quantifier, if the two different cases of the
 3582         character were of different lengths in their UTF-8 codings (there are
 3583         some cases like this - I found 11), and the matching function had to
 3584         back up over a mixture of the two cases, it incorrectly assumed they
 3585         were both the same length.
 3587     (b) When PCRE was configured to use the heap rather than the stack for
 3588         recursion during matching, it was not correctly preserving the data for
 3589         the other case of a UTF-8 character when checking ahead for a match
 3590         while processing a minimizing repeat. If the check also involved
 3591         matching a wide character, but failed, corruption could cause an
 3592         erroneous result when trying to check for a repeat of the original
 3593         character.
 3595 13. Some tidying changes to the testing mechanism:
 3597     (a) The RunTest script now detects the internal link size and whether there
 3598         is UTF-8 and UCP support by running ./pcretest -C instead of relying on
 3599         values substituted by "configure". (The RunGrepTest script already did
 3600         this for UTF-8.) The configure.ac script no longer substitutes the
 3601         relevant variables.
 3603     (b) The debugging options /B and /D in pcretest show the compiled bytecode
 3604         with length and offset values. This means that the output is different
 3605         for different internal link sizes. Test 2 is skipped for link sizes
 3606         other than 2 because of this, bypassing the problem. Unfortunately,
 3607         there was also a test in test 3 (the locale tests) that used /B and
 3608         failed for link sizes other than 2. Rather than cut the whole test out,
 3609         I have added a new /Z option to pcretest that replaces the length and
 3610         offset values with spaces. This is now used to make test 3 independent
 3611         of link size. (Test 2 will be tidied up later.)
 3613 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
 3614     segmentation fault instead of returning the appropriate error message.
 3616 15. In multiline mode when the newline sequence was set to "any", the pattern
 3617     ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
 3618     This doesn't seem right; it now treats the CRLF combination as the line
 3619     ending, and so does not match in that case. It's only a pattern such as ^$
 3620     that would hit this one: something like ^ABC$ would have failed after \r
 3621     and then tried again after \r\n.
 3623 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
 3624     in an attempt to make files that differ only in their line terminators
 3625     compare equal. This works on Linux.
 3627 17. Under certain error circumstances pcregrep might try to free random memory
 3628     as it exited. This is now fixed, thanks to valgrind.
 3630 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
 3631     "abc\r\n\r\n", it found an unwanted second match after the second \r. This
 3632     was because its rules for how to advance for /g after matching an empty
 3633     string at the end of a line did not allow for this case. They now check for
 3634     it specially.
 3636 20. pcretest is supposed to handle patterns and data of any length, by
 3637     extending its buffers when necessary. It was getting this wrong when the
 3638     buffer for a data line had to be extended.
 3640 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
 3641     CRLF as a newline sequence.
 3643 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
 3644     out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
 3645     I have nevertheless tidied it up.
 3647 23. Added some casts to kill warnings from HP-UX ia64 compiler.
 3649 24. Added a man page for pcre-config.
 3652 Version 7.0 19-Dec-06
 3653 ---------------------
 3655  1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
 3656     moving to gcc 4.1.1.
 3658  2. The -S option for pcretest uses setrlimit(); I had omitted to #include
 3659     sys/time.h, which is documented as needed for this function. It doesn't
 3660     seem to matter on Linux, but it showed up on some releases of OS X.
 3662  3. It seems that there are systems where bytes whose values are greater than
 3663     127 match isprint() in the "C" locale. The "C" locale should be the
 3664     default when a C program starts up. In most systems, only ASCII printing
 3665     characters match isprint(). This difference caused the output from pcretest
 3666     to vary, making some of the tests fail. I have changed pcretest so that:
 3668     (a) When it is outputting text in the compiled version of a pattern, bytes
 3669         other than 32-126 are always shown as hex escapes.
 3671     (b) When it is outputting text that is a matched part of a subject string,
 3672         it does the same, unless a different locale has been set for the match
 3673         (using the /L modifier). In this case, it uses isprint() to decide.
 3675  4. Fixed a major bug that caused incorrect computation of the amount of memory
 3676     required for a compiled pattern when options that changed within the
 3677     pattern affected the logic of the preliminary scan that determines the
 3678     length. The relevant options are -x, and -i in UTF-8 mode. The result was
 3679     that the computed length was too small. The symptoms of this bug were
 3680     either the PCRE error "internal error: code overflow" from pcre_compile(),
 3681     or a glibc crash with a message such as "pcretest: free(): invalid next
 3682     size (fast)". Examples of patterns that provoked this bug (shown in
 3683     pcretest format) are:
 3685       /(?-x: )/x
 3686       /(?x)(?-x: \s*#\s*)/
 3687       /((?i)[\x{c0}])/8
 3688       /(?i:[\x{c0}])/8
 3690     HOWEVER: Change 17 below makes this fix obsolete as the memory computation
 3691     is now done differently.
 3693  5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
 3694     wrapper classes; (b) implement a new function in the C++ scanner that is
 3695     more efficient than the old way of doing things because it avoids levels of
 3696     recursion in the regex matching; (c) add a paragraph to the documentation
 3697     for the FullMatch() function.
 3699  6. The escape sequence \n was being treated as whatever was defined as
 3700     "newline". Not only was this contrary to the documentation, which states
 3701     that \n is character 10 (hex 0A), but it also went horribly wrong when
 3702     "newline" was defined as CRLF. This has been fixed.
 3704  7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
 3705     was being set to -1 for the "end of line" case (supposedly a value that no
 3706     character can have). Though this value is never used (the check for end of
 3707     line is "zero bytes in current character"), it caused compiler complaints.
 3708     I've changed it to 0xffffffff.
 3710  8. In pcre_version.c, the version string was being built by a sequence of
 3711     C macros that, in the event of PCRE_PRERELEASE being defined as an empty
 3712     string (as it is for production releases) called a macro with an empty
 3713     argument. The C standard says the result of this is undefined. The gcc
 3714     compiler treats it as an empty string (which was what was wanted) but it is
 3715     reported that Visual C gives an error. The source has been hacked around to
 3716     avoid this problem.
 3718  9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
 3719     builds of pcretest, and changed the call to _setmode() to use _O_BINARY
 3720     instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
 3721     of them did).
 3723 10. Originally, pcretest opened its input and output without "b"; then I was
 3724     told that "b" was needed in some environments, so it was added for release
 3725     5.0 to both the input and output. (It makes no difference on Unix-like
 3726     systems.) Later I was told that it is wrong for the input on Windows. I've
 3727     now abstracted the modes into two macros, to make it easier to fiddle with
 3728     them, and removed "b" from the input mode under Windows.
 3730 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
 3732 12. Added -help and --help to pcretest as an official way of being reminded
 3733     of the options.
 3735 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
 3736     and pcrecpp.cc because they annoy compilers at high warning levels.
 3738 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
 3740 15. Fixed an occurrence of == in configure.ac that should have been = (shell
 3741     scripts are not C programs :-) and which was not noticed because it works
 3742     on Linux.
 3744 16. pcretest is supposed to handle any length of pattern and data line (as one
 3745     line or as a continued sequence of lines) by extending its input buffer if
 3746     necessary. This feature was broken for very long pattern lines, leading to
 3747     a string of junk being passed to pcre_compile() if the pattern was longer
 3748     than about 50K.
 3750 17. I have done a major re-factoring of the way pcre_compile() computes the
 3751     amount of memory needed for a compiled pattern. Previously, there was code
 3752     that made a preliminary scan of the pattern in order to do this. That was
 3753     OK when PCRE was new, but as the facilities have expanded, it has become
 3754     harder and harder to keep it in step with the real compile phase, and there
 3755     have been a number of bugs (see for example, 4 above). I have now found a
 3756     cunning way of running the real compile function in a "fake" mode that
 3757     enables it to compute how much memory it would need, while actually only
 3758     ever using a few hundred bytes of working memory and without too many
 3759     tests of the mode. This should make future maintenance and development
 3760     easier. A side effect of this work is that the limit of 200 on the nesting
 3761     depth of parentheses has been removed (though this was never a serious
 3762     limitation, I suspect). However, there is a downside: pcre_compile() now
 3763     runs more slowly than before (30% or more, depending on the pattern). I
 3764     hope this isn't a big issue. There is no effect on runtime performance.
 3766 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
 3767     newline (only possible for the last line of a file) and it was a
 3768     pattern that set a locale (followed by /Lsomething), pcretest crashed.
 3770 19. Added additional timing features to pcretest. (1) The -tm option now times
 3771     matching only, not compiling. (2) Both -t and -tm can be followed, as a
 3772     separate command line item, by a number that specifies the number of
 3773     repeats to use when timing. The default is 50000; this gives better
 3774     precision, but takes uncomfortably long for very large patterns.
 3776 20. Extended pcre_study() to be more clever in cases where a branch of a
 3777     subpattern has no definite first character. For example, (a*|b*)[cd] would
 3778     previously give no result from pcre_study(). Now it recognizes that the
 3779     first character must be a, b, c, or d.
 3781 21. There was an incorrect error "recursive call could loop indefinitely" if
 3782     a subpattern (or the entire pattern) that was being tested for matching an
 3783     empty string contained only one non-empty item after a nested subpattern.
 3784     For example, the pattern (?>\x{100}*)\d(?R) provoked this error
 3785     incorrectly, because the \d was being skipped in the check.
 3787 22. The pcretest program now has a new pattern option /B and a command line
 3788     option -b, which is equivalent to adding /B to every pattern. This causes
 3789     it to show the compiled bytecode, without the additional information that
 3790     -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
 3791     is the same as /B/I).
 3793 23. A new optimization is now able automatically to treat some sequences such
 3794     as a*b as a*+b. More specifically, if something simple (such as a character
 3795     or a simple class like \d) has an unlimited quantifier, and is followed by
 3796     something that cannot possibly match the quantified thing, the quantifier
 3797     is automatically "possessified".
 3799 24. A recursive reference to a subpattern whose number was greater than 39
 3800     went wrong under certain circumstances in UTF-8 mode. This bug could also
 3801     have affected the operation of pcre_study().
 3803 25. Realized that a little bit of performance could be had by replacing
 3804     (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
 3806 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
 3808 27. Possessive quantifiers such as a++ were previously implemented by turning
 3809     them into atomic groups such as ($>a+). Now they have their own opcodes,
 3810     which improves performance. This includes the automatically created ones
 3811     from 23 above.
 3813 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
 3814     lookahead was broken if it was not anchored. PCRE was mistakenly expecting
 3815     the first matched character to be a colon. This applied both to named and
 3816     numbered groups.
 3818 29. The ucpinternal.h header file was missing its idempotency #ifdef.
 3820 30. I was sent a "project" file called libpcre.a.dev which I understand makes
 3821     building PCRE on Windows easier, so I have included it in the distribution.
 3823 31. There is now a check in pcretest against a ridiculously large number being
 3824     returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
 3825     loop, the loop is abandoned.
 3827 32. Forward references to subpatterns in conditions such as (?(2)...) where
 3828     subpattern 2 is defined later cause pcre_compile() to search forwards in
 3829     the pattern for the relevant set of parentheses. This search went wrong
 3830     when there were unescaped parentheses in a character class, parentheses
 3831     escaped with \Q...\E, or parentheses in a #-comment in /x mode.
 3833 33. "Subroutine" calls and backreferences were previously restricted to
 3834     referencing subpatterns earlier in the regex. This restriction has now
 3835     been removed.
 3837 34. Added a number of extra features that are going to be in Perl 5.10. On the
 3838     whole, these are just syntactic alternatives for features that PCRE had
 3839     previously implemented using the Python syntax or my own invention. The
 3840     other formats are all retained for compatibility.
 3842     (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
 3843         as (?P<name>...). The new forms, as well as being in Perl 5.10, are
 3844         also .NET compatible.
 3846     (b) A recursion or subroutine call to a named group can now be defined as
 3847         (?&name) as well as (?P>name).
 3849     (c) A backreference to a named group can now be defined as \k<name> or
 3850         \k'name' as well as (?P=name). The new forms, as well as being in Perl
 3851         5.10, are also .NET compatible.
 3853     (d) A conditional reference to a named group can now use the syntax
 3854         (?(<name>) or (?('name') as well as (?(name).
 3856     (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
 3857         groups (named and numbered) that are never evaluated inline, but can be
 3858         called as "subroutines" from elsewhere. In effect, the DEFINE condition
 3859         is always false. There may be only one alternative in such a group.
 3861     (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
 3862         as the simple (?(R). The condition is true only if the most recent
 3863         recursion is that of the given number or name. It does not search out
 3864         through the entire recursion stack.
 3866     (g) The escape \gN or \g{N} has been added, where N is a positive or
 3867         negative number, specifying an absolute or relative reference.
 3869 35. Tidied to get rid of some further signed/unsigned compiler warnings and
 3870     some "unreachable code" warnings.
 3872 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
 3873     things, this adds five new scripts.
 3875 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
 3876     There were also incompatibilities regarding the handling of \Q..\E inside
 3877     character classes, for example with patterns like [\Qa\E-\Qz\E] where the
 3878     hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
 3880 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
 3881     matches an empty string, and forcibly breaks the loop. There were bugs in
 3882     this code in non-simple cases. For a pattern such as  ^(a()*)*  matched
 3883     against  aaaa  the result was just "a" rather than "aaaa", for example. Two
 3884     separate and independent bugs (that affected different cases) have been
 3885     fixed.
 3887 39. Refactored the code to abolish the use of different opcodes for small
 3888     capturing bracket numbers. This is a tidy that I avoided doing when I
 3889     removed the limit on the number of capturing brackets for 3.5 back in 2001.
 3890     The new approach is not only tidier, it makes it possible to reduce the
 3891     memory needed to fix the previous bug (38).
 3893 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
 3894     sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
 3895     processing dot, circumflex, or dollar metacharacters, or #-comments in /x
 3896     mode.
 3898 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
 3899     report.
 3901 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
 3902     copy construction and assignment in the C++ wrapper.
 3904 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
 3905     couple of bugs that could have given wrong results in the "--newline=crlf"
 3906     case.
 3908 44. Added a number of casts and did some reorganization of signed/unsigned int
 3909     variables following suggestions from Dair Grant. Also renamed the variable
 3910     "this" as "item" because it is a C++ keyword.
 3912 45. Arranged for dftables to add
 3914       #include "pcre_internal.h"
 3916     to pcre_chartables.c because without it, gcc 4.x may remove the array
 3917     definition from the final binary if PCRE is built into a static library and
 3918     dead code stripping is activated.
 3920 46. For an unanchored pattern, if a match attempt fails at the start of a
 3921     newline sequence, and the newline setting is CRLF or ANY, and the next two
 3922     characters are CRLF, advance by two characters instead of one.
 3925 Version 6.7 04-Jul-06
 3926 ---------------------
 3928  1. In order to handle tests when input lines are enormously long, pcretest has
 3929     been re-factored so that it automatically extends its buffers when
 3930     necessary. The code is crude, but this _is_ just a test program. The
 3931     default size has been increased from 32K to 50K.
 3933  2. The code in pcre_study() was using the value of the re argument before
 3934     testing it for NULL. (Of course, in any sensible call of the function, it
 3935     won't be NULL.)
 3937  3. The memmove() emulation function in pcre_internal.h, which is used on
 3938     systems that lack both memmove() and bcopy() - that is, hardly ever -
 3939     was missing a "static" storage class specifier.
 3941  4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
 3942     containing an extended class (one that cannot be represented by a bitmap
 3943     because it contains high-valued characters or Unicode property items, e.g.
 3944     [\pZ]). Almost always one would set UTF-8 mode when processing such a
 3945     pattern, but PCRE should not loop if you do not (it no longer does).
 3946     [Detail: two cases were found: (a) a repeated subpattern containing an
 3947     extended class; (b) a recursive reference to a subpattern that followed a
 3948     previous extended class. It wasn't skipping over the extended class
 3949     correctly when UTF-8 mode was not set.]
 3951  5. A negated single-character class was not being recognized as fixed-length
 3952     in lookbehind assertions such as (?<=[^f]), leading to an incorrect
 3953     compile error "lookbehind assertion is not fixed length".
 3955  6. The RunPerlTest auxiliary script was showing an unexpected difference
 3956     between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
 3957     write a Perl script that can interpret lines of an input file either as
 3958     byte characters or as UTF-8, which is what "perltest" was being required to
 3959     do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
 3960     can't do is switch easily at run time between having the "use utf8;" pragma
 3961     or not. In the end, I fudged it by using the RunPerlTest script to insert
 3962     "use utf8;" explicitly for the UTF-8 tests.
 3964  7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
 3965     the end of the subject string, contrary to the documentation and to what
 3966     Perl does. This was true of both matching functions. Now it matches only at
 3967     the start of the subject and immediately after *internal* newlines.
 3969  8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
 3970     a pointer to an int instead of a pointer to an unsigned long int. This
 3971     caused problems on 64-bit systems.
 3973  9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
 3974     instance of the 'standard' template library not being so standard".
 3976 10. There was no check on the number of named subpatterns nor the maximum
 3977     length of a subpattern name. The product of these values is used to compute
 3978     the size of the memory block for a compiled pattern. By supplying a very
 3979     long subpattern name and a large number of named subpatterns, the size
 3980     computation could be caused to overflow. This is now prevented by limiting
 3981     the length of names to 32 characters, and the number of named subpatterns
 3982     to 10,000.
 3984 11. Subpatterns that are repeated with specific counts have to be replicated in
 3985     the compiled pattern. The size of memory for this was computed from the
 3986     length of the subpattern and the repeat count. The latter is limited to
 3987     65535, but there was no limit on the former, meaning that integer overflow
 3988     could in principle occur. The compiled length of a repeated subpattern is
 3989     now limited to 30,000 bytes in order to prevent this.
 3991 12. Added the optional facility to have named substrings with the same name.
 3993 13. Added the ability to use a named substring as a condition, using the
 3994     Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
 3995     are numbers (not recommended). Forward references are permitted.
 3997 14. Added forward references in named backreferences (if you see what I mean).
 3999 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
 4000     pattern could run off the end of the subject. For example, the pattern
 4001     "(?s)(.{1,5})"8 did this with the subject "ab".
 4003 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
 4004     PCRE_CASELESS was set when matching characters that were quantified with ?
 4005     or *.
 4007 17. A character class other than a single negated character that had a minimum
 4008     but no maximum quantifier - for example [ab]{6,} - was not handled
 4009     correctly by pce_dfa_exec(). It would match only one character.
 4011 18. A valid (though odd) pattern that looked like a POSIX character
 4012     class but used an invalid character after [ (for example [[,abc,]]) caused
 4013     pcre_compile() to give the error "Failed: internal error: code overflow" or
 4014     in some cases to crash with a glibc free() error. This could even happen if
 4015     the pattern terminated after [[ but there just happened to be a sequence of
 4016     letters, a binary zero, and a closing ] in the memory that followed.
 4018 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
 4019     over the years. Originally (before any Unicode support), just the bottom 8
 4020     bits were taken. Thus, for example, \500 really meant \100. Nowadays the
 4021     output from "man perlunicode" includes this:
 4023       The regular expression compiler produces polymorphic opcodes.  That
 4024       is, the pattern adapts to the data and automatically switches to
 4025       the Unicode character scheme when presented with Unicode data--or
 4026       instead uses a traditional byte scheme when presented with byte
 4027       data.
 4029     Sadly, a wide octal escape does not cause a switch, and in a string with
 4030     no other multibyte characters, these octal escapes are treated as before.
 4031     Thus, in Perl, the pattern  /\500/ actually matches \100 but the pattern
 4032     /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
 4033     Unicode string.
 4035     I have not perpetrated such confusion in PCRE. Up till now, it took just
 4036     the bottom 8 bits, as in old Perl. I have now made octal escapes with
 4037     values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
 4038     translate to the appropriate multibyte character.
 4040 29. Applied some refactoring to reduce the number of warnings from Microsoft
 4041     and Borland compilers. This has included removing the fudge introduced
 4042     seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
 4043     a warning about an unused variable.
 4045 21. PCRE has not included VT (character 0x0b) in the set of whitespace
 4046     characters since release 4.0, because Perl (from release 5.004) does not.
 4047     [Or at least, is documented not to: some releases seem to be in conflict
 4048     with the documentation.] However, when a pattern was studied with
 4049     pcre_study() and all its branches started with \s, PCRE still included VT
 4050     as a possible starting character. Of course, this did no harm; it just
 4051     caused an unnecessary match attempt.
 4053 22. Removed a now-redundant internal flag bit that recorded the fact that case
 4054     dependency changed within the pattern. This was once needed for "required
 4055     byte" processing, but is no longer used. This recovers a now-scarce options
 4056     bit. Also moved the least significant internal flag bit to the most-
 4057     significant bit of the word, which was not previously used (hangover from
 4058     the days when it was an int rather than a uint) to free up another bit for
 4059     the future.
 4061 23. Added support for CRLF line endings as well as CR and LF. As well as the
 4062     default being selectable at build time, it can now be changed at runtime
 4063     via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
 4064     specify that it is scanning data with non-default line endings.
 4066 24. Changed the definition of CXXLINK to make it agree with the definition of
 4067     LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
 4069 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
 4070     recursions. This makes a big different to stack usage for some patterns.
 4072 26. If a subpattern containing a named recursion or subroutine reference such
 4073     as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
 4074     the space required for the compiled pattern went wrong and gave too small a
 4075     value. Depending on the environment, this could lead to "Failed: internal
 4076     error: code overflow at offset 49" or "glibc detected double free or
 4077     corruption" errors.
 4079 27. Applied patches from Google (a) to support the new newline modes and (b) to
 4080     advance over multibyte UTF-8 characters in GlobalReplace.
 4082 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
 4083     difference for some implementation of PCRE in some Windows version.
 4085 29. Added some extra testing facilities to pcretest:
 4087     \q<number>   in a data line sets the "match limit" value
 4088     \Q<number>   in a data line sets the "match recursion limt" value
 4089     -S <number>  sets the stack size, where <number> is in megabytes
 4091     The -S option isn't available for Windows.
 4094 Version 6.6 06-Feb-06
 4095 ---------------------
 4097  1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
 4098     in pcreposix.h. I have copied the definition from pcre.h.
 4100  2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
 4101     because pcre.h is no longer a built file.
 4103  3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
 4104     not normally included in the compiled code.
 4107 Version 6.5 01-Feb-06
 4108 ---------------------
 4110  1. When using the partial match feature with pcre_dfa_exec(), it was not
 4111     anchoring the second and subsequent partial matches at the new starting
 4112     point. This could lead to incorrect results. For example, with the pattern
 4113     /1234/, partially matching against "123" and then "a4" gave a match.
 4115  2. Changes to pcregrep:
 4117     (a) All non-match returns from pcre_exec() were being treated as failures
 4118         to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
 4119         error message is output. Some extra information is given for the
 4121         probably the only errors that are likely to be caused by users (by
 4122         specifying a regex that has nested indefinite repeats, for instance).
 4123         If there are more than 20 of these errors, pcregrep is abandoned.
 4125     (b) A binary zero was treated as data while matching, but terminated the
 4126         output line if it was written out. This has been fixed: binary zeroes
 4127         are now no different to any other data bytes.
 4129     (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
 4130         used to set a locale for matching. The --locale=xxxx long option has
 4131         been added (no short equivalent) to specify a locale explicitly on the
 4132         pcregrep command, overriding the environment variables.
 4134     (d) When -B was used with -n, some line numbers in the output were one less
 4135         than they should have been.
 4137     (e) Added the -o (--only-matching) option.
 4139     (f) If -A or -C was used with -c (count only), some lines of context were
 4140         accidentally printed for the final match.
 4142     (g) Added the -H (--with-filename) option.
 4144     (h) The combination of options -rh failed to suppress file names for files
 4145         that were found from directory arguments.
 4147     (i) Added the -D (--devices) and -d (--directories) options.
 4149     (j) Added the -F (--fixed-strings) option.
 4151     (k) Allow "-" to be used as a file name for -f as well as for a data file.
 4153     (l) Added the --colo(u)r option.
 4155     (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
 4156         is not present by default.
 4158  3. A nasty bug was discovered in the handling of recursive patterns, that is,
 4159     items such as (?R) or (?1), when the recursion could match a number of
 4160     alternatives. If it matched one of the alternatives, but subsequently,
 4161     outside the recursion, there was a failure, the code tried to back up into
 4162     the recursion. However, because of the way PCRE is implemented, this is not
 4163     possible, and the result was an incorrect result from the match.
 4165     In order to prevent this happening, the specification of recursion has
 4166     been changed so that all such subpatterns are automatically treated as
 4167     atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
 4169  4. I had overlooked the fact that, in some locales, there are characters for
 4170     which isalpha() is true but neither isupper() nor islower() are true. In
 4171     the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
 4172     and ordfeminine) are like this. This affected the treatment of \w and \W
 4173     when they appeared in character classes, but not when they appeared outside
 4174     a character class. The bit map for "word" characters is now created
 4175     separately from the results of isalnum() instead of just taking it from the
 4176     upper, lower, and digit maps. (Plus the underscore character, of course.)
 4178  5. The above bug also affected the handling of POSIX character classes such as
 4179     [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
 4180     permanent tables. Instead, the bit maps for such a class were previously
 4181     created as the appropriate unions of the upper, lower, and digit bitmaps.
 4182     Now they are created by subtraction from the [[:word:]] class, which has
 4183     its own bitmap.
 4185  6. The [[:blank:]] character class matches horizontal, but not vertical space.
 4186     It is created by subtracting the vertical space characters (\x09, \x0a,
 4187     \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
 4188     subtraction was done in the overall bitmap for a character class, meaning
 4189     that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
 4190     be recognized. This bug has been fixed.
 4192  7. Patches from the folks at Google:
 4194       (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
 4195       real life, but is still worth protecting against".
 4197       (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
 4198       regular expressions".
 4200       (c) pcre_scanner.cc: avoid use of std::count() because not all systems
 4201       have it.
 4203       (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
 4204       "configure" and the latter not, in order to fix a problem somebody had
 4205       with compiling the Arg class on HP-UX.
 4207       (e) Improve the error-handling of the C++ wrapper a little bit.
 4209       (f) New tests for checking recursion limiting.
 4211  8. The pcre_memmove() function, which is used only if the environment does not
 4212     have a standard memmove() function (and is therefore rarely compiled),
 4213     contained two bugs: (a) use of int instead of size_t, and (b) it was not
 4214     returning a result (though PCRE never actually uses the result).
 4216  9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
 4217     large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
 4218     returned instead of calling malloc() with an overflowing number that would
 4219     most likely cause subsequent chaos.
 4221 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
 4223 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
 4224     with this option is matched, the nmatch and pmatch options of regexec() are
 4225     ignored.
 4227 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
 4228     provided in case anyone wants to the the POSIX interface with UTF-8
 4229     strings.
 4231 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
 4232     C++ linking (needed for some HP-UX environments).
 4234 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
 4235     (unused parameter) and in the pcre_printint() function (omitted "default"
 4236     switch label when the default is to do nothing).
 4238 15. Added some code to make it possible, when PCRE is compiled as a C++
 4239     library, to replace subject pointers for pcre_exec() with a smart pointer
 4240     class, thus making it possible to process discontinuous strings.
 4242 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
 4243     much the same function. They were added by different people who were trying
 4244     to make PCRE easy to compile on non-Unix systems. It has been suggested
 4245     that PCRE_EXPORT be abolished now that there is more automatic apparatus
 4246     for compiling on Windows systems. I have therefore replaced it with
 4247     PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
 4248     defaults to "extern" for C or "extern C" for C++, which works fine on
 4249     Unix-like systems. It is now possible to override the value of PCRE_DATA_
 4250     SCOPE with something explicit in config.h. In addition:
 4252     (a) pcreposix.h still had just "extern" instead of either of these macros;
 4253         I have replaced it with PCRE_DATA_SCOPE.
 4255     (b) Functions such as _pcre_xclass(), which are internal to the library,
 4256         but external in the C sense, all had PCRE_EXPORT in their definitions.
 4257         This is apparently wrong for the Windows case, so I have removed it.
 4258         (It makes no difference on Unix-like systems.)
 4260 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
 4261     of recursive calls to match(). This is different to MATCH_LIMIT because
 4262     that limits the total number of calls to match(), not all of which increase
 4263     the depth of recursion. Limiting the recursion depth limits the amount of
 4264     stack (or heap if NO_RECURSE is set) that is used. The default can be set
 4265     when PCRE is compiled, and changed at run time. A patch from Google adds
 4266     this functionality to the C++ interface.
 4268 18. Changes to the handling of Unicode character properties:
 4270     (a) Updated the table to Unicode 4.1.0.
 4272     (b) Recognize characters that are not in the table as "Cn" (undefined).
 4274     (c) I revised the way the table is implemented to a much improved format
 4275         which includes recognition of ranges. It now supports the ranges that
 4276         are defined in UnicodeData.txt, and it also amalgamates other
 4277         characters into ranges. This has reduced the number of entries in the
 4278         table from around 16,000 to around 3,000, thus reducing its size
 4279         considerably. I realized I did not need to use a tree structure after
 4280         all - a binary chop search is just as efficient. Having reduced the
 4281         number of entries, I extended their size from 6 bytes to 8 bytes to
 4282         allow for more data.
 4284     (d) Added support for Unicode script names via properties such as \p{Han}.
 4286 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
 4287     matching that character.
 4289 20. When matching a repeated Unicode property with a minimum greater than zero,
 4290     (for example \pL{2,}), PCRE could look past the end of the subject if it
 4291     reached it while seeking the minimum number of characters. This could
 4292     happen only if some of the characters were more than one byte long, because
 4293     there is a check for at least the minimum number of bytes.
 4295 21. Refactored the implementation of \p and \P so as to be more general, to
 4296     allow for more different types of property in future. This has changed the
 4297     compiled form incompatibly. Anybody with saved compiled patterns that use
 4298     \p or \P will have to recompile them.
 4300 22. Added "Any" and "L&" to the supported property types.
 4302 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
 4303     but give a compile time error if the value is greater than 0xff.
 4305 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
 4306     accidentally not being installed or uninstalled.
 4308 25. The pcre.h file was built from pcre.h.in, but the only changes that were
 4309     made were to insert the current release number. This seemed silly, because
 4310     it made things harder for people building PCRE on systems that don't run
 4311     "configure". I have turned pcre.h into a distributed file, no longer built
 4312     by "configure", with the version identification directly included. There is
 4313     no longer a pcre.h.in file.
 4315     However, this change necessitated a change to the pcre-config script as
 4316     well. It is built from pcre-config.in, and one of the substitutions was the
 4317     release number. I have updated configure.ac so that ./configure now finds
 4318     the release number by grepping pcre.h.
 4320 26. Added the ability to run the tests under valgrind.
 4323 Version 6.4 05-Sep-05
 4324 ---------------------
 4326  1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
 4327     "--" to be printed when multiple files were scanned, even when none of the
 4328     -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
 4329     consider it to be a bug, and have restored the previous behaviour.
 4331  2. A couple of code tidies to get rid of compiler warnings.
 4333  3. The pcretest program used to cheat by referring to symbols in the library
 4334     whose names begin with _pcre_. These are internal symbols that are not
 4335     really supposed to be visible externally, and in some environments it is
 4336     possible to suppress them. The cheating is now confined to including
 4337     certain files from the library's source, which is a bit cleaner.
 4339  4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
 4340     file's purpose clearer.
 4342  5. Reorganized pcre_ucp_findchar().
 4345 Version 6.3 15-Aug-05
 4346 ---------------------
 4348  1. The file libpcre.pc.in did not have general read permission in the tarball.
 4350  2. There were some problems when building without C++ support:
 4352     (a) If C++ support was not built, "make install" and "make test" still
 4353         tried to test it.
 4355     (b) There were problems when the value of CXX was explicitly set. Some
 4356         changes have been made to try to fix these, and ...
 4358     (c) --disable-cpp can now be used to explicitly disable C++ support.
 4360     (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
 4361         backslash in a target when C++ was disabled. This confuses some
 4362         versions of "make", apparently. Using an intermediate variable solves
 4363         this. (Same for CPP_LOBJ.)
 4365  3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
 4366     (non-Windows) now includes $(CFLAGS) because these flags are sometimes
 4367     necessary on certain architectures.
 4369  4. Added a setting of -export-symbols-regex to the link command to remove
 4370     those symbols that are exported in the C sense, but actually are local
 4371     within the library, and not documented. Their names all begin with
 4372     "_pcre_". This is not a perfect job, because (a) we have to except some
 4373     symbols that pcretest ("illegally") uses, and (b) the facility isn't always
 4374     available (and never for static libraries). I have made a note to try to
 4375     find a way round (a) in the future.
 4378 Version 6.2 01-Aug-05
 4379 ---------------------
 4381  1. There was no test for integer overflow of quantifier values. A construction
 4382     such as {1111111111111111} would give undefined results. What is worse, if
 4383     a minimum quantifier for a parenthesized subpattern overflowed and became
 4384     negative, the calculation of the memory size went wrong. This could have
 4385     led to memory overwriting.
 4387  2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
 4389  3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
 4390     operating environments where this matters.
 4392  4. Applied Giuseppe Maxia's patch to add additional features for controlling
 4393     PCRE options from within the C++ wrapper.
 4395  5. Named capturing subpatterns were not being correctly counted when a pattern
 4396     was compiled. This caused two problems: (a) If there were more than 100
 4397     such subpatterns, the calculation of the memory needed for the whole
 4398     compiled pattern went wrong, leading to an overflow error. (b) Numerical
 4399     back references of the form \12, where the number was greater than 9, were
 4400     not recognized as back references, even though there were sufficient
 4401     previous subpatterns.
 4403  6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
 4404     versions of gcc, e.g. 2.95.4.
 4407 Version 6.1 21-Jun-05
 4408 ---------------------
 4410  1. There was one reference to the variable "posix" in pcretest.c that was not
 4411     surrounded by "#if !defined NOPOSIX".
 4413  2. Make it possible to compile pcretest without DFA support, UTF8 support, or
 4414     the cross-check on the old pcre_info() function, for the benefit of the
 4415     cut-down version of PCRE that is currently imported into Exim.
 4417  3. A (silly) pattern starting with (?i)(?-i) caused an internal space
 4418     allocation error. I've done the easy fix, which wastes 2 bytes for sensible
 4419     patterns that start (?i) but I don't think that matters. The use of (?i) is
 4420     just an example; this all applies to the other options as well.
 4422  4. Since libtool seems to echo the compile commands it is issuing, the output
 4423     from "make" can be reduced a bit by putting "@" in front of each libtool
 4424     compile command.
 4426  5. Patch from the folks at Google for configure.in to be a bit more thorough
 4427     in checking for a suitable C++ installation before trying to compile the
 4428     C++ stuff. This should fix a reported problem when a compiler was present,
 4429     but no suitable headers.
 4431  6. The man pages all had just "PCRE" as their title. I have changed them to
 4432     be the relevant file name. I have also arranged that these names are
 4433     retained in the file doc/pcre.txt, which is a concatenation in text format
 4434     of all the man pages except the little individual ones for each function.
 4436  7. The NON-UNIX-USE file had not been updated for the different set of source
 4437     files that come with release 6. I also added a few comments about the C++
 4438     wrapper.
 4441 Version 6.0 07-Jun-05
 4442 ---------------------
 4444  1. Some minor internal re-organization to help with my DFA experiments.
 4446  2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
 4447     didn't matter for the library itself when fully configured, but did matter
 4448     when compiling without UCP support, or within Exim, where the ucp files are
 4449     not imported.
 4451  3. Refactoring of the library code to split up the various functions into
 4452     different source modules. The addition of the new DFA matching code (see
 4453     below) to a single monolithic source would have made it really too
 4454     unwieldy, quite apart from causing all the code to be include in a
 4455     statically linked application, when only some functions are used. This is
 4456     relevant even without the DFA addition now that patterns can be compiled in
 4457     one application and matched in another.
 4459     The downside of splitting up is that there have to be some external
 4460     functions and data tables that are used internally in different modules of
 4461     the library but which are not part of the API. These have all had their
 4462     names changed to start with "_pcre_" so that they are unlikely to clash
 4463     with other external names.
 4465  4. Added an alternate matching function, pcre_dfa_exec(), which matches using
 4466     a different (DFA) algorithm. Although it is slower than the original
 4467     function, it does have some advantages for certain types of matching
 4468     problem.
 4470  5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
 4471     including restarting after a partial match.
 4473  6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
 4474     defined when compiling for Windows was sent to me. I have put it into the
 4475     code, though I have no means of testing or verifying it.
 4477  7. Added the pcre_refcount() auxiliary function.
 4479  8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
 4480     match before or at the first newline in the subject string. In pcretest,
 4481     the /f option on a pattern can be used to set this.
 4483  9. A repeated \w when used in UTF-8 mode with characters greater than 256
 4484     would behave wrongly. This has been present in PCRE since release 4.0.
 4486 10. A number of changes to the pcregrep command:
 4488     (a) Refactored how -x works; insert ^(...)$ instead of setting
 4489         PCRE_ANCHORED and checking the length, in preparation for adding
 4490         something similar for -w.
 4492     (b) Added the -w (match as a word) option.
 4494     (c) Refactored the way lines are read and buffered so as to have more
 4495         than one at a time available.
 4497     (d) Implemented a pcregrep test script.
 4499     (e) Added the -M (multiline match) option. This allows patterns to match
 4500         over several lines of the subject. The buffering ensures that at least
 4501         8K, or the rest of the document (whichever is the shorter) is available
 4502         for matching (and similarly the previous 8K for lookbehind assertions).
 4504     (f) Changed the --help output so that it now says
 4506           -w, --word-regex(p)
 4508         instead of two lines, one with "regex" and the other with "regexp"
 4509         because that confused at least one person since the short forms are the
 4510         same. (This required a bit of code, as the output is generated
 4511         automatically from a table. It wasn't just a text change.)
 4513     (g) -- can be used to terminate pcregrep options if the next thing isn't an
 4514         option but starts with a hyphen. Could be a pattern or a path name
 4515         starting with a hyphen, for instance.
 4517     (h) "-" can be given as a file name to represent stdin.
 4519     (i) When file names are being printed, "(standard input)" is used for
 4520         the standard input, for compatibility with GNU grep. Previously
 4521         "<stdin>" was used.
 4523     (j) The option --label=xxx can be used to supply a name to be used for
 4524         stdin when file names are being printed. There is no short form.
 4526     (k) Re-factored the options decoding logic because we are going to add
 4527         two more options that take data. Such options can now be given in four
 4528         different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
 4530     (l) Added the -A, -B, and -C options for requesting that lines of context
 4531         around matches be printed.
 4533     (m) Added the -L option to print the names of files that do not contain
 4534         any matching lines, that is, the complement of -l.
 4536     (n) The return code is 2 if any file cannot be opened, but pcregrep does
 4537         continue to scan other files.
 4539     (o) The -s option was incorrectly implemented. For compatibility with other
 4540         greps, it now suppresses the error message for a non-existent or non-
 4541         accessible file (but not the return code). There is a new option called
 4542         -q that suppresses the output of matching lines, which was what -s was
 4543         previously doing.
 4545     (p) Added --include and --exclude options to specify files for inclusion
 4546         and exclusion when recursing.
 4548 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
 4549     Hopefully, it now does.
 4551 12. Missing cast in pcre_study().
 4553 13. Added an "uninstall" target to the makefile.
 4555 14. Replaced "extern" in the function prototypes in Makefile.in with
 4556     "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
 4557     world, but is set differently for Windows.
 4559 15. Added a second compiling function called pcre_compile2(). The only
 4560     difference is that it has an extra argument, which is a pointer to an
 4561     integer error code. When there is a compile-time failure, this is set
 4562     non-zero, in addition to the error test pointer being set to point to an
 4563     error message. The new argument may be NULL if no error number is required
 4564     (but then you may as well call pcre_compile(), which is now just a
 4565     wrapper). This facility is provided because some applications need a
 4566     numeric error indication, but it has also enabled me to tidy up the way
 4567     compile-time errors are handled in the POSIX wrapper.
 4569 16. Added VPATH=.libs to the makefile; this should help when building with one
 4570     prefix path and installing with another. (Or so I'm told by someone who
 4571     knows more about this stuff than I do.)
 4573 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
 4574     passes PCRE_DOTALL to the pcre_compile() function, making the "." character
 4575     match everything, including newlines. This is not POSIX-compatible, but
 4576     somebody wanted the feature. From pcretest it can be activated by using
 4577     both the P and the s flags.
 4579 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
 4581 19. libpcre.pc was being incorrectly installed as executable.
 4583 20. A couple of places in pcretest check for end-of-line by looking for '\n';
 4584     it now also looks for '\r' so that it will work unmodified on Windows.
 4586 21. Added Google's contributed C++ wrapper to the distribution.
 4588 22. Added some untidy missing memory free() calls in pcretest, to keep
 4589     Electric Fence happy when testing.
 4593 Version 5.0 13-Sep-04
 4594 ---------------------
 4596  1. Internal change: literal characters are no longer packed up into items
 4597     containing multiple characters in a single byte-string. Each character
 4598     is now matched using a separate opcode. However, there may be more than one
 4599     byte in the character in UTF-8 mode.
 4601  2. The pcre_callout_block structure has two new fields: pattern_position and
 4602     next_item_length. These contain the offset in the pattern to the next match
 4603     item, and its length, respectively.
 4605  3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
 4606     insertion of callouts before each pattern item. Added the /C option to
 4607     pcretest to make use of this.
 4609  4. On the advice of a Windows user, the lines
 4611       #if defined(_WIN32) || defined(WIN32)
 4612       _setmode( _fileno( stdout ), 0x8000 );
 4613       #endif  /* defined(_WIN32) || defined(WIN32) */
 4615     have been added to the source of pcretest. This apparently does useful
 4616     magic in relation to line terminators.
 4618  5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
 4619     for the benefit of those environments where the "b" makes a difference.
 4621  6. The icc compiler has the same options as gcc, but "configure" doesn't seem
 4622     to know about it. I have put a hack into configure.in that adds in code
 4623     to set GCC=yes if CC=icc. This seems to end up at a point in the
 4624     generated configure script that is early enough to affect the setting of
 4625     compiler options, which is what is needed, but I have no means of testing
 4626     whether it really works. (The user who reported this had patched the
 4627     generated configure script, which of course I cannot do.)
 4629     LATER: After change 22 below (new libtool files), the configure script
 4630     seems to know about icc (and also ecc). Therefore, I have commented out
 4631     this hack in configure.in.
 4633  7. Added support for pkg-config (2 patches were sent in).
 4635  8. Negated POSIX character classes that used a combination of internal tables
 4636     were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
 4637     [[:^ascii]]. Typically, they would match almost any characters. The other
 4638     POSIX classes were not broken in this way.
 4640  9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
 4641     to find the match, as PCRE was deluded into thinking that the match had to
 4642     start at the start point or following a newline. The same bug applied to
 4643     patterns with negative forward assertions or any backward assertions
 4644     preceding ".*" at the start, unless the pattern required a fixed first
 4645     character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
 4647 10. In UTF-8 mode, when moving forwards in the subject after a failed match
 4648     starting at the last subject character, bytes beyond the end of the subject
 4649     string were read.
 4651 11. Renamed the variable "class" as "classbits" to make life easier for C++
 4652     users. (Previously there was a macro definition, but it apparently wasn't
 4653     enough.)
 4655 12. Added the new field "tables" to the extra data so that tables can be passed
 4656     in at exec time, or the internal tables can be re-selected. This allows
 4657     a compiled regex to be saved and re-used at a later time by a different
 4658     program that might have everything at different addresses.
 4660 13. Modified the pcre-config script so that, when run on Solaris, it shows a
 4661     -R library as well as a -L library.
 4663 14. The debugging options of pcretest (-d on the command line or D on a
 4664     pattern) showed incorrect output for anything following an extended class
 4665     that contained multibyte characters and which was followed by a quantifier.
 4667 15. Added optional support for general category Unicode character properties
 4668     via the \p, \P, and \X escapes. Unicode property support implies UTF-8
 4669     support. It adds about 90K to the size of the library. The meanings of the
 4670     inbuilt class escapes such as \d and \s have NOT been changed.
 4672 16. Updated pcredemo.c to include calls to free() to release the memory for the
 4673     compiled pattern.
 4675 17. The generated file chartables.c was being created in the source directory
 4676     instead of in the building directory. This caused the build to fail if the
 4677     source directory was different from the building directory, and was
 4678     read-only.
 4680 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
 4681     file. No doubt somebody will tell me if they don't make sense... Also added
 4682     Dan Mooney's comments about building on OpenVMS.
 4684 19. Added support for partial matching via the PCRE_PARTIAL option for
 4685     pcre_exec() and the \P data escape in pcretest.
 4687 20. Extended pcretest with 3 new pattern features:
 4689     (i)   A pattern option of the form ">rest-of-line" causes pcretest to
 4690           write the compiled pattern to the file whose name is "rest-of-line".
 4691           This is a straight binary dump of the data, with the saved pointer to
 4692           the character tables forced to be NULL. The study data, if any, is
 4693           written too. After writing, pcretest reads a new pattern.
 4695     (ii)  If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
 4696           compiled pattern from the given file. There must not be any
 4697           occurrences of "<" in the file name (pretty unlikely); if there are,
 4698           pcretest will instead treat the initial "<" as a pattern delimiter.
 4699           After reading in the pattern, pcretest goes on to read data lines as
 4700           usual.
 4702     (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
 4703           and 16-bit fields in a compiled pattern, to simulate a pattern that
 4704           was compiled on a host of opposite endianness.
 4706 21. The pcre-exec() function can now cope with patterns that were compiled on
 4707     hosts of opposite endianness, with this restriction:
 4709       As for any compiled expression that is saved and used later, the tables
 4710       pointer field cannot be preserved; the extra_data field in the arguments
 4711       to pcre_exec() should be used to pass in a tables address if a value
 4712       other than the default internal tables were used at compile time.
 4714 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
 4715     now diagnosed as an error. Previously, most of the time, a negative number
 4716     would have been treated as zero, but if in addition "ovector" was passed as
 4717     NULL, a crash could occur.
 4719 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
 4720     new versions from the libtool 1.5 distribution (the last one is a copy of
 4721     a file called libtool.m4). This seems to have fixed the need to patch
 4722     "configure" to support Darwin 1.3 (which I used to do). However, I still
 4723     had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
 4724     workstation).
 4726 24. Changed the PCRE licence to be the more standard "BSD" licence.
 4729 Version 4.5 01-Dec-03
 4730 ---------------------
 4732  1. There has been some re-arrangement of the code for the match() function so
 4733     that it can be compiled in a version that does not call itself recursively.
 4734     Instead, it keeps those local variables that need separate instances for
 4735     each "recursion" in a frame on the heap, and gets/frees frames whenever it
 4736     needs to "recurse". Keeping track of where control must go is done by means
 4737     of setjmp/longjmp. The whole thing is implemented by a set of macros that
 4738     hide most of the details from the main code, and operates only if
 4739     NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
 4740     "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
 4741     operating.
 4743     To make it easier for callers to provide specially tailored get/free
 4744     functions for this usage, two new functions, pcre_stack_malloc, and
 4745     pcre_stack_free, are used. They are always called in strict stacking order,
 4746     and the size of block requested is always the same.
 4748     The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
 4749     PCRE has been compiled to use the stack or the heap for recursion. The
 4750     -C option of pcretest uses this to show which version is compiled.
 4752     A new data escape \S, is added to pcretest; it causes the amounts of store
 4753     obtained and freed by both kinds of malloc/free at match time to be added
 4754     to the output.
 4756  2. Changed the locale test to use "fr_FR" instead of "fr" because that's
 4757     what's available on my current Linux desktop machine.
 4759  3. When matching a UTF-8 string, the test for a valid string at the start has
 4760     been extended. If start_offset is not zero, PCRE now checks that it points
 4761     to a byte that is the start of a UTF-8 character. If not, it returns
 4762     PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
 4763     this is necessary because there may be backward assertions in the pattern.
 4764     When matching the same subject several times, it may save resources to use
 4765     PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
 4767  4. The code for checking the validity of UTF-8 strings has been tightened so
 4768     that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
 4769     containing "overlong sequences".
 4771  5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
 4772     I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
 4773     should have been "&", but it just so happened that all the cases this let
 4774     through by mistake were picked up later in the function.
 4776  6. I had used a variable called "isblank" - this is a C99 function, causing
 4777     some compilers to warn. To avoid this, I renamed it (as "blankclass").
 4779  7. Cosmetic: (a) only output another newline at the end of pcretest if it is
 4780     prompting; (b) run "./pcretest /dev/null" at the start of the test script
 4781     so the version is shown; (c) stop "make test" echoing "./RunTest".
 4783  8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
 4785  9. The prototype for memmove() for systems that don't have it was using
 4786     size_t, but the inclusion of the header that defines size_t was later. I've
 4787     moved the #includes for the C headers earlier to avoid this.
 4789 10. Added some adjustments to the code to make it easier to compiler on certain
 4790     special systems:
 4792       (a) Some "const" qualifiers were missing.
 4793       (b) Added the macro EXPORT before all exported functions; by default this
 4794           is defined to be empty.
 4795       (c) Changed the dftables auxiliary program (that builds chartables.c) so
 4796           that it reads its output file name as an argument instead of writing
 4797           to the standard output and assuming this can be redirected.
 4799 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
 4800     class containing characters with values greater than 255, PCRE compilation
 4801     went into a loop.
 4803 12. A recursive reference to a subpattern that was within another subpattern
 4804     that had a minimum quantifier of zero caused PCRE to crash. For example,
 4805     (x(y(?2))z)? provoked this bug with a subject that got as far as the
 4806     recursion. If the recursively-called subpattern itself had a zero repeat,
 4807     that was OK.
 4809 13. In pcretest, the buffer for reading a data line was set at 30K, but the
 4810     buffer into which it was copied (for escape processing) was still set at
 4811     1024, so long lines caused crashes.
 4813 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
 4814     "internal error: code overflow...". This applied to any character class
 4815     that was followed by a possessive quantifier.
 4817 15. Modified the Makefile to add libpcre.la as a prerequisite for
 4818     libpcreposix.la because I was told this is needed for a parallel build to
 4819     work.
 4821 16. If a pattern that contained .* following optional items at the start was
 4822     studied, the wrong optimizing data was generated, leading to matching
 4823     errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
 4824     matching string must start with a or b or c. The correct conclusion for
 4825     this pattern is that a match can start with any character.
 4828 Version 4.4 13-Aug-03
 4829 ---------------------
 4831  1. In UTF-8 mode, a character class containing characters with values between
 4832     127 and 255 was not handled correctly if the compiled pattern was studied.
 4833     In fixing this, I have also improved the studying algorithm for such
 4834     classes (slightly).
 4836  2. Three internal functions had redundant arguments passed to them. Removal
 4837     might give a very teeny performance improvement.
 4839  3. Documentation bug: the value of the capture_top field in a callout is *one
 4840     more than* the number of the hightest numbered captured substring.
 4842  4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
 4843     in incorrectly linking with a previously installed version. They now link
 4844     explicitly with libpcre.la.
 4846  5. configure.in no longer needs to recognize Cygwin specially.
 4848  6. A problem in pcre.in for Windows platforms is fixed.
 4850  7. If a pattern was successfully studied, and the -d (or /D) flag was given to
 4851     pcretest, it used to include the size of the study block as part of its
 4852     output. Unfortunately, the structure contains a field that has a different
 4853     size on different hardware architectures. This meant that the tests that
 4854     showed this size failed. As the block is currently always of a fixed size,
 4855     this information isn't actually particularly useful in pcretest output, so
 4856     I have just removed it.
 4858  8. Three pre-processor statements accidentally did not start in column 1.
 4859     Sadly, there are *still* compilers around that complain, even though
 4860     standard C has not required this for well over a decade. Sigh.
 4862  9. In pcretest, the code for checking callouts passed small integers in the
 4863     callout_data field, which is a void * field. However, some picky compilers
 4864     complained about the casts involved for this on 64-bit systems. Now
 4865     pcretest passes the address of the small integer instead, which should get
 4866     rid of the warnings.
 4868 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
 4869     both compile and run time, and gives an error if an invalid UTF-8 sequence
 4870     is found. There is a option for disabling this check in cases where the
 4871     string is known to be correct and/or the maximum performance is wanted.
 4873 11. In response to a bug report, I changed one line in Makefile.in from
 4875         -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
 4876     to
 4877         -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
 4879     to look similar to other lines, but I have no way of telling whether this
 4880     is the right thing to do, as I do not use Windows. No doubt I'll get told
 4881     if it's wrong...
 4884 Version 4.3 21-May-03
 4885 ---------------------
 4887 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
 4888    Makefile.
 4890 2. Some refactoring to improve the quality of the code:
 4892    (i)   The utf8_table... variables are now declared "const".
 4894    (ii)  The code for \cx, which used the "case flipping" table to upper case
 4895          lower case letters, now just substracts 32. This is ASCII-specific,
 4896          but the whole concept of \cx is ASCII-specific, so it seems
 4897          reasonable.
 4899    (iii) PCRE was using its character types table to recognize decimal and
 4900          hexadecimal digits in the pattern. This is silly, because it handles
 4901          only 0-9, a-f, and A-F, but the character types table is locale-
 4902          specific, which means strange things might happen. A private
 4903          table is now used for this - though it costs 256 bytes, a table is
 4904          much faster than multiple explicit tests. Of course, the standard
 4905          character types table is still used for matching digits in subject
 4906          strings against \d.
 4908    (iv)  Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
 4909          ending in _t are). So I've renamed it as ESC_tee.
 4911 3. The first argument for regexec() in the POSIX wrapper should have been
 4912    defined as "const".
 4914 4. Changed pcretest to use malloc() for its buffers so that they can be
 4915    Electric Fenced for debugging.
 4917 5. There were several places in the code where, in UTF-8 mode, PCRE would try
 4918    to read one or more bytes before the start of the subject string. Often this
 4919    had no effect on PCRE's behaviour, but in some circumstances it could
 4920    provoke a segmentation fault.
 4922 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
 4923    to try to read one or more bytes before the start of the subject string.
 4925 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
 4926    UTF-8 support could misbehave in various ways if the subject string
 4927    contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
 4928    area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
 4929    back over UTF-8 characters.)
 4932 Version 4.2 14-Apr-03
 4933 ---------------------
 4935 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
 4937 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
 4938      [ON_WINDOWS]: new variable, "#" on non-Windows platforms
 4939      [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
 4940      [WIN_PREFIX]: new variable, "cyg" for Cygwin
 4941      * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
 4942        and BUILD_EXEEXT
 4943      Note: automatic setting of the BUILD variables is not yet working
 4944      set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
 4945        compile-time but not at link-time
 4946      [LINK]: use for linking executables only
 4947      make different versions for Windows and non-Windows
 4948      [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
 4949        libraries
 4950      [LINK_FOR_BUILD]: new variable
 4951      [OBJEXT]: use throughout
 4952      [EXEEXT]: use throughout
 4953      <winshared>: new target
 4954      <wininstall>: new target
 4955      <dftables.o>: use native compiler
 4956      <dftables>: use native linker
 4957      <install>: handle Windows platform correctly
 4958      <clean>: ditto
 4959      <check>: ditto
 4960      copy DLL to top builddir before testing
 4962    As part of these changes, -no-undefined was removed again. This was reported
 4963    to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
 4964    in any case.
 4966 3. Some tidies to get rid of compiler warnings:
 4968    . In the match_data structure, match_limit was an unsigned long int, whereas
 4969      match_call_count was an int. I've made them both unsigned long ints.
 4971    . In pcretest the fact that a const uschar * doesn't automatically cast to
 4972      a void * provoked a warning.
 4974    . Turning on some more compiler warnings threw up some "shadow" variables
 4975      and a few more missing casts.
 4977 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
 4978    option, a class that contained a single character with a value between 128
 4979    and 255 (e.g. /[\xFF]/) caused PCRE to crash.
 4981 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
 4982    option, a class that contained several characters, but with at least one
 4983    whose value was between 128 and 255 caused PCRE to crash.
 4986 Version 4.1 12-Mar-03
 4987 ---------------------
 4989 1. Compiling with gcc -pedantic found a couple of places where casts were
 4990 needed, and a string in dftables.c that was longer than standard compilers are
 4991 required to support.
 4993 2. Compiling with Sun's compiler found a few more places where the code could
 4994 be tidied up in order to avoid warnings.
 4996 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
 4997 first of these names is deprecated in the latest Autoconf in favour of the name
 4998 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
 4999 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
 5000 analogy I have changed it to CFLAGS_FOR_BUILD.
 5002 4. Added -no-undefined to the linking command in the Makefile, because this is
 5003 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
 5004 linking step for the pcreposix library.
 5006 5. PCRE was failing to diagnose the case of two named groups with the same
 5007 name.
 5009 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
 5010 literal character that is needed in the subject for a match, and scans along to
 5011 ensure that it is present before embarking on the full matching process. This
 5012 saves time in cases of nested unlimited repeats that are never going to match.
 5013 Problem: the scan can take a lot of time if the subject is very long (e.g.
 5014 megabytes), thus penalizing straightforward matches. It is now done only if the
 5015 amount of subject to be scanned is less than 1000 bytes.
 5017 7. A lesser problem with the same optimization is that it was recording the
 5018 first character of an anchored pattern as "needed", thus provoking a search
 5019 right along the subject, even when the first match of the pattern was going to
 5020 fail. The "needed" character is now not set for anchored patterns, unless it
 5021 follows something in the pattern that is of non-fixed length. Thus, it still
 5022 fulfils its original purpose of finding quick non-matches in cases of nested
 5023 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
 5026 Version 4.0 17-Feb-03
 5027 ---------------------
 5029 1. If a comment in an extended regex that started immediately after a meta-item
 5030 extended to the end of string, PCRE compiled incorrect data. This could lead to
 5031 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
 5033 2. Moved to autoconf 2.53 and libtool 1.4.2.
 5035 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
 5036 the special perltest8 script is no longer needed - all the tests can be run
 5037 from a single perltest script.
 5039 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
 5040 by \s. It has now been removed in PCRE. This means it isn't recognized as
 5041 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
 5042 class [:space:] *does* include VT, thereby creating a mess.
 5044 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
 5045 space and tab.
 5047 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
 5048 its new features into the main test script, reducing the number of scripts.
 5050 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
 5051 were backward compatible, and made the (?i) apply to the whole pattern, as if
 5052 /i were given. Now it behaves more logically, and applies the option setting
 5053 only to what follows. PCRE has been changed to follow suit. However, if it
 5054 finds options settings right at the start of the pattern, it extracts them into
 5055 the global options, as before. Thus, they show up in the info data.
 5057 8. Added support for the \Q...\E escape sequence. Characters in between are
 5058 treated as literals. This is slightly different from Perl in that $ and @ are
 5059 also handled as literals inside the quotes. In Perl, they will cause variable
 5060 interpolation. Note the following examples:
 5062     Pattern            PCRE matches      Perl matches
 5064     \Qabc$xyz\E        abc$xyz           abc followed by the contents of $xyz
 5065     \Qabc\$xyz\E       abc\$xyz          abc\$xyz
 5066     \Qabc\E\$\Qxyz\E   abc$xyz           abc$xyz
 5068 For compatibility with Perl, \Q...\E sequences are recognized inside character
 5069 classes as well as outside them.
 5071 9. Re-organized 3 code statements in pcretest to avoid "overflow in
 5072 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
 5073 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
 5074 signed/unsigned warnings.
 5076 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
 5077 option for pcretest, so I've replaced it by a simple function that does just
 5078 that job.
 5080 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
 5081 "pcregrep -".
 5083 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
 5084 Java package. This provides some syntactic sugar for simple cases of what my
 5085 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
 5086 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
 5087 item, you can use this simplified notation. Note that only makes sense with
 5088 greedy quantifiers. Consequently, the use of the possessive quantifier forces
 5089 greediness, whatever the setting of the PCRE_UNGREEDY option.
 5091 13. A change of greediness default within a pattern was not taking effect at
 5092 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
 5093 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
 5094 was abstracted outside.
 5096 14. PCRE now supports the \G assertion. It is true when the current matching
 5097 position is at the start point of the match. This differs from \A when the
 5098 starting offset is non-zero. Used with the /g option of pcretest (or similar
 5099 code), it works in the same way as it does for Perl's /g option. If all
 5100 alternatives of a regex begin with \G, the expression is anchored to the start
 5101 match position, and the "anchored" flag is set in the compiled expression.
 5103 15. Some bugs concerning the handling of certain option changes within patterns
 5104 have been fixed. These applied to options other than (?ims). For example,
 5105 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
 5106 been the other way round. Some of this was related to change 7 above.
 5108 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
 5109 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
 5110 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
 5111 POSIX classes only within a class (e.g. /[[:alpha:]]/).
 5113 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
 5114 mode. Unlike ".", it always matches newline, whatever the setting of
 5115 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
 5116 assertions. Perl allows it, but it doesn't (in general) work because it can't
 5117 calculate the length of the lookbehind. At least, that's the case for Perl
 5118 5.8.0 - I've been told they are going to document that it doesn't work in
 5119 future.
 5121 18. Added an error diagnosis for escapes that PCRE does not support: these are
 5122 \L, \l, \N, \P, \p, \U, \u, and \X.
 5124 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
 5125 reading past the end of the pattern in cases such as /[abcd/.
 5127 20. PCRE was getting more memory than necessary for patterns with classes that
 5128 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
 5130 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
 5131 compiling PCRE for use with Virtual Pascal.
 5133 22. Small fix to the Makefile to make it work properly if the build is done
 5134 outside the source tree.
 5136 23. Added a new extension: a condition to go with recursion. If a conditional
 5137 subpattern starts with (?(R) the "true" branch is used if recursion has
 5138 happened, whereas the "false" branch is used only at the top level.
 5140 24. When there was a very long string of literal characters (over 255 bytes
 5141 without UTF support, over 250 bytes with UTF support), the computation of how
 5142 much memory was required could be incorrect, leading to segfaults or other
 5143 strange effects.
 5145 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
 5146 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
 5147 there was a subsequent back reference to those brackets. This meant that, for
 5148 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
 5149 possible to check for precisely this case. All we can do is abandon the
 5150 optimization if .* occurs inside capturing brackets when there are any back
 5151 references whatsoever. (See below for a better fix that came later.)
 5153 26. The handling of the optimization for finding the first character of a
 5154 non-anchored pattern, and for finding a character that is required later in the
 5155 match were failing in some cases. This didn't break the matching; it just
 5156 failed to optimize when it could. The way this is done has been re-implemented.
 5158 27. Fixed typo in error message for invalid (?R item (it said "(?p").
 5160 28. Added a new feature that provides some of the functionality that Perl
 5161 provides with (?{...}). The facility is termed a "callout". The way it is done
 5162 in PCRE is for the caller to provide an optional function, by setting
 5163 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
 5164 global variable. By default it is unset, which disables all calling out. To get
 5165 the function called, the regex must include (?C) at appropriate points. This
 5166 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
 5167 This provides a means of identifying different callout points. When PCRE
 5168 reaches such a point in the regex, if pcre_callout has been set, the external
 5169 function is called. It is provided with data in a structure called
 5170 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
 5171 matching continues; if it returns a non-zero value, the match at the current
 5172 point fails. However, backtracking will occur if possible. [This was changed
 5173 later and other features added - see item 49 below.]
 5175 29. pcretest is upgraded to test the callout functionality. It provides a
 5176 callout function that displays information. By default, it shows the start of
 5177 the match and the current position in the text. There are some new data escapes
 5178 to vary what happens:
 5180     \C+         in addition, show current contents of captured substrings
 5181     \C-         do not supply a callout function
 5182     \C!n        return 1 when callout number n is reached
 5183     \C!n!m      return 1 when callout number n is reached for the mth time
 5185 30. If pcregrep was called with the -l option and just a single file name, it
 5186 output "<stdin>" if a match was found, instead of the file name.
 5188 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
 5189 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
 5190 pcre_exec(). This saves a malloc/free per call. The default value of
 5191 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
 5192 when configuring.
 5194 32. The default maximum size of a compiled pattern is 64K. There have been a
 5195 few cases of people hitting this limit. The code now uses macros to handle the
 5196 storing of links as offsets within the compiled pattern. It defaults to 2-byte
 5197 links, but this can be changed to 3 or 4 bytes by --with-link-size when
 5198 configuring. Tests 2 and 5 work only with 2-byte links because they output
 5199 debugging information about compiled patterns.
 5201 33. Internal code re-arrangements:
 5203 (a) Moved the debugging function for printing out a compiled regex into
 5204     its own source file (printint.c) and used #include to pull it into
 5205     pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
 5206     separate copies.
 5208 (b) Defined the list of op-code names for debugging as a macro in
 5209     internal.h so that it is next to the definition of the opcodes.
 5211 (c) Defined a table of op-code lengths for simpler skipping along compiled
 5212     code. This is again a macro in internal.h so that it is next to the
 5213     definition of the opcodes.
 5215 34. Added support for recursive calls to individual subpatterns, along the
 5216 lines of Robin Houston's patch (but implemented somewhat differently).
 5218 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
 5219 allow it to read and process whole directories in Win32. This code was
 5220 contributed by Lionel Fourquaux; it has not been tested by me.
 5222 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
 5223 used to name a group. Names consist of alphanumerics and underscores, and must
 5224 be unique. Back references use the syntax (?P=name) and recursive calls use
 5225 (?P>name) which is a PCRE extension to the Python extension. Groups still have
 5226 numbers. The function pcre_fullinfo() can be used after compilation to extract
 5227 a name/number map. There are three relevant calls:
 5229   PCRE_INFO_NAMEENTRYSIZE        yields the size of each entry in the map
 5230   PCRE_INFO_NAMECOUNT            yields the number of entries
 5231   PCRE_INFO_NAMETABLE            yields a pointer to the map.
 5233 The map is a vector of fixed-size entries. The size of each entry depends on
 5234 the length of the longest name used. The first two bytes of each entry are the
 5235 group number, most significant byte first. There follows the corresponding
 5236 name, zero terminated. The names are in alphabetical order.
 5238 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
 5239 case instead of 255. Making it the same both with and without UTF-8 support
 5240 means that the same test output works with both.
 5242 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
 5243 calling malloc() with a zero argument.
 5245 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
 5246 optimization. I've improved things by keeping a bitmap of backreferences with
 5247 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
 5248 fact referenced, the optimization can be applied. It is unlikely that a
 5249 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
 5250 the match to follow \n) will appear inside brackets with a number greater than
 5251 31, but if it does, any back reference > 31 suppresses the optimization.
 5253 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
 5254 of disabling numbered capturing parentheses. Any opening parenthesis that is
 5255 not followed by ? behaves as if it were followed by ?: but named parentheses
 5256 can still be used for capturing (and they will acquire numbers in the usual
 5257 way).
 5259 41. Redesigned the return codes from the match() function into yes/no/error so
 5260 that errors can be passed back from deep inside the nested calls. A malloc
 5261 failure while inside a recursive subpattern call now causes the
 5262 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
 5264 42. It is now possible to set a limit on the number of times the match()
 5265 function is called in a call to pcre_exec(). This facility makes it possible to
 5266 limit the amount of recursion and backtracking, though not in a directly
 5267 obvious way, because the match() function is used in a number of different
 5268 circumstances. The count starts from zero for each position in the subject
 5269 string (for non-anchored patterns). The default limit is, for compatibility, a
 5270 large number, namely 10 000 000. You can change this in two ways:
 5272 (a) When configuring PCRE before making, you can use --with-match-limit=n
 5273     to set a default value for the compiled library.
 5275 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
 5276     a different value is set. See 45 below.
 5278 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
 5280 43. Added a new function pcre_config(int, void *) to enable run-time extraction
 5281 of things that can be changed at compile time. The first argument specifies
 5282 what is wanted and the second points to where the information is to be placed.
 5283 The current list of available information is:
 5287 The output is an integer that is set to one if UTF-8 support is available;
 5288 otherwise it is set to zero.
 5292 The output is an integer that it set to the value of the code that is used for
 5293 newline. It is either LF (10) or CR (13).
 5297 The output is an integer that contains the number of bytes used for internal
 5298 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
 5302 The output is an integer that contains the threshold above which the POSIX
 5303 interface uses malloc() for output vectors. See item 31 above.
 5307 The output is an unsigned integer that contains the default limit of the number
 5308 of match() calls in a pcre_exec() execution. See 42 above.
 5310 44. pcretest has been upgraded by the addition of the -C option. This causes it
 5311 to extract all the available output from the new pcre_config() function, and to
 5312 output it. The program then exits immediately.
 5314 45. A need has arisen to pass over additional data with calls to pcre_exec() in
 5315 order to support additional features. One way would have been to define
 5316 pcre_exec2() (for example) with extra arguments, but this would not have been
 5317 extensible, and would also have required all calls to the original function to
 5318 be mapped to the new one. Instead, I have chosen to extend the mechanism that
 5319 is used for passing in "extra" data from pcre_study().
 5321 The pcre_extra structure is now exposed and defined in pcre.h. It currently
 5322 contains the following fields:
 5324   flags         a bitmap indicating which of the following fields are set
 5325   study_data    opaque data from pcre_study()
 5326   match_limit   a way of specifying a limit on match() calls for a specific
 5327                   call to pcre_exec()
 5328   callout_data  data for callouts (see 49 below)
 5330 The flag bits are also defined in pcre.h, and are
 5336 The pcre_study() function now returns one of these new pcre_extra blocks, with
 5337 the actual study data pointed to by the study_data field, and the
 5338 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
 5339 before. That is, this change is entirely upwards-compatible and requires no
 5340 change to existing code.
 5342 If you want to pass in additional data to pcre_exec(), you can either place it
 5343 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
 5344 block.
 5346 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
 5347 data string contains the escape sequence \M, pcretest calls pcre_exec() several
 5348 times with different match limits, until it finds the minimum value needed for
 5349 pcre_exec() to complete. The value is then output. This can be instructive; for
 5350 most simple matches the number is quite small, but for pathological cases it
 5351 gets very large very quickly.
 5353 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
 5354 returns the size of the data block pointed to by the study_data field in a
 5355 pcre_extra block, that is, the value that was passed as the argument to
 5356 pcre_malloc() when PCRE was getting memory in which to place the information
 5357 created by pcre_study(). The fourth argument should point to a size_t variable.
 5358 pcretest has been extended so that this information is shown after a successful
 5359 pcre_study() call when information about the compiled regex is being displayed.
 5361 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
 5362 because what follows is always an absolute path. (Later: it turns out that this
 5363 is more than cosmetic for MinGW, because it doesn't like empty path
 5364 components.)
 5366 49. Some changes have been made to the callout feature (see 28 above):
 5368 (i)  A callout function now has three choices for what it returns:
 5370        0  =>  success, carry on matching
 5371      > 0  =>  failure at this point, but backtrack if possible
 5372      < 0  =>  serious error, return this value from pcre_exec()
 5374      Negative values should normally be chosen from the set of PCRE_ERROR_xxx
 5375      values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
 5376      "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
 5377      use by callout functions. It will never be used by PCRE itself.
 5379 (ii) The pcre_extra structure (see 45 above) has a void * field called
 5380      callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
 5381      pcre_callout_block structure has a field of the same name. The contents of
 5382      the field passed in the pcre_extra structure are passed to the callout
 5383      function in the corresponding field in the callout block. This makes it
 5384      easier to use the same callout-containing regex from multiple threads. For
 5385      testing, the pcretest program has a new data escape
 5387        \C*n        pass the number n (may be negative) as callout_data
 5389      If the callout function in pcretest receives a non-zero value as
 5390      callout_data, it returns that value.
 5392 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
 5393 there were some redundant $(CFLAGS) in commands that are now specified as
 5394 $(LINK), which already includes $(CFLAGS).
 5396 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
 5397 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
 5398 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
 5399 one-byte characters throughout. Note that case-insensitive matching applies
 5400 only to characters whose values are less than 256. PCRE doesn't support the
 5401 notion of cases for higher-valued characters.
 5403 (i)   A character class whose characters are all within 0-255 is handled as
 5404       a bit map, and the map is inverted for negative classes. Previously, a
 5405       character > 255 always failed to match such a class; however it should
 5406       match if the class was a negative one (e.g. [^ab]). This has been fixed.
 5408 (ii)  A negated character class with a single character < 255 is coded as
 5409       "not this character" (OP_NOT). This wasn't working properly when the test
 5410       character was multibyte, either singly or repeated.
 5412 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
 5413       mode, for example: \x{100}{2,3}.
 5415 (iv)  The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
 5416       singly or repeated) now correctly test multibyte characters. However,
 5417       PCRE doesn't recognize any characters with values greater than 255 as
 5418       digits, spaces, or word characters. Such characters always match \D, \S,
 5419       and \W, and never match \d, \s, or \w.
 5421 (v)   Classes may now contain characters and character ranges with values
 5422       greater than 255. For example: [ab\x{100}-\x{400}].
 5424 (vi)  pcregrep now has a --utf-8 option (synonym -u) which makes it call
 5425       PCRE in UTF-8 mode.
 5427 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
 5428 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
 5429 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
 5430 value.)
 5432 53. The single man page has become too large. I have therefore split it up into
 5433 a number of separate man pages. These also give rise to individual HTML pages;
 5434 these are now put in a separate directory, and there is an index.html page that
 5435 lists them all. Some hyperlinking between the pages has been installed.
 5437 54. Added convenience functions for handling named capturing parentheses.
 5439 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
 5440 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
 5441 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
 5442 are faulted.
 5444 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
 5445 calling configure. These values are used when compiling the dftables.c program
 5446 which is run to generate the source of the default character tables. They
 5447 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
 5448 you will need to set these values.
 5450 57. Updated the building process for Windows DLL, as provided by Fred Cox.
 5453 Version 3.9 02-Jan-02
 5454 ---------------------
 5456 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
 5458 2. If --disable-static was given, the building process failed when trying to
 5459 build pcretest and pcregrep. (For some reason it was using libtool to compile
 5460 them, which is not right, as they aren't part of the library.)
 5463 Version 3.8 18-Dec-01
 5464 ---------------------
 5466 1. The experimental UTF-8 code was completely screwed up. It was packing the
 5467 bytes in the wrong order. How dumb can you get?
 5470 Version 3.7 29-Oct-01
 5471 ---------------------
 5473 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
 5474 This caused pcretest, when used on the test data, to segfault. Unfortunately,
 5475 this didn't happen under Solaris 8, where I normally test things.
 5477 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
 5478 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
 5479 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
 5480 this fix an hour or so after the initial 3.7 release.)
 5483 Version 3.6 23-Oct-01
 5484 ---------------------
 5486 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
 5487 offsets passed as NULL with zero offset count.
 5489 2. The config.guess and config.sub files had not been updated when I moved to
 5490 the latest autoconf.
 5493 Version 3.5 15-Aug-01
 5494 ---------------------
 5496 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
 5497 had been forgotten.
 5499 2. By using declared but undefined structures, we can avoid using "void"
 5500 definitions in pcre.h while keeping the internal definitions of the structures
 5501 private.
 5503 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
 5504 user point of view, this means that both static and shared libraries are built
 5505 by default, but this can be individually controlled. More of the work of
 5506 handling this static/shared cases is now inside libtool instead of PCRE's make
 5507 file.
 5509 4. The pcretest utility is now installed along with pcregrep because it is
 5510 useful for users (to test regexs) and by doing this, it automatically gets
 5511 relinked by libtool. The documentation has been turned into a man page, so
 5512 there are now .1, .txt, and .html versions in /doc.
 5514 5. Upgrades to pcregrep:
 5515    (i)   Added long-form option names like gnu grep.
 5516    (ii)  Added --help to list all options with an explanatory phrase.
 5517    (iii) Added -r, --recursive to recurse into sub-directories.
 5518    (iv)  Added -f, --file to read patterns from a file.
 5520 6. pcre_exec() was referring to its "code" argument before testing that
 5521 argument for NULL (and giving an error if it was NULL).
 5523 7. Upgraded Makefile.in to allow for compiling in a different directory from
 5524 the source directory.
 5526 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
 5527 options bits, the pointer it was passed was to an int instead of to an unsigned
 5528 long int. This mattered only on 64-bit systems.
 5530 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
 5531 generated) instead of pcre.in, which it its source. Also made the same change
 5532 in several of the .c files.
 5534 10. A new release of gcc defines printf() as a macro, which broke pcretest
 5535 because it had an ifdef in the middle of a string argument for printf(). Fixed
 5536 by using separate calls to printf().
 5538 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
 5539 script, to force use of CR or LF instead of \n in the source. On non-Unix
 5540 systems, the value can be set in config.h.
 5542 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
 5543 absolute limit. Changed the text of the error message to make this clear, and
 5544 likewise updated the man page.
 5546 13. The limit of 99 on the number of capturing subpatterns has been removed.
 5547 The new limit is 65535, which I hope will not be a "real" limit.
 5550 Version 3.4 22-Aug-00
 5551 ---------------------
 5553 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
 5555 2. Diagnose condition (?(0) as an error instead of crashing on matching.
 5558 Version 3.3 01-Aug-00
 5559 ---------------------
 5561 1. If an octal character was given, but the value was greater than \377, it
 5562 was not getting masked to the least significant bits, as documented. This could
 5563 lead to crashes in some systems.
 5565 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
 5566 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
 5568 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
 5569 These just pass their arguments on to (pcre_free)(), but they are provided
 5570 because some uses of PCRE bind it to non-C systems that can call its functions,
 5571 but cannot call free() or pcre_free() directly.
 5573 4. Add "make test" as a synonym for "make check". Corrected some comments in
 5574 the Makefile.
 5576 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
 5577 Makefile.
 5579 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
 5580 command called pgrep for grepping around the active processes.
 5582 7. Added the beginnings of support for UTF-8 character strings.
 5584 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
 5585 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
 5586 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
 5587 out for the ar command.)
 5590 Version 3.2 12-May-00
 5591 ---------------------
 5593 This is purely a bug fixing release.
 5595 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
 5596 of ZA. This was just one example of several cases that could provoke this bug,
 5597 which was introduced by change 9 of version 2.00. The code for breaking
 5598 infinite loops after an iteration that matches an empty string was't working
 5599 correctly.
 5601 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
 5602 when matched against abbab (for example). After matching an empty string, it
 5603 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
 5604 caused it to match further down the string than it should.
 5606 3. The code contained an inclusion of sys/types.h. It isn't clear why this
 5607 was there because it doesn't seem to be needed, and it causes trouble on some
 5608 systems, as it is not a Standard C header. It has been removed.
 5610 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
 5611 were reported on the Macintosh. The changes were from
 5613   while ((c = *(++ptr)) != 0 && c != '\n');
 5614 to
 5615   while ((c = *(++ptr)) != 0 && c != '\n') ;
 5617 Totally extraordinary, but if that's what it takes...
 5619 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
 5620 available. Added HAVE_BCOPY and an autoconf test for it; if neither
 5621 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
 5622 assumes the way PCRE uses memmove() (always moving upwards).
 5624 6. PCRE is being used in one environment where strchr() is not available. There
 5625 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
 5626 faster code anyway.
 5629 Version 3.1 09-Feb-00
 5630 ---------------------
 5632 The only change in this release is the fixing of some bugs in Makefile.in for
 5633 the "install" target:
 5635 (1) It was failing to install pcreposix.h.
 5637 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
 5640 Version 3.0 01-Feb-00
 5641 ---------------------
 5643 1. Add support for the /+ modifier to perltest (to output $` like it does in
 5644 pcretest).
 5646 2. Add support for the /g modifier to perltest.
 5648 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
 5649 matches null strings.
 5651 4. Fix perltest so that it doesn't do unwanted things when fed an empty
 5652 pattern. Perl treats empty patterns specially - it reuses the most recent
 5653 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
 5654 effect.
 5656 5. The POSIX interface was broken in that it was just handing over the POSIX
 5657 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
 5658 required a bigger vector, with some working space on the end. This means that
 5659 the POSIX wrapper now has to get and free some memory, and copy the results.
 5661 6. Added some simple autoconf support, placing the test data and the
 5662 documentation in separate directories, re-organizing some of the
 5663 information files, and making it build pcre-config (a GNU standard). Also added
 5664 libtool support for building PCRE as a shared library, which is now the
 5665 default.
 5667 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
 5668 09 are not valid octal constants. Single digits will be used for minor values
 5669 less than 10.
 5671 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
 5672 existing programs that set these in the POSIX interface can use PCRE without
 5673 modification.
 5675 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
 5676 return all that pcre_info() returns, plus additional data. The pcre_info()
 5677 function is retained for compatibility, but is considered to be obsolete.
 5679 10. Added experimental recursion feature (?R) to handle one common case that
 5680 Perl 5.6 will be able to do with (?p{...}).
 5682 11. Added support for POSIX character classes like [:alpha:], which Perl is
 5683 adopting.
 5686 Version 2.08 31-Aug-99
 5687 ----------------------
 5689 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
 5690 trying to match at the startoffset position, but instead was moving forward to
 5691 the next newline as if a previous match had failed.
 5693 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
 5694 and could get into a loop if a null string was matched other than at the start
 5695 of the subject.
 5697 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
 5698 be distinguished at compile time, and for completeness also added PCRE_DATE.
 5700 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
 5701 in GnuWin32 environments.
 5704 Version 2.07 29-Jul-99
 5705 ----------------------
 5707 1. The documentation is now supplied in plain text form and HTML as well as in
 5708 the form of man page sources.
 5710 2. C++ compilers don't like assigning (void *) values to other pointer types.
 5711 In particular this affects malloc(). Although there is no problem in Standard
 5712 C, I've put in casts to keep C++ compilers happy.
 5714 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
 5715 should be (const char *).
 5717 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
 5718 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
 5719 However, I haven't made this a standard facility. The documentation doesn't
 5720 mention it, and the Makefile doesn't support it.
 5722 5. The Makefile now contains an "install" target, with editable destinations at
 5723 the top of the file. The pcretest program is not installed.
 5725 6. pgrep -V now gives the PCRE version number and date.
 5727 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
 5728 causing the entire string to be ignored, instead of just the last character.
 5730 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
 5731 non-matching string, it can take a very, very long time, even for strings of
 5732 quite modest length, because of the nested recursion. PCRE now does better in
 5733 some of these cases. It does this by remembering the last required literal
 5734 character in the pattern, and pre-searching the subject to ensure it is present
 5735 before running the real match. In other words, it applies a heuristic to detect
 5736 some types of certain failure quickly, and in the above example, if presented
 5737 with a string that has no trailing " it gives "no match" very quickly.
 5739 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
 5740 other alternatives are tried instead.
 5743 Version 2.06 09-Jun-99
 5744 ----------------------
 5746 1. Change pcretest's output for amount of store used to show just the code
 5747 space, because the remainder (the data block) varies in size between 32-bit and
 5748 64-bit systems.
 5750 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
 5751 start matching at. This allows lookbehinds to work when searching for multiple
 5752 occurrences in a string.
 5754 3. Added additional options to pcretest for testing multiple occurrences:
 5756    /+   outputs the rest of the string that follows a match
 5757    /g   loops for multiple occurrences, using the new startoffset argument
 5758    /G   loops for multiple occurrences by passing an incremented pointer
 5760 4. PCRE wasn't doing the "first character" optimization for patterns starting
 5761 with \b or \B, though it was doing it for other lookbehind assertions. That is,
 5762 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
 5763 the letter 'x'. On long subject strings, this gives a significant speed-up.
 5766 Version 2.05 21-Apr-99
 5767 ----------------------
 5769 1. Changed the type of magic_number from int to long int so that it works
 5770 properly on 16-bit systems.
 5772 2. Fixed a bug which caused patterns starting with .* not to work correctly
 5773 when the subject string contained newline characters. PCRE was assuming
 5774 anchoring for such patterns in all cases, which is not correct because .* will
 5775 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
 5776 DOTALL is set at top level; otherwise it knows that patterns starting with .*
 5777 must be retried after every newline in the subject.
 5780 Version 2.04 18-Feb-99
 5781 ----------------------
 5783 1. For parenthesized subpatterns with repeats whose minimum was zero, the
 5784 computation of the store needed to hold the pattern was incorrect (too large).
 5785 If such patterns were nested a few deep, this could multiply and become a real
 5786 problem.
 5788 2. Added /M option to pcretest to show the memory requirement of a specific
 5789 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
 5791 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
 5792 compiled in such a way that the backtracking after subsequent failure was
 5793 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
 5794 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
 5797 Version 2.03 02-Feb-99
 5798 ----------------------
 5800 1. Fixed typo and small mistake in man page.
 5802 2. Added 4th condition (GPL supersedes if conflict) and created separate
 5803 LICENCE file containing the conditions.
 5805 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
 5806 Perl, that is the internal \ allows the delimiter to be included in the
 5807 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
 5808 the final delimiter, add \ to the end of the pattern (to test the error).
 5810 4. Added the convenience functions for extracting substrings after a successful
 5811 match. Updated pcretest to make it able to test these functions.
 5814 Version 2.02 14-Jan-99
 5815 ----------------------
 5817 1. Initialized the working variables associated with each extraction so that
 5818 their saving and restoring doesn't refer to uninitialized store.
 5820 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
 5821 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
 5822 fix the problem.
 5824 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
 5825 calls, and wasn't printing the correct value for compiling calls. Increased the
 5826 default value of LOOPREPEAT, and the number of significant figures in the
 5827 times.
 5829 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
 5831 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
 5832 a building problem on Windows NT with a FAT file system.
 5835 Version 2.01 21-Oct-98
 5836 ----------------------
 5838 1. Changed the API for pcre_compile() to allow for the provision of a pointer
 5839 to character tables built by pcre_maketables() in the current locale. If NULL
 5840 is passed, the default tables are used.
 5843 Version 2.00 24-Sep-98
 5844 ----------------------
 5846 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
 5847 it any more.
 5849 2. Allow quantification of (?>) groups, and make it work correctly.
 5851 3. The first character computation wasn't working for (?>) groups.
 5853 4. Correct the implementation of \Z (it is permitted to match on the \n at the
 5854 end of the subject) and add 5.005's \z, which really does match only at the
 5855 very end of the subject.
 5857 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
 5859 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
 5860 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
 5861 localized options. All options to pcre_study() were also removed.
 5863 7. Add other new features from 5.005:
 5865    $(?<=           positive lookbehind
 5866    $(?<!           negative lookbehind
 5867    (?imsx-imsx)    added the unsetting capability
 5868                    such a setting is global if at outer level; local otherwise
 5869    (?imsx-imsx:)   non-capturing groups with option setting
 5870    (?(cond)re|re)  conditional pattern matching
 5872    A backreference to itself in a repeated group matches the previous
 5873    captured string.
 5875 8. General tidying up of studying (both automatic and via "study")
 5876 consequential on the addition of new assertions.
 5878 9. As in 5.005, unlimited repeated groups that could match an empty substring
 5879 are no longer faulted at compile time. Instead, the loop is forcibly broken at
 5880 runtime if any iteration does actually match an empty substring.
 5882 10. Include the RunTest script in the distribution.
 5884 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
 5885 discrepancies, some of which were old and were also with respect to 5.004. They
 5886 have now been fixed.
 5889 Version 1.09 28-Apr-98
 5890 ----------------------
 5892 1. A negated single character class followed by a quantifier with a minimum
 5893 value of one (e.g.  [^x]{1,6}  ) was not compiled correctly. This could lead to
 5894 program crashes, or just wrong answers. This did not apply to negated classes
 5895 containing more than one character, or to minima other than one.
 5898 Version 1.08 27-Mar-98
 5899 ----------------------
 5901 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
 5903 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
 5904 latter must appear before anything that relies on it in the pattern.
 5907 Version 1.07 16-Feb-98
 5908 ----------------------
 5910 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
 5911 repeat of a potentially empty string).
 5914 Version 1.06 23-Jan-98
 5915 ----------------------
 5917 1. Added Markus Oberhumer's little patches for C++.
 5919 2. Literal strings longer than 255 characters were broken.
 5922 Version 1.05 23-Dec-97
 5923 ----------------------
 5925 1. Negated character classes containing more than one character were failing if
 5926 PCRE_CASELESS was set at run time.
 5929 Version 1.04 19-Dec-97
 5930 ----------------------
 5932 1. Corrected the man page, where some "const" qualifiers had been omitted.
 5934 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
 5935 input syntax.
 5937 3. Fixed memory leak which occurred when a regex with back references was
 5938 matched with an offsets vector that wasn't big enough. The temporary memory
 5939 that is used in this case wasn't being freed if the match failed.
 5941 4. Tidied pcretest to ensure it frees memory that it gets.
 5943 5. Temporary memory was being obtained in the case where the passed offsets
 5944 vector was exactly big enough.
 5946 6. Corrected definition of offsetof() from change 5 below.
 5948 7. I had screwed up change 6 below and broken the rules for the use of
 5949 setjmp(). Now fixed.
 5952 Version 1.03 18-Dec-97
 5953 ----------------------
 5955 1. A erroneous regex with a missing opening parenthesis was correctly
 5956 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
 5957 on some systems.
 5959 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
 5960 it was reported that one broken compiler failed on the former because "code" is
 5961 also an independent variable.
 5963 3. The erroneous regex a[]b caused an array overrun reference.
 5965 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
 5966 fail on data ending with that character. (It was going on too far, and checking
 5967 the next character, typically a binary zero.) This was specific to the
 5968 optimized code for single-character negative classes.
 5970 5. Added a contributed patch from the TIN world which does the following:
 5972   + Add an undef for memmove, in case the the system defines a macro for it.
 5974   + Add a definition of offsetof(), in case there isn't one. (I don't know
 5975     the reason behind this - offsetof() is part of the ANSI standard - but
 5976     it does no harm).
 5978   + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
 5979     most of the places where whitespace preceded '#'. I have given up and
 5980     allowed the remaining 2 cases to be at the margin.
 5982   + Rename some variables in pcre to eliminate shadowing. This seems very
 5983     pedantic, but does no harm, of course.
 5985 6. Moved the call to setjmp() into its own function, to get rid of warnings
 5986 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
 5988 7. Constructs such as \d{8,} were compiling into the equivalent of
 5989 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
 5990 outcome, but in this particular case used more store than had been allocated,
 5991 which caused the bug to be discovered because it threw up an internal error.
 5993 8. The debugging code in both pcre and pcretest for outputting the compiled
 5994 form of a regex was going wrong in the case of back references followed by
 5995 curly-bracketed repeats.
 5998 Version 1.02 12-Dec-97
 5999 ----------------------
 6001 1. Typos in pcre.3 and comments in the source fixed.
 6003 2. Applied a contributed patch to get rid of places where it used to remove
 6004 'const' from variables, and fixed some signed/unsigned and uninitialized
 6005 variable warnings.
 6007 3. Added the "runtest" target to Makefile.
 6009 4. Set default compiler flag to -O2 rather than just -O.
 6012 Version 1.01 19-Nov-97
 6013 ----------------------
 6015 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
 6016 like /([ab]*)*/, that is, for classes with more than one character in them.
 6018 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
 6019 as /((?>a*))*/ (a PCRE_EXTRA facility).
 6022 Version 1.00 18-Nov-97
 6023 ----------------------
 6025 1. Added compile-time macros to support systems such as SunOS4 which don't have
 6026 memmove() or strerror() but have other things that can be used instead.
 6028 2. Arranged that "make clean" removes the executables.
 6031 Version 0.99 27-Oct-97
 6032 ----------------------
 6034 1. Fixed bug in code for optimizing classes with only one character. It was
 6035 initializing a 32-byte map regardless, which could cause it to run off the end
 6036 of the memory it had got.
 6038 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
 6041 Version 0.98 22-Oct-97
 6042 ----------------------
 6044 1. Fixed bug in code for handling temporary memory usage when there are more
 6045 back references than supplied space in the ovector. This could cause segfaults.
 6048 Version 0.97 21-Oct-97
 6049 ----------------------
 6051 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
 6053 2. Optimized negated single characters not to use a bit map.
 6055 3. Brought error texts together as macro definitions; clarified some of them;
 6056 fixed one that was wrong - it said "range out of order" when it meant "invalid
 6057 escape sequence".
 6059 4. Changed some char * arguments to const char *.
 6061 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
 6063 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
 6064 pcretest.
 6067 Version 0.96 16-Oct-97
 6068 ----------------------
 6070 1. Added a simple "pgrep" utility to the distribution.
 6072 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
 6073 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
 6074 where "ddd" means "one or more decimal digits".
 6076 3. Fixed serious bug. If a pattern had a back reference, but the call to
 6077 pcre_exec() didn't supply a large enough ovector to record the related
 6078 identifying subpattern, the match always failed. PCRE now remembers the number
 6079 of the largest back reference, and gets some temporary memory in which to save
 6080 the offsets during matching if necessary, in order to ensure that
 6081 backreferences always work.
 6083 4. Increased the compatibility with Perl in a number of ways:
 6085   (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
 6086       to request this handling. The option can be set at compile or exec time.
 6088   (b) $ matches before a terminating newline by default; an option
 6089       PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
 6090       mode). The option can be set at compile or exec time.
 6092   (c) The handling of \ followed by a digit other than 0 is now supposed to be
 6093       the same as Perl's. If the decimal number it represents is less than 10
 6094       or there aren't that many previous left capturing parentheses, an octal
 6095       escape is read. Inside a character class, it's always an octal escape,
 6096       even if it is a single digit.
 6098   (d) An escaped but undefined alphabetic character is taken as a literal,
 6099       unless PCRE_EXTRA is set. Currently this just reserves the remaining
 6100       escapes.
 6102   (e) {0} is now permitted. (The previous item is removed from the compiled
 6103       pattern).
 6105 5. Changed all the names of code files so that the basic parts are no longer
 6106 than 10 characters, and abolished the teeny "globals.c" file.
 6108 6. Changed the handling of character classes; they are now done with a 32-byte
 6109 bit map always.
 6111 7. Added the -d and /D options to pcretest to make it possible to look at the
 6112 internals of compilation without having to recompile pcre.
 6115 Version 0.95 23-Sep-97
 6116 ----------------------
 6118 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
 6119 \x20 at the start of a run of normal characters. These were being treated as
 6120 real characters, instead of the source characters being re-checked.
 6123 Version 0.94 18-Sep-97
 6124 ----------------------
 6126 1. The functions are now thread-safe, with the caveat that the global variables
 6127 containing pointers to malloc() and free() or alternative functions are the
 6128 same for all threads.
 6130 2. Get pcre_study() to generate a bitmap of initial characters for non-
 6131 anchored patterns when this is possible, and use it if passed to pcre_exec().
 6134 Version 0.93 15-Sep-97
 6135 ----------------------
 6137 1. /(b)|(:+)/ was computing an incorrect first character.
 6139 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
 6140 but not actually doing anything yet.
 6142 3. Treat "-" characters in classes that cannot be part of ranges as literals,
 6143 as Perl does (e.g. [-az] or [az-]).
 6145 4. Set the anchored flag if a branch starts with .* or .*? because that tests
 6146 all possible positions.
 6148 5. Split up into different modules to avoid including unneeded functions in a
 6149 compiled binary. However, compile and exec are still in one module. The "study"
 6150 function is split off.
 6152 6. The character tables are now in a separate module whose source is generated
 6153 by an auxiliary program - but can then be edited by hand if required. There are
 6154 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
 6155 toupper() in the code.
 6157 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
 6158 make them global. Abolish the function for setting them, as the caller can now
 6159 set them directly.
 6162 Version 0.92 11-Sep-97
 6163 ----------------------
 6165 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
 6166 (e.g. /a{1,3}/) was broken (I mis-optimized it).
 6168 2. Caseless matching was not working in character classes if the characters in
 6169 the pattern were in upper case.
 6171 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
 6173 4. Make PCRE_ANCHORED public and accept as a compile option.
 6175 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
 6176 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
 6177 pass them.
 6179 6. Give an error if bad option bits passed at compile or run time.
 6181 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
 6182 pcretest to cause it to pass that flag.
 6184 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
 6185 options, and the first character, if set.
 6187 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
 6190 Version 0.91 10-Sep-97
 6191 ----------------------
 6193 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
 6194 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
 6196 2. PCRE was looping on encountering an indefinitely repeated back reference to
 6197 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
 6198 Perl does - treats the match as successful.
 6200 ****