"Fossies" - the Fresh Open Source Software Archive

Member "pcre-8.44/ChangeLog" (12 Feb 2020, 299143 Bytes) of package /linux/misc/pcre-8.44.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": 8.43_vs_8.44.

    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 8.44 12 February-2020
    9 -----------------------------
   11 1. Setting --enable-jit=auto for an out-of-tree build failed because the
   12 source directory wasn't in the search path for AC_TRY_COMPILE always. Patch
   13 from Ross Burton.
   15 2. Applied a patch from Michael Shigorin to fix 8.43 build on e2k arch
   16 with lcc compiler (EDG frontend based); the problem it fixes is:
   18   lcc: "pcrecpp.cc", line 74: error: declaration aliased to undefined entity
   19        "_ZN7pcrecpp2RE6no_argE" [-Werror]
   21 3. Change 2 for 8.43 omitted (*LF) from the list of start-of-pattern items. Now
   22 added.
   24 4. Fix ARMv5 JIT improper handling of labels right after a constant pool.
   26 5. Small patch to pcreposix.c to set the erroroffset field to -1 immediately
   27 after a successful compile, instead of at the start of matching to avoid a
   28 sanitizer complaint (regexec is supposed to be thread safe).
   30 6. Check the size of the number after (?C as it is read, in order to avoid
   31 integer overflow.
   33 7. Tidy up left shifts to avoid sanitize warnings; also fix one NULL deference
   34 in pcretest.
   37 Version 8.43 23-February-2019
   38 -----------------------------
   40 1. Some time ago the config macro SUPPORT_UTF8 was changed to SUPPORT_UTF
   41 because it also applies to UTF-16 and UTF-32. However, this change was not made
   42 in the pcre2cpp files; consequently the C++ wrapper has from then been compiled
   43 with a bug in it, which would have been picked up by the unit test except that
   44 it also had its UTF8 code cut out. The bug was in a global replace when moving
   45 forward after matching an empty string.
   47 2. The C++ wrapper got broken a long time ago (version 7.3, August 2007) when
   48 (*CR) was invented (assuming it was the first such start-of-pattern option).
   49 The wrapper could never handle such patterns because it wraps patterns in
   50 (?:...)\z in order to support end anchoring. I have hacked in some code to fix
   51 this, that is, move the wrapping till after any existing start-of-pattern
   52 special settings.
   54 3. "pcre2grep" (sic) was accidentally mentioned in an error message (fix was
   55 ported from PCRE2).
   57 4. Typo LCC_ALL for LC_ALL fixed in pcregrep.
   59 5. In a pattern such as /[^\x{100}-\x{ffff}]*[\x80-\xff]/ which has a repeated
   60 negative class with no characters less than 0x100 followed by a positive class
   61 with only characters less than 0x100, the first class was incorrectly being
   62 auto-possessified, causing incorrect match failures.
   64 6. If the only branch in a conditional subpattern was anchored, the whole
   65 subpattern was treated as anchored, when it should not have been, since the
   66 assumed empty second branch cannot be anchored. Demonstrated by test patterns
   67 such as /(?(1)^())b/ or /(?(?=^))b/.
   69 7. Fix subject buffer overread in JIT when UTF is disabled and \X or \R has
   70 a greater than 1 fixed quantifier. This issue was found by Yunho Kim.
   72 8. If a pattern started with a subroutine call that had a quantifier with a
   73 minimum of zero, an incorrect "match must start with this character" could be
   74 recorded. Example: /(?&xxx)*ABC(?<xxx>XYZ)/ would (incorrectly) expect 'A' to
   75 be the first character of a match.
   77 9. Improve MAP_JIT flag usage on MacOS. Patch by Rich Siegel.
   80 Version 8.42 20-March-2018
   81 --------------------------
   83 1.  Fixed a MIPS issue in the JIT compiler reported by Joshua Kinard.
   85 2.  Fixed outdated real_pcre definitions in pcre.h.in (patch by Evgeny Kotkov).
   87 3.  pcregrep was truncating components of file names to 128 characters when
   88 processing files with the -r option, and also (some very odd code) truncating
   89 path names to 512 characters. There is now a check on the absolute length of
   90 full path file names, which may be up to 2047 characters long.
   92 4.  Using pcre_dfa_exec(), in UTF mode when UCP support was not defined, there
   93 was the possibility of a false positive match when caselessly matching a "not
   94 this character" item such as [^\x{1234}] (with a code point greater than 127)
   95 because the "other case" variable was not being initialized.
   97 5. Although pcre_jit_exec checks whether the pattern is compiled
   98 in a given mode, it was also expected that at least one mode is available.
   99 This is fixed and pcre_jit_exec returns with PCRE_ERROR_JIT_BADOPTION
  100 when the pattern is not optimized by JIT at all.
  102 6. The line number and related variables such as match counts in pcregrep
  103 were all int variables, causing overflow when files with more than 2147483647
  104 lines were processed (assuming 32-bit ints). They have all been changed to
  105 unsigned long ints.
  107 7. If a backreference with a minimum repeat count of zero was first in a
  108 pattern, apart from assertions, an incorrect first matching character could be
  109 recorded. For example, for the pattern /(?=(a))\1?b/, "b" was incorrectly set
  110 as the first character of a match.
  112 8. Fix out-of-bounds read for partial matching of /./ against an empty string
  113 when the newline type is CRLF.
  115 9. When matching using the the REG_STARTEND feature of the POSIX API with a
  116 non-zero starting offset, unset capturing groups with lower numbers than a
  117 group that did capture something were not being correctly returned as "unset"
  118 (that is, with offset values of -1).
  120 10. Matching the pattern /(*UTF)\C[^\v]+\x80/ against an 8-bit string
  121 containing multi-code-unit characters caused bad behaviour and possibly a
  122 crash. This issue was fixed for other kinds of repeat in release 8.37 by change
  123 38, but repeating character classes were overlooked.
  125 11. A small fix to pcregrep to avoid compiler warnings for -Wformat-overflow=2.
  127 12. Added --enable-jit=auto support to configure.ac.
  129 13. Fix misleading error message in configure.ac.
  132 Version 8.41 05-July-2017
  133 -------------------------
  135 1.  Fixed typo in CMakeLists.txt (wrong number of arguments for
  136 PCRE_STATIC_RUNTIME (affects MSVC only).
  138 2.  Issue 1 for 8.40 below was not correctly fixed. If pcregrep in multiline
  139 mode with --only-matching matched several lines, it restarted scanning at the
  140 next line instead of moving on to the end of the matched string, which can be
  141 several lines after the start.
  143 3.  Fix a missing else in the JIT compiler reported by 'idaifish'.
  145 4.  A (?# style comment is now ignored between a basic quantifier and a
  146 following '+' or '?' (example: /X+(?#comment)?Y/.
  148 5.  Avoid use of a potentially overflowing buffer in pcregrep (patch by Petr
  149 Pisar).
  151 6.  Fuzzers have reported issues in pcretest. These are NOT serious (it is,
  152 after all, just a test program). However, to stop the reports, some easy ones
  153 are fixed:
  155     (a) Check for values < 256 when calling isprint() in pcretest.
  156     (b) Give an error for too big a number after \O.
  158 7.  In the 32-bit library in non-UTF mode, an attempt to find a Unicode
  159 property for a character with a code point greater than 0x10ffff (the Unicode
  160 maximum) caused a crash.
  162 8. The alternative matching function, pcre_dfa_exec() misbehaved if it
  163 encountered a character class with a possessive repeat, for example [a-f]{3}+.
  165 9. When pcretest called pcre_copy_substring() in 32-bit mode, it set the buffer
  166 length incorrectly, which could result in buffer overflow.
  168 10. Remove redundant line of code (accidentally left in ages ago).
  170 11. Applied C++ patch from Irfan Adilovic to guard 'using std::' directives
  171 with namespace pcrecpp (Bugzilla #2084).
  173 12. Remove a duplication typo in pcre_tables.c.
  175 13. Fix returned offsets from regexec() when REG_STARTEND is used with a
  176 starting offset greater than zero.
  179 Version 8.40 11-January-2017
  180 ----------------------------
  182 1.  Using -o with -M in pcregrep could cause unnecessary repeated output when
  183     the match extended over a line boundary.
  185 2.  Applied Chris Wilson's second patch (Bugzilla #1681) to CMakeLists.txt for
  186     MSVC static compilation, putting the first patch under a new option.
  188 3.  Fix register overwite in JIT when SSE2 acceleration is enabled.
  190 4.  Ignore "show all captures" (/=) for DFA matching.
  192 5.  Fix JIT unaligned accesses on x86. Patch by Marc Mutz.
  194 6.  In any wide-character mode (8-bit UTF or any 16-bit or 32-bit mode),
  195     without PCRE_UCP set, a negative character type such as \D in a positive
  196     class should cause all characters greater than 255 to match, whatever else
  197     is in the class. There was a bug that caused this not to happen if a
  198     Unicode property item was added to such a class, for example [\D\P{Nd}] or
  199     [\W\pL].
  201 7.  When pcretest was outputing information from a callout, the caret indicator
  202     for the current position in the subject line was incorrect if it was after
  203     an escape sequence for a character whose code point was greater than
  204     \x{ff}.
  206 8.  A pattern such as (?<RA>abc)(?(R)xyz) was incorrectly compiled such that
  207     the conditional was interpreted as a reference to capturing group 1 instead
  208     of a test for recursion. Any group whose name began with R was
  209     misinterpreted in this way. (The reference interpretation should only
  210     happen if the group's name is precisely "R".)
  212 9.  A number of bugs have been mended relating to match start-up optimizations
  213     when the first thing in a pattern is a positive lookahead. These all
  214     applied only when PCRE_NO_START_OPTIMIZE was *not* set:
  216     (a) A pattern such as (?=.*X)X$ was incorrectly optimized as if it needed
  217         both an initial 'X' and a following 'X'.
  218     (b) Some patterns starting with an assertion that started with .* were
  219         incorrectly optimized as having to match at the start of the subject or
  220         after a newline. There are cases where this is not true, for example,
  221         (?=.*[A-Z])(?=.{8,16})(?!.*[\s]) matches after the start in lines that
  222         start with spaces. Starting .* in an assertion is no longer taken as an
  223         indication of matching at the start (or after a newline).
  226 Version 8.39 14-June-2016
  227 -------------------------
  229 1.  If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
  230     an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
  231     misbehaved. This bug was found by the LLVM fuzzer.
  233 2.  Similar to the above, if an isolated \E was present between an item and its
  234     qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
  235     bug was found by the LLVM fuzzer.
  237 3.  Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
  238     working correctly in UCP mode.
  240 4.  The POSIX wrapper function regexec() crashed if the option REG_STARTEND
  241     was set when the pmatch argument was NULL. It now returns REG_INVARG.
  243 5.  Allow for up to 32-bit numbers in the ordin() function in pcregrep.
  245 6.  An empty \Q\E sequence between an item and its qualifier caused
  246     pcre_compile() to misbehave when auto callouts were enabled. This bug was
  247     found by the LLVM fuzzer.
  249 7.  If a pattern that was compiled with PCRE_EXTENDED started with white
  250     space or a #-type comment that was followed by (?-x), which turns off
  251     PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
  252     pcre_compile() assumed that (?-x) applied to the whole pattern and
  253     consequently mis-compiled it. This bug was found by the LLVM fuzzer.
  255 8.  A call of pcre_copy_named_substring() for a named substring whose number
  256     was greater than the space in the ovector could cause a crash.
  258 9.  Yet another buffer overflow bug involved duplicate named groups with a
  259     group that reset capture numbers (compare 8.38/7 below). Once again, I have
  260     just allowed for more memory, even if not needed. (A proper fix is
  261     implemented in PCRE2, but it involves a lot of refactoring.)
  263 10. pcre_get_substring_list() crashed if the use of \K in a match caused the
  264     start of the match to be earlier than the end.
  266 11. Migrating appropriate PCRE2 JIT improvements to PCRE.
  268 12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
  269     assertion, caused pcretest to generate incorrect output, and also to read
  270     uninitialized memory (detected by ASAN or valgrind).
  272 13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
  273     nested set of parentheses of sufficient size caused an overflow of the
  274     compiling workspace (which was diagnosed, but of course is not desirable).
  276 14. And yet another buffer overflow bug involving duplicate named groups, this
  277     time nested, with a nested back reference. Yet again, I have just allowed
  278     for more memory, because anything more needs all the refactoring that has
  279     been done for PCRE2. An example pattern that provoked this bug is:
  280     /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
  281     registered as CVE-2016-1283.
  283 15. pcretest went into a loop if global matching was requested with an ovector
  284     size less than 2. It now gives an error message. This bug was found by
  285     afl-fuzz.
  287 16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
  288     ("assertion expected") when (?(?C) was not followed by an opening
  289     parenthesis.
  291 17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
  292     actually affect anything, by sheer luck.
  294 18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
  295     static compilation.
  297 19. Modified the RunTest script to incorporate a valgrind suppressions file so
  298     that certain errors, provoked by the SSE2 instruction set when JIT is used,
  299     are ignored.
  301 20. A racing condition is fixed in JIT reported by Mozilla.
  303 21. Minor code refactor to avoid "array subscript is below array bounds"
  304     compiler warning.
  306 22. Minor code refactor to avoid "left shift of negative number" warning.
  308 23. Fix typo causing compile error when 16- or 32-bit JIT is compiled without
  309     UCP support.
  311 24. Refactor to avoid compiler warnings in pcrecpp.cc.
  313 25. Refactor to fix a typo in pcre_jit_test.c
  315 26. Patch to support compiling pcrecpp.cc with Intel compiler.
  318 Version 8.38 23-November-2015
  319 -----------------------------
  321 1.  If a group that contained a recursive back reference also contained a
  322     forward reference subroutine call followed by a non-forward-reference
  323     subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
  324     compile correct code, leading to undefined behaviour or an internally
  325     detected error. This bug was discovered by the LLVM fuzzer.
  327 2.  Quantification of certain items (e.g. atomic back references) could cause
  328     incorrect code to be compiled when recursive forward references were
  329     involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
  330     This bug was discovered by the LLVM fuzzer.
  332 3.  A repeated conditional group whose condition was a reference by name caused
  333     a buffer overflow if there was more than one group with the given name.
  334     This bug was discovered by the LLVM fuzzer.
  336 4.  A recursive back reference by name within a group that had the same name as
  337     another group caused a buffer overflow. For example:
  338     /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
  340 5.  A forward reference by name to a group whose number is the same as the
  341     current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
  342     a buffer overflow at compile time. This bug was discovered by the LLVM
  343     fuzzer.
  345 6.  A lookbehind assertion within a set of mutually recursive subpatterns could
  346     provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
  348 7.  Another buffer overflow bug involved duplicate named groups with a
  349     reference between their definition, with a group that reset capture
  350     numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
  351     fixed by always allowing for more memory, even if not needed. (A proper fix
  352     is implemented in PCRE2, but it involves more refactoring.)
  354 8.  There was no check for integer overflow in subroutine calls such as (?123).
  356 9.  The table entry for \l in EBCDIC environments was incorrect, leading to its
  357     being treated as a literal 'l' instead of causing an error.
  359 10. There was a buffer overflow if pcre_exec() was called with an ovector of
  360     size 1. This bug was found by american fuzzy lop.
  362 11. If a non-capturing group containing a conditional group that could match
  363     an empty string was repeated, it was not identified as matching an empty
  364     string itself. For example: /^(?:(?(1)x|)+)+$()/.
  366 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
  367     \a and \e in test subject lines.
  369 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
  370     instead of the EBCDIC value.
  372 14. The handling of \c in an EBCDIC environment has been revised so that it is
  373     now compatible with the specification in Perl's perlebcdic page.
  375 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
  376     ASCII/Unicode. This has now been added to the list of characters that are
  377     recognized as white space in EBCDIC.
  379 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
  380     error (correctly) when used outside a class, but did not give an error
  381     within a class.
  383 17. \h within a class was incorrectly compiled in EBCDIC environments.
  385 18. A pattern with an unmatched closing parenthesis that contained a backward
  386     assertion which itself contained a forward reference caused buffer
  387     overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
  389 19. JIT should return with error when the compiled pattern requires more stack
  390     space than the maximum.
  392 20. A possessively repeated conditional group that could match an empty string,
  393     for example, /(?(R))*+/, was incorrectly compiled.
  395 21. Fix infinite recursion in the JIT compiler when certain patterns such as
  396     /(?:|a|){100}x/ are analysed.
  398 22. Some patterns with character classes involving [: and \\ were incorrectly
  399     compiled and could cause reading from uninitialized memory or an incorrect
  400     error diagnosis.
  402 23. Pathological patterns containing many nested occurrences of [: caused
  403     pcre_compile() to run for a very long time.
  405 24. A conditional group with only one branch has an implicit empty alternative
  406     branch and must therefore be treated as potentially matching an empty
  407     string.
  409 25. If (?R was followed by - or + incorrect behaviour happened instead of a
  410     diagnostic.
  412 26. Arrange to give up on finding the minimum matching length for overly
  413     complex patterns.
  415 27. Similar to (4) above: in a pattern with duplicated named groups and an
  416     occurrence of (?| it is possible for an apparently non-recursive back
  417     reference to become recursive if a later named group with the relevant
  418     number is encountered. This could lead to a buffer overflow. Wen Guanxing
  419     from Venustech ADLAB discovered this bug.
  421 28. If pcregrep was given the -q option with -c or -l, or when handling a
  422     binary file, it incorrectly wrote output to stdout.
  424 29. The JIT compiler did not restore the control verb head in case of *THEN
  425     control verbs. This issue was found by Karl Skomski with a custom LLVM
  426     fuzzer.
  428 30. Error messages for syntax errors following \g and \k were giving inaccurate
  429     offsets in the pattern.
  431 31. Added a check for integer overflow in conditions (?(<digits>) and
  432     (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
  433     fuzzer.
  435 32. Handling recursive references such as (?2) when the reference is to a group
  436     later in the pattern uses code that is very hacked about and error-prone.
  437     It has been re-written for PCRE2. Here in PCRE1, a check has been added to
  438     give an internal error if it is obvious that compiling has gone wrong.
  440 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
  441     This issue was found by Karl Skomski with a custom LLVM fuzzer.
  443 34. The JIT compiler should restore the control chain for empty possessive
  444     repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
  446 35. Match limit check added to JIT recursion. This issue was found by Karl
  447     Skomski with a custom LLVM fuzzer.
  449 36. Yet another case similar to 27 above has been circumvented by an
  450     unconditional allocation of extra memory. This issue is fixed "properly" in
  451     PCRE2 by refactoring the way references are handled. Wen Guanxing
  452     from Venustech ADLAB discovered this bug.
  454 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
  455     with a custom LLVM fuzzer.
  457 38. Fixed a corner case of range optimization in JIT.
  459 39. An incorrect error "overran compiling workspace" was given if there were
  460     exactly enough group forward references such that the last one extended
  461     into the workspace safety margin. The next one would have expanded the
  462     workspace. The test for overflow was not including the safety margin.
  464 40. A match limit issue is fixed in JIT which was found by Karl Skomski
  465     with a custom LLVM fuzzer.
  467 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
  468     work under Windows. (Why has it taken so long for anyone to notice?)
  470 42. In a character class such as [\W\p{Any}] where both a negative-type escape
  471     ("not a word character") and a property escape were present, the property
  472     escape was being ignored.
  474 43. Fix crash caused by very long (*MARK) or (*THEN) names.
  476 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
  477     by a single ASCII character in a class item, was incorrectly compiled in
  478     UCP mode. The POSIX class got lost, but only if the single character
  479     followed it.
  481 45. [:punct:] in UCP mode was matching some characters in the range 128-255
  482     that should not have been matched.
  484 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
  485     class, all characters with code points greater than 255 are in the class.
  486     When a Unicode property was also in the class (if PCRE_UCP is set, escapes
  487     such as \w are turned into Unicode properties), wide characters were not
  488     correctly handled, and could fail to match.
  491 Version 8.37 28-April-2015
  492 --------------------------
  494 1.  When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
  495     for those parentheses to be closed with whatever has been captured so far.
  496     However, it was failing to mark any other groups between the hightest
  497     capture so far and the currrent group as "unset". Thus, the ovector for
  498     those groups contained whatever was previously there. An example is the
  499     pattern /(x)|((*ACCEPT))/ when matched against "abcd".
  501 2.  If an assertion condition was quantified with a minimum of zero (an odd
  502     thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
  504 3.  If a pattern in pcretest input had the P (POSIX) modifier followed by an
  505     unrecognized modifier, a crash could occur.
  507 4.  An attempt to do global matching in pcretest with a zero-length ovector
  508     caused a crash.
  510 5.  Fixed a memory leak during matching that could occur for a subpattern
  511     subroutine call (recursive or otherwise) if the number of captured groups
  512     that had to be saved was greater than ten.
  514 6.  Catch a bad opcode during auto-possessification after compiling a bad UTF
  515     string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
  516     UTF with NO_UTF_CHECK is documented as having an undefined outcome.
  518 7.  A UTF pattern containing a "not" match of a non-ASCII character and a
  519     subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
  521 8. When a pattern is compiled, it remembers the highest back reference so that
  522    when matching, if the ovector is too small, extra memory can be obtained to
  523    use instead. A conditional subpattern whose condition is a check on a
  524    capture having happened, such as, for example in the pattern
  525    /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
  526    setting the highest backreference number. This mattered only if pcre_exec()
  527    was called with an ovector that was too small to hold the capture, and there
  528    was no other kind of back reference (a situation which is probably quite
  529    rare). The effect of the bug was that the condition was always treated as
  530    FALSE when the capture could not be consulted, leading to a incorrect
  531    behaviour by pcre_exec(). This bug has been fixed.
  533 9. A reference to a duplicated named group (either a back reference or a test
  534    for being set in a conditional) that occurred in a part of the pattern where
  535    PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
  536    to be incorrectly calculated, leading to overwriting.
  538 10. A mutually recursive set of back references such as (\2)(\1) caused a
  539     segfault at study time (while trying to find the minimum matching length).
  540     The infinite loop is now broken (with the minimum length unset, that is,
  541     zero).
  543 11. If an assertion that was used as a condition was quantified with a minimum
  544     of zero, matching went wrong. In particular, if the whole group had
  545     unlimited repetition and could match an empty string, a segfault was
  546     likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
  547     assertions to be quantified, but not if they are being used as conditions,
  548     so the above pattern is faulted by Perl. PCRE has now been changed so that
  549     it also rejects such patterns.
  551 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
  552     failed to allow the zero-repeat case if pcre2_exec() was called with an
  553     ovector too small to capture the group.
  555 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
  556     Red Hat Product Security:
  558     (a) A crash if /K and /F were both set with the option to save the compiled
  559     pattern.
  561     (b) Another crash if the option to print captured substrings in a callout
  562     was combined with setting a null ovector, for example \O\C+ as a subject
  563     string.
  565 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
  566     forward reference repeated a large (but limited) number of times within a
  567     repeated outer group that has a zero minimum quantifier, caused incorrect
  568     code to be compiled, leading to the error "internal error:
  569     previously-checked referenced subpattern not found" when an incorrect
  570     memory address was read. This bug was reported as "heap overflow",
  571     discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
  572     CVE-2015-2325.
  574 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
  575     call within a group that also contained a recursive back reference caused
  576     incorrect code to be compiled. This bug was reported as "heap overflow",
  577     discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
  578     number CVE-2015-2326.
  580 24. Computing the size of the JIT read-only data in advance has been a source
  581     of various issues, and new ones are still appear unfortunately. To fix
  582     existing and future issues, size computation is eliminated from the code,
  583     and replaced by on-demand memory allocation.
  585 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
  586     adjacent to the end of the range, and the range contained characters with
  587     more than one other case, caused incorrect behaviour when compiled in UTF
  588     mode. In that example, the range a-j was left out of the class.
  590 26. Fix JIT compilation of conditional blocks, which assertion
  591     is converted to (*FAIL). E.g: /(?(?!))/.
  593 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
  594     discovered by the LLVM fuzzer.
  596 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
  597     when this assertion was used as a condition, for example (?(?!)a|b). In
  598     pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
  599     error about an unsupported item.
  601 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
  602     possessification code could take exponential time to complete. A recursion
  603     depth limit of 1000 has been imposed to limit the resources used by this
  604     optimization.
  606 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
  607     such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
  608     because \S ensures they are all in the class. The code for doing this was
  609     interacting badly with the code for computing the amount of space needed to
  610     compile the pattern, leading to a buffer overflow. This bug was discovered
  611     by the LLVM fuzzer.
  613 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
  614     other kinds of group caused stack overflow at compile time. This bug was
  615     discovered by the LLVM fuzzer.
  617 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
  618     between a subroutine call and its quantifier was incorrectly compiled,
  619     leading to buffer overflow or other errors. This bug was discovered by the
  620     LLVM fuzzer.
  622 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
  623     assertion after (?(. The code was failing to check the character after
  624     (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
  625     was discovered by the LLVM fuzzer.
  627 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
  628     a fixed maximum following a group that contains a subroutine reference was
  629     incorrectly compiled and could trigger buffer overflow. This bug was
  630     discovered by the LLVM fuzzer.
  632 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
  633     caused a stack overflow instead of the diagnosis of a non-fixed length
  634     lookbehind assertion. This bug was discovered by the LLVM fuzzer.
  636 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
  637     (e.g. /(?<=\Ka)/) could make pcregrep loop.
  639 37. There was a similar problem to 36 in pcretest for global matches.
  641 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
  642     and a subsequent item in the pattern caused a non-match, backtracking over
  643     the repeated \X did not stop, but carried on past the start of the subject,
  644     causing reference to random memory and/or a segfault. There were also some
  645     other cases where backtracking after \C could crash. This set of bugs was
  646     discovered by the LLVM fuzzer.
  648 39. The function for finding the minimum length of a matching string could take
  649     a very long time if mutual recursion was present many times in a pattern,
  650     for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
  651     method has been implemented. This infelicity was discovered by the LLVM
  652     fuzzer.
  654 40. Static linking against the PCRE library using the pkg-config module was
  655     failing on missing pthread symbols.
  658 Version 8.36 26-September-2014
  659 ------------------------------
  661 1.  Got rid of some compiler warnings in the C++ modules that were shown up by
  662     -Wmissing-field-initializers and -Wunused-parameter.
  664 2.  The tests for quantifiers being too big (greater than 65535) were being
  665     applied after reading the number, and stupidly assuming that integer
  666     overflow would give a negative number. The tests are now applied as the
  667     numbers are read.
  669 3.  Tidy code in pcre_exec.c where two branches that used to be different are
  670     now the same.
  672 4.  The JIT compiler did not generate match limit checks for certain
  673     bracketed expressions with quantifiers. This may lead to exponential
  674     backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
  675     issue should be resolved now.
  677 5.  Fixed an issue, which occures when nested alternatives are optimized
  678     with table jumps.
  680 6.  Inserted two casts and changed some ints to size_t in the light of some
  681     reported 64-bit compiler warnings (Bugzilla 1477).
  683 7.  Fixed a bug concerned with zero-minimum possessive groups that could match
  684     an empty string, which sometimes were behaving incorrectly in the
  685     interpreter (though correctly in the JIT matcher). This pcretest input is
  686     an example:
  688       '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
  691     the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
  692     JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
  693     for an empty string was breaking the inner loop and carrying on at a lower
  694     level, when possessive repeated groups should always return to a higher
  695     level as they have no backtrack points in them. The empty string test now
  696     occurs at the outer level.
  698 8.  Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
  699     ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
  701 9.  Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
  702     doesn't).
  704 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
  705     to hold a pattern to be incorrectly computed (too small) when there were
  706     named back references to duplicated names. This could cause "internal
  707     error: code overflow" or "double free or corruption" or other memory
  708     handling errors.
  710 11. When named subpatterns had the same prefixes, back references could be
  711     confused. For example, in this pattern:
  713       /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
  715     the reference to 'Name' was incorrectly treated as a reference to a
  716     duplicate name.
  718 12. A pattern such as /^s?c/mi8 where the optional character has more than
  719     one "other case" was incorrectly compiled such that it would only try to
  720     match starting at "c".
  722 13. When a pattern starting with \s was studied, VT was not included in the
  723     list of possible starting characters; this should have been part of the
  724     8.34/18 patch.
  726 14. If a character class started [\Qx]... where x is any character, the class
  727     was incorrectly terminated at the ].
  729 15. If a pattern that started with a caseless match for a character with more
  730     than one "other case" was studied, PCRE did not set up the starting code
  731     unit bit map for the list of possible characters. Now it does. This is an
  732     optimization improvement, not a bug fix.
  734 16. The Unicode data tables have been updated to Unicode 7.0.0.
  736 17. Fixed a number of memory leaks in pcregrep.
  738 18. Avoid a compiler warning (from some compilers) for a function call with
  739     a cast that removes "const" from an lvalue by using an intermediate
  740     variable (to which the compiler does not object).
  742 19. Incorrect code was compiled if a group that contained an internal recursive
  743     back reference was optional (had quantifier with a minimum of zero). This
  744     example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
  745     caused segmentation faults because of stack overflows at compile time.
  747 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
  748     group that is quantified with an indefinite repeat, caused a compile-time
  749     loop which used up all the system stack and provoked a segmentation fault.
  750     This was not the same bug as 19 above.
  752 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
  753     Patch by Mike Frysinger.
  756 Version 8.35 04-April-2014
  757 --------------------------
  759 1.  A new flag is set, when property checks are present in an XCLASS.
  760     When this flag is not set, PCRE can perform certain optimizations
  761     such as studying these XCLASS-es.
  763 2.  The auto-possessification of character sets were improved: a normal
  764     and an extended character set can be compared now. Furthermore
  765     the JIT compiler optimizes more character set checks.
  767 3.  Got rid of some compiler warnings for potentially uninitialized variables
  768     that show up only when compiled with -O2.
  770 4.  A pattern such as (?=ab\K) that uses \K in an assertion can set the start
  771     of a match later then the end of the match. The pcretest program was not
  772     handling the case sensibly - it was outputting from the start to the next
  773     binary zero. It now reports this situation in a message, and outputs the
  774     text from the end to the start.
  776 5.  Fast forward search is improved in JIT. Instead of the first three
  777     characters, any three characters with fixed position can be searched.
  778     Search order: first, last, middle.
  780 6.  Improve character range checks in JIT. Characters are read by an inprecise
  781     function now, which returns with an unknown value if the character code is
  782     above a certain threshold (e.g: 256). The only limitation is that the value
  783     must be bigger than the threshold as well. This function is useful when
  784     the characters above the threshold are handled in the same way.
  786 7.  The macros whose names start with RAWUCHAR are placeholders for a future
  787     mode in which only the bottom 21 bits of 32-bit data items are used. To
  788     make this more memorable for those maintaining the code, the names have
  789     been changed to start with UCHAR21, and an extensive comment has been added
  790     to their definition.
  792 8.  Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
  793     to the export list in Makefile.am (they were accidentally omitted from the
  794     8.34 tarball).
  796 9.  The informational output from pcretest used the phrase "starting byte set"
  797     which is inappropriate for the 16-bit and 32-bit libraries. As the output
  798     for "first char" and "need char" really means "non-UTF-char", I've changed
  799     "byte" to "char", and slightly reworded the output. The documentation about
  800     these values has also been (I hope) clarified.
  802 10. Another JIT related optimization: use table jumps for selecting the correct
  803     backtracking path, when more than four alternatives are present inside a
  804     bracket.
  806 11. Empty match is not possible, when the minimum length is greater than zero,
  807     and there is no \K in the pattern. JIT should avoid empty match checks in
  808     such cases.
  810 12. In a caseless character class with UCP support, when a character with more
  811     than one alternative case was not the first character of a range, not all
  812     the alternative cases were added to the class. For example, s and \x{17f}
  813     are both alternative cases for S: the class [RST] was handled correctly,
  814     but [R-T] was not.
  816 13. The configure.ac file always checked for pthread support when JIT was
  817     enabled. This is not used in Windows, so I have put this test inside a
  818     check for the presence of windows.h (which was already tested for).
  820 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
  821     The algorithm provides a way to skip certain starting offsets, and usually
  822     faster than linear prefix searches.
  824 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
  825     as for 'fr_FR' and 'french'. For some reason, however, it then used the
  826     Windows-specific input and output files, which have 'french' screwed in.
  827     So this could never have worked. One of the problems with locales is that
  828     they aren't always the same. I have now updated RunTest so that it checks
  829     the output of the locale test (test 3) against three different output
  830     files, and it allows the test to pass if any one of them matches. With luck
  831     this should make the test pass on some versions of Solaris where it was
  832     failing. Because of the uncertainty, the script did not used to stop if
  833     test 3 failed; it now does. If further versions of a French locale ever
  834     come to light, they can now easily be added.
  836 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
  837     there was a message during ./configure, but it did not stop. This now
  838     provokes an error. The invalid example in README has been corrected.
  839     If a value less than the minimum is given, the minimum value has always
  840     been used, but now a warning is given.
  842 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
  843     was a bug in the test system, which is now fixed. Also, the list of various
  844     configurations that are tested for each release did not have one with both
  845     16/32 bits and --enable-bar-anycrlf. It now does.
  847 18. pcretest was missing "-C bsr" for displaying the \R default setting.
  849 19. Little endian PowerPC systems are supported now by the JIT compiler.
  851 20. The fast forward newline mechanism could enter to an infinite loop on
  852     certain invalid UTF-8 input. Although we don't support these cases
  853     this issue can be fixed by a performance optimization.
  855 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
  856     not take account if existing stack usage. There is now a new global
  857     variable called pcre_stack_guard that can be set to point to an external
  858     function to check stack availability. It is called at the start of
  859     processing every parenthesized group.
  861 22. A typo in the code meant that in ungreedy mode the max/min qualifier
  862     behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
  863     not match "ab".
  865 23. When UTF was disabled, the JIT program reported some incorrect compile
  866     errors. These messages are silenced now.
  868 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
  869     compiler.
  871 25. Change all the temporary files used in RunGrepTest to be different to those
  872     used by RunTest so that the tests can be run simultaneously, for example by
  873     "make -j check".
  876 Version 8.34 15-December-2013
  877 -----------------------------
  879 1.  Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
  880     executable memory. Patch inspired by Carsten Klein.
  882 2.  ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
  883     this macro is never tested and has no effect, because the work to support
  884     coverage involves only compiling and linking options and special targets in
  885     the Makefile. The comment in config.h implied that defining the macro would
  886     enable coverage support, which is totally false. There was also support for
  887     setting this macro in the CMake files (my fault, I just copied it from
  888     configure). SUPPORT_GCOV has now been removed.
  890 3.  Make a small performance improvement in strlen16() and strlen32() in
  891     pcretest.
  893 4.  Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
  894     detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
  895     these cases). There was also one in pcretest.c.
  897 5.  Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
  899 6.  In UTF mode, the code for checking whether a group could match an empty
  900     string (which is used for indefinitely repeated groups to allow for
  901     breaking an infinite loop) was broken when the group contained a repeated
  902     negated single-character class with a character that occupied more than one
  903     data item and had a minimum repetition of zero (for example, [^\x{100}]* in
  904     UTF-8 mode). The effect was undefined: the group might or might not be
  905     deemed as matching an empty string, or the program might have crashed.
  907 7.  The code for checking whether a group could match an empty string was not
  908     recognizing that \h, \H, \v, \V, and \R must match a character.
  910 8.  Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
  911     an empty string. If it can, pcretest shows this in its information output.
  913 9.  Fixed two related bugs that applied to Unicode extended grapheme clusters
  914     that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
  915     matched by pcre_exec() without using JIT:
  917     (a) If the rest of the pattern did not match after a maximal run of
  918         grapheme clusters, the code for backing up to try with fewer of them
  919         did not always back up over a full grapheme when characters that do not
  920         have the modifier quality were involved, e.g. Hangul syllables.
  922     (b) If the match point in a subject started with modifier character, and
  923         there was no match, the code could incorrectly back up beyond the match
  924         point, and potentially beyond the first character in the subject,
  925         leading to a segfault or an incorrect match result.
  927 10. A conditional group with an assertion condition could lead to PCRE
  928     recording an incorrect first data item for a match if no other first data
  929     item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
  930     first data item, and therefore matched "ca" after "c" instead of at the
  931     start.
  933 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
  934     bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
  936 12. The source of pcregrep now includes z/OS-specific code so that it can be
  937     compiled for z/OS as part of the special z/OS distribution.
  939 13. Added the -T and -TM options to pcretest.
  941 14. The code in pcre_compile.c for creating the table of named capturing groups
  942     has been refactored. Instead of creating the table dynamically during the
  943     actual compiling pass, the information is remembered during the pre-compile
  944     pass (on the stack unless there are more than 20 named groups, in which
  945     case malloc() is used) and the whole table is created before the actual
  946     compile happens. This has simplified the code (it is now nearly 150 lines
  947     shorter) and prepared the way for better handling of references to groups
  948     with duplicate names.
  950 15. A back reference to a named subpattern when there is more than one of the
  951     same name now checks them in the order in which they appear in the pattern.
  952     The first one that is set is used for the reference. Previously only the
  953     first one was inspected. This change makes PCRE more compatible with Perl.
  955 16. Unicode character properties were updated from Unicode 6.3.0.
  957 17. The compile-time code for auto-possessification has been refactored, based
  958     on a patch by Zoltan Herczeg. It now happens after instead of during
  959     compilation. The code is cleaner, and more cases are handled. The option
  960     PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
  961     options in pcretest are provided to set it. It can also be set by
  962     (*NO_AUTO_POSSESS) at the start of a pattern.
  964 18. The character VT has been added to the default ("C" locale) set of
  965     characters that match \s and are generally treated as white space,
  966     following this same change in Perl 5.18. There is now no difference between
  967     "Perl space" and "POSIX space". Whether VT is treated as white space in
  968     other locales depends on the locale.
  970 19. The code for checking named groups as conditions, either for being set or
  971     for being recursed, has been refactored (this is related to 14 and 15
  972     above). Processing unduplicated named groups should now be as fast at
  973     numerical groups, and processing duplicated groups should be faster than
  974     before.
  976 20. Two patches to the CMake build system, by Alexander Barkov:
  978       (1) Replace the "source" command by "." in CMakeLists.txt because
  979           "source" is a bash-ism.
  981       (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
  982           without these the CMake build does not work on Solaris.
  984 21. Perl has changed its handling of \8 and \9. If there is no previously
  985     encountered capturing group of those numbers, they are treated as the
  986     literal characters 8 and 9 instead of a binary zero followed by the
  987     literals. PCRE now does the same.
  989 22. Following Perl, added \o{} to specify codepoints in octal, making it
  990     possible to specify values greater than 0777 and also making them
  991     unambiguous.
  993 23. Perl now gives an error for missing closing braces after \x{... instead of
  994     treating the string as literal. PCRE now does the same.
  996 24. RunTest used to grumble if an inappropriate test was selected explicitly,
  997     but just skip it when running all tests. This make it awkward to run ranges
  998     of tests when one of them was inappropriate. Now it just skips any
  999     inappropriate tests, as it always did when running all tests.
 1001 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
 1002     character types such as \d or \w, too many callouts were inserted, and the
 1003     data that they returned was rubbish.
 1005 26. In UCP mode, \s was not matching two of the characters that Perl matches,
 1006     namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
 1007     were matched by \h. The code has now been refactored so that the lists of
 1008     the horizontal and vertical whitespace characters used for \h and \v (which
 1009     are defined only in one place) are now also used for \s.
 1011 27. Add JIT support for the 64 bit TileGX architecture.
 1012     Patch by Jiong Wang (Tilera Corporation).
 1014 28. Possessive quantifiers for classes (both explicit and automatically
 1015     generated) now use special opcodes instead of wrapping in ONCE brackets.
 1017 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
 1018     (because it's meaningless), this was not happening when PCRE_CASELESS was
 1019     set. Not wrong, but inefficient.
 1021 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
 1022     properties for \w, \d, etc) is present in a test regex. Otherwise if the
 1023     test contains no characters greater than 255, Perl doesn't realise it
 1024     should be using Unicode semantics.
 1026 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
 1027     [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
 1028     does in Unicode mode.
 1030 32. Added the "forbid" facility to pcretest so that putting tests into the
 1031     wrong test files can sometimes be quickly detected.
 1033 33. There is now a limit (default 250) on the depth of nesting of parentheses.
 1034     This limit is imposed to control the amount of system stack used at compile
 1035     time. It can be changed at build time by --with-parens-nest-limit=xxx or
 1036     the equivalent in CMake.
 1038 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
 1039     errors. Perl warns for these when in warning mode, but PCRE has no facility
 1040     for giving warnings.
 1042 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
 1043     However, this was not working for (?!) because it is optimized to (*FAIL),
 1044     for which PCRE does not allow quantifiers. The optimization is now disabled
 1045     when a quantifier follows (?!). I can't see any use for this, but it makes
 1046     things uniform.
 1048 36. Perl no longer allows group names to start with digits, so I have made this
 1049     change also in PCRE. It simplifies the code a bit.
 1051 37. In extended mode, Perl ignores spaces before a + that indicates a
 1052     possessive quantifier. PCRE allowed a space before the quantifier, but not
 1053     before the possessive +. It now does.
 1055 38. The use of \K (reset reported match start) within a repeated possessive
 1056     group such as (a\Kb)*+ was not working.
 1058 40. Document that the same character tables must be used at compile time and
 1059     run time, and that the facility to pass tables to pcre_exec() and
 1060     pcre_dfa_exec() is for use only with saved/restored patterns.
 1062 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
 1063     features for Builds with MSVC:
 1065     1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
 1066        builds). The .rc files can be used to set FileDescription and many other
 1067        attributes.
 1069     2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
 1070        This allows higher-level build scripts which want .pdb files to avoid
 1071        hard-coding the exact files needed."
 1073 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
 1074     mean "start of word" and "end of word", respectively, as a transition aid.
 1076 43. A minimizing repeat of a class containing codepoints greater than 255 in
 1077     non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
 1078     compiled to use the heap for recursion.
 1080 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
 1081     is configured.
 1084 Version 8.33 28-May-2013
 1085 ------------------------
 1087 1.  Added 'U' to some constants that are compared to unsigned integers, to
 1088     avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
 1089     variables that are added to signed variables, to ensure the result is
 1090     signed and can be negated.
 1092 2.  Applied patch by Daniel Richard G for quashing MSVC warnings to the
 1093     CMake config files.
 1095 3.  Revise the creation of config.h.generic so that all boolean macros are
 1096     #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
 1097     overriding via -D on the command line possible.
 1099 4.  Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
 1100     to unsigned int is reported to make a quite noticeable speed difference in
 1101     a specific Windows environment. Testing on Linux did also appear to show
 1102     some benefit (and it is clearly not harmful). Also fixed the definition of
 1103     Xop which should be unsigned.
 1105 5.  Related to (4), changing the definition of the intermediate variable cc
 1106     in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
 1107     improvements.
 1109 6.  Fix forward search in JIT when link size is 3 or greater. Also removed some
 1110     unnecessary spaces.
 1112 7.  Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
 1113     and later.
 1115 8.  Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
 1117 9.  Optimizing fast_forward_start_bits in JIT.
 1119 10. Adding support for callouts in JIT, and fixing some issues revealed
 1120     during this work. Namely:
 1122     (a) Unoptimized capturing brackets incorrectly reset on backtrack.
 1124     (b) Minimum length was not checked before the matching is started.
 1126 11. The value of capture_last that is passed to callouts was incorrect in some
 1127     cases when there was a capture on one path that was subsequently abandoned
 1128     after a backtrack. Also, the capture_last value is now reset after a
 1129     recursion, since all captures are also reset in this case.
 1131 12. The interpreter no longer returns the "too many substrings" error in the
 1132     case when an overflowing capture is in a branch that is subsequently
 1133     abandoned after a backtrack.
 1135 13. In the pathological case when an offset vector of size 2 is used, pcretest
 1136     now prints out the matched string after a yield of 0 or 1.
 1138 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
 1139     Only supported by the JIT compiler at the moment.
 1141 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
 1143 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
 1144     offset of the starting point of the matching process, provided the offsets
 1145     vector is large enough.
 1147 17. The \A escape now records a lookbehind value of 1, though its execution
 1148     does not actually inspect the previous character. This is to ensure that,
 1149     in partial multi-segment matching, at least one character from the old
 1150     segment is retained when a new segment is processed. Otherwise, if there
 1151     are no lookbehinds in the pattern, \A might match incorrectly at the start
 1152     of a new segment.
 1154 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
 1156 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
 1157     gives some modest performance improvement in 8-bit mode.
 1159 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
 1160     be expressed in certain programming languages using Universal Character
 1161     Names.
 1163 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
 1164     which points out that "non characters" are not "characters that may not
 1165     appear in Unicode strings" but rather "characters that are reserved for
 1166     internal use and have only local meaning".
 1168 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
 1169     there was a conditional group that depended on an assertion, if the
 1170     assertion was false, the callout that immediately followed the alternation
 1171     in the condition was skipped when pcre_exec() was used for matching.
 1173 23. Allow an explicit callout to be inserted before an assertion that is the
 1174     condition for a conditional group, for compatibility with automatic
 1175     callouts, which always insert a callout at this point.
 1177 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
 1178     confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
 1180 25. (*PRUNE) is now supported by the JIT compiler.
 1182 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
 1184 27. Fix the case where there are two or more SKIPs with arguments that may be
 1185     ignored.
 1187 28. (*SKIP) is now supported by the JIT compiler.
 1189 29. (*THEN) is now supported by the JIT compiler.
 1191 30. Update RunTest with additional test selector options.
 1193 31. The way PCRE handles backtracking verbs has been changed in two ways.
 1195     (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
 1196     SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
 1197     backtracking. In some cases this makes it more Perl-compatible, but Perl's
 1198     rather obscure rules do not always do the same thing.
 1200     (2) Previously, backtracking verbs were confined within assertions. This is
 1201     no longer the case for positive assertions, except for (*ACCEPT). Again,
 1202     this sometimes improves Perl compatibility, and sometimes does not.
 1204 32. A number of tests that were in test 2 because Perl did things differently
 1205     have been moved to test 1, because either Perl or PCRE has changed, and
 1206     these tests are now compatible.
 1208 32. Backtracking control verbs are now handled in the same way in JIT and
 1209     interpreter.
 1211 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
 1212     contained a forward subroutine reference caused a compile error.
 1214 34. Auto-detect and optimize limited repetitions in JIT.
 1216 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
 1217     blocking (*UTF) etc.
 1219 36. In the interpreter, maximizing pattern repetitions for characters and
 1220     character types now use tail recursion, which reduces stack usage.
 1222 37. The value of the max lookbehind was not correctly preserved if a compiled
 1223     and saved regex was reloaded on a host of different endianness.
 1225 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
 1226     of the compiled pattern block, expand the flags field from 16 to 32 bits
 1227     because it was almost full.
 1229 39. Try madvise first before posix_madvise.
 1231 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
 1232     with a pattern such as ^$. It has taken 4 years for anybody to notice! The
 1233     original change locked out all matches of empty strings. This has been
 1234     changed so that one match of an empty string per line is recognized.
 1235     Subsequent searches on the same line (for colouring or for --only-matching,
 1236     for example) do not recognize empty strings.
 1238 41. Applied a user patch to fix a number of spelling mistakes in comments.
 1240 42. Data lines longer than 65536 caused pcretest to crash.
 1242 43. Clarified the data type for length and startoffset arguments for pcre_exec
 1243     and pcre_dfa_exec in the function-specific man pages, where they were
 1244     explicitly stated to be in bytes, never having been updated. I also added
 1245     some clarification to the pcreapi man page.
 1247 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
 1248     a segmentation fault.
 1251 Version 8.32 30-November-2012
 1252 -----------------------------
 1254 1.  Improved JIT compiler optimizations for first character search and single
 1255     character iterators.
 1257 2.  Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
 1258     Patch by Daniel Richard G.
 1260 3.  Single character iterator optimizations in the JIT compiler.
 1262 4.  Improved JIT compiler optimizations for character ranges.
 1264 5.  Rename the "leave" variable names to "quit" to improve WinCE compatibility.
 1265     Reported by Giuseppe D'Angelo.
 1267 6.  The PCRE_STARTLINE bit, indicating that a match can occur only at the start
 1268     of a line, was being set incorrectly in cases where .* appeared inside
 1269     atomic brackets at the start of a pattern, or where there was a subsequent
 1270     *PRUNE or *SKIP.
 1272 7.  Improved instruction cache flush for POWER/PowerPC.
 1273     Patch by Daniel Richard G.
 1275 8.  Fixed a number of issues in pcregrep, making it more compatible with GNU
 1276     grep:
 1278     (a) There is now no limit to the number of patterns to be matched.
 1280     (b) An error is given if a pattern is too long.
 1282     (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
 1283         are now supported.
 1285     (d) --exclude-from and --include-from (multiple use) have been added.
 1287     (e) Exclusions and inclusions now apply to all files and directories, not
 1288         just to those obtained from scanning a directory recursively.
 1290     (f) Multiple uses of -f and --file-list are now supported.
 1292     (g) In a Windows environment, the default for -d has been changed from
 1293         "read" (the GNU grep default) to "skip", because otherwise the presence
 1294         of a directory in the file list provokes an error.
 1296     (h) The documentation has been revised and clarified in places.
 1298 9.  Improve the matching speed of capturing brackets.
 1300 10. Changed the meaning of \X so that it now matches a Unicode extended
 1301     grapheme cluster.
 1303 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
 1304     out POSIX threads when JIT support is configured.
 1306 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
 1308 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
 1309     structure could go wrong in environments where size_t is not the same size
 1310     as int.
 1312 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
 1313     set.
 1315 15. The EBCDIC support had decayed; later updates to the code had included
 1316     explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
 1317     general tidy up of EBCDIC-related issues, and the documentation was also
 1318     not quite right. There is now a test that can be run on ASCII systems to
 1319     check some of the EBCDIC-related things (but is it not a full test).
 1321 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
 1322     in a small tidy to the code.
 1324 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
 1326 18. If the --only-matching (-o) option in pcregrep is specified multiple
 1327     times, each one causes appropriate output. For example, -o1 -o2 outputs the
 1328     substrings matched by the 1st and 2nd capturing parentheses. A separating
 1329     string can be specified by --om-separator (default empty).
 1331 19. Improving the first n character searches.
 1333 20. Turn case lists for horizontal and vertical white space into macros so that
 1334     they are defined only once.
 1336 21. This set of changes together give more compatible Unicode case-folding
 1337     behaviour for characters that have more than one other case when UCP
 1338     support is available.
 1340     (a) The Unicode property table now has offsets into a new table of sets of
 1341         three or more characters that are case-equivalent. The MultiStage2.py
 1342         script that generates these tables (the pcre_ucd.c file) now scans
 1343         CaseFolding.txt instead of UnicodeData.txt for character case
 1344         information.
 1346     (b) The code for adding characters or ranges of characters to a character
 1347         class has been abstracted into a generalized function that also handles
 1348         case-independence. In UTF-mode with UCP support, this uses the new data
 1349         to handle characters with more than one other case.
 1351     (c) A bug that is fixed as a result of (b) is that codepoints less than 256
 1352         whose other case is greater than 256 are now correctly matched
 1353         caselessly. Previously, the high codepoint matched the low one, but not
 1354         vice versa.
 1356     (d) The processing of \h, \H, \v, and \ in character classes now makes use
 1357         of the new class addition function, using character lists defined as
 1358         macros alongside the case definitions of 20 above.
 1360     (e) Caseless back references now work with characters that have more than
 1361         one other case.
 1363     (f) General caseless matching of characters with more than one other case
 1364         is supported.
 1366 22. Unicode character properties were updated from Unicode 6.2.0
 1368 23. Improved CMake support under Windows. Patch by Daniel Richard G.
 1370 24. Add support for 32-bit character strings, and UTF-32
 1372 25. Major JIT compiler update (code refactoring and bugfixing).
 1373     Experimental Sparc 32 support is added.
 1375 26. Applied a modified version of Daniel Richard G's patch to create
 1376     pcre.h.generic and config.h.generic by "make" instead of in the
 1377     PrepareRelease script.
 1379 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
 1380     pcre_compile.c when checking for a zero character.
 1382 28. Introducing a native interface for JIT. Through this interface, the compiled
 1383     machine code can be directly executed. The purpose of this interface is to
 1384     provide fast pattern matching, so several sanity checks are not performed.
 1385     However, feature tests are still performed. The new interface provides
 1386     1.4x speedup compared to the old one.
 1388 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
 1389     the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
 1390     was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
 1392 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
 1393     greater than 0x7fffffff (which cannot be represented in UTF-8, even under
 1394     the "old" RFC 2279). Instead, it ended up passing a negative length to
 1395     pcre_exec().
 1397 31. Add support for GCC's visibility feature to hide internal functions.
 1399 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
 1400     "unknown -C option" after outputting 0 or 1.
 1402 33. There is now support for generating a code coverage report for the test
 1403     suite in environments where gcc is the compiler and lcov is installed. This
 1404     is mainly for the benefit of the developers.
 1406 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
 1407     unaddressable using valgrind annotations, allowing valgrind to detect
 1408     invalid memory accesses. This is mainly for the benefit of the developers.
 1410 25. (*UTF) can now be used to start a pattern in any of the three libraries.
 1412 26. Give configure error if --enable-cpp but no C++ compiler found.
 1415 Version 8.31 06-July-2012
 1416 -------------------------
 1418 1.  Fixing a wrong JIT test case and some compiler warnings.
 1420 2.  Removed a bashism from the RunTest script.
 1422 3.  Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
 1423     to unsigned type, result still unsigned" that was given by an MS compiler
 1424     on encountering the code "-sizeof(xxx)".
 1426 4.  Partial matching support is added to the JIT compiler.
 1428 5.  Fixed several bugs concerned with partial matching of items that consist
 1429     of more than one character:
 1431     (a) /^(..)\1/ did not partially match "aba" because checking references was
 1432         done on an "all or nothing" basis. This also applied to repeated
 1433         references.
 1435     (b) \R did not give a hard partial match if \r was found at the end of the
 1436         subject.
 1438     (c) \X did not give a hard partial match after matching one or more
 1439         characters at the end of the subject.
 1441     (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
 1442         a partial match for the string "\r".
 1444     (e) When newline was set to CRLF, the metacharacter "." did not recognize
 1445         a partial match for a CR character at the end of the subject string.
 1447 6.  If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
 1448     running pcretest, the text "(JIT)" added to the output whenever JIT is
 1449     actually used to run the match.
 1451 7.  Individual JIT compile options can be set in pcretest by following -s+[+]
 1452     or /S+[+] with a digit between 1 and 7.
 1454 8.  OP_NOT now supports any UTF character not just single-byte ones.
 1456 9.  (*MARK) control verb is now supported by the JIT compiler.
 1458 10. The command "./RunTest list" lists the available tests without actually
 1459     running any of them. (Because I keep forgetting what they all are.)
 1463 12. Applied a (slightly modified) user-supplied patch that improves performance
 1464     when the heap is used for recursion (compiled with --disable-stack-for-
 1465     recursion). Instead of malloc and free for each heap frame each time a
 1466     logical recursion happens, frames are retained on a chain and re-used where
 1467     possible. This sometimes gives as much as 30% improvement.
 1469 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
 1470     call.
 1472 14. As documented, (*COMMIT) is now confined to within a positive assertion.
 1474 15. It is now possible to link pcretest with libedit as an alternative to
 1475     libreadline.
 1477 16. (*COMMIT) control verb is now supported by the JIT compiler.
 1479 17. The Unicode data tables have been updated to Unicode 6.1.0.
 1481 18. Added --file-list option to pcregrep.
 1483 19. Added binary file support to pcregrep, including the -a, --binary-files,
 1484     -I, and --text options.
 1486 20. The madvise function is renamed for posix_madvise for QNX compatibility
 1487     reasons. Fixed by Giuseppe D'Angelo.
 1489 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
 1491 22. Changed the option for creating symbolic links for 16-bit man pages from
 1492     -s to -sf so that re-installing does not cause issues.
 1494 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
 1496 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
 1497     matches in certain environments (the workspace was not being correctly
 1498     retained). Also added to pcre_dfa_exec() a simple plausibility check on
 1499     some of the workspace data at the beginning of a restart.
 1501 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
 1502     was not doing so when it should - probably a typo introduced by SVN 528
 1503     (change 8.10/14).
 1505 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
 1506     \w+ when the character tables indicated that \x{c4} was a word character.
 1507     There were several related cases, all because the tests for doing a table
 1508     lookup were testing for characters less than 127 instead of 255.
 1510 27. If a pattern contains capturing parentheses that are not used in a match,
 1511     their slots in the ovector are set to -1. For those that are higher than
 1512     any matched groups, this happens at the end of processing. In the case when
 1513     there were back references that the ovector was too small to contain
 1514     (causing temporary malloc'd memory to be used during matching), and the
 1515     highest capturing number was not used, memory off the end of the ovector
 1516     was incorrectly being set to -1. (It was using the size of the temporary
 1517     memory instead of the true size.)
 1519 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
 1520     ovector size, it uses memory at the end of the block that it has got.
 1522 29. Check for an overlong MARK name and give an error at compile time. The
 1523     limit is 255 for the 8-bit library and 65535 for the 16-bit library.
 1525 30. JIT compiler update.
 1527 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
 1528     Rill for the patch.
 1530 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
 1532 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
 1534 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
 1535     SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
 1536     was enabled.
 1538 35. Improve JIT code generation for greedy plus quantifier.
 1540 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
 1541     1 to "aa" instead of to an empty string. The bug affected repeated groups
 1542     that could potentially match an empty string.
 1544 37. Optimizing single character iterators in JIT.
 1546 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
 1547     the same checks as \x{...} characters in non-JavaScript mode. Specifically,
 1548     codepoints that are too big for the mode are faulted, and in a UTF mode,
 1549     disallowed codepoints are also faulted.
 1551 39. If PCRE was compiled with UTF support, in three places in the DFA
 1552     matcher there was code that should only have been obeyed in UTF mode, but
 1553     was being obeyed unconditionally. In 8-bit mode this could cause incorrect
 1554     processing when bytes with values greater than 127 were present. In 16-bit
 1555     mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
 1556     both cases the values are those that cannot be the first data item in a UTF
 1557     character. The three items that might have provoked this were recursions,
 1558     possessively repeated groups, and atomic groups.
 1560 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
 1561     and pcregrep, because some OS require shared objects to be explicitly
 1562     passed to ld, causing the link step to fail if they are not.
 1564 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
 1565     mode, patterns that started with \h* or \R* might be incorrectly matched.
 1568 Version 8.30 04-February-2012
 1569 -----------------------------
 1571 1.  Renamed "isnumber" as "is_a_number" because in some Mac environments this
 1572     name is defined in ctype.h.
 1574 2.  Fixed a bug in fixed-length calculation for lookbehinds that would show up
 1575     only in quite long subpatterns.
 1577 3.  Removed the function pcre_info(), which has been obsolete and deprecated
 1578     since it was replaced by pcre_fullinfo() in February 2000.
 1580 4.  For a non-anchored pattern, if (*SKIP) was given with a name that did not
 1581     match a (*MARK), and the match failed at the start of the subject, a
 1582     reference to memory before the start of the subject could occur. This bug
 1583     was introduced by fix 17 of release 8.21.
 1585 5.  A reference to an unset group with zero minimum repetition was giving
 1586     totally wrong answers (in non-JavaScript-compatibility mode). For example,
 1587     /(another)?(\1?)test/ matched against "hello world test". This bug was
 1588     introduced in release 8.13.
 1590 6.  Add support for 16-bit character strings (a large amount of work involving
 1591     many changes and refactorings).
 1593 7.  RunGrepTest failed on msys because \r\n was replaced by whitespace when the
 1594     command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
 1595     from a file.
 1597 8.  Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
 1598     rounding is not applied in this particular case).
 1600 9.  The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
 1601     if they appear, or are escaped, in patterns.
 1603 10. Get rid of a number of -Wunused-but-set-variable warnings.
 1605 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
 1606     "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
 1607     Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
 1608     also returns the mark "x". This bug applied to capturing parentheses,
 1609     non-capturing parentheses, and atomic parentheses. It also applied to some
 1610     assertions.
 1612 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
 1613     information out of configure.ac instead of relying on pcre.h.generic, which
 1614     is not stored in the repository.
 1616 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
 1617     -lreadline.
 1619 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
 1621 15. Applied Graycode's patch to put the top-level frame on the stack rather
 1622     than the heap when not using the stack for recursion. This gives a
 1623     performance improvement in many cases when recursion is not deep.
 1625 16. Experimental code added to "pcretest -C" to output the stack frame size.
 1628 Version 8.21 12-Dec-2011
 1629 ------------------------
 1631 1.  Updating the JIT compiler.
 1633 2.  JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
 1634     are added as well.
 1636 3.  Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
 1637     PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
 1638     calling _pcre_jit_exec. Some extra comments are added.
 1640 4.  (*MARK) settings inside atomic groups that do not contain any capturing
 1641     parentheses, for example, (?>a(*:m)), were not being passed out. This bug
 1642     was introduced by change 18 for 8.20.
 1644 5.  Supporting of \x, \U and \u in JavaScript compatibility mode based on the
 1645     ECMA-262 standard.
 1647 6.  Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
 1648     erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
 1649     This bug was probably introduced by change 9 of 8.13.
 1651 7.  While fixing 6 above, I noticed that a number of other items were being
 1652     incorrectly rejected as "not fixed length". This arose partly because newer
 1653     opcodes had not been added to the fixed-length checking code. I have (a)
 1654     corrected the bug and added tests for these items, and (b) arranged for an
 1655     error to occur if an unknown opcode is encountered while checking for fixed
 1656     length instead of just assuming "not fixed length". The items that were
 1657     rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
 1658     (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
 1659     repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
 1661 8.  A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
 1662     being incorrectly compiled and would have given unpredicatble results.
 1664 9.  A possessively repeated subpattern with minimum repeat count greater than
 1665     one behaved incorrectly. For example, (A){2,}+ behaved as if it was
 1666     (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
 1667     the first (A) could occur when it should not.
 1669 10. Add a cast and remove a redundant test from the code.
 1671 11. JIT should use pcre_malloc/pcre_free for allocation.
 1673 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
 1674     best practice nowadays, and helps with cross-compiling. (If the exec_prefix
 1675     is anything other than /usr, -L is still shown).
 1677 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
 1679 14. Perl does not support \N without a following name in a [] class; PCRE now
 1680     also gives an error.
 1682 15. If a forward reference was repeated with an upper limit of around 2000,
 1683     it caused the error "internal error: overran compiling workspace". The
 1684     maximum number of forward references (including repeats) was limited by the
 1685     internal workspace, and dependent on the LINK_SIZE. The code has been
 1686     rewritten so that the workspace expands (via pcre_malloc) if necessary, and
 1687     the default depends on LINK_SIZE. There is a new upper limit (for safety)
 1688     of around 200,000 forward references. While doing this, I also speeded up
 1689     the filling in of repeated forward references.
 1691 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
 1692     incorrectly expecting the subject to contain another "a" after the start.
 1694 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
 1695     in the match, the SKIP should be ignored. This was not happening; instead
 1696     the SKIP was being treated as NOMATCH. For patterns such as
 1697     /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
 1698     tested.
 1700 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
 1701     now much more compatible with Perl, in particular in cases where the result
 1702     is a non-match for a non-anchored pattern. For example, if
 1703     /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
 1704     "m", where previously it did not return a name. A side effect of this
 1705     change is that for partial matches, the last encountered mark name is
 1706     returned, as for non matches. A number of tests that were previously not
 1707     Perl-compatible have been moved into the Perl-compatible test files. The
 1708     refactoring has had the pleasing side effect of removing one argument from
 1709     the match() function, thus reducing its stack requirements.
 1711 19. If the /S+ option was used in pcretest to study a pattern using JIT,
 1712     subsequent uses of /S (without +) incorrectly behaved like /S+.
 1714 21. Retrieve executable code size support for the JIT compiler and fixing
 1715     some warnings.
 1717 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
 1718     not work when the shorter character appeared right at the end of the
 1719     subject string.
 1721 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
 1722     systems.
 1724 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
 1725     output it when the /M option is used in pcretest.
 1727 25. The CheckMan script was not being included in the distribution. Also, added
 1728     an explicit "perl" to run Perl scripts from the PrepareRelease script
 1729     because this is reportedly needed in Windows.
 1731 26. If study data was being save in a file and studying had not found a set of
 1732     "starts with" bytes for the pattern, the data written to the file (though
 1733     never used) was taken from uninitialized memory and so caused valgrind to
 1734     complain.
 1736 27. Updated RunTest.bat as provided by Sheri Pierce.
 1738 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
 1740 29. Computation of memory usage for the table of capturing group names was
 1741     giving an unnecessarily large value.
 1744 Version 8.20 21-Oct-2011
 1745 ------------------------
 1747 1.  Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
 1748     a POSIX class. After further experiments with Perl, which convinced me that
 1749     Perl has bugs and confusions, a closing square bracket is no longer allowed
 1750     in a POSIX name. This bug also affected patterns with classes that started
 1751     with full stops.
 1753 2.  If a pattern such as /(a)b|ac/ is matched against "ac", there is no
 1754     captured substring, but while checking the failing first alternative,
 1755     substring 1 is temporarily captured. If the output vector supplied to
 1756     pcre_exec() was not big enough for this capture, the yield of the function
 1757     was still zero ("insufficient space for captured substrings"). This cannot
 1758     be totally fixed without adding another stack variable, which seems a lot
 1759     of expense for a edge case. However, I have improved the situation in cases
 1760     such as /(a)(b)x|abc/ matched against "abc", where the return code
 1761     indicates that fewer than the maximum number of slots in the ovector have
 1762     been set.
 1764 3.  Related to (2) above: when there are more back references in a pattern than
 1765     slots in the output vector, pcre_exec() uses temporary memory during
 1766     matching, and copies in the captures as far as possible afterwards. It was
 1767     using the entire output vector, but this conflicts with the specification
 1768     that only 2/3 is used for passing back captured substrings. Now it uses
 1769     only the first 2/3, for compatibility. This is, of course, another edge
 1770     case.
 1772 4.  Zoltan Herczeg's just-in-time compiler support has been integrated into the
 1773     main code base, and can be used by building with --enable-jit. When this is
 1774     done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
 1775     runtime --no-jit option is given.
 1777 5.  When the number of matches in a pcre_dfa_exec() run exactly filled the
 1778     ovector, the return from the function was zero, implying that there were
 1779     other matches that did not fit. The correct "exactly full" value is now
 1780     returned.
 1782 6.  If a subpattern that was called recursively or as a subroutine contained
 1783     (*PRUNE) or any other control that caused it to give a non-standard return,
 1784     invalid errors such as "Error -26 (nested recursion at the same subject
 1785     position)" or even infinite loops could occur.
 1787 7.  If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
 1788     computing the minimum length on reaching *ACCEPT, and so ended up with the
 1789     wrong value of 1 rather than 0. Further investigation indicates that
 1790     computing a minimum subject length in the presence of *ACCEPT is difficult
 1791     (think back references, subroutine calls), and so I have changed the code
 1792     so that no minimum is registered for a pattern that contains *ACCEPT.
 1794 8.  If (*THEN) was present in the first (true) branch of a conditional group,
 1795     it was not handled as intended. [But see 16 below.]
 1797 9.  Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
 1798     Sheri Pierce.
 1800 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
 1801     the first byte in a match must be "a".
 1803 11. Change 17 for 8.13 increased the recursion depth for patterns like
 1804     /a(?:.)*?a/ drastically. I've improved things by remembering whether a
 1805     pattern contains any instances of (*THEN). If it does not, the old
 1806     optimizations are restored. It would be nice to do this on a per-group
 1807     basis, but at the moment that is not feasible.
 1809 12. In some environments, the output of pcretest -C is CRLF terminated. This
 1810     broke RunTest's code that checks for the link size. A single white space
 1811     character after the value is now allowed for.
 1813 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
 1814     For "fr", it uses the Windows-specific input and output files.
 1816 14. If (*THEN) appeared in a group that was called recursively or as a
 1817     subroutine, it did not work as intended. [But see next item.]
 1819 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
 1820     pattern fragments (but not containing any | characters). If A and B are
 1821     matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
 1822     was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
 1823     D. In other words, Perl considers parentheses that do not contain any |
 1824     characters to be part of a surrounding alternative, whereas PCRE was
 1825     treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
 1826     differently. PCRE now behaves in the same way as Perl, except in the case
 1827     of subroutine/recursion calls such as (?1) which have in any case always
 1828     been different (but PCRE had them first :-).
 1830 16. Related to 15 above: Perl does not treat the | in a conditional group as
 1831     creating alternatives. Such a group is treated in the same way as an
 1832     ordinary group without any | characters when processing (*THEN). PCRE has
 1833     been changed to match Perl's behaviour.
 1835 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
 1836     RunGrepTest script failed.
 1838 18. Change 22 for version 13 caused atomic groups to use more stack. This is
 1839     inevitable for groups that contain captures, but it can lead to a lot of
 1840     stack use in large patterns. The old behaviour has been restored for atomic
 1841     groups that do not contain any capturing parentheses.
 1843 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
 1844     suppress the check for a minimum subject length at run time. (If it was
 1845     given to pcre_exec() or pcre_dfa_exec() it did work.)
 1847 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
 1848     fail to work when decoding hex characters in data strings in EBCDIC
 1849     environments.
 1851 21. It appears that in at least one Mac OS environment, the isxdigit() function
 1852     is implemented as a macro that evaluates to its argument more than once,
 1853     contravening the C 90 Standard (I haven't checked a later standard). There
 1854     was an instance in pcretest which caused it to go wrong when processing
 1855     \x{...} escapes in subject strings. The has been rewritten to avoid using
 1856     things like p++ in the argument of isxdigit().
 1859 Version 8.13 16-Aug-2011
 1860 ------------------------
 1862 1.  The Unicode data tables have been updated to Unicode 6.0.0.
 1864 2.  Two minor typos in pcre_internal.h have been fixed.
 1866 3.  Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
 1867     pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
 1868     in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
 1870 4.  There were a number of related bugs in the code for matching backrefences
 1871     caselessly in UTF-8 mode when codes for the characters concerned were
 1872     different numbers of bytes. For example, U+023A and U+2C65 are an upper
 1873     and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
 1874     (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
 1875     code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
 1876     2-byte code at the end of the subject (it thought there wasn't enough data
 1877     left).
 1879 5.  Comprehensive information about what went wrong is now returned by
 1880     pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
 1881     as the output vector has at least 2 elements. The offset of the start of
 1882     the failing character and a reason code are placed in the vector.
 1884 6.  When the UTF-8 string check fails for pcre_compile(), the offset that is
 1885     now returned is for the first byte of the failing character, instead of the
 1886     last byte inspected. This is an incompatible change, but I hope it is small
 1887     enough not to be a problem. It makes the returned offset consistent with
 1888     pcre_exec() and pcre_dfa_exec().
 1890 7.  pcretest now gives a text phrase as well as the error number when
 1891     pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
 1892     failure, the offset and reason code are output.
 1894 8.  When \R was used with a maximizing quantifier it failed to skip backwards
 1895     over a \r\n pair if the subsequent match failed. Instead, it just skipped
 1896     back over a single character (\n). This seems wrong (because it treated the
 1897     two characters as a single entity when going forwards), conflicts with the
 1898     documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
 1899     behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
 1900     has been changed.
 1902 9.  Some internal refactoring has changed the processing so that the handling
 1903     of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
 1904     time (the PCRE_DOTALL option was changed this way some time ago: version
 1905     7.7 change 16). This has made it possible to abolish the OP_OPT op code,
 1906     which was always a bit of a fudge. It also means that there is one less
 1907     argument for the match() function, which reduces its stack requirements
 1908     slightly. This change also fixes an incompatibility with Perl: the pattern
 1909     (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
 1911 10. More internal refactoring has drastically reduced the number of recursive
 1912     calls to match() for possessively repeated groups such as (abc)++ when
 1913     using pcre_exec().
 1915 11. While implementing 10, a number of bugs in the handling of groups were
 1916     discovered and fixed:
 1918     (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
 1919     (a|)*(?1) gave a compile-time internal error.
 1920     ((a|)+)+  did not notice that the outer group could match an empty string.
 1921     (^a|^)+   was not marked as anchored.
 1922     (.*a|.*)+ was not marked as matching at start or after a newline.
 1924 12. Yet more internal refactoring has removed another argument from the match()
 1925     function. Special calls to this function are now indicated by setting a
 1926     value in a variable in the "match data" data block.
 1928 13. Be more explicit in pcre_study() instead of relying on "default" for
 1929     opcodes that mean there is no starting character; this means that when new
 1930     ones are added and accidentally left out of pcre_study(), testing should
 1931     pick them up.
 1933 14. The -s option of pcretest has been documented for ages as being an old
 1934     synonym of -m (show memory usage). I have changed it to mean "force study
 1935     for every regex", that is, assume /S for every regex. This is similar to -i
 1936     and -d etc. It's slightly incompatible, but I'm hoping nobody is still
 1937     using it. It makes it easier to run collections of tests with and without
 1938     study enabled, and thereby test pcre_study() more easily. All the standard
 1939     tests are now run with and without -s (but some patterns can be marked as
 1940     "never study" - see 20 below).
 1942 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
 1943     restoration of the capturing data to the outer values was not happening
 1944     correctly.
 1946 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
 1947     empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
 1948     pattern had matched an empty string, and so incorrectly returned a no
 1949     match.
 1951 17. There was optimizing code for the last branch of non-capturing parentheses,
 1952     and also for the obeyed branch of a conditional subexpression, which used
 1953     tail recursion to cut down on stack usage. Unfortunately, now that there is
 1954     the possibility of (*THEN) occurring in these branches, tail recursion is
 1955     no longer possible because the return has to be checked for (*THEN). These
 1956     two optimizations have therefore been removed. [But see 8.20/11 above.]
 1958 18. If a pattern containing \R was studied, it was assumed that \R always
 1959     matched two bytes, thus causing the minimum subject length to be
 1960     incorrectly computed because \R can also match just one byte.
 1962 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
 1963     was incorrectly computed.
 1965 20. If /S is present twice on a test pattern in pcretest input, it now
 1966     *disables* studying, thereby overriding the use of -s on the command line
 1967     (see 14 above). This is necessary for one or two tests to keep the output
 1968     identical in both cases.
 1970 21. When (*ACCEPT) was used in an assertion that matched an empty string and
 1971     PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
 1973 22. When an atomic group that contained a capturing parenthesis was
 1974     successfully matched, but the branch in which it appeared failed, the
 1975     capturing was not being forgotten if a higher numbered group was later
 1976     captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
 1977     group 1 to "a", when in fact it should be unset. This applied to multi-
 1978     branched capturing and non-capturing groups, repeated or not, and also to
 1979     positive assertions (capturing in negative assertions does not happen
 1980     in PCRE) and also to nested atomic groups.
 1982 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
 1983     subject after a captured substring, to make it easier to tell which of a
 1984     number of identical substrings has been captured.
 1986 24. The way atomic groups are processed by pcre_exec() has been changed so that
 1987     if they are repeated, backtracking one repetition now resets captured
 1988     values correctly. For example, if ((?>(a+)b)+aabab) is matched against
 1989     "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
 1990     "aaa". Previously, it would have been "a". As part of this code
 1991     refactoring, the way recursive calls are handled has also been changed.
 1993 25. If an assertion condition captured any substrings, they were not passed
 1994     back unless some other capturing happened later. For example, if
 1995     (?(?=(a))a) was matched against "a", no capturing was returned.
 1997 26. When studying a pattern that contained subroutine calls or assertions,
 1998     the code for finding the minimum length of a possible match was handling
 1999     direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
 2000     group 1 called group 2 while simultaneously a separate group 2 called group
 2001     1). A stack overflow occurred in this case. I have fixed this by limiting
 2002     the recursion depth to 10.
 2004 27. Updated RunTest.bat in the distribution to the version supplied by Tom
 2005     Fortmann. This supports explicit test numbers on the command line, and has
 2006     argument validation and error reporting.
 2008 28. An instance of \X with an unlimited repeat could fail if at any point the
 2009     first character it looked at was a mark character.
 2011 29. Some minor code refactoring concerning Unicode properties and scripts
 2012     should reduce the stack requirement of match() slightly.
 2014 30. Added the '=' option to pcretest to check the setting of unused capturing
 2015     slots at the end of the pattern, which are documented as being -1, but are
 2016     not included in the return count.
 2018 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
 2019     compiled something random. Now it gives a compile-time error (as does
 2020     Perl).
 2022 32. A *MARK encountered during the processing of a positive assertion is now
 2023     recorded and passed back (compatible with Perl).
 2025 33. If --only-matching or --colour was set on a pcregrep call whose pattern
 2026     had alternative anchored branches, the search for a second match in a line
 2027     was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
 2028     matched the line "0102" twice. The same bug affected patterns that started
 2029     with a backwards assertion. For example /\b01|\b02/ also matched "0102"
 2030     twice.
 2032 34. Previously, PCRE did not allow quantification of assertions. However, Perl
 2033     does, and because of capturing effects, quantifying parenthesized
 2034     assertions may at times be useful. Quantifiers are now allowed for
 2035     parenthesized assertions.
 2037 35. A minor code tidy in pcre_compile() when checking options for \R usage.
 2039 36. \g was being checked for fancy things in a character class, when it should
 2040     just be a literal "g".
 2042 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
 2043     appearance of a nested POSIX class supersedes an apparent external class.
 2044     For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
 2045     unescaped square brackets may also appear as part of class names. For
 2046     example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
 2047     more like Perl. (But see 8.20/1 above.)
 2049 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
 2050     was because it thought it was \N{name}, which is not supported).
 2052 39. Add minix to OS list not supporting the -S option in pcretest.
 2054 40. PCRE tries to detect cases of infinite recursion at compile time, but it
 2055     cannot analyze patterns in sufficient detail to catch mutual recursions
 2056     such as ((?1))((?2)). There is now a runtime test that gives an error if a
 2057     subgroup is called recursively as a subpattern for a second time at the
 2058     same position in the subject string. In previous releases this might have
 2059     been caught by the recursion limit, or it might have run out of stack.
 2061 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
 2062     happen only once. PCRE was, however incorrectly giving a compile time error
 2063     "recursive call could loop indefinitely" because it cannot analyze the
 2064     pattern in sufficient detail. The compile time test no longer happens when
 2065     PCRE is compiling a conditional subpattern, but actual runaway loops are
 2066     now caught at runtime (see 40 above).
 2068 42. It seems that Perl allows any characters other than a closing parenthesis
 2069     to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
 2070     has been changed to be the same.
 2072 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
 2073     as not to get warnings when autogen.sh is called. Also changed
 2074     AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
 2076 44. To help people who use pcregrep to scan files containing exceedingly long
 2077     lines, the following changes have been made:
 2079     (a) The default value of the buffer size parameter has been increased from
 2080         8K to 20K. (The actual buffer used is three times this size.)
 2082     (b) The default can be changed by ./configure --with-pcregrep-bufsize when
 2083         PCRE is built.
 2085     (c) A --buffer-size=n option has been added to pcregrep, to allow the size
 2086         to be set at run time.
 2088     (d) Numerical values in pcregrep options can be followed by K or M, for
 2089         example --buffer-size=50K.
 2091     (e) If a line being scanned overflows pcregrep's buffer, an error is now
 2092         given and the return code is set to 2.
 2094 45. Add a pointer to the latest mark to the callout data block.
 2096 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
 2097     partial match of an empty string instead of no match. This was specific to
 2098     the use of ".".
 2100 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
 2101     complete match instead of a partial match. This bug was dependent on both
 2102     the PCRE_UTF8 and PCRE_DOTALL options being set.
 2104 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
 2105     starting byte set, because \b was not being ignored.
 2108 Version 8.12 15-Jan-2011
 2109 ------------------------
 2111 1.  Fixed some typos in the markup of the man pages, and wrote a script that
 2112     checks for such things as part of the documentation building process.
 2114 2.  On a big-endian 64-bit system, pcregrep did not correctly process the
 2115     --match-limit and --recursion-limit options (added for 8.11). In
 2116     particular, this made one of the standard tests fail. (The integer value
 2117     went into the wrong half of a long int.)
 2119 3.  If the --colour option was given to pcregrep with -v (invert match), it
 2120     did strange things, either producing crazy output, or crashing. It should,
 2121     of course, ignore a request for colour when reporting lines that do not
 2122     match.
 2124 4.  Another pcregrep bug caused similar problems if --colour was specified with
 2125     -M (multiline) and the pattern match finished with a line ending.
 2127 5.  In pcregrep, when a pattern that ended with a literal newline sequence was
 2128     matched in multiline mode, the following line was shown as part of the
 2129     match. This seems wrong, so I have changed it.
 2131 6.  Another pcregrep bug in multiline mode, when --colour was specified, caused
 2132     the check for further matches in the same line (so they could be coloured)
 2133     to overrun the end of the current line. If another match was found, it was
 2134     incorrectly shown (and then shown again when found in the next line).
 2136 7.  If pcregrep was compiled under Windows, there was a reference to the
 2137     function pcregrep_exit() before it was defined. I am assuming this was
 2138     the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
 2139     reported by a user. I've moved the definition above the reference.
 2142 Version 8.11 10-Dec-2010
 2143 ------------------------
 2145 1.  (*THEN) was not working properly if there were untried alternatives prior
 2146     to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
 2147     backtracked to try for "b" instead of moving to the next alternative branch
 2148     at the same level (in this case, to look for "c"). The Perl documentation
 2149     is clear that when (*THEN) is backtracked onto, it goes to the "next
 2150     alternative in the innermost enclosing group".
 2152 2.  (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
 2153     such as   (A(*COMMIT)B(*THEN)C|D)  any failure after matching A should
 2154     result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
 2155     (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
 2156     (*THEN).
 2158 3.  If \s appeared in a character class, it removed the VT character from
 2159     the class, even if it had been included by some previous item, for example
 2160     in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
 2161     of \s, but is part of the POSIX "space" class.)
 2163 4.  A partial match never returns an empty string (because you can always
 2164     match an empty string at the end of the subject); however the checking for
 2165     an empty string was starting at the "start of match" point. This has been
 2166     changed to the "earliest inspected character" point, because the returned
 2167     data for a partial match starts at this character. This means that, for
 2168     example, /(?<=abc)def/ gives a partial match for the subject "abc"
 2169     (previously it gave "no match").
 2171 5.  Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
 2172     of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
 2173     previously a full match would be given. However, setting PCRE_PARTIAL_HARD
 2174     has an implication that the given string is incomplete (because a partial
 2175     match is preferred over a full match). For this reason, these items now
 2176     give a partial match in this situation. [Aside: previously, the one case
 2177     /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
 2178     match rather than a full match, which was wrong by the old rules, but is
 2179     now correct.]
 2181 6.  There was a bug in the handling of #-introduced comments, recognized when
 2182     PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
 2183     If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
 2184     UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
 2185     scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
 2186     but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
 2187     places in pcre_compile().
 2189 7.  Related to (6) above, when pcre_compile() was skipping #-introduced
 2190     comments when looking ahead for named forward references to subpatterns,
 2191     the only newline sequence it recognized was NL. It now handles newlines
 2192     according to the set newline convention.
 2194 8.  SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
 2195     former, but used strtoul(), whereas pcretest avoided strtoul() but did not
 2196     cater for a lack of strerror(). These oversights have been fixed.
 2198 9.  Added --match-limit and --recursion-limit to pcregrep.
 2200 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
 2202 11. When the -o option was used, pcregrep was setting a return code of 1, even
 2203     when matches were found, and --line-buffered was not being honoured.
 2205 12. Added an optional parentheses number to the -o and --only-matching options
 2206     of pcregrep.
 2208 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
 2209     can match an empty string. The code to do it in pcretest and pcredemo
 2210     needed fixing:
 2212     (a) When the newline convention was "crlf", pcretest got it wrong, skipping
 2213         only one byte after an empty string match just before CRLF (this case
 2214         just got forgotten; "any" and "anycrlf" were OK).
 2216     (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
 2217         mode when an empty string match preceded an ASCII character followed by
 2218         a non-ASCII character. (The code for advancing by one character rather
 2219         than one byte was nonsense.)
 2221     (c) The pcredemo.c sample program did not have any code at all to handle
 2222         the cases when CRLF is a valid newline sequence.
 2224 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
 2225     as a starting offset was within the subject string. There is now a new
 2226     error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
 2227     negative or greater than the length of the string. In order to test this,
 2228     pcretest is extended to allow the setting of negative starting offsets.
 2230 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
 2231     starting offset points to the beginning of a UTF-8 character was
 2232     unnecessarily clumsy. I tidied it up.
 2234 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
 2235     bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
 2237 17. Nobody had reported that the --include_dir option, which was added in
 2238     release 7.7 should have been called --include-dir (hyphen, not underscore)
 2239     for compatibility with GNU grep. I have changed it to --include-dir, but
 2240     left --include_dir as an undocumented synonym, and the same for
 2241     --exclude-dir, though that is not available in GNU grep, at least as of
 2242     release 2.5.4.
 2244 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
 2245     characters from a string of bytes) have been redefined so as not to use
 2246     loops, in order to improve performance in some environments. At the same
 2247     time, I abstracted some of the common code into auxiliary macros to save
 2248     repetition (this should not affect the compiled code).
 2250 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
 2251     compile-time error is now given if \c is not followed by an ASCII
 2252     character, that is, a byte less than 128. (In EBCDIC mode, the code is
 2253     different, and any byte value is allowed.)
 2255 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
 2256     START_OPTIMIZE option, which is now allowed at compile time - but just
 2257     passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
 2258     to pcregrep and other applications that have no direct access to PCRE
 2259     options. The new /Y option in pcretest sets this option when calling
 2260     pcre_compile().
 2262 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
 2263     back references. Groups containing recursive back references were forced to
 2264     be atomic by that change, but in the case of named groups, the amount of
 2265     memory required was incorrectly computed, leading to "Failed: internal
 2266     error: code overflow". This has been fixed.
 2268 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
 2269     pcretest.c, to avoid build problems in some Borland environments.
 2272 Version 8.10 25-Jun-2010
 2273 ------------------------
 2275 1.  Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
 2276     THEN.
 2278 2.  (*ACCEPT) was not working when inside an atomic group.
 2280 3.  Inside a character class, \B is treated as a literal by default, but
 2281     faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
 2282     causes the error). The code is unchanged, but I tidied the documentation.
 2284 4.  Inside a character class, PCRE always treated \R and \X as literals,
 2285     whereas Perl faults them if its -w option is set. I have changed PCRE so
 2286     that it faults them when PCRE_EXTRA is set.
 2288 5.  Added support for \N, which always matches any character other than
 2289     newline. (It is the same as "." when PCRE_DOTALL is not set.)
 2291 6.  When compiling pcregrep with newer versions of gcc which may have
 2292     FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
 2293     declared with attribute warn_unused_result" were given. Just casting the
 2294     result to (void) does not stop the warnings; a more elaborate fudge is
 2295     needed. I've used a macro to implement this.
 2297 7.  Minor change to pcretest.c to avoid a compiler warning.
 2299 8.  Added four artifical Unicode properties to help with an option to make
 2300     \s etc use properties (see next item). The new properties are: Xan
 2301     (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
 2303 9.  Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
 2304     use Unicode properties. (*UCP) at the start of a pattern can be used to set
 2305     this option. Modified pcretest to add /W to test this facility. Added
 2306     REG_UCP to make it available via the POSIX interface.
 2308 10. Added --line-buffered to pcregrep.
 2310 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
 2311     studied, and the match started with a letter with a code point greater than
 2312     127 whose first byte was different to the first byte of the other case of
 2313     the letter, the other case of this starting letter was not recognized
 2314     (#976).
 2316 12. If a pattern that was studied started with a repeated Unicode property
 2317     test, for example, \p{Nd}+, there was the theoretical possibility of
 2318     setting up an incorrect bitmap of starting bytes, but fortunately it could
 2319     not have actually happened in practice until change 8 above was made (it
 2320     added property types that matched character-matching opcodes).
 2322 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
 2323     possible starting bytes for non-anchored patterns.
 2325 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
 2326     \R, and also a number of cases that involve Unicode properties, both
 2327     explicit and implicit when PCRE_UCP is set.
 2329 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
 2330     input, it could crash or give wrong results if characters with values
 2331     greater than 0xc0 were present in the subject string. (Detail: it assumed
 2332     UTF-8 input when processing these items.)
 2334 16. Added a lot of (int) casts to avoid compiler warnings in systems where
 2335     size_t is 64-bit (#991).
 2337 17. Added a check for running out of memory when PCRE is compiled with
 2338     --disable-stack-for-recursion (#990).
 2340 18. If the last data line in a file for pcretest does not have a newline on
 2341     the end, a newline was missing in the output.
 2343 19. The default pcre_chartables.c file recognizes only ASCII characters (values
 2344     less than 128) in its various bitmaps. However, there is a facility for
 2345     generating tables according to the current locale when PCRE is compiled. It
 2346     turns out that in some environments, 0x85 and 0xa0, which are Unicode space
 2347     characters, are recognized by isspace() and therefore were getting set in
 2348     these tables, and indeed these tables seem to approximate to ISO 8859. This
 2349     caused a problem in UTF-8 mode when pcre_study() was used to create a list
 2350     of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
 2351     which of course cannot start UTF-8 characters. I have changed the code so
 2352     that only real ASCII characters (less than 128) and the correct starting
 2353     bytes for UTF-8 encodings are set for characters greater than 127 when in
 2354     UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
 2355     altogether.)
 2357 20. Added the /T option to pcretest so as to be able to run tests with non-
 2358     standard character tables, thus making it possible to include the tests
 2359     used for 19 above in the standard set of tests.
 2361 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
 2362     reference to a subpattern the other side of a comment that contains an
 2363     opening parenthesis caused either an internal compiling error, or a
 2364     reference to the wrong subpattern.
 2367 Version 8.02 19-Mar-2010
 2368 ------------------------
 2370 1.  The Unicode data tables have been updated to Unicode 5.2.0.
 2372 2.  Added the option --libs-cpp to pcre-config, but only when C++ support is
 2373     configured.
 2375 3.  Updated the licensing terms in the pcregexp.pas file, as agreed with the
 2376     original author of that file, following a query about its status.
 2378 4.  On systems that do not have stdint.h (e.g. Solaris), check for and include
 2379     inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
 2381 5.  A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
 2382     quantifier applied to a forward-referencing subroutine call, could compile
 2383     incorrect code or give the error "internal error: previously-checked
 2384     referenced subpattern not found".
 2386 6.  Both MS Visual Studio and Symbian OS have problems with initializing
 2387     variables to point to external functions. For these systems, therefore,
 2388     pcre_malloc etc. are now initialized to local functions that call the
 2389     relevant global functions.
 2391 7.  There were two entries missing in the vectors called coptable and poptable
 2392     in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
 2393     I've fixed the data, and added a kludgy way of testing at compile time that
 2394     the lengths are correct (equal to the number of opcodes).
 2396 8.  Following on from 7, I added a similar kludge to check the length of the
 2397     eint vector in pcreposix.c.
 2399 9.  Error texts for pcre_compile() are held as one long string to avoid too
 2400     much relocation at load time. To find a text, the string is searched,
 2401     counting zeros. There was no check for running off the end of the string,
 2402     which could happen if a new error number was added without updating the
 2403     string.
 2405 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
 2407 11. \K was not working if it appeared in an atomic group or in a group that
 2408     was called as a "subroutine", or in an assertion. Perl 5.11 documents that
 2409     \K is "not well defined" if used in an assertion. PCRE now accepts it if
 2410     the assertion is positive, but not if it is negative.
 2412 12. Change 11 fortuitously reduced the size of the stack frame used in the
 2413     "match()" function of pcre_exec.c by one pointer. Forthcoming
 2414     implementation of support for (*MARK) will need an extra pointer on the
 2415     stack; I have reserved it now, so that the stack frame size does not
 2416     decrease.
 2418 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
 2419     item in branch that calls a recursion is a subroutine call - as in the
 2420     second branch in the above example - was incorrectly given the compile-
 2421     time error "recursive call could loop indefinitely" because pcre_compile()
 2422     was not correctly checking the subroutine for matching a non-empty string.
 2424 14. The checks for overrunning compiling workspace could trigger after an
 2425     overrun had occurred. This is a "should never occur" error, but it can be
 2426     triggered by pathological patterns such as hundreds of nested parentheses.
 2427     The checks now trigger 100 bytes before the end of the workspace.
 2429 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
 2432 Version 8.01 19-Jan-2010
 2433 ------------------------
 2435 1.  If a pattern contained a conditional subpattern with only one branch (in
 2436     particular, this includes all (*DEFINE) patterns), a call to pcre_study()
 2437     computed the wrong minimum data length (which is of course zero for such
 2438     subpatterns). This could cause incorrect "no match" results.
 2440 2.  For patterns such as (?i)a(?-i)b|c where an option setting at the start of
 2441     the pattern is reset in the first branch, pcre_compile() failed with
 2442     "internal error: code overflow at offset...". This happened only when
 2443     the reset was to the original external option setting. (An optimization
 2444     abstracts leading options settings into an external setting, which was the
 2445     cause of this.)
 2447 3.  A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
 2448     of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
 2449     assertion pattern did not match (meaning that the assertion was true), it
 2450     was incorrectly treated as false if the SKIP had been reached during the
 2451     matching. This also applied to assertions used as conditions.
 2453 4.  If an item that is not supported by pcre_dfa_exec() was encountered in an
 2454     assertion subpattern, including such a pattern used as a condition,
 2455     unpredictable results occurred, instead of the error return
 2458 5.  The C++ GlobalReplace function was not working like Perl for the special
 2459     situation when an empty string is matched. It now does the fancy magic
 2460     stuff that is necessary.
 2462 6.  In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
 2463     removed. (These were left over from very, very early versions of PCRE.)
 2465 7.  Some cosmetic changes to the code to make life easier when compiling it
 2466     as part of something else:
 2468     (a) Change DEBUG to PCRE_DEBUG.
 2470     (b) In pcre_compile(), rename the member of the "branch_chain" structure
 2471         called "current" as "current_branch", to prevent a collision with the
 2472         Linux macro when compiled as a kernel module.
 2474     (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
 2475         prevent a collision with the Linux macro when compiled as a kernel
 2476         module.
 2478 8.  In pcre_compile() there are some checks for integer overflows that used to
 2479     cast potentially large values to (double). This has been changed to that
 2480     when building, a check for int64_t is made, and if it is found, it is used
 2481     instead, thus avoiding the use of floating point arithmetic. (There is no
 2482     other use of FP in PCRE.) If int64_t is not found, the fallback is to
 2483     double.
 2485 9.  Added two casts to avoid signed/unsigned warnings from VS Studio Express
 2486     2005 (difference between two addresses compared to an unsigned value).
 2488 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
 2489     custom one, because of the following reported problem in Windows:
 2491       - libbz2 uses the Pascal calling convention (WINAPI) for the functions
 2492           under Win32.
 2493       - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
 2494           therefore missing the function definition.
 2495       - The compiler thus generates a "C" signature for the test function.
 2496       - The linker fails to find the "C" function.
 2497       - PCRE fails to configure if asked to do so against libbz2.
 2499 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
 2500     messages were output:
 2502       Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
 2503       rerunning libtoolize, to keep the correct libtool macros in-tree.
 2504       Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
 2506     I have done both of these things.
 2508 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
 2509     most of the time, it *can* run out if it is given a pattern that contains a
 2510     runaway infinite recursion. I updated the discussion in the pcrestack man
 2511     page.
 2513 13. Now that we have gone to the x.xx style of version numbers, the minor
 2514     version may start with zero. Using 08 or 09 is a bad idea because users
 2515     might check the value of PCRE_MINOR in their code, and 08 or 09 may be
 2516     interpreted as invalid octal numbers. I've updated the previous comment in
 2517     configure.ac, and also added a check that gives an error if 08 or 09 are
 2518     used.
 2520 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
 2521     causing partial matching to fail when the end of the subject matched \W
 2522     in a UTF-8 pattern where \W was quantified with a minimum of 3.
 2524 15. There were some discrepancies between the declarations in pcre_internal.h
 2525     of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
 2526     their definitions. The declarations used "const uschar *" and the
 2527     definitions used USPTR. Even though USPTR is normally defined as "const
 2528     unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
 2529     reported that: "This difference in casting confuses some C++ compilers, for
 2530     example, SunCC recognizes above declarations as different functions and
 2531     generates broken code for hbpcre." I have changed the declarations to use
 2532     USPTR.
 2534 16. GNU libtool is named differently on some systems. The autogen.sh script now
 2535     tries several variants such as glibtoolize (MacOSX) and libtoolize1x
 2536     (FreeBSD).
 2538 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
 2539     (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
 2540     comment: "Figure out how to create a longlong from a string: strtoll and
 2541     equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
 2542     instance, but it only takes 2 args instead of 3!"
 2544 18. A subtle bug concerned with back references has been fixed by a change of
 2545     specification, with a corresponding code fix. A pattern such as
 2546     ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
 2547     refers, was giving matches when it shouldn't. For example, xa=xaaa would
 2548     match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
 2549     same bug. Such groups have to be quantified to be useful, or contained
 2550     inside another quantified group. (If there's no repetition, the reference
 2551     can never match.) The problem arises because, having left the group and
 2552     moved on to the rest of the pattern, a later failure that backtracks into
 2553     the group uses the captured value from the final iteration of the group
 2554     rather than the correct earlier one. I have fixed this in PCRE by forcing
 2555     any group that contains a reference to itself to be an atomic group; that
 2556     is, there cannot be any backtracking into it once it has completed. This is
 2557     similar to recursive and subroutine calls.
 2560 Version 8.00 19-Oct-09
 2561 ----------------------
 2563 1.  The table for translating pcre_compile() error codes into POSIX error codes
 2564     was out-of-date, and there was no check on the pcre_compile() error code
 2565     being within the table. This could lead to an OK return being given in
 2566     error.
 2568 2.  Changed the call to open a subject file in pcregrep from fopen(pathname,
 2569     "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
 2570     in a Windows environment.
 2572 3.  The pcregrep --count option prints the count for each file even when it is
 2573     zero, as does GNU grep. However, pcregrep was also printing all files when
 2574     --files-with-matches was added. Now, when both options are given, it prints
 2575     counts only for those files that have at least one match. (GNU grep just
 2576     prints the file name in this circumstance, but including the count seems
 2577     more useful - otherwise, why use --count?) Also ensured that the
 2578     combination -clh just lists non-zero counts, with no names.
 2580 4.  The long form of the pcregrep -F option was incorrectly implemented as
 2581     --fixed_strings instead of --fixed-strings. This is an incompatible change,
 2582     but it seems right to fix it, and I didn't think it was worth preserving
 2583     the old behaviour.
 2585 5.  The command line items --regex=pattern and --regexp=pattern were not
 2586     recognized by pcregrep, which required --regex pattern or --regexp pattern
 2587     (with a space rather than an '='). The man page documented the '=' forms,
 2588     which are compatible with GNU grep; these now work.
 2590 6.  No libpcreposix.pc file was created for pkg-config; there was just
 2591     libpcre.pc and libpcrecpp.pc. The omission has been rectified.
 2593 7.  Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
 2594     when UCP support is not needed, by modifying the Python script that
 2595     generates it from Unicode data files. This should not matter if the module
 2596     is correctly used as a library, but I received one complaint about 50K of
 2597     unwanted data. My guess is that the person linked everything into his
 2598     program rather than using a library. Anyway, it does no harm.
 2600 8.  A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
 2601     was a minimum greater than 1 for a wide character in a possessive
 2602     repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
 2603     which had an unlimited repeat of a nested, fixed maximum repeat of a wide
 2604     character. Chaos in the form of incorrect output or a compiling loop could
 2605     result.
 2607 9.  The restrictions on what a pattern can contain when partial matching is
 2608     requested for pcre_exec() have been removed. All patterns can now be
 2609     partially matched by this function. In addition, if there are at least two
 2610     slots in the offset vector, the offset of the earliest inspected character
 2611     for the match and the offset of the end of the subject are set in them when
 2612     PCRE_ERROR_PARTIAL is returned.
 2614 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
 2615     synonymous with PCRE_PARTIAL, for backwards compatibility, and
 2616     PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
 2617     and may be more useful for multi-segment matching.
 2619 11. Partial matching with pcre_exec() is now more intuitive. A partial match
 2620     used to be given if ever the end of the subject was reached; now it is
 2621     given only if matching could not proceed because another character was
 2622     needed. This makes a difference in some odd cases such as Z(*FAIL) with the
 2623     string "Z", which now yields "no match" instead of "partial match". In the
 2624     case of pcre_dfa_exec(), "no match" is given if every matching path for the
 2625     final character ended with (*FAIL).
 2627 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
 2628     if the pattern had a "must contain" character that was already found in the
 2629     earlier partial match, unless partial matching was again requested. For
 2630     example, with the pattern /dog.(body)?/, the "must contain" character is
 2631     "g". If the first part-match was for the string "dog", restarting with
 2632     "sbody" failed. This bug has been fixed.
 2634 13. The string returned by pcre_dfa_exec() after a partial match has been
 2635     changed so that it starts at the first inspected character rather than the
 2636     first character of the match. This makes a difference only if the pattern
 2637     starts with a lookbehind assertion or \b or \B (\K is not supported by
 2638     pcre_dfa_exec()). It's an incompatible change, but it makes the two
 2639     matching functions compatible, and I think it's the right thing to do.
 2641 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
 2642     so that the demonstration program is easily available in environments where
 2643     PCRE has not been installed from source.
 2645 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
 2646     libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
 2647     library.
 2649 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
 2650     It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
 2651     is not the first non-POSIX option to be added. Clearly some people find
 2652     these options useful.
 2654 17. If a caller to the POSIX matching function regexec() passes a non-zero
 2655     value for nmatch with a NULL value for pmatch, the value of
 2656     nmatch is forced to zero.
 2658 18. RunGrepTest did not have a test for the availability of the -u option of
 2659     the diff command, as RunTest does. It now checks in the same way as
 2660     RunTest, and also checks for the -b option.
 2662 19. If an odd number of negated classes containing just a single character
 2663     interposed, within parentheses, between a forward reference to a named
 2664     subpattern and the definition of the subpattern, compilation crashed with
 2665     an internal error, complaining that it could not find the referenced
 2666     subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
 2667     [The bug was that it was starting one character too far in when skipping
 2668     over the character class, thus treating the ] as data rather than
 2669     terminating the class. This meant it could skip too much.]
 2671 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
 2672     /g option in pcretest when the pattern contains \K, which makes it possible
 2673     to have an empty string match not at the start, even when the pattern is
 2674     anchored. Updated pcretest and pcredemo to use this option.
 2676 21. If the maximum number of capturing subpatterns in a recursion was greater
 2677     than the maximum at the outer level, the higher number was returned, but
 2678     with unset values at the outer level. The correct (outer level) value is
 2679     now given.
 2681 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
 2682     PCRE did not set those parentheses (unlike Perl). I have now found a way to
 2683     make it do so. The string so far is captured, making this feature
 2684     compatible with Perl.
 2686 23. The tests have been re-organized, adding tests 11 and 12, to make it
 2687     possible to check the Perl 5.10 features against Perl 5.10.
 2689 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
 2690     pattern matches a fixed length string. PCRE did not allow this; now it
 2691     does. Neither allows recursion.
 2693 25. I finally figured out how to implement a request to provide the minimum
 2694     length of subject string that was needed in order to match a given pattern.
 2695     (It was back references and recursion that I had previously got hung up
 2696     on.) This code has now been added to pcre_study(); it finds a lower bound
 2697     to the length of subject needed. It is not necessarily the greatest lower
 2698     bound, but using it to avoid searching strings that are too short does give
 2699     some useful speed-ups. The value is available to calling programs via
 2700     pcre_fullinfo().
 2702 26. While implementing 25, I discovered to my embarrassment that pcretest had
 2703     not been passing the result of pcre_study() to pcre_dfa_exec(), so the
 2704     study optimizations had never been tested with that matching function.
 2705     Oops. What is worse, even when it was passed study data, there was a bug in
 2706     pcre_dfa_exec() that meant it never actually used it. Double oops. There
 2707     were also very few tests of studied patterns with pcre_dfa_exec().
 2709 27. If (?| is used to create subpatterns with duplicate numbers, they are now
 2710     allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
 2711     on the other side of the coin, they are no longer allowed to have different
 2712     names, because these cannot be distinguished in PCRE, and this has caused
 2713     confusion. (This is a difference from Perl.)
 2715 28. When duplicate subpattern names are present (necessarily with different
 2716     numbers, as required by 27 above), and a test is made by name in a
 2717     conditional pattern, either for a subpattern having been matched, or for
 2718     recursion in such a pattern, all the associated numbered subpatterns are
 2719     tested, and the overall condition is true if the condition is true for any
 2720     one of them. This is the way Perl works, and is also more like the way
 2721     testing by number works.
 2724 Version 7.9 11-Apr-09
 2725 ---------------------
 2727 1.  When building with support for bzlib/zlib (pcregrep) and/or readline
 2728     (pcretest), all targets were linked against these libraries. This included
 2729     libpcre, libpcreposix, and libpcrecpp, even though they do not use these
 2730     libraries. This caused unwanted dependencies to be created. This problem
 2731     has been fixed, and now only pcregrep is linked with bzlib/zlib and only
 2732     pcretest is linked with readline.
 2734 2.  The "typedef int BOOL" in pcre_internal.h that was included inside the
 2735     "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
 2736     moved outside it again, because FALSE and TRUE are already defined in AIX,
 2737     but BOOL is not.
 2739 3.  The pcre_config() function was treating the PCRE_MATCH_LIMIT and
 2740     PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
 2742 4.  The pcregrep documentation said spaces were inserted as well as colons (or
 2743     hyphens) following file names and line numbers when outputting matching
 2744     lines. This is not true; no spaces are inserted. I have also clarified the
 2745     wording for the --colour (or --color) option.
 2747 5.  In pcregrep, when --colour was used with -o, the list of matching strings
 2748     was not coloured; this is different to GNU grep, so I have changed it to be
 2749     the same.
 2751 6.  When --colo(u)r was used in pcregrep, only the first matching substring in
 2752     each matching line was coloured. Now it goes on to look for further matches
 2753     of any of the test patterns, which is the same behaviour as GNU grep.
 2755 7.  A pattern that could match an empty string could cause pcregrep to loop; it
 2756     doesn't make sense to accept an empty string match in pcregrep, so I have
 2757     locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
 2758     seems to be how GNU grep behaves. [But see later change 40 for release
 2759     8.33.]
 2761 8.  The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
 2762     start or after a newline", because the conditional assertion was not being
 2763     correctly handled. The rule now is that both the assertion and what follows
 2764     in the first alternative must satisfy the test.
 2766 9.  If auto-callout was enabled in a pattern with a conditional group whose
 2767     condition was an assertion, PCRE could crash during matching, both with
 2768     pcre_exec() and pcre_dfa_exec().
 2770 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
 2771     used for matching.
 2773 11. Unicode property support in character classes was not working for
 2774     characters (bytes) greater than 127 when not in UTF-8 mode.
 2776 12. Added the -M command line option to pcretest.
 2778 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
 2780 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
 2782 16. Added comments and documentation about mis-use of no_arg in the C++
 2783     wrapper.
 2785 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
 2786     from Martin Jerabek that uses macro names for all relevant character and
 2787     string constants.
 2789 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
 2790     SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
 2791     SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
 2792     these, but not everybody uses configure.
 2794 19. A conditional group that had only one branch was not being correctly
 2795     recognized as an item that could match an empty string. This meant that an
 2796     enclosing group might also not be so recognized, causing infinite looping
 2797     (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
 2798     with the subject "ab", where knowledge that the repeated group can match
 2799     nothing is needed in order to break the loop.
 2801 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
 2802     exec(), but without supplying a callout function, matching went wrong.
 2804 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
 2805     leak if the size of the offset vector was greater than 30. When the vector
 2806     is smaller, the saved offsets during recursion go onto a local stack
 2807     vector, but for larger vectors malloc() is used. It was failing to free
 2808     when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
 2809     error, in fact).
 2811 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
 2812     heapframe that is used only when UTF-8 support is enabled. This caused no
 2813     problem, but was untidy.
 2815 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
 2816     CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
 2817     included within another project.
 2819 24. Steven Van Ingelgem's patches to add more options to the CMake support,
 2820     slightly modified by me:
 2822       (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
 2823           not building pcregrep.
 2825       (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
 2826           if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
 2828 25. Forward references, both numeric and by name, in patterns that made use of
 2829     duplicate group numbers, could behave incorrectly or give incorrect errors,
 2830     because when scanning forward to find the reference group, PCRE was not
 2831     taking into account the duplicate group numbers. A pattern such as
 2832     ^X(?3)(a)(?|(b)|(q))(Y) is an example.
 2834 26. Changed a few more instances of "const unsigned char *" to USPTR, making
 2835     the feature of a custom pointer more persuasive (as requested by a user).
 2837 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
 2838     pcretest.c, inside #ifndefs, because it seems they are sometimes already
 2839     pre-defined.
 2841 28. Added support for (*UTF8) at the start of a pattern.
 2843 29. Arrange for flags added by the "release type" setting in CMake to be shown
 2844     in the configuration summary.
 2847 Version 7.8 05-Sep-08
 2848 ---------------------
 2850 1.  Replaced UCP searching code with optimized version as implemented for Ad
 2851     Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
 2852     stage table and inline lookup instead of a function, giving speed ups of 2
 2853     to 5 times on some simple patterns that I tested. Permission was given to
 2854     distribute the MultiStage2.py script that generates the tables (it's not in
 2855     the tarball, but is in the Subversion repository).
 2857 2.  Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
 2858     scripts.
 2860 3.  Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
 2861     a group with a zero qualifier. The result of the study could be incorrect,
 2862     or the function might crash, depending on the pattern.
 2864 4.  Caseless matching was not working for non-ASCII characters in back
 2865     references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
 2866     It now works when Unicode Property Support is available.
 2868 5.  In pcretest, an escape such as \x{de} in the data was always generating
 2869     a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
 2870     non-UTF-8 mode. If the value is greater than 255, it gives a warning about
 2871     truncation.
 2873 6.  Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
 2875 7.  Added two (int) casts to pcregrep when printing the difference of two
 2876     pointers, in case they are 64-bit values.
 2878 8.  Added comments about Mac OS X stack usage to the pcrestack man page and to
 2879     test 2 if it fails.
 2881 9.  Added PCRE_CALL_CONVENTION just before the names of all exported functions,
 2882     and a #define of that name to empty if it is not externally set. This is to
 2883     allow users of MSVC to set it if necessary.
 2885 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
 2886     the convenience functions in the pcre_get.c source file.
 2888 11. An option change at the start of a pattern that had top-level alternatives
 2889     could cause overwriting and/or a crash. This command provoked a crash in
 2890     some environments:
 2892       printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
 2894     This potential security problem was recorded as CVE-2008-2371.
 2896 12. For a pattern where the match had to start at the beginning or immediately
 2897     after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
 2898     pcre_dfa_exec() could read past the end of the passed subject if there was
 2899     no match. To help with detecting such bugs (e.g. with valgrind), I modified
 2900     pcretest so that it places the subject at the end of its malloc-ed buffer.
 2902 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
 2903     exec() might read past the end of the data buffer in UTF-8 mode.
 2905 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
 2906     the data contained the byte 0x85 as part of a UTF-8 character within its
 2907     first line. This applied both to normal and DFA matching.
 2909 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
 2910     /^[^d]*?$/8 failed to match "abc".
 2912 16. Added a missing copyright notice to pcrecpp_internal.h.
 2914 17. Make it more clear in the documentation that values returned from
 2915     pcre_exec() in ovector are byte offsets, not character counts.
 2917 18. Tidied a few places to stop certain compilers from issuing warnings.
 2919 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
 2920     supplied by Stefan Weber. I made a further small update for 7.8 because
 2921     there is a change of source arrangements: the pcre_searchfuncs.c module is
 2922     replaced by pcre_ucd.c.
 2925 Version 7.7 07-May-08
 2926 ---------------------
 2928 1.  Applied Craig's patch to sort out a long long problem: "If we can't convert
 2929     a string to a long long, pretend we don't even have a long long." This is
 2930     done by checking for the strtoq, strtoll, and _strtoi64 functions.
 2932 2.  Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
 2933     pre-7.6 versions, which defined a global no_arg variable instead of putting
 2934     it in the RE class. (See also #8 below.)
 2936 3.  Remove a line of dead code, identified by coverity and reported by Nuno
 2937     Lopes.
 2939 4.  Fixed two related pcregrep bugs involving -r with --include or --exclude:
 2941     (1) The include/exclude patterns were being applied to the whole pathnames
 2942         of files, instead of just to the final components.
 2944     (2) If there was more than one level of directory, the subdirectories were
 2945         skipped unless they satisfied the include/exclude conditions. This is
 2946         inconsistent with GNU grep (and could even be seen as contrary to the
 2947         pcregrep specification - which I improved to make it absolutely clear).
 2948         The action now is always to scan all levels of directory, and just
 2949         apply the include/exclude patterns to regular files.
 2951 5.  Added the --include_dir and --exclude_dir patterns to pcregrep, and used
 2952     --exclude_dir in the tests to avoid scanning .svn directories.
 2954 6.  Applied Craig's patch to the QuoteMeta function so that it escapes the
 2955     NUL character as backslash + 0 rather than backslash + NUL, because PCRE
 2956     doesn't support NULs in patterns.
 2958 7.  Added some missing "const"s to declarations of static tables in
 2959     pcre_compile.c and pcre_dfa_exec.c.
 2961 8.  Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
 2962     caused by fix #2  above. (Subsequently also a second patch to fix the
 2963     first patch. And a third patch - this was a messy problem.)
 2965 9.  Applied Craig's patch to remove the use of push_back().
 2967 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
 2968     matching function regexec().
 2970 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
 2971     which, however, unlike Perl's \g{...}, are subroutine calls, not back
 2972     references. PCRE supports relative numbers with this syntax (I don't think
 2973     Oniguruma does).
 2975 12. Previously, a group with a zero repeat such as (...){0} was completely
 2976     omitted from the compiled regex. However, this means that if the group
 2977     was called as a subroutine from elsewhere in the pattern, things went wrong
 2978     (an internal error was given). Such groups are now left in the compiled
 2979     pattern, with a new opcode that causes them to be skipped at execution
 2980     time.
 2982 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
 2983     to the way PCRE behaves:
 2985     (a) A lone ] character is dis-allowed (Perl treats it as data).
 2987     (b) A back reference to an unmatched subpattern matches an empty string
 2988         (Perl fails the current match path).
 2990     (c) A data ] in a character class must be notated as \] because if the
 2991         first data character in a class is ], it defines an empty class. (In
 2992         Perl it is not possible to have an empty class.) The empty class []
 2993         never matches; it forces failure and is equivalent to (*FAIL) or (?!).
 2994         The negative empty class [^] matches any one character, independently
 2995         of the DOTALL setting.
 2997 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
 2998     non-existent subpattern following a character class starting with ']' and
 2999     containing () gave an internal compiling error instead of "reference to
 3000     non-existent subpattern". Fortunately, when the pattern did exist, the
 3001     compiled code was correct. (When scanning forwards to check for the
 3002     existence of the subpattern, it was treating the data ']' as terminating
 3003     the class, so got the count wrong. When actually compiling, the reference
 3004     was subsequently set up correctly.)
 3006 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
 3007     it was being rejected as not supported by pcre_dfa_exec(), even though
 3008     other assertions are supported. I have made pcre_dfa_exec() support
 3009     (*FAIL).
 3011 16. The implementation of 13c above involved the invention of a new opcode,
 3012     OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
 3013     cannot be changed at match time, I realized I could make a small
 3014     improvement to matching performance by compiling OP_ALLANY instead of
 3015     OP_ANY for "." when DOTALL was set, and then removing the runtime tests
 3016     on the OP_ANY path.
 3018 17. Compiling pcretest on Windows with readline support failed without the
 3019     following two fixes: (1) Make the unistd.h include conditional on
 3020     HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
 3022 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
 3023     ncurses library to be included for pcretest when ReadLine support is
 3024     requested, but also to allow for it to be overridden. This patch came from
 3025     Daniel Bergström.
 3027 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
 3028     as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
 3029     any errors with the current Unicode tables. Thanks to Peter Kankowski for
 3030     spotting this.
 3033 Version 7.6 28-Jan-08
 3034 ---------------------
 3036 1.  A character class containing a very large number of characters with
 3037     codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
 3038     overflow.
 3040 2.  Patch to cut out the "long long" test in pcrecpp_unittest when
 3041     HAVE_LONG_LONG is not defined.
 3043 3.  Applied Christian Ehrlicher's patch to update the CMake build files to
 3044     bring them up to date and include new features. This patch includes:
 3046     - Fixed PH's badly added libz and libbz2 support.
 3047     - Fixed a problem with static linking.
 3048     - Added pcredemo. [But later removed - see 7 below.]
 3049     - Fixed dftables problem and added an option.
 3050     - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
 3051         HAVE_LONG_LONG.
 3052     - Added readline support for pcretest.
 3053     - Added an listing of the option settings after cmake has run.
 3055 4.  A user submitted a patch to Makefile that makes it easy to create
 3056     "pcre.dll" under mingw when using Configure/Make. I added stuff to
 3057     Makefile.am that cause it to include this special target, without
 3058     affecting anything else. Note that the same mingw target plus all
 3059     the other distribution libraries and programs are now supported
 3060     when configuring with CMake (see 6 below) instead of with
 3061     Configure/Make.
 3063 5.  Applied Craig's patch that moves no_arg into the RE class in the C++ code.
 3064     This is an attempt to solve the reported problem "pcrecpp::no_arg is not
 3065     exported in the Windows port". It has not yet been confirmed that the patch
 3066     solves the problem, but it does no harm.
 3068 6.  Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
 3069     NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
 3070     with CMake, and also correct the comment about stack recursion.
 3072 7.  Remove the automatic building of pcredemo from the ./configure system and
 3073     from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
 3074     of a program that users should build themselves after PCRE is installed, so
 3075     building it automatically is not really right. What is more, it gave
 3076     trouble in some build environments.
 3078 8.  Further tidies to CMakeLists.txt from Sheri and Christian.
 3081 Version 7.5 10-Jan-08
 3082 ---------------------
 3084 1.  Applied a patch from Craig: "This patch makes it possible to 'ignore'
 3085     values in parens when parsing an RE using the C++ wrapper."
 3087 2.  Negative specials like \S did not work in character classes in UTF-8 mode.
 3088     Characters greater than 255 were excluded from the class instead of being
 3089     included.
 3091 3.  The same bug as (2) above applied to negated POSIX classes such as
 3092     [:^space:].
 3094 4.  PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
 3095     defined or documented. It seems to have been a typo for PCRE_STATIC, so
 3096     I have changed it.
 3098 5.  The construct (?&) was not diagnosed as a syntax error (it referenced the
 3099     first named subpattern) and a construct such as (?&a) would reference the
 3100     first named subpattern whose name started with "a" (in other words, the
 3101     length check was missing). Both these problems are fixed. "Subpattern name
 3102     expected" is now given for (?&) (a zero-length name), and this patch also
 3103     makes it give the same error for \k'' (previously it complained that that
 3104     was a reference to a non-existent subpattern).
 3106 6.  The erroneous patterns (?+-a) and (?-+a) give different error messages;
 3107     this is right because (?- can be followed by option settings as well as by
 3108     digits. I have, however, made the messages clearer.
 3110 7.  Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
 3111     than the number used in the conditional) now cause a compile-time error.
 3112     This is actually not compatible with Perl, which accepts such patterns, but
 3113     treats the conditional as always being FALSE (as PCRE used to), but it
 3114     seems to me that giving a diagnostic is better.
 3116 8.  Change "alphameric" to the more common word "alphanumeric" in comments
 3117     and messages.
 3119 9.  Fix two occurrences of "backslash" in comments that should have been
 3120     "backspace".
 3122 10. Remove two redundant lines of code that can never be obeyed (their function
 3123     was moved elsewhere).
 3125 11. The program that makes PCRE's Unicode character property table had a bug
 3126     which caused it to generate incorrect table entries for sequences of
 3127     characters that have the same character type, but are in different scripts.
 3128     It amalgamated them into a single range, with the script of the first of
 3129     them. In other words, some characters were in the wrong script. There were
 3130     thirteen such cases, affecting characters in the following ranges:
 3132       U+002b0 - U+002c1
 3133       U+0060c - U+0060d
 3134       U+0061e - U+00612
 3135       U+0064b - U+0065e
 3136       U+0074d - U+0076d
 3137       U+01800 - U+01805
 3138       U+01d00 - U+01d77
 3139       U+01d9b - U+01dbf
 3140       U+0200b - U+0200f
 3141       U+030fc - U+030fe
 3142       U+03260 - U+0327f
 3143       U+0fb46 - U+0fbb1
 3144       U+10450 - U+1049d
 3146 12. The -o option (show only the matching part of a line) for pcregrep was not
 3147     compatible with GNU grep in that, if there was more than one match in a
 3148     line, it showed only the first of them. It now behaves in the same way as
 3149     GNU grep.
 3151 13. If the -o and -v options were combined for pcregrep, it printed a blank
 3152     line for every non-matching line. GNU grep prints nothing, and pcregrep now
 3153     does the same. The return code can be used to tell if there were any
 3154     non-matching lines.
 3156 14. Added --file-offsets and --line-offsets to pcregrep.
 3158 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
 3159     infinitely looping recursion. The bug was that positive lookaheads were not
 3160     being skipped when checking for a possible empty match (negative lookaheads
 3161     and both kinds of lookbehind were skipped).
 3163 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
 3164     inclusion of <windows.h> to before rather than after the definition of
 3165     INVALID_FILE_ATTRIBUTES (patch from David Byron).
 3167 17. Specifying a possessive quantifier with a specific limit for a Unicode
 3168     character property caused pcre_compile() to compile bad code, which led at
 3169     runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
 3170     are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
 3171     caused the error; without that there was no problem.
 3173 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
 3175 19. Added --enable-pcretest-libreadline.
 3177 20. In pcrecpp.cc, the variable 'count' was incremented twice in
 3178     RE::GlobalReplace(). As a result, the number of replacements returned was
 3179     double what it should be. I removed one of the increments, but Craig sent a
 3180     later patch that removed the other one (the right fix) and added unit tests
 3181     that check the return values (which was not done before).
 3183 21. Several CMake things:
 3185     (1) Arranged that, when cmake is used on Unix, the libraries end up with
 3186         the names libpcre and libpcreposix, not just pcre and pcreposix.
 3188     (2) The above change means that pcretest and pcregrep are now correctly
 3189         linked with the newly-built libraries, not previously installed ones.
 3193 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
 3194     crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
 3195     UTF-8 newline character). The key issue is that the pattern starts .*;
 3196     this means that the match must be either at the beginning, or after a
 3197     newline. The bug was in the code for advancing after a failed match and
 3198     checking that the new position followed a newline. It was not taking
 3199     account of UTF-8 characters correctly.
 3201 23. PCRE was behaving differently from Perl in the way it recognized POSIX
 3202     character classes. PCRE was not treating the sequence [:...:] as a
 3203     character class unless the ... were all letters. Perl, however, seems to
 3204     allow any characters between [: and :], though of course it rejects as
 3205     unknown any "names" that contain non-letters, because all the known class
 3206     names consist only of letters. Thus, Perl gives an error for [[:1234:]],
 3207     for example, whereas PCRE did not - it did not recognize a POSIX character
 3208     class. This seemed a bit dangerous, so the code has been changed to be
 3209     closer to Perl. The behaviour is not identical to Perl, because PCRE will
 3210     diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
 3211     treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
 3212     Perl does, and where it didn't before.
 3214 24. Rewrite so as to remove the single use of %n from pcregrep because in some
 3215     Windows environments %n is disabled by default.
 3218 Version 7.4 21-Sep-07
 3219 ---------------------
 3221 1.  Change 7.3/28 was implemented for classes by looking at the bitmap. This
 3222     means that a class such as [\s] counted as "explicit reference to CR or
 3223     LF". That isn't really right - the whole point of the change was to try to
 3224     help when there was an actual mention of one of the two characters. So now
 3225     the change happens only if \r or \n (or a literal CR or LF) character is
 3226     encountered.
 3228 2.  The 32-bit options word was also used for 6 internal flags, but the numbers
 3229     of both had grown to the point where there were only 3 bits left.
 3230     Fortunately, there was spare space in the data structure, and so I have
 3231     moved the internal flags into a new 16-bit field to free up more option
 3232     bits.
 3234 3.  The appearance of (?J) at the start of a pattern set the DUPNAMES option,
 3235     but did not set the internal JCHANGED flag - either of these is enough to
 3236     control the way the "get" function works - but the PCRE_INFO_JCHANGED
 3237     facility is supposed to tell if (?J) was ever used, so now (?J) at the
 3238     start sets both bits.
 3240 4.  Added options (at build time, compile time, exec time) to change \R from
 3241     matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
 3243 5.  doc/pcresyntax.html was missing from the distribution.
 3245 6.  Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
 3246     compatibility, even though it is no longer used.
 3248 7.  Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
 3249     strtoull to pcrecpp.cc to select the available functions in WIN32 when the
 3250     windows.h file is present (where different names are used). [This was
 3251     reversed later after testing - see 16 below.]
 3253 8.  Changed all #include <config.h> to #include "config.h". There were also
 3254     some further <pcre.h> cases that I changed to "pcre.h".
 3256 9.  When pcregrep was used with the --colour option, it missed the line ending
 3257     sequence off the lines that it output.
 3259 10. It was pointed out to me that arrays of string pointers cause lots of
 3260     relocations when a shared library is dynamically loaded. A technique of
 3261     using a single long string with a table of offsets can drastically reduce
 3262     these. I have refactored PCRE in four places to do this. The result is
 3263     dramatic:
 3265       Originally:                          290
 3266       After changing UCP table:            187
 3267       After changing error message table:   43
 3268       After changing table of "verbs"       36
 3269       After changing table of Posix names   22
 3271     Thanks to the folks working on Gregex for glib for this insight.
 3273 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
 3274     unicode-properties was also set.
 3276 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
 3278 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
 3279     checked only for CRLF.
 3281 14. Added casts to pcretest.c to avoid compiler warnings.
 3283 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
 3285 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
 3286     and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
 3287     entirely. This removes changes made in 7 above.
 3289 17. The CMake files have been updated, and there is now more information about
 3290     building with CMake in the NON-UNIX-USE document.
 3293 Version 7.3 28-Aug-07
 3294 ---------------------
 3296  1. In the rejigging of the build system that eventually resulted in 7.1, the
 3297     line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
 3298     brackets there is not right, since it causes compilers to look for an
 3299     installed pcre.h, not the version that is in the source that is being
 3300     compiled (which of course may be different). I have changed it back to:
 3302       #include "pcre.h"
 3304     I have a vague recollection that the change was concerned with compiling in
 3305     different directories, but in the new build system, that is taken care of
 3306     by the VPATH setting the Makefile.
 3308  2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
 3309     when the subject happened to end in the byte 0x85 (e.g. if the last
 3310     character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
 3311     characters but of course it shouldn't be taken as a newline when it is part
 3312     of another character. The bug was that, for an unlimited repeat of . in
 3313     not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
 3314     characters when looking for a newline.
 3316  3. A small performance improvement in the DOTALL UTF-8 mode .* case.
 3318  4. Debugging: adjusted the names of opcodes for different kinds of parentheses
 3319     in debug output.
 3321  5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
 3322     long printing in the pcrecpp unittest when running under MinGW.
 3324  6. ESC_K was left out of the EBCDIC table.
 3326  7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
 3327     parentheses; I made it 1000, which seemed large enough. Unfortunately, the
 3328     limit also applies to "virtual nesting" when a pattern is recursive, and in
 3329     this case 1000 isn't so big. I have been able to remove this limit at the
 3330     expense of backing off one optimization in certain circumstances. Normally,
 3331     when pcre_exec() would call its internal match() function recursively and
 3332     immediately return the result unconditionally, it uses a "tail recursion"
 3333     feature to save stack. However, when a subpattern that can match an empty
 3334     string has an unlimited repetition quantifier, it no longer makes this
 3335     optimization. That gives it a stack frame in which to save the data for
 3336     checking that an empty string has been matched. Previously this was taken
 3337     from the 1000-entry workspace that had been reserved. So now there is no
 3338     explicit limit, but more stack is used.
 3340  8. Applied Daniel's patches to solve problems with the import/export magic
 3341     syntax that is required for Windows, and which was going wrong for the
 3342     pcreposix and pcrecpp parts of the library. These were overlooked when this
 3343     problem was solved for the main library.
 3345  9. There were some crude static tests to avoid integer overflow when computing
 3346     the size of patterns that contain repeated groups with explicit upper
 3347     limits. As the maximum quantifier is 65535, the maximum group length was
 3348     set at 30,000 so that the product of these two numbers did not overflow a
 3349     32-bit integer. However, it turns out that people want to use groups that
 3350     are longer than 30,000 bytes (though not repeat them that many times).
 3351     Change 7.0/17 (the refactoring of the way the pattern size is computed) has
 3352     made it possible to implement the integer overflow checks in a much more
 3353     dynamic way, which I have now done. The artificial limitation on group
 3354     length has been removed - we now have only the limit on the total length of
 3355     the compiled pattern, which depends on the LINK_SIZE setting.
 3357 10. Fixed a bug in the documentation for get/copy named substring when
 3358     duplicate names are permitted. If none of the named substrings are set, the
 3359     functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
 3360     empty string.
 3362 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
 3363     instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
 3364     because the ] is interpreted as the first data character and the
 3365     terminating ] is not found. PCRE has been made compatible with Perl in this
 3366     regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
 3367     cause memory overwriting.
 3369 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
 3370     string has been matched (to stop an infinite loop). It was not recognizing
 3371     a conditional subpattern that could match an empty string if that
 3372     subpattern was within another subpattern. For example, it looped when
 3373     trying to match  (((?(1)X|))*)  but it was OK with  ((?(1)X|)*)  where the
 3374     condition was not nested. This bug has been fixed.
 3376 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
 3377     past the start of the subject in the presence of bytes with the top bit
 3378     set, for example "\x8aBCD".
 3380 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
 3381     (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
 3383 14. Optimized (?!) to (*FAIL).
 3385 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
 3386     This restricts code points to be within the range 0 to 0x10FFFF, excluding
 3387     the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
 3388     full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
 3389     does: it's just the validity check that is more restrictive.
 3391 16. Inserted checks for integer overflows during escape sequence (backslash)
 3392     processing, and also fixed erroneous offset values for syntax errors during
 3393     backslash processing.
 3395 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
 3396     for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
 3398 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
 3399     caused an overrun.
 3401 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
 3402     something other than just ASCII characters) inside a group that had an
 3403     unlimited repeat caused a loop at compile time (while checking to see
 3404     whether the group could match an empty string).
 3406 20. Debugging a pattern containing \p or \P could cause a crash. For example,
 3407     [\P{Any}] did so. (Error in the code for printing property names.)
 3409 21. An orphan \E inside a character class could cause a crash.
 3411 22. A repeated capturing bracket such as (A)? could cause a wild memory
 3412     reference during compilation.
 3414 23. There are several functions in pcre_compile() that scan along a compiled
 3415     expression for various reasons (e.g. to see if it's fixed length for look
 3416     behind). There were bugs in these functions when a repeated \p or \P was
 3417     present in the pattern. These operators have additional parameters compared
 3418     with \d, etc, and these were not being taken into account when moving along
 3419     the compiled data. Specifically:
 3421     (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
 3422         length.
 3424     (b) An item such as \pL+ within a repeated group could cause crashes or
 3425         loops.
 3427     (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
 3428         "reference to non-existent subpattern" error.
 3430     (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
 3432 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
 3433     characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
 3435 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
 3437 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
 3438     character were causing crashes (broken optimization).
 3440 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
 3441     \p or \P) caused a compile-time loop.
 3443 28. More problems have arisen in unanchored patterns when CRLF is a valid line
 3444     break. For example, the unstudied pattern [\r\n]A does not match the string
 3445     "\r\nA" because change 7.0/46 below moves the current point on by two
 3446     characters after failing to match at the start. However, the pattern \nA
 3447     *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
 3448     the same is true. There doesn't seem any very clean way out of this, but
 3449     what I have chosen to do makes the common cases work: PCRE now takes note
 3450     of whether there can be an explicit match for \r or \n anywhere in the
 3451     pattern, and if so, 7.0/46 no longer applies. As part of this change,
 3452     there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
 3453     pattern has explicit CR or LF references.
 3455 29. Added (*CR) etc for changing newline setting at start of pattern.
 3458 Version 7.2 19-Jun-07
 3459 ---------------------
 3461  1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
 3462     which is apparently normally available under Windows.
 3464  2. Re-jig the pcregrep tests with different newline settings in an attempt
 3465     to make them independent of the local environment's newline setting.
 3467  3. Add code to configure.ac to remove -g from the CFLAGS default settings.
 3469  4. Some of the "internals" tests were previously cut out when the link size
 3470     was not 2, because the output contained actual offsets. The recent new
 3471     "Z" feature of pcretest means that these can be cut out, making the tests
 3472     usable with all link sizes.
 3474  5. Implemented Stan Switzer's goto replacement for longjmp() when not using
 3475     stack recursion. This gives a massive performance boost under BSD, but just
 3476     a small improvement under Linux. However, it saves one field in the frame
 3477     in all cases.
 3479  6. Added more features from the forthcoming Perl 5.10:
 3481     (a) (?-n) (where n is a string of digits) is a relative subroutine or
 3482         recursion call. It refers to the nth most recently opened parentheses.
 3484     (b) (?+n) is also a relative subroutine call; it refers to the nth next
 3485         to be opened parentheses.
 3487     (c) Conditions that refer to capturing parentheses can be specified
 3488         relatively, for example, (?(-2)... or (?(+3)...
 3490     (d) \K resets the start of the current match so that everything before
 3491         is not part of it.
 3493     (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
 3495     (f) \g{name} is another synonym - part of Perl 5.10's unification of
 3496         reference syntax.
 3498     (g) (?| introduces a group in which the numbering of parentheses in each
 3499         alternative starts with the same number.
 3501     (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
 3503  7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
 3506  8. A pattern such as  (.*(.)?)*  caused pcre_exec() to fail by either not
 3507     terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
 3508     for detecting groups that can match an empty string.
 3510  9. A pattern with a very large number of alternatives (more than several
 3511     hundred) was running out of internal workspace during the pre-compile
 3512     phase, where pcre_compile() figures out how much memory will be needed. A
 3513     bit of new cunning has reduced the workspace needed for groups with
 3514     alternatives. The 1000-alternative test pattern now uses 12 bytes of
 3515     workspace instead of running out of the 4096 that are available.
 3517 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
 3519 11. Applied patch from Google to remove an optimization that didn't quite work.
 3520     The report of the bug said:
 3522       pcrecpp::RE("a*").FullMatch("aaa") matches, while
 3523       pcrecpp::RE("a*?").FullMatch("aaa") does not, and
 3524       pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
 3526 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
 3527     it matched the wrong number of bytes.
 3530 Version 7.1 24-Apr-07
 3531 ---------------------
 3533  1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
 3534     that is more "standard", making use of automake and other Autotools. There
 3535     is some re-arrangement of the files and adjustment of comments consequent
 3536     on this.
 3538  2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
 3539     for recursive directory scanning broke on some systems because the files
 3540     are not scanned in any specific order and on different systems the order
 3541     was different. A call to "sort" has been inserted into RunGrepTest for the
 3542     approprate test as a short-term fix. In the longer term there may be an
 3543     alternative.
 3545  3. I had an email from Eric Raymond about problems translating some of PCRE's
 3546     man pages to HTML (despite the fact that I distribute HTML pages, some
 3547     people do their own conversions for various reasons). The problems
 3548     concerned the use of low-level troff macros .br and .in. I have therefore
 3549     removed all such uses from the man pages (some were redundant, some could
 3550     be replaced by .nf/.fi pairs). The 132html script that I use to generate
 3551     HTML has been updated to handle .nf/.fi and to complain if it encounters
 3552     .br or .in.
 3554  4. Updated comments in configure.ac that get placed in config.h.in and also
 3555     arranged for config.h to be included in the distribution, with the name
 3556     config.h.generic, for the benefit of those who have to compile without
 3557     Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
 3559  5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
 3560     Weber: (1) pcre_internal.h was missing some function renames; (2) updated
 3561     makevp.bat for the current PCRE, using the additional files
 3562     makevp_c.txt, makevp_l.txt, and pcregexp.pas.
 3564  6. A Windows user reported a minor discrepancy with test 2, which turned out
 3565     to be caused by a trailing space on an input line that had got lost in his
 3566     copy. The trailing space was an accident, so I've just removed it.
 3568  7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
 3569     that is needed.
 3571  8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
 3572     as "const" (a) because they are and (b) because it helps the PHP
 3573     maintainers who have recently made a script to detect big data structures
 3574     in the php code that should be moved to the .rodata section. I remembered
 3575     to update Builducptable as well, so it won't revert if ucptable.h is ever
 3576     re-created.
 3578  9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
 3579     pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
 3580     order to be able to cut out the UTF-8 tables in the latter when UTF-8
 3581     support is not required. This saves 1.5-2K of code, which is important in
 3582     some applications.
 3584     Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
 3585     so as not to refer to the tables, even though these functions will never be
 3586     called when UTF-8 support is disabled. Otherwise there are problems with a
 3587     shared library.
 3589 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
 3591     (a) It was defining its arguments as char * instead of void *.
 3593     (b) It was assuming that all moves were upwards in memory; this was true
 3594         a long time ago when I wrote it, but is no longer the case.
 3596     The emulated memove() is provided for those environments that have neither
 3597     memmove() nor bcopy(). I didn't think anyone used it these days, but that
 3598     is clearly not the case, as these two bugs were recently reported.
 3600 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
 3601     and Detrail to create the HTML documentation, the .txt form of the man
 3602     pages, and it removes trailing spaces from listed files. It also creates
 3603     pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
 3604     case, it wraps all the #defines with #ifndefs. This script should be run
 3605     before "make dist".
 3607 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
 3608     with Unicode property support.
 3610     (a) For a maximizing quantifier, if the two different cases of the
 3611         character were of different lengths in their UTF-8 codings (there are
 3612         some cases like this - I found 11), and the matching function had to
 3613         back up over a mixture of the two cases, it incorrectly assumed they
 3614         were both the same length.
 3616     (b) When PCRE was configured to use the heap rather than the stack for
 3617         recursion during matching, it was not correctly preserving the data for
 3618         the other case of a UTF-8 character when checking ahead for a match
 3619         while processing a minimizing repeat. If the check also involved
 3620         matching a wide character, but failed, corruption could cause an
 3621         erroneous result when trying to check for a repeat of the original
 3622         character.
 3624 13. Some tidying changes to the testing mechanism:
 3626     (a) The RunTest script now detects the internal link size and whether there
 3627         is UTF-8 and UCP support by running ./pcretest -C instead of relying on
 3628         values substituted by "configure". (The RunGrepTest script already did
 3629         this for UTF-8.) The configure.ac script no longer substitutes the
 3630         relevant variables.
 3632     (b) The debugging options /B and /D in pcretest show the compiled bytecode
 3633         with length and offset values. This means that the output is different
 3634         for different internal link sizes. Test 2 is skipped for link sizes
 3635         other than 2 because of this, bypassing the problem. Unfortunately,
 3636         there was also a test in test 3 (the locale tests) that used /B and
 3637         failed for link sizes other than 2. Rather than cut the whole test out,
 3638         I have added a new /Z option to pcretest that replaces the length and
 3639         offset values with spaces. This is now used to make test 3 independent
 3640         of link size. (Test 2 will be tidied up later.)
 3642 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
 3643     segmentation fault instead of returning the appropriate error message.
 3645 15. In multiline mode when the newline sequence was set to "any", the pattern
 3646     ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
 3647     This doesn't seem right; it now treats the CRLF combination as the line
 3648     ending, and so does not match in that case. It's only a pattern such as ^$
 3649     that would hit this one: something like ^ABC$ would have failed after \r
 3650     and then tried again after \r\n.
 3652 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
 3653     in an attempt to make files that differ only in their line terminators
 3654     compare equal. This works on Linux.
 3656 17. Under certain error circumstances pcregrep might try to free random memory
 3657     as it exited. This is now fixed, thanks to valgrind.
 3659 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
 3660     "abc\r\n\r\n", it found an unwanted second match after the second \r. This
 3661     was because its rules for how to advance for /g after matching an empty
 3662     string at the end of a line did not allow for this case. They now check for
 3663     it specially.
 3665 20. pcretest is supposed to handle patterns and data of any length, by
 3666     extending its buffers when necessary. It was getting this wrong when the
 3667     buffer for a data line had to be extended.
 3669 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
 3670     CRLF as a newline sequence.
 3672 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
 3673     out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
 3674     I have nevertheless tidied it up.
 3676 23. Added some casts to kill warnings from HP-UX ia64 compiler.
 3678 24. Added a man page for pcre-config.
 3681 Version 7.0 19-Dec-06
 3682 ---------------------
 3684  1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
 3685     moving to gcc 4.1.1.
 3687  2. The -S option for pcretest uses setrlimit(); I had omitted to #include
 3688     sys/time.h, which is documented as needed for this function. It doesn't
 3689     seem to matter on Linux, but it showed up on some releases of OS X.
 3691  3. It seems that there are systems where bytes whose values are greater than
 3692     127 match isprint() in the "C" locale. The "C" locale should be the
 3693     default when a C program starts up. In most systems, only ASCII printing
 3694     characters match isprint(). This difference caused the output from pcretest
 3695     to vary, making some of the tests fail. I have changed pcretest so that:
 3697     (a) When it is outputting text in the compiled version of a pattern, bytes
 3698         other than 32-126 are always shown as hex escapes.
 3700     (b) When it is outputting text that is a matched part of a subject string,
 3701         it does the same, unless a different locale has been set for the match
 3702         (using the /L modifier). In this case, it uses isprint() to decide.
 3704  4. Fixed a major bug that caused incorrect computation of the amount of memory
 3705     required for a compiled pattern when options that changed within the
 3706     pattern affected the logic of the preliminary scan that determines the
 3707     length. The relevant options are -x, and -i in UTF-8 mode. The result was
 3708     that the computed length was too small. The symptoms of this bug were
 3709     either the PCRE error "internal error: code overflow" from pcre_compile(),
 3710     or a glibc crash with a message such as "pcretest: free(): invalid next
 3711     size (fast)". Examples of patterns that provoked this bug (shown in
 3712     pcretest format) are:
 3714       /(?-x: )/x
 3715       /(?x)(?-x: \s*#\s*)/
 3716       /((?i)[\x{c0}])/8
 3717       /(?i:[\x{c0}])/8
 3719     HOWEVER: Change 17 below makes this fix obsolete as the memory computation
 3720     is now done differently.
 3722  5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
 3723     wrapper classes; (b) implement a new function in the C++ scanner that is
 3724     more efficient than the old way of doing things because it avoids levels of
 3725     recursion in the regex matching; (c) add a paragraph to the documentation
 3726     for the FullMatch() function.
 3728  6. The escape sequence \n was being treated as whatever was defined as
 3729     "newline". Not only was this contrary to the documentation, which states
 3730     that \n is character 10 (hex 0A), but it also went horribly wrong when
 3731     "newline" was defined as CRLF. This has been fixed.
 3733  7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
 3734     was being set to -1 for the "end of line" case (supposedly a value that no
 3735     character can have). Though this value is never used (the check for end of
 3736     line is "zero bytes in current character"), it caused compiler complaints.
 3737     I've changed it to 0xffffffff.
 3739  8. In pcre_version.c, the version string was being built by a sequence of
 3740     C macros that, in the event of PCRE_PRERELEASE being defined as an empty
 3741     string (as it is for production releases) called a macro with an empty
 3742     argument. The C standard says the result of this is undefined. The gcc
 3743     compiler treats it as an empty string (which was what was wanted) but it is
 3744     reported that Visual C gives an error. The source has been hacked around to
 3745     avoid this problem.
 3747  9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
 3748     builds of pcretest, and changed the call to _setmode() to use _O_BINARY
 3749     instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
 3750     of them did).
 3752 10. Originally, pcretest opened its input and output without "b"; then I was
 3753     told that "b" was needed in some environments, so it was added for release
 3754     5.0 to both the input and output. (It makes no difference on Unix-like
 3755     systems.) Later I was told that it is wrong for the input on Windows. I've
 3756     now abstracted the modes into two macros, to make it easier to fiddle with
 3757     them, and removed "b" from the input mode under Windows.
 3759 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
 3761 12. Added -help and --help to pcretest as an official way of being reminded
 3762     of the options.
 3764 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
 3765     and pcrecpp.cc because they annoy compilers at high warning levels.
 3767 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
 3769 15. Fixed an occurrence of == in configure.ac that should have been = (shell
 3770     scripts are not C programs :-) and which was not noticed because it works
 3771     on Linux.
 3773 16. pcretest is supposed to handle any length of pattern and data line (as one
 3774     line or as a continued sequence of lines) by extending its input buffer if
 3775     necessary. This feature was broken for very long pattern lines, leading to
 3776     a string of junk being passed to pcre_compile() if the pattern was longer
 3777     than about 50K.
 3779 17. I have done a major re-factoring of the way pcre_compile() computes the
 3780     amount of memory needed for a compiled pattern. Previously, there was code
 3781     that made a preliminary scan of the pattern in order to do this. That was
 3782     OK when PCRE was new, but as the facilities have expanded, it has become
 3783     harder and harder to keep it in step with the real compile phase, and there
 3784     have been a number of bugs (see for example, 4 above). I have now found a
 3785     cunning way of running the real compile function in a "fake" mode that
 3786     enables it to compute how much memory it would need, while actually only
 3787     ever using a few hundred bytes of working memory and without too many
 3788     tests of the mode. This should make future maintenance and development
 3789     easier. A side effect of this work is that the limit of 200 on the nesting
 3790     depth of parentheses has been removed (though this was never a serious
 3791     limitation, I suspect). However, there is a downside: pcre_compile() now
 3792     runs more slowly than before (30% or more, depending on the pattern). I
 3793     hope this isn't a big issue. There is no effect on runtime performance.
 3795 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
 3796     newline (only possible for the last line of a file) and it was a
 3797     pattern that set a locale (followed by /Lsomething), pcretest crashed.
 3799 19. Added additional timing features to pcretest. (1) The -tm option now times
 3800     matching only, not compiling. (2) Both -t and -tm can be followed, as a
 3801     separate command line item, by a number that specifies the number of
 3802     repeats to use when timing. The default is 50000; this gives better
 3803     precision, but takes uncomfortably long for very large patterns.
 3805 20. Extended pcre_study() to be more clever in cases where a branch of a
 3806     subpattern has no definite first character. For example, (a*|b*)[cd] would
 3807     previously give no result from pcre_study(). Now it recognizes that the
 3808     first character must be a, b, c, or d.
 3810 21. There was an incorrect error "recursive call could loop indefinitely" if
 3811     a subpattern (or the entire pattern) that was being tested for matching an
 3812     empty string contained only one non-empty item after a nested subpattern.
 3813     For example, the pattern (?>\x{100}*)\d(?R) provoked this error
 3814     incorrectly, because the \d was being skipped in the check.
 3816 22. The pcretest program now has a new pattern option /B and a command line
 3817     option -b, which is equivalent to adding /B to every pattern. This causes
 3818     it to show the compiled bytecode, without the additional information that
 3819     -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
 3820     is the same as /B/I).
 3822 23. A new optimization is now able automatically to treat some sequences such
 3823     as a*b as a*+b. More specifically, if something simple (such as a character
 3824     or a simple class like \d) has an unlimited quantifier, and is followed by
 3825     something that cannot possibly match the quantified thing, the quantifier
 3826     is automatically "possessified".
 3828 24. A recursive reference to a subpattern whose number was greater than 39
 3829     went wrong under certain circumstances in UTF-8 mode. This bug could also
 3830     have affected the operation of pcre_study().
 3832 25. Realized that a little bit of performance could be had by replacing
 3833     (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
 3835 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
 3837 27. Possessive quantifiers such as a++ were previously implemented by turning
 3838     them into atomic groups such as ($>a+). Now they have their own opcodes,
 3839     which improves performance. This includes the automatically created ones
 3840     from 23 above.
 3842 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
 3843     lookahead was broken if it was not anchored. PCRE was mistakenly expecting
 3844     the first matched character to be a colon. This applied both to named and
 3845     numbered groups.
 3847 29. The ucpinternal.h header file was missing its idempotency #ifdef.
 3849 30. I was sent a "project" file called libpcre.a.dev which I understand makes
 3850     building PCRE on Windows easier, so I have included it in the distribution.
 3852 31. There is now a check in pcretest against a ridiculously large number being
 3853     returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
 3854     loop, the loop is abandoned.
 3856 32. Forward references to subpatterns in conditions such as (?(2)...) where
 3857     subpattern 2 is defined later cause pcre_compile() to search forwards in
 3858     the pattern for the relevant set of parentheses. This search went wrong
 3859     when there were unescaped parentheses in a character class, parentheses
 3860     escaped with \Q...\E, or parentheses in a #-comment in /x mode.
 3862 33. "Subroutine" calls and backreferences were previously restricted to
 3863     referencing subpatterns earlier in the regex. This restriction has now
 3864     been removed.
 3866 34. Added a number of extra features that are going to be in Perl 5.10. On the
 3867     whole, these are just syntactic alternatives for features that PCRE had
 3868     previously implemented using the Python syntax or my own invention. The
 3869     other formats are all retained for compatibility.
 3871     (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
 3872         as (?P<name>...). The new forms, as well as being in Perl 5.10, are
 3873         also .NET compatible.
 3875     (b) A recursion or subroutine call to a named group can now be defined as
 3876         (?&name) as well as (?P>name).
 3878     (c) A backreference to a named group can now be defined as \k<name> or
 3879         \k'name' as well as (?P=name). The new forms, as well as being in Perl
 3880         5.10, are also .NET compatible.
 3882     (d) A conditional reference to a named group can now use the syntax
 3883         (?(<name>) or (?('name') as well as (?(name).
 3885     (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
 3886         groups (named and numbered) that are never evaluated inline, but can be
 3887         called as "subroutines" from elsewhere. In effect, the DEFINE condition
 3888         is always false. There may be only one alternative in such a group.
 3890     (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
 3891         as the simple (?(R). The condition is true only if the most recent
 3892         recursion is that of the given number or name. It does not search out
 3893         through the entire recursion stack.
 3895     (g) The escape \gN or \g{N} has been added, where N is a positive or
 3896         negative number, specifying an absolute or relative reference.
 3898 35. Tidied to get rid of some further signed/unsigned compiler warnings and
 3899     some "unreachable code" warnings.
 3901 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
 3902     things, this adds five new scripts.
 3904 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
 3905     There were also incompatibilities regarding the handling of \Q..\E inside
 3906     character classes, for example with patterns like [\Qa\E-\Qz\E] where the
 3907     hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
 3909 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
 3910     matches an empty string, and forcibly breaks the loop. There were bugs in
 3911     this code in non-simple cases. For a pattern such as  ^(a()*)*  matched
 3912     against  aaaa  the result was just "a" rather than "aaaa", for example. Two
 3913     separate and independent bugs (that affected different cases) have been
 3914     fixed.
 3916 39. Refactored the code to abolish the use of different opcodes for small
 3917     capturing bracket numbers. This is a tidy that I avoided doing when I
 3918     removed the limit on the number of capturing brackets for 3.5 back in 2001.
 3919     The new approach is not only tidier, it makes it possible to reduce the
 3920     memory needed to fix the previous bug (38).
 3922 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
 3923     sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
 3924     processing dot, circumflex, or dollar metacharacters, or #-comments in /x
 3925     mode.
 3927 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
 3928     report.
 3930 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
 3931     copy construction and assignment in the C++ wrapper.
 3933 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
 3934     couple of bugs that could have given wrong results in the "--newline=crlf"
 3935     case.
 3937 44. Added a number of casts and did some reorganization of signed/unsigned int
 3938     variables following suggestions from Dair Grant. Also renamed the variable
 3939     "this" as "item" because it is a C++ keyword.
 3941 45. Arranged for dftables to add
 3943       #include "pcre_internal.h"
 3945     to pcre_chartables.c because without it, gcc 4.x may remove the array
 3946     definition from the final binary if PCRE is built into a static library and
 3947     dead code stripping is activated.
 3949 46. For an unanchored pattern, if a match attempt fails at the start of a
 3950     newline sequence, and the newline setting is CRLF or ANY, and the next two
 3951     characters are CRLF, advance by two characters instead of one.
 3954 Version 6.7 04-Jul-06
 3955 ---------------------
 3957  1. In order to handle tests when input lines are enormously long, pcretest has
 3958     been re-factored so that it automatically extends its buffers when
 3959     necessary. The code is crude, but this _is_ just a test program. The
 3960     default size has been increased from 32K to 50K.
 3962  2. The code in pcre_study() was using the value of the re argument before
 3963     testing it for NULL. (Of course, in any sensible call of the function, it
 3964     won't be NULL.)
 3966  3. The memmove() emulation function in pcre_internal.h, which is used on
 3967     systems that lack both memmove() and bcopy() - that is, hardly ever -
 3968     was missing a "static" storage class specifier.
 3970  4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
 3971     containing an extended class (one that cannot be represented by a bitmap
 3972     because it contains high-valued characters or Unicode property items, e.g.
 3973     [\pZ]). Almost always one would set UTF-8 mode when processing such a
 3974     pattern, but PCRE should not loop if you do not (it no longer does).
 3975     [Detail: two cases were found: (a) a repeated subpattern containing an
 3976     extended class; (b) a recursive reference to a subpattern that followed a
 3977     previous extended class. It wasn't skipping over the extended class
 3978     correctly when UTF-8 mode was not set.]
 3980  5. A negated single-character class was not being recognized as fixed-length
 3981     in lookbehind assertions such as (?<=[^f]), leading to an incorrect
 3982     compile error "lookbehind assertion is not fixed length".
 3984  6. The RunPerlTest auxiliary script was showing an unexpected difference
 3985     between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
 3986     write a Perl script that can interpret lines of an input file either as
 3987     byte characters or as UTF-8, which is what "perltest" was being required to
 3988     do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
 3989     can't do is switch easily at run time between having the "use utf8;" pragma
 3990     or not. In the end, I fudged it by using the RunPerlTest script to insert
 3991     "use utf8;" explicitly for the UTF-8 tests.
 3993  7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
 3994     the end of the subject string, contrary to the documentation and to what
 3995     Perl does. This was true of both matching functions. Now it matches only at
 3996     the start of the subject and immediately after *internal* newlines.
 3998  8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
 3999     a pointer to an int instead of a pointer to an unsigned long int. This
 4000     caused problems on 64-bit systems.
 4002  9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
 4003     instance of the 'standard' template library not being so standard".
 4005 10. There was no check on the number of named subpatterns nor the maximum
 4006     length of a subpattern name. The product of these values is used to compute
 4007     the size of the memory block for a compiled pattern. By supplying a very
 4008     long subpattern name and a large number of named subpatterns, the size
 4009     computation could be caused to overflow. This is now prevented by limiting
 4010     the length of names to 32 characters, and the number of named subpatterns
 4011     to 10,000.
 4013 11. Subpatterns that are repeated with specific counts have to be replicated in
 4014     the compiled pattern. The size of memory for this was computed from the
 4015     length of the subpattern and the repeat count. The latter is limited to
 4016     65535, but there was no limit on the former, meaning that integer overflow
 4017     could in principle occur. The compiled length of a repeated subpattern is
 4018     now limited to 30,000 bytes in order to prevent this.
 4020 12. Added the optional facility to have named substrings with the same name.
 4022 13. Added the ability to use a named substring as a condition, using the
 4023     Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
 4024     are numbers (not recommended). Forward references are permitted.
 4026 14. Added forward references in named backreferences (if you see what I mean).
 4028 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
 4029     pattern could run off the end of the subject. For example, the pattern
 4030     "(?s)(.{1,5})"8 did this with the subject "ab".
 4032 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
 4033     PCRE_CASELESS was set when matching characters that were quantified with ?
 4034     or *.
 4036 17. A character class other than a single negated character that had a minimum
 4037     but no maximum quantifier - for example [ab]{6,} - was not handled
 4038     correctly by pce_dfa_exec(). It would match only one character.
 4040 18. A valid (though odd) pattern that looked like a POSIX character
 4041     class but used an invalid character after [ (for example [[,abc,]]) caused
 4042     pcre_compile() to give the error "Failed: internal error: code overflow" or
 4043     in some cases to crash with a glibc free() error. This could even happen if
 4044     the pattern terminated after [[ but there just happened to be a sequence of
 4045     letters, a binary zero, and a closing ] in the memory that followed.
 4047 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
 4048     over the years. Originally (before any Unicode support), just the bottom 8
 4049     bits were taken. Thus, for example, \500 really meant \100. Nowadays the
 4050     output from "man perlunicode" includes this:
 4052       The regular expression compiler produces polymorphic opcodes.  That
 4053       is, the pattern adapts to the data and automatically switches to
 4054       the Unicode character scheme when presented with Unicode data--or
 4055       instead uses a traditional byte scheme when presented with byte
 4056       data.
 4058     Sadly, a wide octal escape does not cause a switch, and in a string with
 4059     no other multibyte characters, these octal escapes are treated as before.
 4060     Thus, in Perl, the pattern  /\500/ actually matches \100 but the pattern
 4061     /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
 4062     Unicode string.
 4064     I have not perpetrated such confusion in PCRE. Up till now, it took just
 4065     the bottom 8 bits, as in old Perl. I have now made octal escapes with
 4066     values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
 4067     translate to the appropriate multibyte character.
 4069 29. Applied some refactoring to reduce the number of warnings from Microsoft
 4070     and Borland compilers. This has included removing the fudge introduced
 4071     seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
 4072     a warning about an unused variable.
 4074 21. PCRE has not included VT (character 0x0b) in the set of whitespace
 4075     characters since release 4.0, because Perl (from release 5.004) does not.
 4076     [Or at least, is documented not to: some releases seem to be in conflict
 4077     with the documentation.] However, when a pattern was studied with
 4078     pcre_study() and all its branches started with \s, PCRE still included VT
 4079     as a possible starting character. Of course, this did no harm; it just
 4080     caused an unnecessary match attempt.
 4082 22. Removed a now-redundant internal flag bit that recorded the fact that case
 4083     dependency changed within the pattern. This was once needed for "required
 4084     byte" processing, but is no longer used. This recovers a now-scarce options
 4085     bit. Also moved the least significant internal flag bit to the most-
 4086     significant bit of the word, which was not previously used (hangover from
 4087     the days when it was an int rather than a uint) to free up another bit for
 4088     the future.
 4090 23. Added support for CRLF line endings as well as CR and LF. As well as the
 4091     default being selectable at build time, it can now be changed at runtime
 4092     via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
 4093     specify that it is scanning data with non-default line endings.
 4095 24. Changed the definition of CXXLINK to make it agree with the definition of
 4096     LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
 4098 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
 4099     recursions. This makes a big different to stack usage for some patterns.
 4101 26. If a subpattern containing a named recursion or subroutine reference such
 4102     as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
 4103     the space required for the compiled pattern went wrong and gave too small a
 4104     value. Depending on the environment, this could lead to "Failed: internal
 4105     error: code overflow at offset 49" or "glibc detected double free or
 4106     corruption" errors.
 4108 27. Applied patches from Google (a) to support the new newline modes and (b) to
 4109     advance over multibyte UTF-8 characters in GlobalReplace.
 4111 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
 4112     difference for some implementation of PCRE in some Windows version.
 4114 29. Added some extra testing facilities to pcretest:
 4116     \q<number>   in a data line sets the "match limit" value
 4117     \Q<number>   in a data line sets the "match recursion limt" value
 4118     -S <number>  sets the stack size, where <number> is in megabytes
 4120     The -S option isn't available for Windows.
 4123 Version 6.6 06-Feb-06
 4124 ---------------------
 4126  1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
 4127     in pcreposix.h. I have copied the definition from pcre.h.
 4129  2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
 4130     because pcre.h is no longer a built file.
 4132  3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
 4133     not normally included in the compiled code.
 4136 Version 6.5 01-Feb-06
 4137 ---------------------
 4139  1. When using the partial match feature with pcre_dfa_exec(), it was not
 4140     anchoring the second and subsequent partial matches at the new starting
 4141     point. This could lead to incorrect results. For example, with the pattern
 4142     /1234/, partially matching against "123" and then "a4" gave a match.
 4144  2. Changes to pcregrep:
 4146     (a) All non-match returns from pcre_exec() were being treated as failures
 4147         to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
 4148         error message is output. Some extra information is given for the
 4150         probably the only errors that are likely to be caused by users (by
 4151         specifying a regex that has nested indefinite repeats, for instance).
 4152         If there are more than 20 of these errors, pcregrep is abandoned.
 4154     (b) A binary zero was treated as data while matching, but terminated the
 4155         output line if it was written out. This has been fixed: binary zeroes
 4156         are now no different to any other data bytes.
 4158     (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
 4159         used to set a locale for matching. The --locale=xxxx long option has
 4160         been added (no short equivalent) to specify a locale explicitly on the
 4161         pcregrep command, overriding the environment variables.
 4163     (d) When -B was used with -n, some line numbers in the output were one less
 4164         than they should have been.
 4166     (e) Added the -o (--only-matching) option.
 4168     (f) If -A or -C was used with -c (count only), some lines of context were
 4169         accidentally printed for the final match.
 4171     (g) Added the -H (--with-filename) option.
 4173     (h) The combination of options -rh failed to suppress file names for files
 4174         that were found from directory arguments.
 4176     (i) Added the -D (--devices) and -d (--directories) options.
 4178     (j) Added the -F (--fixed-strings) option.
 4180     (k) Allow "-" to be used as a file name for -f as well as for a data file.
 4182     (l) Added the --colo(u)r option.
 4184     (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
 4185         is not present by default.
 4187  3. A nasty bug was discovered in the handling of recursive patterns, that is,
 4188     items such as (?R) or (?1), when the recursion could match a number of
 4189     alternatives. If it matched one of the alternatives, but subsequently,
 4190     outside the recursion, there was a failure, the code tried to back up into
 4191     the recursion. However, because of the way PCRE is implemented, this is not
 4192     possible, and the result was an incorrect result from the match.
 4194     In order to prevent this happening, the specification of recursion has
 4195     been changed so that all such subpatterns are automatically treated as
 4196     atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
 4198  4. I had overlooked the fact that, in some locales, there are characters for
 4199     which isalpha() is true but neither isupper() nor islower() are true. In
 4200     the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
 4201     and ordfeminine) are like this. This affected the treatment of \w and \W
 4202     when they appeared in character classes, but not when they appeared outside
 4203     a character class. The bit map for "word" characters is now created
 4204     separately from the results of isalnum() instead of just taking it from the
 4205     upper, lower, and digit maps. (Plus the underscore character, of course.)
 4207  5. The above bug also affected the handling of POSIX character classes such as
 4208     [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
 4209     permanent tables. Instead, the bit maps for such a class were previously
 4210     created as the appropriate unions of the upper, lower, and digit bitmaps.
 4211     Now they are created by subtraction from the [[:word:]] class, which has
 4212     its own bitmap.
 4214  6. The [[:blank:]] character class matches horizontal, but not vertical space.
 4215     It is created by subtracting the vertical space characters (\x09, \x0a,
 4216     \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
 4217     subtraction was done in the overall bitmap for a character class, meaning
 4218     that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
 4219     be recognized. This bug has been fixed.
 4221  7. Patches from the folks at Google:
 4223       (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
 4224       real life, but is still worth protecting against".
 4226       (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
 4227       regular expressions".
 4229       (c) pcre_scanner.cc: avoid use of std::count() because not all systems
 4230       have it.
 4232       (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
 4233       "configure" and the latter not, in order to fix a problem somebody had
 4234       with compiling the Arg class on HP-UX.
 4236       (e) Improve the error-handling of the C++ wrapper a little bit.
 4238       (f) New tests for checking recursion limiting.
 4240  8. The pcre_memmove() function, which is used only if the environment does not
 4241     have a standard memmove() function (and is therefore rarely compiled),
 4242     contained two bugs: (a) use of int instead of size_t, and (b) it was not
 4243     returning a result (though PCRE never actually uses the result).
 4245  9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
 4246     large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
 4247     returned instead of calling malloc() with an overflowing number that would
 4248     most likely cause subsequent chaos.
 4250 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
 4252 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
 4253     with this option is matched, the nmatch and pmatch options of regexec() are
 4254     ignored.
 4256 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
 4257     provided in case anyone wants to the the POSIX interface with UTF-8
 4258     strings.
 4260 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
 4261     C++ linking (needed for some HP-UX environments).
 4263 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
 4264     (unused parameter) and in the pcre_printint() function (omitted "default"
 4265     switch label when the default is to do nothing).
 4267 15. Added some code to make it possible, when PCRE is compiled as a C++
 4268     library, to replace subject pointers for pcre_exec() with a smart pointer
 4269     class, thus making it possible to process discontinuous strings.
 4271 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
 4272     much the same function. They were added by different people who were trying
 4273     to make PCRE easy to compile on non-Unix systems. It has been suggested
 4274     that PCRE_EXPORT be abolished now that there is more automatic apparatus
 4275     for compiling on Windows systems. I have therefore replaced it with
 4276     PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
 4277     defaults to "extern" for C or "extern C" for C++, which works fine on
 4278     Unix-like systems. It is now possible to override the value of PCRE_DATA_
 4279     SCOPE with something explicit in config.h. In addition:
 4281     (a) pcreposix.h still had just "extern" instead of either of these macros;
 4282         I have replaced it with PCRE_DATA_SCOPE.
 4284     (b) Functions such as _pcre_xclass(), which are internal to the library,
 4285         but external in the C sense, all had PCRE_EXPORT in their definitions.
 4286         This is apparently wrong for the Windows case, so I have removed it.
 4287         (It makes no difference on Unix-like systems.)
 4289 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
 4290     of recursive calls to match(). This is different to MATCH_LIMIT because
 4291     that limits the total number of calls to match(), not all of which increase
 4292     the depth of recursion. Limiting the recursion depth limits the amount of
 4293     stack (or heap if NO_RECURSE is set) that is used. The default can be set
 4294     when PCRE is compiled, and changed at run time. A patch from Google adds
 4295     this functionality to the C++ interface.
 4297 18. Changes to the handling of Unicode character properties:
 4299     (a) Updated the table to Unicode 4.1.0.
 4301     (b) Recognize characters that are not in the table as "Cn" (undefined).
 4303     (c) I revised the way the table is implemented to a much improved format
 4304         which includes recognition of ranges. It now supports the ranges that
 4305         are defined in UnicodeData.txt, and it also amalgamates other
 4306         characters into ranges. This has reduced the number of entries in the
 4307         table from around 16,000 to around 3,000, thus reducing its size
 4308         considerably. I realized I did not need to use a tree structure after
 4309         all - a binary chop search is just as efficient. Having reduced the
 4310         number of entries, I extended their size from 6 bytes to 8 bytes to
 4311         allow for more data.
 4313     (d) Added support for Unicode script names via properties such as \p{Han}.
 4315 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
 4316     matching that character.
 4318 20. When matching a repeated Unicode property with a minimum greater than zero,
 4319     (for example \pL{2,}), PCRE could look past the end of the subject if it
 4320     reached it while seeking the minimum number of characters. This could
 4321     happen only if some of the characters were more than one byte long, because
 4322     there is a check for at least the minimum number of bytes.
 4324 21. Refactored the implementation of \p and \P so as to be more general, to
 4325     allow for more different types of property in future. This has changed the
 4326     compiled form incompatibly. Anybody with saved compiled patterns that use
 4327     \p or \P will have to recompile them.
 4329 22. Added "Any" and "L&" to the supported property types.
 4331 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
 4332     but give a compile time error if the value is greater than 0xff.
 4334 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
 4335     accidentally not being installed or uninstalled.
 4337 25. The pcre.h file was built from pcre.h.in, but the only changes that were
 4338     made were to insert the current release number. This seemed silly, because
 4339     it made things harder for people building PCRE on systems that don't run
 4340     "configure". I have turned pcre.h into a distributed file, no longer built
 4341     by "configure", with the version identification directly included. There is
 4342     no longer a pcre.h.in file.
 4344     However, this change necessitated a change to the pcre-config script as
 4345     well. It is built from pcre-config.in, and one of the substitutions was the
 4346     release number. I have updated configure.ac so that ./configure now finds
 4347     the release number by grepping pcre.h.
 4349 26. Added the ability to run the tests under valgrind.
 4352 Version 6.4 05-Sep-05
 4353 ---------------------
 4355  1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
 4356     "--" to be printed when multiple files were scanned, even when none of the
 4357     -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
 4358     consider it to be a bug, and have restored the previous behaviour.
 4360  2. A couple of code tidies to get rid of compiler warnings.
 4362  3. The pcretest program used to cheat by referring to symbols in the library
 4363     whose names begin with _pcre_. These are internal symbols that are not
 4364     really supposed to be visible externally, and in some environments it is
 4365     possible to suppress them. The cheating is now confined to including
 4366     certain files from the library's source, which is a bit cleaner.
 4368  4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
 4369     file's purpose clearer.
 4371  5. Reorganized pcre_ucp_findchar().
 4374 Version 6.3 15-Aug-05
 4375 ---------------------
 4377  1. The file libpcre.pc.in did not have general read permission in the tarball.
 4379  2. There were some problems when building without C++ support:
 4381     (a) If C++ support was not built, "make install" and "make test" still
 4382         tried to test it.
 4384     (b) There were problems when the value of CXX was explicitly set. Some
 4385         changes have been made to try to fix these, and ...
 4387     (c) --disable-cpp can now be used to explicitly disable C++ support.
 4389     (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
 4390         backslash in a target when C++ was disabled. This confuses some
 4391         versions of "make", apparently. Using an intermediate variable solves
 4392         this. (Same for CPP_LOBJ.)
 4394  3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
 4395     (non-Windows) now includes $(CFLAGS) because these flags are sometimes
 4396     necessary on certain architectures.
 4398  4. Added a setting of -export-symbols-regex to the link command to remove
 4399     those symbols that are exported in the C sense, but actually are local
 4400     within the library, and not documented. Their names all begin with
 4401     "_pcre_". This is not a perfect job, because (a) we have to except some
 4402     symbols that pcretest ("illegally") uses, and (b) the facility isn't always
 4403     available (and never for static libraries). I have made a note to try to
 4404     find a way round (a) in the future.
 4407 Version 6.2 01-Aug-05
 4408 ---------------------
 4410  1. There was no test for integer overflow of quantifier values. A construction
 4411     such as {1111111111111111} would give undefined results. What is worse, if
 4412     a minimum quantifier for a parenthesized subpattern overflowed and became
 4413     negative, the calculation of the memory size went wrong. This could have
 4414     led to memory overwriting.
 4416  2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
 4418  3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
 4419     operating environments where this matters.
 4421  4. Applied Giuseppe Maxia's patch to add additional features for controlling
 4422     PCRE options from within the C++ wrapper.
 4424  5. Named capturing subpatterns were not being correctly counted when a pattern
 4425     was compiled. This caused two problems: (a) If there were more than 100
 4426     such subpatterns, the calculation of the memory needed for the whole
 4427     compiled pattern went wrong, leading to an overflow error. (b) Numerical
 4428     back references of the form \12, where the number was greater than 9, were
 4429     not recognized as back references, even though there were sufficient
 4430     previous subpatterns.
 4432  6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
 4433     versions of gcc, e.g. 2.95.4.
 4436 Version 6.1 21-Jun-05
 4437 ---------------------
 4439  1. There was one reference to the variable "posix" in pcretest.c that was not
 4440     surrounded by "#if !defined NOPOSIX".
 4442  2. Make it possible to compile pcretest without DFA support, UTF8 support, or
 4443     the cross-check on the old pcre_info() function, for the benefit of the
 4444     cut-down version of PCRE that is currently imported into Exim.
 4446  3. A (silly) pattern starting with (?i)(?-i) caused an internal space
 4447     allocation error. I've done the easy fix, which wastes 2 bytes for sensible
 4448     patterns that start (?i) but I don't think that matters. The use of (?i) is
 4449     just an example; this all applies to the other options as well.
 4451  4. Since libtool seems to echo the compile commands it is issuing, the output
 4452     from "make" can be reduced a bit by putting "@" in front of each libtool
 4453     compile command.
 4455  5. Patch from the folks at Google for configure.in to be a bit more thorough
 4456     in checking for a suitable C++ installation before trying to compile the
 4457     C++ stuff. This should fix a reported problem when a compiler was present,
 4458     but no suitable headers.
 4460  6. The man pages all had just "PCRE" as their title. I have changed them to
 4461     be the relevant file name. I have also arranged that these names are
 4462     retained in the file doc/pcre.txt, which is a concatenation in text format
 4463     of all the man pages except the little individual ones for each function.
 4465  7. The NON-UNIX-USE file had not been updated for the different set of source
 4466     files that come with release 6. I also added a few comments about the C++
 4467     wrapper.
 4470 Version 6.0 07-Jun-05
 4471 ---------------------
 4473  1. Some minor internal re-organization to help with my DFA experiments.
 4475  2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
 4476     didn't matter for the library itself when fully configured, but did matter
 4477     when compiling without UCP support, or within Exim, where the ucp files are
 4478     not imported.
 4480  3. Refactoring of the library code to split up the various functions into
 4481     different source modules. The addition of the new DFA matching code (see
 4482     below) to a single monolithic source would have made it really too
 4483     unwieldy, quite apart from causing all the code to be include in a
 4484     statically linked application, when only some functions are used. This is
 4485     relevant even without the DFA addition now that patterns can be compiled in
 4486     one application and matched in another.
 4488     The downside of splitting up is that there have to be some external
 4489     functions and data tables that are used internally in different modules of
 4490     the library but which are not part of the API. These have all had their
 4491     names changed to start with "_pcre_" so that they are unlikely to clash
 4492     with other external names.
 4494  4. Added an alternate matching function, pcre_dfa_exec(), which matches using
 4495     a different (DFA) algorithm. Although it is slower than the original
 4496     function, it does have some advantages for certain types of matching
 4497     problem.
 4499  5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
 4500     including restarting after a partial match.
 4502  6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
 4503     defined when compiling for Windows was sent to me. I have put it into the
 4504     code, though I have no means of testing or verifying it.
 4506  7. Added the pcre_refcount() auxiliary function.
 4508  8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
 4509     match before or at the first newline in the subject string. In pcretest,
 4510     the /f option on a pattern can be used to set this.
 4512  9. A repeated \w when used in UTF-8 mode with characters greater than 256
 4513     would behave wrongly. This has been present in PCRE since release 4.0.
 4515 10. A number of changes to the pcregrep command:
 4517     (a) Refactored how -x works; insert ^(...)$ instead of setting
 4518         PCRE_ANCHORED and checking the length, in preparation for adding
 4519         something similar for -w.
 4521     (b) Added the -w (match as a word) option.
 4523     (c) Refactored the way lines are read and buffered so as to have more
 4524         than one at a time available.
 4526     (d) Implemented a pcregrep test script.
 4528     (e) Added the -M (multiline match) option. This allows patterns to match
 4529         over several lines of the subject. The buffering ensures that at least
 4530         8K, or the rest of the document (whichever is the shorter) is available
 4531         for matching (and similarly the previous 8K for lookbehind assertions).
 4533     (f) Changed the --help output so that it now says
 4535           -w, --word-regex(p)
 4537         instead of two lines, one with "regex" and the other with "regexp"
 4538         because that confused at least one person since the short forms are the
 4539         same. (This required a bit of code, as the output is generated
 4540         automatically from a table. It wasn't just a text change.)
 4542     (g) -- can be used to terminate pcregrep options if the next thing isn't an
 4543         option but starts with a hyphen. Could be a pattern or a path name
 4544         starting with a hyphen, for instance.
 4546     (h) "-" can be given as a file name to represent stdin.
 4548     (i) When file names are being printed, "(standard input)" is used for
 4549         the standard input, for compatibility with GNU grep. Previously
 4550         "<stdin>" was used.
 4552     (j) The option --label=xxx can be used to supply a name to be used for
 4553         stdin when file names are being printed. There is no short form.
 4555     (k) Re-factored the options decoding logic because we are going to add
 4556         two more options that take data. Such options can now be given in four
 4557         different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
 4559     (l) Added the -A, -B, and -C options for requesting that lines of context
 4560         around matches be printed.
 4562     (m) Added the -L option to print the names of files that do not contain
 4563         any matching lines, that is, the complement of -l.
 4565     (n) The return code is 2 if any file cannot be opened, but pcregrep does
 4566         continue to scan other files.
 4568     (o) The -s option was incorrectly implemented. For compatibility with other
 4569         greps, it now suppresses the error message for a non-existent or non-
 4570         accessible file (but not the return code). There is a new option called
 4571         -q that suppresses the output of matching lines, which was what -s was
 4572         previously doing.
 4574     (p) Added --include and --exclude options to specify files for inclusion
 4575         and exclusion when recursing.
 4577 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
 4578     Hopefully, it now does.
 4580 12. Missing cast in pcre_study().
 4582 13. Added an "uninstall" target to the makefile.
 4584 14. Replaced "extern" in the function prototypes in Makefile.in with
 4585     "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
 4586     world, but is set differently for Windows.
 4588 15. Added a second compiling function called pcre_compile2(). The only
 4589     difference is that it has an extra argument, which is a pointer to an
 4590     integer error code. When there is a compile-time failure, this is set
 4591     non-zero, in addition to the error test pointer being set to point to an
 4592     error message. The new argument may be NULL if no error number is required
 4593     (but then you may as well call pcre_compile(), which is now just a
 4594     wrapper). This facility is provided because some applications need a
 4595     numeric error indication, but it has also enabled me to tidy up the way
 4596     compile-time errors are handled in the POSIX wrapper.
 4598 16. Added VPATH=.libs to the makefile; this should help when building with one
 4599     prefix path and installing with another. (Or so I'm told by someone who
 4600     knows more about this stuff than I do.)
 4602 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
 4603     passes PCRE_DOTALL to the pcre_compile() function, making the "." character
 4604     match everything, including newlines. This is not POSIX-compatible, but
 4605     somebody wanted the feature. From pcretest it can be activated by using
 4606     both the P and the s flags.
 4608 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
 4610 19. libpcre.pc was being incorrectly installed as executable.
 4612 20. A couple of places in pcretest check for end-of-line by looking for '\n';
 4613     it now also looks for '\r' so that it will work unmodified on Windows.
 4615 21. Added Google's contributed C++ wrapper to the distribution.
 4617 22. Added some untidy missing memory free() calls in pcretest, to keep
 4618     Electric Fence happy when testing.
 4622 Version 5.0 13-Sep-04
 4623 ---------------------
 4625  1. Internal change: literal characters are no longer packed up into items
 4626     containing multiple characters in a single byte-string. Each character
 4627     is now matched using a separate opcode. However, there may be more than one
 4628     byte in the character in UTF-8 mode.
 4630  2. The pcre_callout_block structure has two new fields: pattern_position and
 4631     next_item_length. These contain the offset in the pattern to the next match
 4632     item, and its length, respectively.
 4634  3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
 4635     insertion of callouts before each pattern item. Added the /C option to
 4636     pcretest to make use of this.
 4638  4. On the advice of a Windows user, the lines
 4640       #if defined(_WIN32) || defined(WIN32)
 4641       _setmode( _fileno( stdout ), 0x8000 );
 4642       #endif  /* defined(_WIN32) || defined(WIN32) */
 4644     have been added to the source of pcretest. This apparently does useful
 4645     magic in relation to line terminators.
 4647  5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
 4648     for the benefit of those environments where the "b" makes a difference.
 4650  6. The icc compiler has the same options as gcc, but "configure" doesn't seem
 4651     to know about it. I have put a hack into configure.in that adds in code
 4652     to set GCC=yes if CC=icc. This seems to end up at a point in the
 4653     generated configure script that is early enough to affect the setting of
 4654     compiler options, which is what is needed, but I have no means of testing
 4655     whether it really works. (The user who reported this had patched the
 4656     generated configure script, which of course I cannot do.)
 4658     LATER: After change 22 below (new libtool files), the configure script
 4659     seems to know about icc (and also ecc). Therefore, I have commented out
 4660     this hack in configure.in.
 4662  7. Added support for pkg-config (2 patches were sent in).
 4664  8. Negated POSIX character classes that used a combination of internal tables
 4665     were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
 4666     [[:^ascii]]. Typically, they would match almost any characters. The other
 4667     POSIX classes were not broken in this way.
 4669  9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
 4670     to find the match, as PCRE was deluded into thinking that the match had to
 4671     start at the start point or following a newline. The same bug applied to
 4672     patterns with negative forward assertions or any backward assertions
 4673     preceding ".*" at the start, unless the pattern required a fixed first
 4674     character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
 4676 10. In UTF-8 mode, when moving forwards in the subject after a failed match
 4677     starting at the last subject character, bytes beyond the end of the subject
 4678     string were read.
 4680 11. Renamed the variable "class" as "classbits" to make life easier for C++
 4681     users. (Previously there was a macro definition, but it apparently wasn't
 4682     enough.)
 4684 12. Added the new field "tables" to the extra data so that tables can be passed
 4685     in at exec time, or the internal tables can be re-selected. This allows
 4686     a compiled regex to be saved and re-used at a later time by a different
 4687     program that might have everything at different addresses.
 4689 13. Modified the pcre-config script so that, when run on Solaris, it shows a
 4690     -R library as well as a -L library.
 4692 14. The debugging options of pcretest (-d on the command line or D on a
 4693     pattern) showed incorrect output for anything following an extended class
 4694     that contained multibyte characters and which was followed by a quantifier.
 4696 15. Added optional support for general category Unicode character properties
 4697     via the \p, \P, and \X escapes. Unicode property support implies UTF-8
 4698     support. It adds about 90K to the size of the library. The meanings of the
 4699     inbuilt class escapes such as \d and \s have NOT been changed.
 4701 16. Updated pcredemo.c to include calls to free() to release the memory for the
 4702     compiled pattern.
 4704 17. The generated file chartables.c was being created in the source directory
 4705     instead of in the building directory. This caused the build to fail if the
 4706     source directory was different from the building directory, and was
 4707     read-only.
 4709 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
 4710     file. No doubt somebody will tell me if they don't make sense... Also added
 4711     Dan Mooney's comments about building on OpenVMS.
 4713 19. Added support for partial matching via the PCRE_PARTIAL option for
 4714     pcre_exec() and the \P data escape in pcretest.
 4716 20. Extended pcretest with 3 new pattern features:
 4718     (i)   A pattern option of the form ">rest-of-line" causes pcretest to
 4719           write the compiled pattern to the file whose name is "rest-of-line".
 4720           This is a straight binary dump of the data, with the saved pointer to
 4721           the character tables forced to be NULL. The study data, if any, is
 4722           written too. After writing, pcretest reads a new pattern.
 4724     (ii)  If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
 4725           compiled pattern from the given file. There must not be any
 4726           occurrences of "<" in the file name (pretty unlikely); if there are,
 4727           pcretest will instead treat the initial "<" as a pattern delimiter.
 4728           After reading in the pattern, pcretest goes on to read data lines as
 4729           usual.
 4731     (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
 4732           and 16-bit fields in a compiled pattern, to simulate a pattern that
 4733           was compiled on a host of opposite endianness.
 4735 21. The pcre-exec() function can now cope with patterns that were compiled on
 4736     hosts of opposite endianness, with this restriction:
 4738       As for any compiled expression that is saved and used later, the tables
 4739       pointer field cannot be preserved; the extra_data field in the arguments
 4740       to pcre_exec() should be used to pass in a tables address if a value
 4741       other than the default internal tables were used at compile time.
 4743 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
 4744     now diagnosed as an error. Previously, most of the time, a negative number
 4745     would have been treated as zero, but if in addition "ovector" was passed as
 4746     NULL, a crash could occur.
 4748 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
 4749     new versions from the libtool 1.5 distribution (the last one is a copy of
 4750     a file called libtool.m4). This seems to have fixed the need to patch
 4751     "configure" to support Darwin 1.3 (which I used to do). However, I still
 4752     had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
 4753     workstation).
 4755 24. Changed the PCRE licence to be the more standard "BSD" licence.
 4758 Version 4.5 01-Dec-03
 4759 ---------------------
 4761  1. There has been some re-arrangement of the code for the match() function so
 4762     that it can be compiled in a version that does not call itself recursively.
 4763     Instead, it keeps those local variables that need separate instances for
 4764     each "recursion" in a frame on the heap, and gets/frees frames whenever it
 4765     needs to "recurse". Keeping track of where control must go is done by means
 4766     of setjmp/longjmp. The whole thing is implemented by a set of macros that
 4767     hide most of the details from the main code, and operates only if
 4768     NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
 4769     "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
 4770     operating.
 4772     To make it easier for callers to provide specially tailored get/free
 4773     functions for this usage, two new functions, pcre_stack_malloc, and
 4774     pcre_stack_free, are used. They are always called in strict stacking order,
 4775     and the size of block requested is always the same.
 4777     The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
 4778     PCRE has been compiled to use the stack or the heap for recursion. The
 4779     -C option of pcretest uses this to show which version is compiled.
 4781     A new data escape \S, is added to pcretest; it causes the amounts of store
 4782     obtained and freed by both kinds of malloc/free at match time to be added
 4783     to the output.
 4785  2. Changed the locale test to use "fr_FR" instead of "fr" because that's
 4786     what's available on my current Linux desktop machine.
 4788  3. When matching a UTF-8 string, the test for a valid string at the start has
 4789     been extended. If start_offset is not zero, PCRE now checks that it points
 4790     to a byte that is the start of a UTF-8 character. If not, it returns
 4791     PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
 4792     this is necessary because there may be backward assertions in the pattern.
 4793     When matching the same subject several times, it may save resources to use
 4794     PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
 4796  4. The code for checking the validity of UTF-8 strings has been tightened so
 4797     that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
 4798     containing "overlong sequences".
 4800  5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
 4801     I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
 4802     should have been "&", but it just so happened that all the cases this let
 4803     through by mistake were picked up later in the function.
 4805  6. I had used a variable called "isblank" - this is a C99 function, causing
 4806     some compilers to warn. To avoid this, I renamed it (as "blankclass").
 4808  7. Cosmetic: (a) only output another newline at the end of pcretest if it is
 4809     prompting; (b) run "./pcretest /dev/null" at the start of the test script
 4810     so the version is shown; (c) stop "make test" echoing "./RunTest".
 4812  8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
 4814  9. The prototype for memmove() for systems that don't have it was using
 4815     size_t, but the inclusion of the header that defines size_t was later. I've
 4816     moved the #includes for the C headers earlier to avoid this.
 4818 10. Added some adjustments to the code to make it easier to compiler on certain
 4819     special systems:
 4821       (a) Some "const" qualifiers were missing.
 4822       (b) Added the macro EXPORT before all exported functions; by default this
 4823           is defined to be empty.
 4824       (c) Changed the dftables auxiliary program (that builds chartables.c) so
 4825           that it reads its output file name as an argument instead of writing
 4826           to the standard output and assuming this can be redirected.
 4828 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
 4829     class containing characters with values greater than 255, PCRE compilation
 4830     went into a loop.
 4832 12. A recursive reference to a subpattern that was within another subpattern
 4833     that had a minimum quantifier of zero caused PCRE to crash. For example,
 4834     (x(y(?2))z)? provoked this bug with a subject that got as far as the
 4835     recursion. If the recursively-called subpattern itself had a zero repeat,
 4836     that was OK.
 4838 13. In pcretest, the buffer for reading a data line was set at 30K, but the
 4839     buffer into which it was copied (for escape processing) was still set at
 4840     1024, so long lines caused crashes.
 4842 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
 4843     "internal error: code overflow...". This applied to any character class
 4844     that was followed by a possessive quantifier.
 4846 15. Modified the Makefile to add libpcre.la as a prerequisite for
 4847     libpcreposix.la because I was told this is needed for a parallel build to
 4848     work.
 4850 16. If a pattern that contained .* following optional items at the start was
 4851     studied, the wrong optimizing data was generated, leading to matching
 4852     errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
 4853     matching string must start with a or b or c. The correct conclusion for
 4854     this pattern is that a match can start with any character.
 4857 Version 4.4 13-Aug-03
 4858 ---------------------
 4860  1. In UTF-8 mode, a character class containing characters with values between
 4861     127 and 255 was not handled correctly if the compiled pattern was studied.
 4862     In fixing this, I have also improved the studying algorithm for such
 4863     classes (slightly).
 4865  2. Three internal functions had redundant arguments passed to them. Removal
 4866     might give a very teeny performance improvement.
 4868  3. Documentation bug: the value of the capture_top field in a callout is *one
 4869     more than* the number of the hightest numbered captured substring.
 4871  4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
 4872     in incorrectly linking with a previously installed version. They now link
 4873     explicitly with libpcre.la.
 4875  5. configure.in no longer needs to recognize Cygwin specially.
 4877  6. A problem in pcre.in for Windows platforms is fixed.
 4879  7. If a pattern was successfully studied, and the -d (or /D) flag was given to
 4880     pcretest, it used to include the size of the study block as part of its
 4881     output. Unfortunately, the structure contains a field that has a different
 4882     size on different hardware architectures. This meant that the tests that
 4883     showed this size failed. As the block is currently always of a fixed size,
 4884     this information isn't actually particularly useful in pcretest output, so
 4885     I have just removed it.
 4887  8. Three pre-processor statements accidentally did not start in column 1.
 4888     Sadly, there are *still* compilers around that complain, even though
 4889     standard C has not required this for well over a decade. Sigh.
 4891  9. In pcretest, the code for checking callouts passed small integers in the
 4892     callout_data field, which is a void * field. However, some picky compilers
 4893     complained about the casts involved for this on 64-bit systems. Now
 4894     pcretest passes the address of the small integer instead, which should get
 4895     rid of the warnings.
 4897 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
 4898     both compile and run time, and gives an error if an invalid UTF-8 sequence
 4899     is found. There is a option for disabling this check in cases where the
 4900     string is known to be correct and/or the maximum performance is wanted.
 4902 11. In response to a bug report, I changed one line in Makefile.in from
 4904         -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
 4905     to
 4906         -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
 4908     to look similar to other lines, but I have no way of telling whether this
 4909     is the right thing to do, as I do not use Windows. No doubt I'll get told
 4910     if it's wrong...
 4913 Version 4.3 21-May-03
 4914 ---------------------
 4916 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
 4917    Makefile.
 4919 2. Some refactoring to improve the quality of the code:
 4921    (i)   The utf8_table... variables are now declared "const".
 4923    (ii)  The code for \cx, which used the "case flipping" table to upper case
 4924          lower case letters, now just substracts 32. This is ASCII-specific,
 4925          but the whole concept of \cx is ASCII-specific, so it seems
 4926          reasonable.
 4928    (iii) PCRE was using its character types table to recognize decimal and
 4929          hexadecimal digits in the pattern. This is silly, because it handles
 4930          only 0-9, a-f, and A-F, but the character types table is locale-
 4931          specific, which means strange things might happen. A private
 4932          table is now used for this - though it costs 256 bytes, a table is
 4933          much faster than multiple explicit tests. Of course, the standard
 4934          character types table is still used for matching digits in subject
 4935          strings against \d.
 4937    (iv)  Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
 4938          ending in _t are). So I've renamed it as ESC_tee.
 4940 3. The first argument for regexec() in the POSIX wrapper should have been
 4941    defined as "const".
 4943 4. Changed pcretest to use malloc() for its buffers so that they can be
 4944    Electric Fenced for debugging.
 4946 5. There were several places in the code where, in UTF-8 mode, PCRE would try
 4947    to read one or more bytes before the start of the subject string. Often this
 4948    had no effect on PCRE's behaviour, but in some circumstances it could
 4949    provoke a segmentation fault.
 4951 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
 4952    to try to read one or more bytes before the start of the subject string.
 4954 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
 4955    UTF-8 support could misbehave in various ways if the subject string
 4956    contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
 4957    area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
 4958    back over UTF-8 characters.)
 4961 Version 4.2 14-Apr-03
 4962 ---------------------
 4964 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
 4966 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
 4967      [ON_WINDOWS]: new variable, "#" on non-Windows platforms
 4968      [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
 4969      [WIN_PREFIX]: new variable, "cyg" for Cygwin
 4970      * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
 4971        and BUILD_EXEEXT
 4972      Note: automatic setting of the BUILD variables is not yet working
 4973      set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
 4974        compile-time but not at link-time
 4975      [LINK]: use for linking executables only
 4976      make different versions for Windows and non-Windows
 4977      [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
 4978        libraries
 4979      [LINK_FOR_BUILD]: new variable
 4980      [OBJEXT]: use throughout
 4981      [EXEEXT]: use throughout
 4982      <winshared>: new target
 4983      <wininstall>: new target
 4984      <dftables.o>: use native compiler
 4985      <dftables>: use native linker
 4986      <install>: handle Windows platform correctly
 4987      <clean>: ditto
 4988      <check>: ditto
 4989      copy DLL to top builddir before testing
 4991    As part of these changes, -no-undefined was removed again. This was reported
 4992    to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
 4993    in any case.
 4995 3. Some tidies to get rid of compiler warnings:
 4997    . In the match_data structure, match_limit was an unsigned long int, whereas
 4998      match_call_count was an int. I've made them both unsigned long ints.
 5000    . In pcretest the fact that a const uschar * doesn't automatically cast to
 5001      a void * provoked a warning.
 5003    . Turning on some more compiler warnings threw up some "shadow" variables
 5004      and a few more missing casts.
 5006 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
 5007    option, a class that contained a single character with a value between 128
 5008    and 255 (e.g. /[\xFF]/) caused PCRE to crash.
 5010 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
 5011    option, a class that contained several characters, but with at least one
 5012    whose value was between 128 and 255 caused PCRE to crash.
 5015 Version 4.1 12-Mar-03
 5016 ---------------------
 5018 1. Compiling with gcc -pedantic found a couple of places where casts were
 5019 needed, and a string in dftables.c that was longer than standard compilers are
 5020 required to support.
 5022 2. Compiling with Sun's compiler found a few more places where the code could
 5023 be tidied up in order to avoid warnings.
 5025 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
 5026 first of these names is deprecated in the latest Autoconf in favour of the name
 5027 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
 5028 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
 5029 analogy I have changed it to CFLAGS_FOR_BUILD.
 5031 4. Added -no-undefined to the linking command in the Makefile, because this is
 5032 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
 5033 linking step for the pcreposix library.
 5035 5. PCRE was failing to diagnose the case of two named groups with the same
 5036 name.
 5038 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
 5039 literal character that is needed in the subject for a match, and scans along to
 5040 ensure that it is present before embarking on the full matching process. This
 5041 saves time in cases of nested unlimited repeats that are never going to match.
 5042 Problem: the scan can take a lot of time if the subject is very long (e.g.
 5043 megabytes), thus penalizing straightforward matches. It is now done only if the
 5044 amount of subject to be scanned is less than 1000 bytes.
 5046 7. A lesser problem with the same optimization is that it was recording the
 5047 first character of an anchored pattern as "needed", thus provoking a search
 5048 right along the subject, even when the first match of the pattern was going to
 5049 fail. The "needed" character is now not set for anchored patterns, unless it
 5050 follows something in the pattern that is of non-fixed length. Thus, it still
 5051 fulfils its original purpose of finding quick non-matches in cases of nested
 5052 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
 5055 Version 4.0 17-Feb-03
 5056 ---------------------
 5058 1. If a comment in an extended regex that started immediately after a meta-item
 5059 extended to the end of string, PCRE compiled incorrect data. This could lead to
 5060 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
 5062 2. Moved to autoconf 2.53 and libtool 1.4.2.
 5064 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
 5065 the special perltest8 script is no longer needed - all the tests can be run
 5066 from a single perltest script.
 5068 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
 5069 by \s. It has now been removed in PCRE. This means it isn't recognized as
 5070 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
 5071 class [:space:] *does* include VT, thereby creating a mess.
 5073 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
 5074 space and tab.
 5076 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
 5077 its new features into the main test script, reducing the number of scripts.
 5079 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
 5080 were backward compatible, and made the (?i) apply to the whole pattern, as if
 5081 /i were given. Now it behaves more logically, and applies the option setting
 5082 only to what follows. PCRE has been changed to follow suit. However, if it
 5083 finds options settings right at the start of the pattern, it extracts them into
 5084 the global options, as before. Thus, they show up in the info data.
 5086 8. Added support for the \Q...\E escape sequence. Characters in between are
 5087 treated as literals. This is slightly different from Perl in that $ and @ are
 5088 also handled as literals inside the quotes. In Perl, they will cause variable
 5089 interpolation. Note the following examples:
 5091     Pattern            PCRE matches      Perl matches
 5093     \Qabc$xyz\E        abc$xyz           abc followed by the contents of $xyz
 5094     \Qabc\$xyz\E       abc\$xyz          abc\$xyz
 5095     \Qabc\E\$\Qxyz\E   abc$xyz           abc$xyz
 5097 For compatibility with Perl, \Q...\E sequences are recognized inside character
 5098 classes as well as outside them.
 5100 9. Re-organized 3 code statements in pcretest to avoid "overflow in
 5101 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
 5102 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
 5103 signed/unsigned warnings.
 5105 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
 5106 option for pcretest, so I've replaced it by a simple function that does just
 5107 that job.
 5109 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
 5110 "pcregrep -".
 5112 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
 5113 Java package. This provides some syntactic sugar for simple cases of what my
 5114 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
 5115 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
 5116 item, you can use this simplified notation. Note that only makes sense with
 5117 greedy quantifiers. Consequently, the use of the possessive quantifier forces
 5118 greediness, whatever the setting of the PCRE_UNGREEDY option.
 5120 13. A change of greediness default within a pattern was not taking effect at
 5121 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
 5122 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
 5123 was abstracted outside.
 5125 14. PCRE now supports the \G assertion. It is true when the current matching
 5126 position is at the start point of the match. This differs from \A when the
 5127 starting offset is non-zero. Used with the /g option of pcretest (or similar
 5128 code), it works in the same way as it does for Perl's /g option. If all
 5129 alternatives of a regex begin with \G, the expression is anchored to the start
 5130 match position, and the "anchored" flag is set in the compiled expression.
 5132 15. Some bugs concerning the handling of certain option changes within patterns
 5133 have been fixed. These applied to options other than (?ims). For example,
 5134 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
 5135 been the other way round. Some of this was related to change 7 above.
 5137 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
 5138 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
 5139 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
 5140 POSIX classes only within a class (e.g. /[[:alpha:]]/).
 5142 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
 5143 mode. Unlike ".", it always matches newline, whatever the setting of
 5144 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
 5145 assertions. Perl allows it, but it doesn't (in general) work because it can't
 5146 calculate the length of the lookbehind. At least, that's the case for Perl
 5147 5.8.0 - I've been told they are going to document that it doesn't work in
 5148 future.
 5150 18. Added an error diagnosis for escapes that PCRE does not support: these are
 5151 \L, \l, \N, \P, \p, \U, \u, and \X.
 5153 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
 5154 reading past the end of the pattern in cases such as /[abcd/.
 5156 20. PCRE was getting more memory than necessary for patterns with classes that
 5157 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
 5159 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
 5160 compiling PCRE for use with Virtual Pascal.
 5162 22. Small fix to the Makefile to make it work properly if the build is done
 5163 outside the source tree.
 5165 23. Added a new extension: a condition to go with recursion. If a conditional
 5166 subpattern starts with (?(R) the "true" branch is used if recursion has
 5167 happened, whereas the "false" branch is used only at the top level.
 5169 24. When there was a very long string of literal characters (over 255 bytes
 5170 without UTF support, over 250 bytes with UTF support), the computation of how
 5171 much memory was required could be incorrect, leading to segfaults or other
 5172 strange effects.
 5174 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
 5175 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
 5176 there was a subsequent back reference to those brackets. This meant that, for
 5177 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
 5178 possible to check for precisely this case. All we can do is abandon the
 5179 optimization if .* occurs inside capturing brackets when there are any back
 5180 references whatsoever. (See below for a better fix that came later.)
 5182 26. The handling of the optimization for finding the first character of a
 5183 non-anchored pattern, and for finding a character that is required later in the
 5184 match were failing in some cases. This didn't break the matching; it just
 5185 failed to optimize when it could. The way this is done has been re-implemented.
 5187 27. Fixed typo in error message for invalid (?R item (it said "(?p").
 5189 28. Added a new feature that provides some of the functionality that Perl
 5190 provides with (?{...}). The facility is termed a "callout". The way it is done
 5191 in PCRE is for the caller to provide an optional function, by setting
 5192 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
 5193 global variable. By default it is unset, which disables all calling out. To get
 5194 the function called, the regex must include (?C) at appropriate points. This
 5195 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
 5196 This provides a means of identifying different callout points. When PCRE
 5197 reaches such a point in the regex, if pcre_callout has been set, the external
 5198 function is called. It is provided with data in a structure called
 5199 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
 5200 matching continues; if it returns a non-zero value, the match at the current
 5201 point fails. However, backtracking will occur if possible. [This was changed
 5202 later and other features added - see item 49 below.]
 5204 29. pcretest is upgraded to test the callout functionality. It provides a
 5205 callout function that displays information. By default, it shows the start of
 5206 the match and the current position in the text. There are some new data escapes
 5207 to vary what happens:
 5209     \C+         in addition, show current contents of captured substrings
 5210     \C-         do not supply a callout function
 5211     \C!n        return 1 when callout number n is reached
 5212     \C!n!m      return 1 when callout number n is reached for the mth time
 5214 30. If pcregrep was called with the -l option and just a single file name, it
 5215 output "<stdin>" if a match was found, instead of the file name.
 5217 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
 5218 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
 5219 pcre_exec(). This saves a malloc/free per call. The default value of
 5220 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
 5221 when configuring.
 5223 32. The default maximum size of a compiled pattern is 64K. There have been a
 5224 few cases of people hitting this limit. The code now uses macros to handle the
 5225 storing of links as offsets within the compiled pattern. It defaults to 2-byte
 5226 links, but this can be changed to 3 or 4 bytes by --with-link-size when
 5227 configuring. Tests 2 and 5 work only with 2-byte links because they output
 5228 debugging information about compiled patterns.
 5230 33. Internal code re-arrangements:
 5232 (a) Moved the debugging function for printing out a compiled regex into
 5233     its own source file (printint.c) and used #include to pull it into
 5234     pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
 5235     separate copies.
 5237 (b) Defined the list of op-code names for debugging as a macro in
 5238     internal.h so that it is next to the definition of the opcodes.
 5240 (c) Defined a table of op-code lengths for simpler skipping along compiled
 5241     code. This is again a macro in internal.h so that it is next to the
 5242     definition of the opcodes.
 5244 34. Added support for recursive calls to individual subpatterns, along the
 5245 lines of Robin Houston's patch (but implemented somewhat differently).
 5247 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
 5248 allow it to read and process whole directories in Win32. This code was
 5249 contributed by Lionel Fourquaux; it has not been tested by me.
 5251 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
 5252 used to name a group. Names consist of alphanumerics and underscores, and must
 5253 be unique. Back references use the syntax (?P=name) and recursive calls use
 5254 (?P>name) which is a PCRE extension to the Python extension. Groups still have
 5255 numbers. The function pcre_fullinfo() can be used after compilation to extract
 5256 a name/number map. There are three relevant calls:
 5258   PCRE_INFO_NAMEENTRYSIZE        yields the size of each entry in the map
 5259   PCRE_INFO_NAMECOUNT            yields the number of entries
 5260   PCRE_INFO_NAMETABLE            yields a pointer to the map.
 5262 The map is a vector of fixed-size entries. The size of each entry depends on
 5263 the length of the longest name used. The first two bytes of each entry are the
 5264 group number, most significant byte first. There follows the corresponding
 5265 name, zero terminated. The names are in alphabetical order.
 5267 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
 5268 case instead of 255. Making it the same both with and without UTF-8 support
 5269 means that the same test output works with both.
 5271 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
 5272 calling malloc() with a zero argument.
 5274 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
 5275 optimization. I've improved things by keeping a bitmap of backreferences with
 5276 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
 5277 fact referenced, the optimization can be applied. It is unlikely that a
 5278 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
 5279 the match to follow \n) will appear inside brackets with a number greater than
 5280 31, but if it does, any back reference > 31 suppresses the optimization.
 5282 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
 5283 of disabling numbered capturing parentheses. Any opening parenthesis that is
 5284 not followed by ? behaves as if it were followed by ?: but named parentheses
 5285 can still be used for capturing (and they will acquire numbers in the usual
 5286 way).
 5288 41. Redesigned the return codes from the match() function into yes/no/error so
 5289 that errors can be passed back from deep inside the nested calls. A malloc
 5290 failure while inside a recursive subpattern call now causes the
 5291 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
 5293 42. It is now possible to set a limit on the number of times the match()
 5294 function is called in a call to pcre_exec(). This facility makes it possible to
 5295 limit the amount of recursion and backtracking, though not in a directly
 5296 obvious way, because the match() function is used in a number of different
 5297 circumstances. The count starts from zero for each position in the subject
 5298 string (for non-anchored patterns). The default limit is, for compatibility, a
 5299 large number, namely 10 000 000. You can change this in two ways:
 5301 (a) When configuring PCRE before making, you can use --with-match-limit=n
 5302     to set a default value for the compiled library.
 5304 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
 5305     a different value is set. See 45 below.
 5307 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
 5309 43. Added a new function pcre_config(int, void *) to enable run-time extraction
 5310 of things that can be changed at compile time. The first argument specifies
 5311 what is wanted and the second points to where the information is to be placed.
 5312 The current list of available information is:
 5316 The output is an integer that is set to one if UTF-8 support is available;
 5317 otherwise it is set to zero.
 5321 The output is an integer that it set to the value of the code that is used for
 5322 newline. It is either LF (10) or CR (13).
 5326 The output is an integer that contains the number of bytes used for internal
 5327 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
 5331 The output is an integer that contains the threshold above which the POSIX
 5332 interface uses malloc() for output vectors. See item 31 above.
 5336 The output is an unsigned integer that contains the default limit of the number
 5337 of match() calls in a pcre_exec() execution. See 42 above.
 5339 44. pcretest has been upgraded by the addition of the -C option. This causes it
 5340 to extract all the available output from the new pcre_config() function, and to
 5341 output it. The program then exits immediately.
 5343 45. A need has arisen to pass over additional data with calls to pcre_exec() in
 5344 order to support additional features. One way would have been to define
 5345 pcre_exec2() (for example) with extra arguments, but this would not have been
 5346 extensible, and would also have required all calls to the original function to
 5347 be mapped to the new one. Instead, I have chosen to extend the mechanism that
 5348 is used for passing in "extra" data from pcre_study().
 5350 The pcre_extra structure is now exposed and defined in pcre.h. It currently
 5351 contains the following fields:
 5353   flags         a bitmap indicating which of the following fields are set
 5354   study_data    opaque data from pcre_study()
 5355   match_limit   a way of specifying a limit on match() calls for a specific
 5356                   call to pcre_exec()
 5357   callout_data  data for callouts (see 49 below)
 5359 The flag bits are also defined in pcre.h, and are
 5365 The pcre_study() function now returns one of these new pcre_extra blocks, with
 5366 the actual study data pointed to by the study_data field, and the
 5367 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
 5368 before. That is, this change is entirely upwards-compatible and requires no
 5369 change to existing code.
 5371 If you want to pass in additional data to pcre_exec(), you can either place it
 5372 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
 5373 block.
 5375 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
 5376 data string contains the escape sequence \M, pcretest calls pcre_exec() several
 5377 times with different match limits, until it finds the minimum value needed for
 5378 pcre_exec() to complete. The value is then output. This can be instructive; for
 5379 most simple matches the number is quite small, but for pathological cases it
 5380 gets very large very quickly.
 5382 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
 5383 returns the size of the data block pointed to by the study_data field in a
 5384 pcre_extra block, that is, the value that was passed as the argument to
 5385 pcre_malloc() when PCRE was getting memory in which to place the information
 5386 created by pcre_study(). The fourth argument should point to a size_t variable.
 5387 pcretest has been extended so that this information is shown after a successful
 5388 pcre_study() call when information about the compiled regex is being displayed.
 5390 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
 5391 because what follows is always an absolute path. (Later: it turns out that this
 5392 is more than cosmetic for MinGW, because it doesn't like empty path
 5393 components.)
 5395 49. Some changes have been made to the callout feature (see 28 above):
 5397 (i)  A callout function now has three choices for what it returns:
 5399        0  =>  success, carry on matching
 5400      > 0  =>  failure at this point, but backtrack if possible
 5401      < 0  =>  serious error, return this value from pcre_exec()
 5403      Negative values should normally be chosen from the set of PCRE_ERROR_xxx
 5404      values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
 5405      "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
 5406      use by callout functions. It will never be used by PCRE itself.
 5408 (ii) The pcre_extra structure (see 45 above) has a void * field called
 5409      callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
 5410      pcre_callout_block structure has a field of the same name. The contents of
 5411      the field passed in the pcre_extra structure are passed to the callout
 5412      function in the corresponding field in the callout block. This makes it
 5413      easier to use the same callout-containing regex from multiple threads. For
 5414      testing, the pcretest program has a new data escape
 5416        \C*n        pass the number n (may be negative) as callout_data
 5418      If the callout function in pcretest receives a non-zero value as
 5419      callout_data, it returns that value.
 5421 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
 5422 there were some redundant $(CFLAGS) in commands that are now specified as
 5423 $(LINK), which already includes $(CFLAGS).
 5425 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
 5426 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
 5427 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
 5428 one-byte characters throughout. Note that case-insensitive matching applies
 5429 only to characters whose values are less than 256. PCRE doesn't support the
 5430 notion of cases for higher-valued characters.
 5432 (i)   A character class whose characters are all within 0-255 is handled as
 5433       a bit map, and the map is inverted for negative classes. Previously, a
 5434       character > 255 always failed to match such a class; however it should
 5435       match if the class was a negative one (e.g. [^ab]). This has been fixed.
 5437 (ii)  A negated character class with a single character < 255 is coded as
 5438       "not this character" (OP_NOT). This wasn't working properly when the test
 5439       character was multibyte, either singly or repeated.
 5441 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
 5442       mode, for example: \x{100}{2,3}.
 5444 (iv)  The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
 5445       singly or repeated) now correctly test multibyte characters. However,
 5446       PCRE doesn't recognize any characters with values greater than 255 as
 5447       digits, spaces, or word characters. Such characters always match \D, \S,
 5448       and \W, and never match \d, \s, or \w.
 5450 (v)   Classes may now contain characters and character ranges with values
 5451       greater than 255. For example: [ab\x{100}-\x{400}].
 5453 (vi)  pcregrep now has a --utf-8 option (synonym -u) which makes it call
 5454       PCRE in UTF-8 mode.
 5456 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
 5457 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
 5458 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
 5459 value.)
 5461 53. The single man page has become too large. I have therefore split it up into
 5462 a number of separate man pages. These also give rise to individual HTML pages;
 5463 these are now put in a separate directory, and there is an index.html page that
 5464 lists them all. Some hyperlinking between the pages has been installed.
 5466 54. Added convenience functions for handling named capturing parentheses.
 5468 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
 5469 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
 5470 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
 5471 are faulted.
 5473 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
 5474 calling configure. These values are used when compiling the dftables.c program
 5475 which is run to generate the source of the default character tables. They
 5476 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
 5477 you will need to set these values.
 5479 57. Updated the building process for Windows DLL, as provided by Fred Cox.
 5482 Version 3.9 02-Jan-02
 5483 ---------------------
 5485 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
 5487 2. If --disable-static was given, the building process failed when trying to
 5488 build pcretest and pcregrep. (For some reason it was using libtool to compile
 5489 them, which is not right, as they aren't part of the library.)
 5492 Version 3.8 18-Dec-01
 5493 ---------------------
 5495 1. The experimental UTF-8 code was completely screwed up. It was packing the
 5496 bytes in the wrong order. How dumb can you get?
 5499 Version 3.7 29-Oct-01
 5500 ---------------------
 5502 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
 5503 This caused pcretest, when used on the test data, to segfault. Unfortunately,
 5504 this didn't happen under Solaris 8, where I normally test things.
 5506 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
 5507 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
 5508 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
 5509 this fix an hour or so after the initial 3.7 release.)
 5512 Version 3.6 23-Oct-01
 5513 ---------------------
 5515 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
 5516 offsets passed as NULL with zero offset count.
 5518 2. The config.guess and config.sub files had not been updated when I moved to
 5519 the latest autoconf.
 5522 Version 3.5 15-Aug-01
 5523 ---------------------
 5525 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
 5526 had been forgotten.
 5528 2. By using declared but undefined structures, we can avoid using "void"
 5529 definitions in pcre.h while keeping the internal definitions of the structures
 5530 private.
 5532 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
 5533 user point of view, this means that both static and shared libraries are built
 5534 by default, but this can be individually controlled. More of the work of
 5535 handling this static/shared cases is now inside libtool instead of PCRE's make
 5536 file.
 5538 4. The pcretest utility is now installed along with pcregrep because it is
 5539 useful for users (to test regexs) and by doing this, it automatically gets
 5540 relinked by libtool. The documentation has been turned into a man page, so
 5541 there are now .1, .txt, and .html versions in /doc.
 5543 5. Upgrades to pcregrep:
 5544    (i)   Added long-form option names like gnu grep.
 5545    (ii)  Added --help to list all options with an explanatory phrase.
 5546    (iii) Added -r, --recursive to recurse into sub-directories.
 5547    (iv)  Added -f, --file to read patterns from a file.
 5549 6. pcre_exec() was referring to its "code" argument before testing that
 5550 argument for NULL (and giving an error if it was NULL).
 5552 7. Upgraded Makefile.in to allow for compiling in a different directory from
 5553 the source directory.
 5555 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
 5556 options bits, the pointer it was passed was to an int instead of to an unsigned
 5557 long int. This mattered only on 64-bit systems.
 5559 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
 5560 generated) instead of pcre.in, which it its source. Also made the same change
 5561 in several of the .c files.
 5563 10. A new release of gcc defines printf() as a macro, which broke pcretest
 5564 because it had an ifdef in the middle of a string argument for printf(). Fixed
 5565 by using separate calls to printf().
 5567 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
 5568 script, to force use of CR or LF instead of \n in the source. On non-Unix
 5569 systems, the value can be set in config.h.
 5571 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
 5572 absolute limit. Changed the text of the error message to make this clear, and
 5573 likewise updated the man page.
 5575 13. The limit of 99 on the number of capturing subpatterns has been removed.
 5576 The new limit is 65535, which I hope will not be a "real" limit.
 5579 Version 3.4 22-Aug-00
 5580 ---------------------
 5582 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
 5584 2. Diagnose condition (?(0) as an error instead of crashing on matching.
 5587 Version 3.3 01-Aug-00
 5588 ---------------------
 5590 1. If an octal character was given, but the value was greater than \377, it
 5591 was not getting masked to the least significant bits, as documented. This could
 5592 lead to crashes in some systems.
 5594 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
 5595 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
 5597 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
 5598 These just pass their arguments on to (pcre_free)(), but they are provided
 5599 because some uses of PCRE bind it to non-C systems that can call its functions,
 5600 but cannot call free() or pcre_free() directly.
 5602 4. Add "make test" as a synonym for "make check". Corrected some comments in
 5603 the Makefile.
 5605 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
 5606 Makefile.
 5608 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
 5609 command called pgrep for grepping around the active processes.
 5611 7. Added the beginnings of support for UTF-8 character strings.
 5613 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
 5614 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
 5615 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
 5616 out for the ar command.)
 5619 Version 3.2 12-May-00
 5620 ---------------------
 5622 This is purely a bug fixing release.
 5624 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
 5625 of ZA. This was just one example of several cases that could provoke this bug,
 5626 which was introduced by change 9 of version 2.00. The code for breaking
 5627 infinite loops after an iteration that matches an empty string was't working
 5628 correctly.
 5630 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
 5631 when matched against abbab (for example). After matching an empty string, it
 5632 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
 5633 caused it to match further down the string than it should.
 5635 3. The code contained an inclusion of sys/types.h. It isn't clear why this
 5636 was there because it doesn't seem to be needed, and it causes trouble on some
 5637 systems, as it is not a Standard C header. It has been removed.
 5639 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
 5640 were reported on the Macintosh. The changes were from
 5642   while ((c = *(++ptr)) != 0 && c != '\n');
 5643 to
 5644   while ((c = *(++ptr)) != 0 && c != '\n') ;
 5646 Totally extraordinary, but if that's what it takes...
 5648 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
 5649 available. Added HAVE_BCOPY and an autoconf test for it; if neither
 5650 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
 5651 assumes the way PCRE uses memmove() (always moving upwards).
 5653 6. PCRE is being used in one environment where strchr() is not available. There
 5654 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
 5655 faster code anyway.
 5658 Version 3.1 09-Feb-00
 5659 ---------------------
 5661 The only change in this release is the fixing of some bugs in Makefile.in for
 5662 the "install" target:
 5664 (1) It was failing to install pcreposix.h.
 5666 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
 5669 Version 3.0 01-Feb-00
 5670 ---------------------
 5672 1. Add support for the /+ modifier to perltest (to output $` like it does in
 5673 pcretest).
 5675 2. Add support for the /g modifier to perltest.
 5677 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
 5678 matches null strings.
 5680 4. Fix perltest so that it doesn't do unwanted things when fed an empty
 5681 pattern. Perl treats empty patterns specially - it reuses the most recent
 5682 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
 5683 effect.
 5685 5. The POSIX interface was broken in that it was just handing over the POSIX
 5686 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
 5687 required a bigger vector, with some working space on the end. This means that
 5688 the POSIX wrapper now has to get and free some memory, and copy the results.
 5690 6. Added some simple autoconf support, placing the test data and the
 5691 documentation in separate directories, re-organizing some of the
 5692 information files, and making it build pcre-config (a GNU standard). Also added
 5693 libtool support for building PCRE as a shared library, which is now the
 5694 default.
 5696 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
 5697 09 are not valid octal constants. Single digits will be used for minor values
 5698 less than 10.
 5700 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
 5701 existing programs that set these in the POSIX interface can use PCRE without
 5702 modification.
 5704 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
 5705 return all that pcre_info() returns, plus additional data. The pcre_info()
 5706 function is retained for compatibility, but is considered to be obsolete.
 5708 10. Added experimental recursion feature (?R) to handle one common case that
 5709 Perl 5.6 will be able to do with (?p{...}).
 5711 11. Added support for POSIX character classes like [:alpha:], which Perl is
 5712 adopting.
 5715 Version 2.08 31-Aug-99
 5716 ----------------------
 5718 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
 5719 trying to match at the startoffset position, but instead was moving forward to
 5720 the next newline as if a previous match had failed.
 5722 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
 5723 and could get into a loop if a null string was matched other than at the start
 5724 of the subject.
 5726 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
 5727 be distinguished at compile time, and for completeness also added PCRE_DATE.
 5729 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
 5730 in GnuWin32 environments.
 5733 Version 2.07 29-Jul-99
 5734 ----------------------
 5736 1. The documentation is now supplied in plain text form and HTML as well as in
 5737 the form of man page sources.
 5739 2. C++ compilers don't like assigning (void *) values to other pointer types.
 5740 In particular this affects malloc(). Although there is no problem in Standard
 5741 C, I've put in casts to keep C++ compilers happy.
 5743 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
 5744 should be (const char *).
 5746 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
 5747 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
 5748 However, I haven't made this a standard facility. The documentation doesn't
 5749 mention it, and the Makefile doesn't support it.
 5751 5. The Makefile now contains an "install" target, with editable destinations at
 5752 the top of the file. The pcretest program is not installed.
 5754 6. pgrep -V now gives the PCRE version number and date.
 5756 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
 5757 causing the entire string to be ignored, instead of just the last character.
 5759 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
 5760 non-matching string, it can take a very, very long time, even for strings of
 5761 quite modest length, because of the nested recursion. PCRE now does better in
 5762 some of these cases. It does this by remembering the last required literal
 5763 character in the pattern, and pre-searching the subject to ensure it is present
 5764 before running the real match. In other words, it applies a heuristic to detect
 5765 some types of certain failure quickly, and in the above example, if presented
 5766 with a string that has no trailing " it gives "no match" very quickly.
 5768 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
 5769 other alternatives are tried instead.
 5772 Version 2.06 09-Jun-99
 5773 ----------------------
 5775 1. Change pcretest's output for amount of store used to show just the code
 5776 space, because the remainder (the data block) varies in size between 32-bit and
 5777 64-bit systems.
 5779 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
 5780 start matching at. This allows lookbehinds to work when searching for multiple
 5781 occurrences in a string.
 5783 3. Added additional options to pcretest for testing multiple occurrences:
 5785    /+   outputs the rest of the string that follows a match
 5786    /g   loops for multiple occurrences, using the new startoffset argument
 5787    /G   loops for multiple occurrences by passing an incremented pointer
 5789 4. PCRE wasn't doing the "first character" optimization for patterns starting
 5790 with \b or \B, though it was doing it for other lookbehind assertions. That is,
 5791 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
 5792 the letter 'x'. On long subject strings, this gives a significant speed-up.
 5795 Version 2.05 21-Apr-99
 5796 ----------------------
 5798 1. Changed the type of magic_number from int to long int so that it works
 5799 properly on 16-bit systems.
 5801 2. Fixed a bug which caused patterns starting with .* not to work correctly
 5802 when the subject string contained newline characters. PCRE was assuming
 5803 anchoring for such patterns in all cases, which is not correct because .* will
 5804 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
 5805 DOTALL is set at top level; otherwise it knows that patterns starting with .*
 5806 must be retried after every newline in the subject.
 5809 Version 2.04 18-Feb-99
 5810 ----------------------
 5812 1. For parenthesized subpatterns with repeats whose minimum was zero, the
 5813 computation of the store needed to hold the pattern was incorrect (too large).
 5814 If such patterns were nested a few deep, this could multiply and become a real
 5815 problem.
 5817 2. Added /M option to pcretest to show the memory requirement of a specific
 5818 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
 5820 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
 5821 compiled in such a way that the backtracking after subsequent failure was
 5822 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
 5823 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
 5826 Version 2.03 02-Feb-99
 5827 ----------------------
 5829 1. Fixed typo and small mistake in man page.
 5831 2. Added 4th condition (GPL supersedes if conflict) and created separate
 5832 LICENCE file containing the conditions.
 5834 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
 5835 Perl, that is the internal \ allows the delimiter to be included in the
 5836 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
 5837 the final delimiter, add \ to the end of the pattern (to test the error).
 5839 4. Added the convenience functions for extracting substrings after a successful
 5840 match. Updated pcretest to make it able to test these functions.
 5843 Version 2.02 14-Jan-99
 5844 ----------------------
 5846 1. Initialized the working variables associated with each extraction so that
 5847 their saving and restoring doesn't refer to uninitialized store.
 5849 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
 5850 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
 5851 fix the problem.
 5853 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
 5854 calls, and wasn't printing the correct value for compiling calls. Increased the
 5855 default value of LOOPREPEAT, and the number of significant figures in the
 5856 times.
 5858 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
 5860 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
 5861 a building problem on Windows NT with a FAT file system.
 5864 Version 2.01 21-Oct-98
 5865 ----------------------
 5867 1. Changed the API for pcre_compile() to allow for the provision of a pointer
 5868 to character tables built by pcre_maketables() in the current locale. If NULL
 5869 is passed, the default tables are used.
 5872 Version 2.00 24-Sep-98
 5873 ----------------------
 5875 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
 5876 it any more.
 5878 2. Allow quantification of (?>) groups, and make it work correctly.
 5880 3. The first character computation wasn't working for (?>) groups.
 5882 4. Correct the implementation of \Z (it is permitted to match on the \n at the
 5883 end of the subject) and add 5.005's \z, which really does match only at the
 5884 very end of the subject.
 5886 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
 5888 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
 5889 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
 5890 localized options. All options to pcre_study() were also removed.
 5892 7. Add other new features from 5.005:
 5894    $(?<=           positive lookbehind
 5895    $(?<!           negative lookbehind
 5896    (?imsx-imsx)    added the unsetting capability
 5897                    such a setting is global if at outer level; local otherwise
 5898    (?imsx-imsx:)   non-capturing groups with option setting
 5899    (?(cond)re|re)  conditional pattern matching
 5901    A backreference to itself in a repeated group matches the previous
 5902    captured string.
 5904 8. General tidying up of studying (both automatic and via "study")
 5905 consequential on the addition of new assertions.
 5907 9. As in 5.005, unlimited repeated groups that could match an empty substring
 5908 are no longer faulted at compile time. Instead, the loop is forcibly broken at
 5909 runtime if any iteration does actually match an empty substring.
 5911 10. Include the RunTest script in the distribution.
 5913 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
 5914 discrepancies, some of which were old and were also with respect to 5.004. They
 5915 have now been fixed.
 5918 Version 1.09 28-Apr-98
 5919 ----------------------
 5921 1. A negated single character class followed by a quantifier with a minimum
 5922 value of one (e.g.  [^x]{1,6}  ) was not compiled correctly. This could lead to
 5923 program crashes, or just wrong answers. This did not apply to negated classes
 5924 containing more than one character, or to minima other than one.
 5927 Version 1.08 27-Mar-98
 5928 ----------------------
 5930 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
 5932 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
 5933 latter must appear before anything that relies on it in the pattern.
 5936 Version 1.07 16-Feb-98
 5937 ----------------------
 5939 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
 5940 repeat of a potentially empty string).
 5943 Version 1.06 23-Jan-98
 5944 ----------------------
 5946 1. Added Markus Oberhumer's little patches for C++.
 5948 2. Literal strings longer than 255 characters were broken.
 5951 Version 1.05 23-Dec-97
 5952 ----------------------
 5954 1. Negated character classes containing more than one character were failing if
 5955 PCRE_CASELESS was set at run time.
 5958 Version 1.04 19-Dec-97
 5959 ----------------------
 5961 1. Corrected the man page, where some "const" qualifiers had been omitted.
 5963 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
 5964 input syntax.
 5966 3. Fixed memory leak which occurred when a regex with back references was
 5967 matched with an offsets vector that wasn't big enough. The temporary memory
 5968 that is used in this case wasn't being freed if the match failed.
 5970 4. Tidied pcretest to ensure it frees memory that it gets.
 5972 5. Temporary memory was being obtained in the case where the passed offsets
 5973 vector was exactly big enough.
 5975 6. Corrected definition of offsetof() from change 5 below.
 5977 7. I had screwed up change 6 below and broken the rules for the use of
 5978 setjmp(). Now fixed.
 5981 Version 1.03 18-Dec-97
 5982 ----------------------
 5984 1. A erroneous regex with a missing opening parenthesis was correctly
 5985 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
 5986 on some systems.
 5988 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
 5989 it was reported that one broken compiler failed on the former because "code" is
 5990 also an independent variable.
 5992 3. The erroneous regex a[]b caused an array overrun reference.
 5994 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
 5995 fail on data ending with that character. (It was going on too far, and checking
 5996 the next character, typically a binary zero.) This was specific to the
 5997 optimized code for single-character negative classes.
 5999 5. Added a contributed patch from the TIN world which does the following:
 6001   + Add an undef for memmove, in case the the system defines a macro for it.
 6003   + Add a definition of offsetof(), in case there isn't one. (I don't know
 6004     the reason behind this - offsetof() is part of the ANSI standard - but
 6005     it does no harm).
 6007   + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
 6008     most of the places where whitespace preceded '#'. I have given up and
 6009     allowed the remaining 2 cases to be at the margin.
 6011   + Rename some variables in pcre to eliminate shadowing. This seems very
 6012     pedantic, but does no harm, of course.
 6014 6. Moved the call to setjmp() into its own function, to get rid of warnings
 6015 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
 6017 7. Constructs such as \d{8,} were compiling into the equivalent of
 6018 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
 6019 outcome, but in this particular case used more store than had been allocated,
 6020 which caused the bug to be discovered because it threw up an internal error.
 6022 8. The debugging code in both pcre and pcretest for outputting the compiled
 6023 form of a regex was going wrong in the case of back references followed by
 6024 curly-bracketed repeats.
 6027 Version 1.02 12-Dec-97
 6028 ----------------------
 6030 1. Typos in pcre.3 and comments in the source fixed.
 6032 2. Applied a contributed patch to get rid of places where it used to remove
 6033 'const' from variables, and fixed some signed/unsigned and uninitialized
 6034 variable warnings.
 6036 3. Added the "runtest" target to Makefile.
 6038 4. Set default compiler flag to -O2 rather than just -O.
 6041 Version 1.01 19-Nov-97
 6042 ----------------------
 6044 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
 6045 like /([ab]*)*/, that is, for classes with more than one character in them.
 6047 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
 6048 as /((?>a*))*/ (a PCRE_EXTRA facility).
 6051 Version 1.00 18-Nov-97
 6052 ----------------------
 6054 1. Added compile-time macros to support systems such as SunOS4 which don't have
 6055 memmove() or strerror() but have other things that can be used instead.
 6057 2. Arranged that "make clean" removes the executables.
 6060 Version 0.99 27-Oct-97
 6061 ----------------------
 6063 1. Fixed bug in code for optimizing classes with only one character. It was
 6064 initializing a 32-byte map regardless, which could cause it to run off the end
 6065 of the memory it had got.
 6067 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
 6070 Version 0.98 22-Oct-97
 6071 ----------------------
 6073 1. Fixed bug in code for handling temporary memory usage when there are more
 6074 back references than supplied space in the ovector. This could cause segfaults.
 6077 Version 0.97 21-Oct-97
 6078 ----------------------
 6080 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
 6082 2. Optimized negated single characters not to use a bit map.
 6084 3. Brought error texts together as macro definitions; clarified some of them;
 6085 fixed one that was wrong - it said "range out of order" when it meant "invalid
 6086 escape sequence".
 6088 4. Changed some char * arguments to const char *.
 6090 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
 6092 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
 6093 pcretest.
 6096 Version 0.96 16-Oct-97
 6097 ----------------------
 6099 1. Added a simple "pgrep" utility to the distribution.
 6101 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
 6102 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
 6103 where "ddd" means "one or more decimal digits".
 6105 3. Fixed serious bug. If a pattern had a back reference, but the call to
 6106 pcre_exec() didn't supply a large enough ovector to record the related
 6107 identifying subpattern, the match always failed. PCRE now remembers the number
 6108 of the largest back reference, and gets some temporary memory in which to save
 6109 the offsets during matching if necessary, in order to ensure that
 6110 backreferences always work.
 6112 4. Increased the compatibility with Perl in a number of ways:
 6114   (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
 6115       to request this handling. The option can be set at compile or exec time.
 6117   (b) $ matches before a terminating newline by default; an option
 6118       PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
 6119       mode). The option can be set at compile or exec time.
 6121   (c) The handling of \ followed by a digit other than 0 is now supposed to be
 6122       the same as Perl's. If the decimal number it represents is less than 10
 6123       or there aren't that many previous left capturing parentheses, an octal
 6124       escape is read. Inside a character class, it's always an octal escape,
 6125       even if it is a single digit.
 6127   (d) An escaped but undefined alphabetic character is taken as a literal,
 6128       unless PCRE_EXTRA is set. Currently this just reserves the remaining
 6129       escapes.
 6131   (e) {0} is now permitted. (The previous item is removed from the compiled
 6132       pattern).
 6134 5. Changed all the names of code files so that the basic parts are no longer
 6135 than 10 characters, and abolished the teeny "globals.c" file.
 6137 6. Changed the handling of character classes; they are now done with a 32-byte
 6138 bit map always.
 6140 7. Added the -d and /D options to pcretest to make it possible to look at the
 6141 internals of compilation without having to recompile pcre.
 6144 Version 0.95 23-Sep-97
 6145 ----------------------
 6147 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
 6148 \x20 at the start of a run of normal characters. These were being treated as
 6149 real characters, instead of the source characters being re-checked.
 6152 Version 0.94 18-Sep-97
 6153 ----------------------
 6155 1. The functions are now thread-safe, with the caveat that the global variables
 6156 containing pointers to malloc() and free() or alternative functions are the
 6157 same for all threads.
 6159 2. Get pcre_study() to generate a bitmap of initial characters for non-
 6160 anchored patterns when this is possible, and use it if passed to pcre_exec().
 6163 Version 0.93 15-Sep-97
 6164 ----------------------
 6166 1. /(b)|(:+)/ was computing an incorrect first character.
 6168 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
 6169 but not actually doing anything yet.
 6171 3. Treat "-" characters in classes that cannot be part of ranges as literals,
 6172 as Perl does (e.g. [-az] or [az-]).
 6174 4. Set the anchored flag if a branch starts with .* or .*? because that tests
 6175 all possible positions.
 6177 5. Split up into different modules to avoid including unneeded functions in a
 6178 compiled binary. However, compile and exec are still in one module. The "study"
 6179 function is split off.
 6181 6. The character tables are now in a separate module whose source is generated
 6182 by an auxiliary program - but can then be edited by hand if required. There are
 6183 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
 6184 toupper() in the code.
 6186 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
 6187 make them global. Abolish the function for setting them, as the caller can now
 6188 set them directly.
 6191 Version 0.92 11-Sep-97
 6192 ----------------------
 6194 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
 6195 (e.g. /a{1,3}/) was broken (I mis-optimized it).
 6197 2. Caseless matching was not working in character classes if the characters in
 6198 the pattern were in upper case.
 6200 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
 6202 4. Make PCRE_ANCHORED public and accept as a compile option.
 6204 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
 6205 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
 6206 pass them.
 6208 6. Give an error if bad option bits passed at compile or run time.
 6210 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
 6211 pcretest to cause it to pass that flag.
 6213 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
 6214 options, and the first character, if set.
 6216 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
 6219 Version 0.91 10-Sep-97
 6220 ----------------------
 6222 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
 6223 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
 6225 2. PCRE was looping on encountering an indefinitely repeated back reference to
 6226 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
 6227 Perl does - treats the match as successful.
 6229 ****