1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/tools/pcre/ChangeLog Tue Jul 28 17:51:34 2009 -0400
1.3 @@ -0,0 +1,3508 @@
1.4 +ChangeLog for PCRE
1.5 +------------------
1.6 +
1.7 +Version 7.9 11-Apr-09
1.8 +---------------------
1.9 +
1.10 +1. When building with support for bzlib/zlib (pcregrep) and/or readline
1.11 + (pcretest), all targets were linked against these libraries. This included
1.12 + libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1.13 + libraries. This caused unwanted dependencies to be created. This problem
1.14 + has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1.15 + pcretest is linked with readline.
1.16 +
1.17 +2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1.18 + "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1.19 + moved outside it again, because FALSE and TRUE are already defined in AIX,
1.20 + but BOOL is not.
1.21 +
1.22 +3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1.23 + PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1.24 +
1.25 +4. The pcregrep documentation said spaces were inserted as well as colons (or
1.26 + hyphens) following file names and line numbers when outputting matching
1.27 + lines. This is not true; no spaces are inserted. I have also clarified the
1.28 + wording for the --colour (or --color) option.
1.29 +
1.30 +5. In pcregrep, when --colour was used with -o, the list of matching strings
1.31 + was not coloured; this is different to GNU grep, so I have changed it to be
1.32 + the same.
1.33 +
1.34 +6. When --colo(u)r was used in pcregrep, only the first matching substring in
1.35 + each matching line was coloured. Now it goes on to look for further matches
1.36 + of any of the test patterns, which is the same behaviour as GNU grep.
1.37 +
1.38 +7. A pattern that could match an empty string could cause pcregrep to loop; it
1.39 + doesn't make sense to accept an empty string match in pcregrep, so I have
1.40 + locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1.41 + seems to be how GNU grep behaves.
1.42 +
1.43 +8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1.44 + start or after a newline", because the conditional assertion was not being
1.45 + correctly handled. The rule now is that both the assertion and what follows
1.46 + in the first alternative must satisfy the test.
1.47 +
1.48 +9. If auto-callout was enabled in a pattern with a conditional group whose
1.49 + condition was an assertion, PCRE could crash during matching, both with
1.50 + pcre_exec() and pcre_dfa_exec().
1.51 +
1.52 +10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1.53 + used for matching.
1.54 +
1.55 +11. Unicode property support in character classes was not working for
1.56 + characters (bytes) greater than 127 when not in UTF-8 mode.
1.57 +
1.58 +12. Added the -M command line option to pcretest.
1.59 +
1.60 +14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1.61 +
1.62 +15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1.63 +
1.64 +16. Added comments and documentation about mis-use of no_arg in the C++
1.65 + wrapper.
1.66 +
1.67 +17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1.68 + from Martin Jerabek that uses macro names for all relevant character and
1.69 + string constants.
1.70 +
1.71 +18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1.72 + SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1.73 + SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1.74 + these, but not everybody uses configure.
1.75 +
1.76 +19. A conditional group that had only one branch was not being correctly
1.77 + recognized as an item that could match an empty string. This meant that an
1.78 + enclosing group might also not be so recognized, causing infinite looping
1.79 + (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1.80 + with the subject "ab", where knowledge that the repeated group can match
1.81 + nothing is needed in order to break the loop.
1.82 +
1.83 +20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1.84 + exec(), but without supplying a callout function, matching went wrong.
1.85 +
1.86 +21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1.87 + leak if the size of the offset vector was greater than 30. When the vector
1.88 + is smaller, the saved offsets during recursion go onto a local stack
1.89 + vector, but for larger vectors malloc() is used. It was failing to free
1.90 + when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1.91 + error, in fact).
1.92 +
1.93 +22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1.94 + heapframe that is used only when UTF-8 support is enabled. This caused no
1.95 + problem, but was untidy.
1.96 +
1.97 +23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1.98 + CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1.99 + included within another project.
1.100 +
1.101 +24. Steven Van Ingelgem's patches to add more options to the CMake support,
1.102 + slightly modified by me:
1.103 +
1.104 + (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1.105 + not building pcregrep.
1.106 +
1.107 + (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1.108 + if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1.109 +
1.110 +25. Forward references, both numeric and by name, in patterns that made use of
1.111 + duplicate group numbers, could behave incorrectly or give incorrect errors,
1.112 + because when scanning forward to find the reference group, PCRE was not
1.113 + taking into account the duplicate group numbers. A pattern such as
1.114 + ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1.115 +
1.116 +26. Changed a few more instances of "const unsigned char *" to USPTR, making
1.117 + the feature of a custom pointer more persuasive (as requested by a user).
1.118 +
1.119 +27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1.120 + pcretest.c, inside #ifndefs, because it seems they are sometimes already
1.121 + pre-defined.
1.122 +
1.123 +28. Added support for (*UTF8) at the start of a pattern.
1.124 +
1.125 +29. Arrange for flags added by the "release type" setting in CMake to be shown
1.126 + in the configuration summary.
1.127 +
1.128 +
1.129 +Version 7.8 05-Sep-08
1.130 +---------------------
1.131 +
1.132 +1. Replaced UCP searching code with optimized version as implemented for Ad
1.133 + Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1.134 + stage table and inline lookup instead of a function, giving speed ups of 2
1.135 + to 5 times on some simple patterns that I tested. Permission was given to
1.136 + distribute the MultiStage2.py script that generates the tables (it's not in
1.137 + the tarball, but is in the Subversion repository).
1.138 +
1.139 +2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1.140 + scripts.
1.141 +
1.142 +3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1.143 + a group with a zero qualifier. The result of the study could be incorrect,
1.144 + or the function might crash, depending on the pattern.
1.145 +
1.146 +4. Caseless matching was not working for non-ASCII characters in back
1.147 + references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1.148 + It now works when Unicode Property Support is available.
1.149 +
1.150 +5. In pcretest, an escape such as \x{de} in the data was always generating
1.151 + a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1.152 + non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1.153 + truncation.
1.154 +
1.155 +6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1.156 +
1.157 +7. Added two (int) casts to pcregrep when printing the difference of two
1.158 + pointers, in case they are 64-bit values.
1.159 +
1.160 +8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1.161 + test 2 if it fails.
1.162 +
1.163 +9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1.164 + and a #define of that name to empty if it is not externally set. This is to
1.165 + allow users of MSVC to set it if necessary.
1.166 +
1.167 +10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1.168 + the convenience functions in the pcre_get.c source file.
1.169 +
1.170 +11. An option change at the start of a pattern that had top-level alternatives
1.171 + could cause overwriting and/or a crash. This command provoked a crash in
1.172 + some environments:
1.173 +
1.174 + printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1.175 +
1.176 + This potential security problem was recorded as CVE-2008-2371.
1.177 +
1.178 +12. For a pattern where the match had to start at the beginning or immediately
1.179 + after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1.180 + pcre_dfa_exec() could read past the end of the passed subject if there was
1.181 + no match. To help with detecting such bugs (e.g. with valgrind), I modified
1.182 + pcretest so that it places the subject at the end of its malloc-ed buffer.
1.183 +
1.184 +13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1.185 + exec() might read past the end of the data buffer in UTF-8 mode.
1.186 +
1.187 +14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1.188 + the data contained the byte 0x85 as part of a UTF-8 character within its
1.189 + first line. This applied both to normal and DFA matching.
1.190 +
1.191 +15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1.192 + /^[^d]*?$/8 failed to match "abc".
1.193 +
1.194 +16. Added a missing copyright notice to pcrecpp_internal.h.
1.195 +
1.196 +17. Make it more clear in the documentation that values returned from
1.197 + pcre_exec() in ovector are byte offsets, not character counts.
1.198 +
1.199 +18. Tidied a few places to stop certain compilers from issuing warnings.
1.200 +
1.201 +19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1.202 + supplied by Stefan Weber. I made a further small update for 7.8 because
1.203 + there is a change of source arrangements: the pcre_searchfuncs.c module is
1.204 + replaced by pcre_ucd.c.
1.205 +
1.206 +
1.207 +Version 7.7 07-May-08
1.208 +---------------------
1.209 +
1.210 +1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1.211 + a string to a long long, pretend we don't even have a long long." This is
1.212 + done by checking for the strtoq, strtoll, and _strtoi64 functions.
1.213 +
1.214 +2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1.215 + pre-7.6 versions, which defined a global no_arg variable instead of putting
1.216 + it in the RE class. (See also #8 below.)
1.217 +
1.218 +3. Remove a line of dead code, identified by coverity and reported by Nuno
1.219 + Lopes.
1.220 +
1.221 +4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1.222 +
1.223 + (1) The include/exclude patterns were being applied to the whole pathnames
1.224 + of files, instead of just to the final components.
1.225 +
1.226 + (2) If there was more than one level of directory, the subdirectories were
1.227 + skipped unless they satisfied the include/exclude conditions. This is
1.228 + inconsistent with GNU grep (and could even be seen as contrary to the
1.229 + pcregrep specification - which I improved to make it absolutely clear).
1.230 + The action now is always to scan all levels of directory, and just
1.231 + apply the include/exclude patterns to regular files.
1.232 +
1.233 +5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1.234 + --exclude_dir in the tests to avoid scanning .svn directories.
1.235 +
1.236 +6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1.237 + NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1.238 + doesn't support NULs in patterns.
1.239 +
1.240 +7. Added some missing "const"s to declarations of static tables in
1.241 + pcre_compile.c and pcre_dfa_exec.c.
1.242 +
1.243 +8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1.244 + caused by fix #2 above. (Subsequently also a second patch to fix the
1.245 + first patch. And a third patch - this was a messy problem.)
1.246 +
1.247 +9. Applied Craig's patch to remove the use of push_back().
1.248 +
1.249 +10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1.250 + matching function regexec().
1.251 +
1.252 +11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1.253 + which, however, unlike Perl's \g{...}, are subroutine calls, not back
1.254 + references. PCRE supports relative numbers with this syntax (I don't think
1.255 + Oniguruma does).
1.256 +
1.257 +12. Previously, a group with a zero repeat such as (...){0} was completely
1.258 + omitted from the compiled regex. However, this means that if the group
1.259 + was called as a subroutine from elsewhere in the pattern, things went wrong
1.260 + (an internal error was given). Such groups are now left in the compiled
1.261 + pattern, with a new opcode that causes them to be skipped at execution
1.262 + time.
1.263 +
1.264 +13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1.265 + to the way PCRE behaves:
1.266 +
1.267 + (a) A lone ] character is dis-allowed (Perl treats it as data).
1.268 +
1.269 + (b) A back reference to an unmatched subpattern matches an empty string
1.270 + (Perl fails the current match path).
1.271 +
1.272 + (c) A data ] in a character class must be notated as \] because if the
1.273 + first data character in a class is ], it defines an empty class. (In
1.274 + Perl it is not possible to have an empty class.) The empty class []
1.275 + never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1.276 + The negative empty class [^] matches any one character, independently
1.277 + of the DOTALL setting.
1.278 +
1.279 +14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1.280 + non-existent subpattern following a character class starting with ']' and
1.281 + containing () gave an internal compiling error instead of "reference to
1.282 + non-existent subpattern". Fortunately, when the pattern did exist, the
1.283 + compiled code was correct. (When scanning forwards to check for the
1.284 + existencd of the subpattern, it was treating the data ']' as terminating
1.285 + the class, so got the count wrong. When actually compiling, the reference
1.286 + was subsequently set up correctly.)
1.287 +
1.288 +15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1.289 + it was being rejected as not supported by pcre_dfa_exec(), even though
1.290 + other assertions are supported. I have made pcre_dfa_exec() support
1.291 + (*FAIL).
1.292 +
1.293 +16. The implementation of 13c above involved the invention of a new opcode,
1.294 + OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1.295 + cannot be changed at match time, I realized I could make a small
1.296 + improvement to matching performance by compiling OP_ALLANY instead of
1.297 + OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1.298 + on the OP_ANY path.
1.299 +
1.300 +17. Compiling pcretest on Windows with readline support failed without the
1.301 + following two fixes: (1) Make the unistd.h include conditional on
1.302 + HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1.303 +
1.304 +18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1.305 + ncurses library to be included for pcretest when ReadLine support is
1.306 + requested, but also to allow for it to be overridden. This patch came from
1.307 + Daniel Bergstr�m.
1.308 +
1.309 +19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1.310 + as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1.311 + any errors with the current Unicode tables. Thanks to Peter Kankowski for
1.312 + spotting this.
1.313 +
1.314 +
1.315 +Version 7.6 28-Jan-08
1.316 +---------------------
1.317 +
1.318 +1. A character class containing a very large number of characters with
1.319 + codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1.320 + overflow.
1.321 +
1.322 +2. Patch to cut out the "long long" test in pcrecpp_unittest when
1.323 + HAVE_LONG_LONG is not defined.
1.324 +
1.325 +3. Applied Christian Ehrlicher's patch to update the CMake build files to
1.326 + bring them up to date and include new features. This patch includes:
1.327 +
1.328 + - Fixed PH's badly added libz and libbz2 support.
1.329 + - Fixed a problem with static linking.
1.330 + - Added pcredemo. [But later removed - see 7 below.]
1.331 + - Fixed dftables problem and added an option.
1.332 + - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1.333 + HAVE_LONG_LONG.
1.334 + - Added readline support for pcretest.
1.335 + - Added an listing of the option settings after cmake has run.
1.336 +
1.337 +4. A user submitted a patch to Makefile that makes it easy to create
1.338 + "pcre.dll" under mingw when using Configure/Make. I added stuff to
1.339 + Makefile.am that cause it to include this special target, without
1.340 + affecting anything else. Note that the same mingw target plus all
1.341 + the other distribution libraries and programs are now supported
1.342 + when configuring with CMake (see 6 below) instead of with
1.343 + Configure/Make.
1.344 +
1.345 +5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1.346 + This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1.347 + exported in the Windows port". It has not yet been confirmed that the patch
1.348 + solves the problem, but it does no harm.
1.349 +
1.350 +6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1.351 + NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1.352 + with CMake, and also correct the comment about stack recursion.
1.353 +
1.354 +7. Remove the automatic building of pcredemo from the ./configure system and
1.355 + from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1.356 + of a program that users should build themselves after PCRE is installed, so
1.357 + building it automatically is not really right. What is more, it gave
1.358 + trouble in some build environments.
1.359 +
1.360 +8. Further tidies to CMakeLists.txt from Sheri and Christian.
1.361 +
1.362 +
1.363 +Version 7.5 10-Jan-08
1.364 +---------------------
1.365 +
1.366 +1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1.367 + values in parens when parsing an RE using the C++ wrapper."
1.368 +
1.369 +2. Negative specials like \S did not work in character classes in UTF-8 mode.
1.370 + Characters greater than 255 were excluded from the class instead of being
1.371 + included.
1.372 +
1.373 +3. The same bug as (2) above applied to negated POSIX classes such as
1.374 + [:^space:].
1.375 +
1.376 +4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1.377 + defined or documented. It seems to have been a typo for PCRE_STATIC, so
1.378 + I have changed it.
1.379 +
1.380 +5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1.381 + first named subpattern) and a construct such as (?&a) would reference the
1.382 + first named subpattern whose name started with "a" (in other words, the
1.383 + length check was missing). Both these problems are fixed. "Subpattern name
1.384 + expected" is now given for (?&) (a zero-length name), and this patch also
1.385 + makes it give the same error for \k'' (previously it complained that that
1.386 + was a reference to a non-existent subpattern).
1.387 +
1.388 +6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1.389 + this is right because (?- can be followed by option settings as well as by
1.390 + digits. I have, however, made the messages clearer.
1.391 +
1.392 +7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1.393 + than the number used in the conditional) now cause a compile-time error.
1.394 + This is actually not compatible with Perl, which accepts such patterns, but
1.395 + treats the conditional as always being FALSE (as PCRE used to), but it
1.396 + seems to me that giving a diagnostic is better.
1.397 +
1.398 +8. Change "alphameric" to the more common word "alphanumeric" in comments
1.399 + and messages.
1.400 +
1.401 +9. Fix two occurrences of "backslash" in comments that should have been
1.402 + "backspace".
1.403 +
1.404 +10. Remove two redundant lines of code that can never be obeyed (their function
1.405 + was moved elsewhere).
1.406 +
1.407 +11. The program that makes PCRE's Unicode character property table had a bug
1.408 + which caused it to generate incorrect table entries for sequences of
1.409 + characters that have the same character type, but are in different scripts.
1.410 + It amalgamated them into a single range, with the script of the first of
1.411 + them. In other words, some characters were in the wrong script. There were
1.412 + thirteen such cases, affecting characters in the following ranges:
1.413 +
1.414 + U+002b0 - U+002c1
1.415 + U+0060c - U+0060d
1.416 + U+0061e - U+00612
1.417 + U+0064b - U+0065e
1.418 + U+0074d - U+0076d
1.419 + U+01800 - U+01805
1.420 + U+01d00 - U+01d77
1.421 + U+01d9b - U+01dbf
1.422 + U+0200b - U+0200f
1.423 + U+030fc - U+030fe
1.424 + U+03260 - U+0327f
1.425 + U+0fb46 - U+0fbb1
1.426 + U+10450 - U+1049d
1.427 +
1.428 +12. The -o option (show only the matching part of a line) for pcregrep was not
1.429 + compatible with GNU grep in that, if there was more than one match in a
1.430 + line, it showed only the first of them. It now behaves in the same way as
1.431 + GNU grep.
1.432 +
1.433 +13. If the -o and -v options were combined for pcregrep, it printed a blank
1.434 + line for every non-matching line. GNU grep prints nothing, and pcregrep now
1.435 + does the same. The return code can be used to tell if there were any
1.436 + non-matching lines.
1.437 +
1.438 +14. Added --file-offsets and --line-offsets to pcregrep.
1.439 +
1.440 +15. The pattern (?=something)(?R) was not being diagnosed as a potentially
1.441 + infinitely looping recursion. The bug was that positive lookaheads were not
1.442 + being skipped when checking for a possible empty match (negative lookaheads
1.443 + and both kinds of lookbehind were skipped).
1.444 +
1.445 +16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
1.446 + inclusion of <windows.h> to before rather than after the definition of
1.447 + INVALID_FILE_ATTRIBUTES (patch from David Byron).
1.448 +
1.449 +17. Specifying a possessive quantifier with a specific limit for a Unicode
1.450 + character property caused pcre_compile() to compile bad code, which led at
1.451 + runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
1.452 + are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
1.453 + caused the error; without that there was no problem.
1.454 +
1.455 +18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
1.456 +
1.457 +19. Added --enable-pcretest-libreadline.
1.458 +
1.459 +20. In pcrecpp.cc, the variable 'count' was incremented twice in
1.460 + RE::GlobalReplace(). As a result, the number of replacements returned was
1.461 + double what it should be. I removed one of the increments, but Craig sent a
1.462 + later patch that removed the other one (the right fix) and added unit tests
1.463 + that check the return values (which was not done before).
1.464 +
1.465 +21. Several CMake things:
1.466 +
1.467 + (1) Arranged that, when cmake is used on Unix, the libraries end up with
1.468 + the names libpcre and libpcreposix, not just pcre and pcreposix.
1.469 +
1.470 + (2) The above change means that pcretest and pcregrep are now correctly
1.471 + linked with the newly-built libraries, not previously installed ones.
1.472 +
1.473 + (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
1.474 +
1.475 +22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
1.476 + crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
1.477 + UTF-8 newline character). The key issue is that the pattern starts .*;
1.478 + this means that the match must be either at the beginning, or after a
1.479 + newline. The bug was in the code for advancing after a failed match and
1.480 + checking that the new position followed a newline. It was not taking
1.481 + account of UTF-8 characters correctly.
1.482 +
1.483 +23. PCRE was behaving differently from Perl in the way it recognized POSIX
1.484 + character classes. PCRE was not treating the sequence [:...:] as a
1.485 + character class unless the ... were all letters. Perl, however, seems to
1.486 + allow any characters between [: and :], though of course it rejects as
1.487 + unknown any "names" that contain non-letters, because all the known class
1.488 + names consist only of letters. Thus, Perl gives an error for [[:1234:]],
1.489 + for example, whereas PCRE did not - it did not recognize a POSIX character
1.490 + class. This seemed a bit dangerous, so the code has been changed to be
1.491 + closer to Perl. The behaviour is not identical to Perl, because PCRE will
1.492 + diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
1.493 + treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
1.494 + Perl does, and where it didn't before.
1.495 +
1.496 +24. Rewrite so as to remove the single use of %n from pcregrep because in some
1.497 + Windows environments %n is disabled by default.
1.498 +
1.499 +
1.500 +Version 7.4 21-Sep-07
1.501 +---------------------
1.502 +
1.503 +1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
1.504 + means that a class such as [\s] counted as "explicit reference to CR or
1.505 + LF". That isn't really right - the whole point of the change was to try to
1.506 + help when there was an actual mention of one of the two characters. So now
1.507 + the change happens only if \r or \n (or a literal CR or LF) character is
1.508 + encountered.
1.509 +
1.510 +2. The 32-bit options word was also used for 6 internal flags, but the numbers
1.511 + of both had grown to the point where there were only 3 bits left.
1.512 + Fortunately, there was spare space in the data structure, and so I have
1.513 + moved the internal flags into a new 16-bit field to free up more option
1.514 + bits.
1.515 +
1.516 +3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
1.517 + but did not set the internal JCHANGED flag - either of these is enough to
1.518 + control the way the "get" function works - but the PCRE_INFO_JCHANGED
1.519 + facility is supposed to tell if (?J) was ever used, so now (?J) at the
1.520 + start sets both bits.
1.521 +
1.522 +4. Added options (at build time, compile time, exec time) to change \R from
1.523 + matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
1.524 +
1.525 +5. doc/pcresyntax.html was missing from the distribution.
1.526 +
1.527 +6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
1.528 + compatibility, even though it is no longer used.
1.529 +
1.530 +7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
1.531 + strtoull to pcrecpp.cc to select the available functions in WIN32 when the
1.532 + windows.h file is present (where different names are used). [This was
1.533 + reversed later after testing - see 16 below.]
1.534 +
1.535 +8. Changed all #include <config.h> to #include "config.h". There were also
1.536 + some further <pcre.h> cases that I changed to "pcre.h".
1.537 +
1.538 +9. When pcregrep was used with the --colour option, it missed the line ending
1.539 + sequence off the lines that it output.
1.540 +
1.541 +10. It was pointed out to me that arrays of string pointers cause lots of
1.542 + relocations when a shared library is dynamically loaded. A technique of
1.543 + using a single long string with a table of offsets can drastically reduce
1.544 + these. I have refactored PCRE in four places to do this. The result is
1.545 + dramatic:
1.546 +
1.547 + Originally: 290
1.548 + After changing UCP table: 187
1.549 + After changing error message table: 43
1.550 + After changing table of "verbs" 36
1.551 + After changing table of Posix names 22
1.552 +
1.553 + Thanks to the folks working on Gregex for glib for this insight.
1.554 +
1.555 +11. --disable-stack-for-recursion caused compiling to fail unless -enable-
1.556 + unicode-properties was also set.
1.557 +
1.558 +12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
1.559 +
1.560 +13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
1.561 + checked only for CRLF.
1.562 +
1.563 +14. Added casts to pcretest.c to avoid compiler warnings.
1.564 +
1.565 +15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
1.566 +
1.567 +16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
1.568 + and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
1.569 + entirely. This removes changes made in 7 above.
1.570 +
1.571 +17. The CMake files have been updated, and there is now more information about
1.572 + building with CMake in the NON-UNIX-USE document.
1.573 +
1.574 +
1.575 +Version 7.3 28-Aug-07
1.576 +---------------------
1.577 +
1.578 + 1. In the rejigging of the build system that eventually resulted in 7.1, the
1.579 + line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
1.580 + brackets there is not right, since it causes compilers to look for an
1.581 + installed pcre.h, not the version that is in the source that is being
1.582 + compiled (which of course may be different). I have changed it back to:
1.583 +
1.584 + #include "pcre.h"
1.585 +
1.586 + I have a vague recollection that the change was concerned with compiling in
1.587 + different directories, but in the new build system, that is taken care of
1.588 + by the VPATH setting the Makefile.
1.589 +
1.590 + 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
1.591 + when the subject happened to end in the byte 0x85 (e.g. if the last
1.592 + character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
1.593 + characters but of course it shouldn't be taken as a newline when it is part
1.594 + of another character. The bug was that, for an unlimited repeat of . in
1.595 + not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
1.596 + characters when looking for a newline.
1.597 +
1.598 + 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
1.599 +
1.600 + 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
1.601 + in debug output.
1.602 +
1.603 + 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
1.604 + long printing in the pcrecpp unittest when running under MinGW.
1.605 +
1.606 + 6. ESC_K was left out of the EBCDIC table.
1.607 +
1.608 + 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
1.609 + parentheses; I made it 1000, which seemed large enough. Unfortunately, the
1.610 + limit also applies to "virtual nesting" when a pattern is recursive, and in
1.611 + this case 1000 isn't so big. I have been able to remove this limit at the
1.612 + expense of backing off one optimization in certain circumstances. Normally,
1.613 + when pcre_exec() would call its internal match() function recursively and
1.614 + immediately return the result unconditionally, it uses a "tail recursion"
1.615 + feature to save stack. However, when a subpattern that can match an empty
1.616 + string has an unlimited repetition quantifier, it no longer makes this
1.617 + optimization. That gives it a stack frame in which to save the data for
1.618 + checking that an empty string has been matched. Previously this was taken
1.619 + from the 1000-entry workspace that had been reserved. So now there is no
1.620 + explicit limit, but more stack is used.
1.621 +
1.622 + 8. Applied Daniel's patches to solve problems with the import/export magic
1.623 + syntax that is required for Windows, and which was going wrong for the
1.624 + pcreposix and pcrecpp parts of the library. These were overlooked when this
1.625 + problem was solved for the main library.
1.626 +
1.627 + 9. There were some crude static tests to avoid integer overflow when computing
1.628 + the size of patterns that contain repeated groups with explicit upper
1.629 + limits. As the maximum quantifier is 65535, the maximum group length was
1.630 + set at 30,000 so that the product of these two numbers did not overflow a
1.631 + 32-bit integer. However, it turns out that people want to use groups that
1.632 + are longer than 30,000 bytes (though not repeat them that many times).
1.633 + Change 7.0/17 (the refactoring of the way the pattern size is computed) has
1.634 + made it possible to implement the integer overflow checks in a much more
1.635 + dynamic way, which I have now done. The artificial limitation on group
1.636 + length has been removed - we now have only the limit on the total length of
1.637 + the compiled pattern, which depends on the LINK_SIZE setting.
1.638 +
1.639 +10. Fixed a bug in the documentation for get/copy named substring when
1.640 + duplicate names are permitted. If none of the named substrings are set, the
1.641 + functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
1.642 + empty string.
1.643 +
1.644 +11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
1.645 + instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
1.646 + because the ] is interpreted as the first data character and the
1.647 + terminating ] is not found. PCRE has been made compatible with Perl in this
1.648 + regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
1.649 + cause memory overwriting.
1.650 +
1.651 +10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
1.652 + string has been matched (to stop an infinite loop). It was not recognizing
1.653 + a conditional subpattern that could match an empty string if that
1.654 + subpattern was within another subpattern. For example, it looped when
1.655 + trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
1.656 + condition was not nested. This bug has been fixed.
1.657 +
1.658 +12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
1.659 + past the start of the subject in the presence of bytes with the top bit
1.660 + set, for example "\x8aBCD".
1.661 +
1.662 +13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
1.663 + (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
1.664 +
1.665 +14. Optimized (?!) to (*FAIL).
1.666 +
1.667 +15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
1.668 + This restricts code points to be within the range 0 to 0x10FFFF, excluding
1.669 + the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
1.670 + full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
1.671 + does: it's just the validity check that is more restrictive.
1.672 +
1.673 +16. Inserted checks for integer overflows during escape sequence (backslash)
1.674 + processing, and also fixed erroneous offset values for syntax errors during
1.675 + backslash processing.
1.676 +
1.677 +17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
1.678 + for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
1.679 +
1.680 +18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
1.681 + caused an overrun.
1.682 +
1.683 +19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
1.684 + something other than just ASCII characters) inside a group that had an
1.685 + unlimited repeat caused a loop at compile time (while checking to see
1.686 + whether the group could match an empty string).
1.687 +
1.688 +20. Debugging a pattern containing \p or \P could cause a crash. For example,
1.689 + [\P{Any}] did so. (Error in the code for printing property names.)
1.690 +
1.691 +21. An orphan \E inside a character class could cause a crash.
1.692 +
1.693 +22. A repeated capturing bracket such as (A)? could cause a wild memory
1.694 + reference during compilation.
1.695 +
1.696 +23. There are several functions in pcre_compile() that scan along a compiled
1.697 + expression for various reasons (e.g. to see if it's fixed length for look
1.698 + behind). There were bugs in these functions when a repeated \p or \P was
1.699 + present in the pattern. These operators have additional parameters compared
1.700 + with \d, etc, and these were not being taken into account when moving along
1.701 + the compiled data. Specifically:
1.702 +
1.703 + (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
1.704 + length.
1.705 +
1.706 + (b) An item such as \pL+ within a repeated group could cause crashes or
1.707 + loops.
1.708 +
1.709 + (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
1.710 + "reference to non-existent subpattern" error.
1.711 +
1.712 + (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
1.713 +
1.714 +24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
1.715 + characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
1.716 +
1.717 +25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
1.718 +
1.719 +26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
1.720 + character were causing crashes (broken optimization).
1.721 +
1.722 +27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
1.723 + \p or \P) caused a compile-time loop.
1.724 +
1.725 +28. More problems have arisen in unanchored patterns when CRLF is a valid line
1.726 + break. For example, the unstudied pattern [\r\n]A does not match the string
1.727 + "\r\nA" because change 7.0/46 below moves the current point on by two
1.728 + characters after failing to match at the start. However, the pattern \nA
1.729 + *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
1.730 + the same is true. There doesn't seem any very clean way out of this, but
1.731 + what I have chosen to do makes the common cases work: PCRE now takes note
1.732 + of whether there can be an explicit match for \r or \n anywhere in the
1.733 + pattern, and if so, 7.0/46 no longer applies. As part of this change,
1.734 + there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
1.735 + pattern has explicit CR or LF references.
1.736 +
1.737 +29. Added (*CR) etc for changing newline setting at start of pattern.
1.738 +
1.739 +
1.740 +Version 7.2 19-Jun-07
1.741 +---------------------
1.742 +
1.743 + 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
1.744 + which is apparently normally available under Windows.
1.745 +
1.746 + 2. Re-jig the pcregrep tests with different newline settings in an attempt
1.747 + to make them independent of the local environment's newline setting.
1.748 +
1.749 + 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
1.750 +
1.751 + 4. Some of the "internals" tests were previously cut out when the link size
1.752 + was not 2, because the output contained actual offsets. The recent new
1.753 + "Z" feature of pcretest means that these can be cut out, making the tests
1.754 + usable with all link sizes.
1.755 +
1.756 + 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
1.757 + stack recursion. This gives a massive performance boost under BSD, but just
1.758 + a small improvement under Linux. However, it saves one field in the frame
1.759 + in all cases.
1.760 +
1.761 + 6. Added more features from the forthcoming Perl 5.10:
1.762 +
1.763 + (a) (?-n) (where n is a string of digits) is a relative subroutine or
1.764 + recursion call. It refers to the nth most recently opened parentheses.
1.765 +
1.766 + (b) (?+n) is also a relative subroutine call; it refers to the nth next
1.767 + to be opened parentheses.
1.768 +
1.769 + (c) Conditions that refer to capturing parentheses can be specified
1.770 + relatively, for example, (?(-2)... or (?(+3)...
1.771 +
1.772 + (d) \K resets the start of the current match so that everything before
1.773 + is not part of it.
1.774 +
1.775 + (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
1.776 +
1.777 + (f) \g{name} is another synonym - part of Perl 5.10's unification of
1.778 + reference syntax.
1.779 +
1.780 + (g) (?| introduces a group in which the numbering of parentheses in each
1.781 + alternative starts with the same number.
1.782 +
1.783 + (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
1.784 +
1.785 + 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
1.786 + PCRE_INFO_JCHANGED.
1.787 +
1.788 + 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
1.789 + terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
1.790 + for detecting groups that can match an empty string.
1.791 +
1.792 + 9. A pattern with a very large number of alternatives (more than several
1.793 + hundred) was running out of internal workspace during the pre-compile
1.794 + phase, where pcre_compile() figures out how much memory will be needed. A
1.795 + bit of new cunning has reduced the workspace needed for groups with
1.796 + alternatives. The 1000-alternative test pattern now uses 12 bytes of
1.797 + workspace instead of running out of the 4096 that are available.
1.798 +
1.799 +10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
1.800 +
1.801 +11. Applied patch from Google to remove an optimization that didn't quite work.
1.802 + The report of the bug said:
1.803 +
1.804 + pcrecpp::RE("a*").FullMatch("aaa") matches, while
1.805 + pcrecpp::RE("a*?").FullMatch("aaa") does not, and
1.806 + pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
1.807 +
1.808 +12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
1.809 + it matched the wrong number of bytes.
1.810 +
1.811 +
1.812 +Version 7.1 24-Apr-07
1.813 +---------------------
1.814 +
1.815 + 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
1.816 + that is more "standard", making use of automake and other Autotools. There
1.817 + is some re-arrangement of the files and adjustment of comments consequent
1.818 + on this.
1.819 +
1.820 + 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
1.821 + for recursive directory scanning broke on some systems because the files
1.822 + are not scanned in any specific order and on different systems the order
1.823 + was different. A call to "sort" has been inserted into RunGrepTest for the
1.824 + approprate test as a short-term fix. In the longer term there may be an
1.825 + alternative.
1.826 +
1.827 + 3. I had an email from Eric Raymond about problems translating some of PCRE's
1.828 + man pages to HTML (despite the fact that I distribute HTML pages, some
1.829 + people do their own conversions for various reasons). The problems
1.830 + concerned the use of low-level troff macros .br and .in. I have therefore
1.831 + removed all such uses from the man pages (some were redundant, some could
1.832 + be replaced by .nf/.fi pairs). The 132html script that I use to generate
1.833 + HTML has been updated to handle .nf/.fi and to complain if it encounters
1.834 + .br or .in.
1.835 +
1.836 + 4. Updated comments in configure.ac that get placed in config.h.in and also
1.837 + arranged for config.h to be included in the distribution, with the name
1.838 + config.h.generic, for the benefit of those who have to compile without
1.839 + Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
1.840 +
1.841 + 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
1.842 + Weber: (1) pcre_internal.h was missing some function renames; (2) updated
1.843 + makevp.bat for the current PCRE, using the additional files
1.844 + makevp_c.txt, makevp_l.txt, and pcregexp.pas.
1.845 +
1.846 + 6. A Windows user reported a minor discrepancy with test 2, which turned out
1.847 + to be caused by a trailing space on an input line that had got lost in his
1.848 + copy. The trailing space was an accident, so I've just removed it.
1.849 +
1.850 + 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
1.851 + that is needed.
1.852 +
1.853 + 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
1.854 + as "const" (a) because they are and (b) because it helps the PHP
1.855 + maintainers who have recently made a script to detect big data structures
1.856 + in the php code that should be moved to the .rodata section. I remembered
1.857 + to update Builducptable as well, so it won't revert if ucptable.h is ever
1.858 + re-created.
1.859 +
1.860 + 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
1.861 + pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
1.862 + order to be able to cut out the UTF-8 tables in the latter when UTF-8
1.863 + support is not required. This saves 1.5-2K of code, which is important in
1.864 + some applications.
1.865 +
1.866 + Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
1.867 + so as not to refer to the tables, even though these functions will never be
1.868 + called when UTF-8 support is disabled. Otherwise there are problems with a
1.869 + shared library.
1.870 +
1.871 +10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
1.872 +
1.873 + (a) It was defining its arguments as char * instead of void *.
1.874 +
1.875 + (b) It was assuming that all moves were upwards in memory; this was true
1.876 + a long time ago when I wrote it, but is no longer the case.
1.877 +
1.878 + The emulated memove() is provided for those environments that have neither
1.879 + memmove() nor bcopy(). I didn't think anyone used it these days, but that
1.880 + is clearly not the case, as these two bugs were recently reported.
1.881 +
1.882 +11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
1.883 + and Detrail to create the HTML documentation, the .txt form of the man
1.884 + pages, and it removes trailing spaces from listed files. It also creates
1.885 + pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
1.886 + case, it wraps all the #defines with #ifndefs. This script should be run
1.887 + before "make dist".
1.888 +
1.889 +12. Fixed two fairly obscure bugs concerned with quantified caseless matching
1.890 + with Unicode property support.
1.891 +
1.892 + (a) For a maximizing quantifier, if the two different cases of the
1.893 + character were of different lengths in their UTF-8 codings (there are
1.894 + some cases like this - I found 11), and the matching function had to
1.895 + back up over a mixture of the two cases, it incorrectly assumed they
1.896 + were both the same length.
1.897 +
1.898 + (b) When PCRE was configured to use the heap rather than the stack for
1.899 + recursion during matching, it was not correctly preserving the data for
1.900 + the other case of a UTF-8 character when checking ahead for a match
1.901 + while processing a minimizing repeat. If the check also involved
1.902 + matching a wide character, but failed, corruption could cause an
1.903 + erroneous result when trying to check for a repeat of the original
1.904 + character.
1.905 +
1.906 +13. Some tidying changes to the testing mechanism:
1.907 +
1.908 + (a) The RunTest script now detects the internal link size and whether there
1.909 + is UTF-8 and UCP support by running ./pcretest -C instead of relying on
1.910 + values substituted by "configure". (The RunGrepTest script already did
1.911 + this for UTF-8.) The configure.ac script no longer substitutes the
1.912 + relevant variables.
1.913 +
1.914 + (b) The debugging options /B and /D in pcretest show the compiled bytecode
1.915 + with length and offset values. This means that the output is different
1.916 + for different internal link sizes. Test 2 is skipped for link sizes
1.917 + other than 2 because of this, bypassing the problem. Unfortunately,
1.918 + there was also a test in test 3 (the locale tests) that used /B and
1.919 + failed for link sizes other than 2. Rather than cut the whole test out,
1.920 + I have added a new /Z option to pcretest that replaces the length and
1.921 + offset values with spaces. This is now used to make test 3 independent
1.922 + of link size. (Test 2 will be tidied up later.)
1.923 +
1.924 +14. If erroroffset was passed as NULL to pcre_compile, it provoked a
1.925 + segmentation fault instead of returning the appropriate error message.
1.926 +
1.927 +15. In multiline mode when the newline sequence was set to "any", the pattern
1.928 + ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
1.929 + This doesn't seem right; it now treats the CRLF combination as the line
1.930 + ending, and so does not match in that case. It's only a pattern such as ^$
1.931 + that would hit this one: something like ^ABC$ would have failed after \r
1.932 + and then tried again after \r\n.
1.933 +
1.934 +16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
1.935 + in an attempt to make files that differ only in their line terminators
1.936 + compare equal. This works on Linux.
1.937 +
1.938 +17. Under certain error circumstances pcregrep might try to free random memory
1.939 + as it exited. This is now fixed, thanks to valgrind.
1.940 +
1.941 +19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
1.942 + "abc\r\n\r\n", it found an unwanted second match after the second \r. This
1.943 + was because its rules for how to advance for /g after matching an empty
1.944 + string at the end of a line did not allow for this case. They now check for
1.945 + it specially.
1.946 +
1.947 +20. pcretest is supposed to handle patterns and data of any length, by
1.948 + extending its buffers when necessary. It was getting this wrong when the
1.949 + buffer for a data line had to be extended.
1.950 +
1.951 +21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
1.952 + CRLF as a newline sequence.
1.953 +
1.954 +22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
1.955 + out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
1.956 + I have nevertheless tidied it up.
1.957 +
1.958 +23. Added some casts to kill warnings from HP-UX ia64 compiler.
1.959 +
1.960 +24. Added a man page for pcre-config.
1.961 +
1.962 +
1.963 +Version 7.0 19-Dec-06
1.964 +---------------------
1.965 +
1.966 + 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
1.967 + moving to gcc 4.1.1.
1.968 +
1.969 + 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
1.970 + sys/time.h, which is documented as needed for this function. It doesn't
1.971 + seem to matter on Linux, but it showed up on some releases of OS X.
1.972 +
1.973 + 3. It seems that there are systems where bytes whose values are greater than
1.974 + 127 match isprint() in the "C" locale. The "C" locale should be the
1.975 + default when a C program starts up. In most systems, only ASCII printing
1.976 + characters match isprint(). This difference caused the output from pcretest
1.977 + to vary, making some of the tests fail. I have changed pcretest so that:
1.978 +
1.979 + (a) When it is outputting text in the compiled version of a pattern, bytes
1.980 + other than 32-126 are always shown as hex escapes.
1.981 +
1.982 + (b) When it is outputting text that is a matched part of a subject string,
1.983 + it does the same, unless a different locale has been set for the match
1.984 + (using the /L modifier). In this case, it uses isprint() to decide.
1.985 +
1.986 + 4. Fixed a major bug that caused incorrect computation of the amount of memory
1.987 + required for a compiled pattern when options that changed within the
1.988 + pattern affected the logic of the preliminary scan that determines the
1.989 + length. The relevant options are -x, and -i in UTF-8 mode. The result was
1.990 + that the computed length was too small. The symptoms of this bug were
1.991 + either the PCRE error "internal error: code overflow" from pcre_compile(),
1.992 + or a glibc crash with a message such as "pcretest: free(): invalid next
1.993 + size (fast)". Examples of patterns that provoked this bug (shown in
1.994 + pcretest format) are:
1.995 +
1.996 + /(?-x: )/x
1.997 + /(?x)(?-x: \s*#\s*)/
1.998 + /((?i)[\x{c0}])/8
1.999 + /(?i:[\x{c0}])/8
1.1000 +
1.1001 + HOWEVER: Change 17 below makes this fix obsolete as the memory computation
1.1002 + is now done differently.
1.1003 +
1.1004 + 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
1.1005 + wrapper classes; (b) implement a new function in the C++ scanner that is
1.1006 + more efficient than the old way of doing things because it avoids levels of
1.1007 + recursion in the regex matching; (c) add a paragraph to the documentation
1.1008 + for the FullMatch() function.
1.1009 +
1.1010 + 6. The escape sequence \n was being treated as whatever was defined as
1.1011 + "newline". Not only was this contrary to the documentation, which states
1.1012 + that \n is character 10 (hex 0A), but it also went horribly wrong when
1.1013 + "newline" was defined as CRLF. This has been fixed.
1.1014 +
1.1015 + 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
1.1016 + was being set to -1 for the "end of line" case (supposedly a value that no
1.1017 + character can have). Though this value is never used (the check for end of
1.1018 + line is "zero bytes in current character"), it caused compiler complaints.
1.1019 + I've changed it to 0xffffffff.
1.1020 +
1.1021 + 8. In pcre_version.c, the version string was being built by a sequence of
1.1022 + C macros that, in the event of PCRE_PRERELEASE being defined as an empty
1.1023 + string (as it is for production releases) called a macro with an empty
1.1024 + argument. The C standard says the result of this is undefined. The gcc
1.1025 + compiler treats it as an empty string (which was what was wanted) but it is
1.1026 + reported that Visual C gives an error. The source has been hacked around to
1.1027 + avoid this problem.
1.1028 +
1.1029 + 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
1.1030 + builds of pcretest, and changed the call to _setmode() to use _O_BINARY
1.1031 + instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
1.1032 + of them did).
1.1033 +
1.1034 +10. Originally, pcretest opened its input and output without "b"; then I was
1.1035 + told that "b" was needed in some environments, so it was added for release
1.1036 + 5.0 to both the input and output. (It makes no difference on Unix-like
1.1037 + systems.) Later I was told that it is wrong for the input on Windows. I've
1.1038 + now abstracted the modes into two macros, to make it easier to fiddle with
1.1039 + them, and removed "b" from the input mode under Windows.
1.1040 +
1.1041 +11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
1.1042 +
1.1043 +12. Added -help and --help to pcretest as an official way of being reminded
1.1044 + of the options.
1.1045 +
1.1046 +13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
1.1047 + and pcrecpp.cc because they annoy compilers at high warning levels.
1.1048 +
1.1049 +14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
1.1050 +
1.1051 +15. Fixed an occurrence of == in configure.ac that should have been = (shell
1.1052 + scripts are not C programs :-) and which was not noticed because it works
1.1053 + on Linux.
1.1054 +
1.1055 +16. pcretest is supposed to handle any length of pattern and data line (as one
1.1056 + line or as a continued sequence of lines) by extending its input buffer if
1.1057 + necessary. This feature was broken for very long pattern lines, leading to
1.1058 + a string of junk being passed to pcre_compile() if the pattern was longer
1.1059 + than about 50K.
1.1060 +
1.1061 +17. I have done a major re-factoring of the way pcre_compile() computes the
1.1062 + amount of memory needed for a compiled pattern. Previously, there was code
1.1063 + that made a preliminary scan of the pattern in order to do this. That was
1.1064 + OK when PCRE was new, but as the facilities have expanded, it has become
1.1065 + harder and harder to keep it in step with the real compile phase, and there
1.1066 + have been a number of bugs (see for example, 4 above). I have now found a
1.1067 + cunning way of running the real compile function in a "fake" mode that
1.1068 + enables it to compute how much memory it would need, while actually only
1.1069 + ever using a few hundred bytes of working memory and without too many
1.1070 + tests of the mode. This should make future maintenance and development
1.1071 + easier. A side effect of this work is that the limit of 200 on the nesting
1.1072 + depth of parentheses has been removed (though this was never a serious
1.1073 + limitation, I suspect). However, there is a downside: pcre_compile() now
1.1074 + runs more slowly than before (30% or more, depending on the pattern). I
1.1075 + hope this isn't a big issue. There is no effect on runtime performance.
1.1076 +
1.1077 +18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
1.1078 + newline (only possible for the last line of a file) and it was a
1.1079 + pattern that set a locale (followed by /Lsomething), pcretest crashed.
1.1080 +
1.1081 +19. Added additional timing features to pcretest. (1) The -tm option now times
1.1082 + matching only, not compiling. (2) Both -t and -tm can be followed, as a
1.1083 + separate command line item, by a number that specifies the number of
1.1084 + repeats to use when timing. The default is 50000; this gives better
1.1085 + precision, but takes uncomfortably long for very large patterns.
1.1086 +
1.1087 +20. Extended pcre_study() to be more clever in cases where a branch of a
1.1088 + subpattern has no definite first character. For example, (a*|b*)[cd] would
1.1089 + previously give no result from pcre_study(). Now it recognizes that the
1.1090 + first character must be a, b, c, or d.
1.1091 +
1.1092 +21. There was an incorrect error "recursive call could loop indefinitely" if
1.1093 + a subpattern (or the entire pattern) that was being tested for matching an
1.1094 + empty string contained only one non-empty item after a nested subpattern.
1.1095 + For example, the pattern (?>\x{100}*)\d(?R) provoked this error
1.1096 + incorrectly, because the \d was being skipped in the check.
1.1097 +
1.1098 +22. The pcretest program now has a new pattern option /B and a command line
1.1099 + option -b, which is equivalent to adding /B to every pattern. This causes
1.1100 + it to show the compiled bytecode, without the additional information that
1.1101 + -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
1.1102 + is the same as /B/I).
1.1103 +
1.1104 +23. A new optimization is now able automatically to treat some sequences such
1.1105 + as a*b as a*+b. More specifically, if something simple (such as a character
1.1106 + or a simple class like \d) has an unlimited quantifier, and is followed by
1.1107 + something that cannot possibly match the quantified thing, the quantifier
1.1108 + is automatically "possessified".
1.1109 +
1.1110 +24. A recursive reference to a subpattern whose number was greater than 39
1.1111 + went wrong under certain circumstances in UTF-8 mode. This bug could also
1.1112 + have affected the operation of pcre_study().
1.1113 +
1.1114 +25. Realized that a little bit of performance could be had by replacing
1.1115 + (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
1.1116 +
1.1117 +26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
1.1118 +
1.1119 +27. Possessive quantifiers such as a++ were previously implemented by turning
1.1120 + them into atomic groups such as ($>a+). Now they have their own opcodes,
1.1121 + which improves performance. This includes the automatically created ones
1.1122 + from 23 above.
1.1123 +
1.1124 +28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
1.1125 + lookahead was broken if it was not anchored. PCRE was mistakenly expecting
1.1126 + the first matched character to be a colon. This applied both to named and
1.1127 + numbered groups.
1.1128 +
1.1129 +29. The ucpinternal.h header file was missing its idempotency #ifdef.
1.1130 +
1.1131 +30. I was sent a "project" file called libpcre.a.dev which I understand makes
1.1132 + building PCRE on Windows easier, so I have included it in the distribution.
1.1133 +
1.1134 +31. There is now a check in pcretest against a ridiculously large number being
1.1135 + returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
1.1136 + loop, the loop is abandoned.
1.1137 +
1.1138 +32. Forward references to subpatterns in conditions such as (?(2)...) where
1.1139 + subpattern 2 is defined later cause pcre_compile() to search forwards in
1.1140 + the pattern for the relevant set of parentheses. This search went wrong
1.1141 + when there were unescaped parentheses in a character class, parentheses
1.1142 + escaped with \Q...\E, or parentheses in a #-comment in /x mode.
1.1143 +
1.1144 +33. "Subroutine" calls and backreferences were previously restricted to
1.1145 + referencing subpatterns earlier in the regex. This restriction has now
1.1146 + been removed.
1.1147 +
1.1148 +34. Added a number of extra features that are going to be in Perl 5.10. On the
1.1149 + whole, these are just syntactic alternatives for features that PCRE had
1.1150 + previously implemented using the Python syntax or my own invention. The
1.1151 + other formats are all retained for compatibility.
1.1152 +
1.1153 + (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
1.1154 + as (?P<name>...). The new forms, as well as being in Perl 5.10, are
1.1155 + also .NET compatible.
1.1156 +
1.1157 + (b) A recursion or subroutine call to a named group can now be defined as
1.1158 + (?&name) as well as (?P>name).
1.1159 +
1.1160 + (c) A backreference to a named group can now be defined as \k<name> or
1.1161 + \k'name' as well as (?P=name). The new forms, as well as being in Perl
1.1162 + 5.10, are also .NET compatible.
1.1163 +
1.1164 + (d) A conditional reference to a named group can now use the syntax
1.1165 + (?(<name>) or (?('name') as well as (?(name).
1.1166 +
1.1167 + (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
1.1168 + groups (named and numbered) that are never evaluated inline, but can be
1.1169 + called as "subroutines" from elsewhere. In effect, the DEFINE condition
1.1170 + is always false. There may be only one alternative in such a group.
1.1171 +
1.1172 + (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
1.1173 + as the simple (?(R). The condition is true only if the most recent
1.1174 + recursion is that of the given number or name. It does not search out
1.1175 + through the entire recursion stack.
1.1176 +
1.1177 + (g) The escape \gN or \g{N} has been added, where N is a positive or
1.1178 + negative number, specifying an absolute or relative reference.
1.1179 +
1.1180 +35. Tidied to get rid of some further signed/unsigned compiler warnings and
1.1181 + some "unreachable code" warnings.
1.1182 +
1.1183 +36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
1.1184 + things, this adds five new scripts.
1.1185 +
1.1186 +37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
1.1187 + There were also incompatibilities regarding the handling of \Q..\E inside
1.1188 + character classes, for example with patterns like [\Qa\E-\Qz\E] where the
1.1189 + hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
1.1190 +
1.1191 +38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
1.1192 + matches an empty string, and forcibly breaks the loop. There were bugs in
1.1193 + this code in non-simple cases. For a pattern such as ^(a()*)* matched
1.1194 + against aaaa the result was just "a" rather than "aaaa", for example. Two
1.1195 + separate and independent bugs (that affected different cases) have been
1.1196 + fixed.
1.1197 +
1.1198 +39. Refactored the code to abolish the use of different opcodes for small
1.1199 + capturing bracket numbers. This is a tidy that I avoided doing when I
1.1200 + removed the limit on the number of capturing brackets for 3.5 back in 2001.
1.1201 + The new approach is not only tidier, it makes it possible to reduce the
1.1202 + memory needed to fix the previous bug (38).
1.1203 +
1.1204 +40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
1.1205 + sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
1.1206 + processing dot, circumflex, or dollar metacharacters, or #-comments in /x
1.1207 + mode.
1.1208 +
1.1209 +41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
1.1210 + report.
1.1211 +
1.1212 +42. Applied patch, originally from Ari Pollak, modified by Google, to allow
1.1213 + copy construction and assignment in the C++ wrapper.
1.1214 +
1.1215 +43. Updated pcregrep to support "--newline=any". In the process, I fixed a
1.1216 + couple of bugs that could have given wrong results in the "--newline=crlf"
1.1217 + case.
1.1218 +
1.1219 +44. Added a number of casts and did some reorganization of signed/unsigned int
1.1220 + variables following suggestions from Dair Grant. Also renamed the variable
1.1221 + "this" as "item" because it is a C++ keyword.
1.1222 +
1.1223 +45. Arranged for dftables to add
1.1224 +
1.1225 + #include "pcre_internal.h"
1.1226 +
1.1227 + to pcre_chartables.c because without it, gcc 4.x may remove the array
1.1228 + definition from the final binary if PCRE is built into a static library and
1.1229 + dead code stripping is activated.
1.1230 +
1.1231 +46. For an unanchored pattern, if a match attempt fails at the start of a
1.1232 + newline sequence, and the newline setting is CRLF or ANY, and the next two
1.1233 + characters are CRLF, advance by two characters instead of one.
1.1234 +
1.1235 +
1.1236 +Version 6.7 04-Jul-06
1.1237 +---------------------
1.1238 +
1.1239 + 1. In order to handle tests when input lines are enormously long, pcretest has
1.1240 + been re-factored so that it automatically extends its buffers when
1.1241 + necessary. The code is crude, but this _is_ just a test program. The
1.1242 + default size has been increased from 32K to 50K.
1.1243 +
1.1244 + 2. The code in pcre_study() was using the value of the re argument before
1.1245 + testing it for NULL. (Of course, in any sensible call of the function, it
1.1246 + won't be NULL.)
1.1247 +
1.1248 + 3. The memmove() emulation function in pcre_internal.h, which is used on
1.1249 + systems that lack both memmove() and bcopy() - that is, hardly ever -
1.1250 + was missing a "static" storage class specifier.
1.1251 +
1.1252 + 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
1.1253 + containing an extended class (one that cannot be represented by a bitmap
1.1254 + because it contains high-valued characters or Unicode property items, e.g.
1.1255 + [\pZ]). Almost always one would set UTF-8 mode when processing such a
1.1256 + pattern, but PCRE should not loop if you do not (it no longer does).
1.1257 + [Detail: two cases were found: (a) a repeated subpattern containing an
1.1258 + extended class; (b) a recursive reference to a subpattern that followed a
1.1259 + previous extended class. It wasn't skipping over the extended class
1.1260 + correctly when UTF-8 mode was not set.]
1.1261 +
1.1262 + 5. A negated single-character class was not being recognized as fixed-length
1.1263 + in lookbehind assertions such as (?<=[^f]), leading to an incorrect
1.1264 + compile error "lookbehind assertion is not fixed length".
1.1265 +
1.1266 + 6. The RunPerlTest auxiliary script was showing an unexpected difference
1.1267 + between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
1.1268 + write a Perl script that can interpret lines of an input file either as
1.1269 + byte characters or as UTF-8, which is what "perltest" was being required to
1.1270 + do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
1.1271 + can't do is switch easily at run time between having the "use utf8;" pragma
1.1272 + or not. In the end, I fudged it by using the RunPerlTest script to insert
1.1273 + "use utf8;" explicitly for the UTF-8 tests.
1.1274 +
1.1275 + 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
1.1276 + the end of the subject string, contrary to the documentation and to what
1.1277 + Perl does. This was true of both matching functions. Now it matches only at
1.1278 + the start of the subject and immediately after *internal* newlines.
1.1279 +
1.1280 + 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
1.1281 + a pointer to an int instead of a pointer to an unsigned long int. This
1.1282 + caused problems on 64-bit systems.
1.1283 +
1.1284 + 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
1.1285 + instance of the 'standard' template library not being so standard".
1.1286 +
1.1287 +10. There was no check on the number of named subpatterns nor the maximum
1.1288 + length of a subpattern name. The product of these values is used to compute
1.1289 + the size of the memory block for a compiled pattern. By supplying a very
1.1290 + long subpattern name and a large number of named subpatterns, the size
1.1291 + computation could be caused to overflow. This is now prevented by limiting
1.1292 + the length of names to 32 characters, and the number of named subpatterns
1.1293 + to 10,000.
1.1294 +
1.1295 +11. Subpatterns that are repeated with specific counts have to be replicated in
1.1296 + the compiled pattern. The size of memory for this was computed from the
1.1297 + length of the subpattern and the repeat count. The latter is limited to
1.1298 + 65535, but there was no limit on the former, meaning that integer overflow
1.1299 + could in principle occur. The compiled length of a repeated subpattern is
1.1300 + now limited to 30,000 bytes in order to prevent this.
1.1301 +
1.1302 +12. Added the optional facility to have named substrings with the same name.
1.1303 +
1.1304 +13. Added the ability to use a named substring as a condition, using the
1.1305 + Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
1.1306 + are numbers (not recommended). Forward references are permitted.
1.1307 +
1.1308 +14. Added forward references in named backreferences (if you see what I mean).
1.1309 +
1.1310 +15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
1.1311 + pattern could run off the end of the subject. For example, the pattern
1.1312 + "(?s)(.{1,5})"8 did this with the subject "ab".
1.1313 +
1.1314 +16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
1.1315 + PCRE_CASELESS was set when matching characters that were quantified with ?
1.1316 + or *.
1.1317 +
1.1318 +17. A character class other than a single negated character that had a minimum
1.1319 + but no maximum quantifier - for example [ab]{6,} - was not handled
1.1320 + correctly by pce_dfa_exec(). It would match only one character.
1.1321 +
1.1322 +18. A valid (though odd) pattern that looked like a POSIX character
1.1323 + class but used an invalid character after [ (for example [[,abc,]]) caused
1.1324 + pcre_compile() to give the error "Failed: internal error: code overflow" or
1.1325 + in some cases to crash with a glibc free() error. This could even happen if
1.1326 + the pattern terminated after [[ but there just happened to be a sequence of
1.1327 + letters, a binary zero, and a closing ] in the memory that followed.
1.1328 +
1.1329 +19. Perl's treatment of octal escapes in the range \400 to \777 has changed
1.1330 + over the years. Originally (before any Unicode support), just the bottom 8
1.1331 + bits were taken. Thus, for example, \500 really meant \100. Nowadays the
1.1332 + output from "man perlunicode" includes this:
1.1333 +
1.1334 + The regular expression compiler produces polymorphic opcodes. That
1.1335 + is, the pattern adapts to the data and automatically switches to
1.1336 + the Unicode character scheme when presented with Unicode data--or
1.1337 + instead uses a traditional byte scheme when presented with byte
1.1338 + data.
1.1339 +
1.1340 + Sadly, a wide octal escape does not cause a switch, and in a string with
1.1341 + no other multibyte characters, these octal escapes are treated as before.
1.1342 + Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
1.1343 + /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
1.1344 + Unicode string.
1.1345 +
1.1346 + I have not perpetrated such confusion in PCRE. Up till now, it took just
1.1347 + the bottom 8 bits, as in old Perl. I have now made octal escapes with
1.1348 + values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
1.1349 + translate to the appropriate multibyte character.
1.1350 +
1.1351 +29. Applied some refactoring to reduce the number of warnings from Microsoft
1.1352 + and Borland compilers. This has included removing the fudge introduced
1.1353 + seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
1.1354 + a warning about an unused variable.
1.1355 +
1.1356 +21. PCRE has not included VT (character 0x0b) in the set of whitespace
1.1357 + characters since release 4.0, because Perl (from release 5.004) does not.
1.1358 + [Or at least, is documented not to: some releases seem to be in conflict
1.1359 + with the documentation.] However, when a pattern was studied with
1.1360 + pcre_study() and all its branches started with \s, PCRE still included VT
1.1361 + as a possible starting character. Of course, this did no harm; it just
1.1362 + caused an unnecessary match attempt.
1.1363 +
1.1364 +22. Removed a now-redundant internal flag bit that recorded the fact that case
1.1365 + dependency changed within the pattern. This was once needed for "required
1.1366 + byte" processing, but is no longer used. This recovers a now-scarce options
1.1367 + bit. Also moved the least significant internal flag bit to the most-
1.1368 + significant bit of the word, which was not previously used (hangover from
1.1369 + the days when it was an int rather than a uint) to free up another bit for
1.1370 + the future.
1.1371 +
1.1372 +23. Added support for CRLF line endings as well as CR and LF. As well as the
1.1373 + default being selectable at build time, it can now be changed at runtime
1.1374 + via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
1.1375 + specify that it is scanning data with non-default line endings.
1.1376 +
1.1377 +24. Changed the definition of CXXLINK to make it agree with the definition of
1.1378 + LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
1.1379 +
1.1380 +25. Applied Ian Taylor's patches to avoid using another stack frame for tail
1.1381 + recursions. This makes a big different to stack usage for some patterns.
1.1382 +
1.1383 +26. If a subpattern containing a named recursion or subroutine reference such
1.1384 + as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
1.1385 + the space required for the compiled pattern went wrong and gave too small a
1.1386 + value. Depending on the environment, this could lead to "Failed: internal
1.1387 + error: code overflow at offset 49" or "glibc detected double free or
1.1388 + corruption" errors.
1.1389 +
1.1390 +27. Applied patches from Google (a) to support the new newline modes and (b) to
1.1391 + advance over multibyte UTF-8 characters in GlobalReplace.
1.1392 +
1.1393 +28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
1.1394 + difference for some implementation of PCRE in some Windows version.
1.1395 +
1.1396 +29. Added some extra testing facilities to pcretest:
1.1397 +
1.1398 + \q<number> in a data line sets the "match limit" value
1.1399 + \Q<number> in a data line sets the "match recursion limt" value
1.1400 + -S <number> sets the stack size, where <number> is in megabytes
1.1401 +
1.1402 + The -S option isn't available for Windows.
1.1403 +
1.1404 +
1.1405 +Version 6.6 06-Feb-06
1.1406 +---------------------
1.1407 +
1.1408 + 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
1.1409 + in pcreposix.h. I have copied the definition from pcre.h.
1.1410 +
1.1411 + 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
1.1412 + because pcre.h is no longer a built file.
1.1413 +
1.1414 + 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
1.1415 + not normally included in the compiled code.
1.1416 +
1.1417 +
1.1418 +Version 6.5 01-Feb-06
1.1419 +---------------------
1.1420 +
1.1421 + 1. When using the partial match feature with pcre_dfa_exec(), it was not
1.1422 + anchoring the second and subsequent partial matches at the new starting
1.1423 + point. This could lead to incorrect results. For example, with the pattern
1.1424 + /1234/, partially matching against "123" and then "a4" gave a match.
1.1425 +
1.1426 + 2. Changes to pcregrep:
1.1427 +
1.1428 + (a) All non-match returns from pcre_exec() were being treated as failures
1.1429 + to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
1.1430 + error message is output. Some extra information is given for the
1.1431 + PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
1.1432 + probably the only errors that are likely to be caused by users (by
1.1433 + specifying a regex that has nested indefinite repeats, for instance).
1.1434 + If there are more than 20 of these errors, pcregrep is abandoned.
1.1435 +
1.1436 + (b) A binary zero was treated as data while matching, but terminated the
1.1437 + output line if it was written out. This has been fixed: binary zeroes
1.1438 + are now no different to any other data bytes.
1.1439 +
1.1440 + (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
1.1441 + used to set a locale for matching. The --locale=xxxx long option has
1.1442 + been added (no short equivalent) to specify a locale explicitly on the
1.1443 + pcregrep command, overriding the environment variables.
1.1444 +
1.1445 + (d) When -B was used with -n, some line numbers in the output were one less
1.1446 + than they should have been.
1.1447 +
1.1448 + (e) Added the -o (--only-matching) option.
1.1449 +
1.1450 + (f) If -A or -C was used with -c (count only), some lines of context were
1.1451 + accidentally printed for the final match.
1.1452 +
1.1453 + (g) Added the -H (--with-filename) option.
1.1454 +
1.1455 + (h) The combination of options -rh failed to suppress file names for files
1.1456 + that were found from directory arguments.
1.1457 +
1.1458 + (i) Added the -D (--devices) and -d (--directories) options.
1.1459 +
1.1460 + (j) Added the -F (--fixed-strings) option.
1.1461 +
1.1462 + (k) Allow "-" to be used as a file name for -f as well as for a data file.
1.1463 +
1.1464 + (l) Added the --colo(u)r option.
1.1465 +
1.1466 + (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
1.1467 + is not present by default.
1.1468 +
1.1469 + 3. A nasty bug was discovered in the handling of recursive patterns, that is,
1.1470 + items such as (?R) or (?1), when the recursion could match a number of
1.1471 + alternatives. If it matched one of the alternatives, but subsequently,
1.1472 + outside the recursion, there was a failure, the code tried to back up into
1.1473 + the recursion. However, because of the way PCRE is implemented, this is not
1.1474 + possible, and the result was an incorrect result from the match.
1.1475 +
1.1476 + In order to prevent this happening, the specification of recursion has
1.1477 + been changed so that all such subpatterns are automatically treated as
1.1478 + atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
1.1479 +
1.1480 + 4. I had overlooked the fact that, in some locales, there are characters for
1.1481 + which isalpha() is true but neither isupper() nor islower() are true. In
1.1482 + the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
1.1483 + and ordfeminine) are like this. This affected the treatment of \w and \W
1.1484 + when they appeared in character classes, but not when they appeared outside
1.1485 + a character class. The bit map for "word" characters is now created
1.1486 + separately from the results of isalnum() instead of just taking it from the
1.1487 + upper, lower, and digit maps. (Plus the underscore character, of course.)
1.1488 +
1.1489 + 5. The above bug also affected the handling of POSIX character classes such as
1.1490 + [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
1.1491 + permanent tables. Instead, the bit maps for such a class were previously
1.1492 + created as the appropriate unions of the upper, lower, and digit bitmaps.
1.1493 + Now they are created by subtraction from the [[:word:]] class, which has
1.1494 + its own bitmap.
1.1495 +
1.1496 + 6. The [[:blank:]] character class matches horizontal, but not vertical space.
1.1497 + It is created by subtracting the vertical space characters (\x09, \x0a,
1.1498 + \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
1.1499 + subtraction was done in the overall bitmap for a character class, meaning
1.1500 + that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
1.1501 + be recognized. This bug has been fixed.
1.1502 +
1.1503 + 7. Patches from the folks at Google:
1.1504 +
1.1505 + (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
1.1506 + real life, but is still worth protecting against".
1.1507 +
1.1508 + (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
1.1509 + regular expressions".
1.1510 +
1.1511 + (c) pcre_scanner.cc: avoid use of std::count() because not all systems
1.1512 + have it.
1.1513 +
1.1514 + (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
1.1515 + "configure" and the latter not, in order to fix a problem somebody had
1.1516 + with compiling the Arg class on HP-UX.
1.1517 +
1.1518 + (e) Improve the error-handling of the C++ wrapper a little bit.
1.1519 +
1.1520 + (f) New tests for checking recursion limiting.
1.1521 +
1.1522 + 8. The pcre_memmove() function, which is used only if the environment does not
1.1523 + have a standard memmove() function (and is therefore rarely compiled),
1.1524 + contained two bugs: (a) use of int instead of size_t, and (b) it was not
1.1525 + returning a result (though PCRE never actually uses the result).
1.1526 +
1.1527 + 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
1.1528 + large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
1.1529 + returned instead of calling malloc() with an overflowing number that would
1.1530 + most likely cause subsequent chaos.
1.1531 +
1.1532 +10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
1.1533 +
1.1534 +11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
1.1535 + with this option is matched, the nmatch and pmatch options of regexec() are
1.1536 + ignored.
1.1537 +
1.1538 +12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
1.1539 + provided in case anyone wants to the the POSIX interface with UTF-8
1.1540 + strings.
1.1541 +
1.1542 +13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
1.1543 + C++ linking (needed for some HP-UX environments).
1.1544 +
1.1545 +14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
1.1546 + (unused parameter) and in the pcre_printint() function (omitted "default"
1.1547 + switch label when the default is to do nothing).
1.1548 +
1.1549 +15. Added some code to make it possible, when PCRE is compiled as a C++
1.1550 + library, to replace subject pointers for pcre_exec() with a smart pointer
1.1551 + class, thus making it possible to process discontinuous strings.
1.1552 +
1.1553 +16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
1.1554 + much the same function. They were added by different people who were trying
1.1555 + to make PCRE easy to compile on non-Unix systems. It has been suggested
1.1556 + that PCRE_EXPORT be abolished now that there is more automatic apparatus
1.1557 + for compiling on Windows systems. I have therefore replaced it with
1.1558 + PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
1.1559 + defaults to "extern" for C or "extern C" for C++, which works fine on
1.1560 + Unix-like systems. It is now possible to override the value of PCRE_DATA_
1.1561 + SCOPE with something explicit in config.h. In addition:
1.1562 +
1.1563 + (a) pcreposix.h still had just "extern" instead of either of these macros;
1.1564 + I have replaced it with PCRE_DATA_SCOPE.
1.1565 +
1.1566 + (b) Functions such as _pcre_xclass(), which are internal to the library,
1.1567 + but external in the C sense, all had PCRE_EXPORT in their definitions.
1.1568 + This is apparently wrong for the Windows case, so I have removed it.
1.1569 + (It makes no difference on Unix-like systems.)
1.1570 +
1.1571 +17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
1.1572 + of recursive calls to match(). This is different to MATCH_LIMIT because
1.1573 + that limits the total number of calls to match(), not all of which increase
1.1574 + the depth of recursion. Limiting the recursion depth limits the amount of
1.1575 + stack (or heap if NO_RECURSE is set) that is used. The default can be set
1.1576 + when PCRE is compiled, and changed at run time. A patch from Google adds
1.1577 + this functionality to the C++ interface.
1.1578 +
1.1579 +18. Changes to the handling of Unicode character properties:
1.1580 +
1.1581 + (a) Updated the table to Unicode 4.1.0.
1.1582 +
1.1583 + (b) Recognize characters that are not in the table as "Cn" (undefined).
1.1584 +
1.1585 + (c) I revised the way the table is implemented to a much improved format
1.1586 + which includes recognition of ranges. It now supports the ranges that
1.1587 + are defined in UnicodeData.txt, and it also amalgamates other
1.1588 + characters into ranges. This has reduced the number of entries in the
1.1589 + table from around 16,000 to around 3,000, thus reducing its size
1.1590 + considerably. I realized I did not need to use a tree structure after
1.1591 + all - a binary chop search is just as efficient. Having reduced the
1.1592 + number of entries, I extended their size from 6 bytes to 8 bytes to
1.1593 + allow for more data.
1.1594 +
1.1595 + (d) Added support for Unicode script names via properties such as \p{Han}.
1.1596 +
1.1597 +19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
1.1598 + matching that character.
1.1599 +
1.1600 +20. When matching a repeated Unicode property with a minimum greater than zero,
1.1601 + (for example \pL{2,}), PCRE could look past the end of the subject if it
1.1602 + reached it while seeking the minimum number of characters. This could
1.1603 + happen only if some of the characters were more than one byte long, because
1.1604 + there is a check for at least the minimum number of bytes.
1.1605 +
1.1606 +21. Refactored the implementation of \p and \P so as to be more general, to
1.1607 + allow for more different types of property in future. This has changed the
1.1608 + compiled form incompatibly. Anybody with saved compiled patterns that use
1.1609 + \p or \P will have to recompile them.
1.1610 +
1.1611 +22. Added "Any" and "L&" to the supported property types.
1.1612 +
1.1613 +23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
1.1614 + but give a compile time error if the value is greater than 0xff.
1.1615 +
1.1616 +24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
1.1617 + accidentally not being installed or uninstalled.
1.1618 +
1.1619 +25. The pcre.h file was built from pcre.h.in, but the only changes that were
1.1620 + made were to insert the current release number. This seemed silly, because
1.1621 + it made things harder for people building PCRE on systems that don't run
1.1622 + "configure". I have turned pcre.h into a distributed file, no longer built
1.1623 + by "configure", with the version identification directly included. There is
1.1624 + no longer a pcre.h.in file.
1.1625 +
1.1626 + However, this change necessitated a change to the pcre-config script as
1.1627 + well. It is built from pcre-config.in, and one of the substitutions was the
1.1628 + release number. I have updated configure.ac so that ./configure now finds
1.1629 + the release number by grepping pcre.h.
1.1630 +
1.1631 +26. Added the ability to run the tests under valgrind.
1.1632 +
1.1633 +
1.1634 +Version 6.4 05-Sep-05
1.1635 +---------------------
1.1636 +
1.1637 + 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
1.1638 + "--" to be printed when multiple files were scanned, even when none of the
1.1639 + -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
1.1640 + consider it to be a bug, and have restored the previous behaviour.
1.1641 +
1.1642 + 2. A couple of code tidies to get rid of compiler warnings.
1.1643 +
1.1644 + 3. The pcretest program used to cheat by referring to symbols in the library
1.1645 + whose names begin with _pcre_. These are internal symbols that are not
1.1646 + really supposed to be visible externally, and in some environments it is
1.1647 + possible to suppress them. The cheating is now confined to including
1.1648 + certain files from the library's source, which is a bit cleaner.
1.1649 +
1.1650 + 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
1.1651 + file's purpose clearer.
1.1652 +
1.1653 + 5. Reorganized pcre_ucp_findchar().
1.1654 +
1.1655 +
1.1656 +Version 6.3 15-Aug-05
1.1657 +---------------------
1.1658 +
1.1659 + 1. The file libpcre.pc.in did not have general read permission in the tarball.
1.1660 +
1.1661 + 2. There were some problems when building without C++ support:
1.1662 +
1.1663 + (a) If C++ support was not built, "make install" and "make test" still
1.1664 + tried to test it.
1.1665 +
1.1666 + (b) There were problems when the value of CXX was explicitly set. Some
1.1667 + changes have been made to try to fix these, and ...
1.1668 +
1.1669 + (c) --disable-cpp can now be used to explicitly disable C++ support.
1.1670 +
1.1671 + (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
1.1672 + backslash in a target when C++ was disabled. This confuses some
1.1673 + versions of "make", apparently. Using an intermediate variable solves
1.1674 + this. (Same for CPP_LOBJ.)
1.1675 +
1.1676 + 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
1.1677 + (non-Windows) now includes $(CFLAGS) because these flags are sometimes
1.1678 + necessary on certain architectures.
1.1679 +
1.1680 + 4. Added a setting of -export-symbols-regex to the link command to remove
1.1681 + those symbols that are exported in the C sense, but actually are local
1.1682 + within the library, and not documented. Their names all begin with
1.1683 + "_pcre_". This is not a perfect job, because (a) we have to except some
1.1684 + symbols that pcretest ("illegally") uses, and (b) the facility isn't always
1.1685 + available (and never for static libraries). I have made a note to try to
1.1686 + find a way round (a) in the future.
1.1687 +
1.1688 +
1.1689 +Version 6.2 01-Aug-05
1.1690 +---------------------
1.1691 +
1.1692 + 1. There was no test for integer overflow of quantifier values. A construction
1.1693 + such as {1111111111111111} would give undefined results. What is worse, if
1.1694 + a minimum quantifier for a parenthesized subpattern overflowed and became
1.1695 + negative, the calculation of the memory size went wrong. This could have
1.1696 + led to memory overwriting.
1.1697 +
1.1698 + 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
1.1699 +
1.1700 + 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
1.1701 + operating environments where this matters.
1.1702 +
1.1703 + 4. Applied Giuseppe Maxia's patch to add additional features for controlling
1.1704 + PCRE options from within the C++ wrapper.
1.1705 +
1.1706 + 5. Named capturing subpatterns were not being correctly counted when a pattern
1.1707 + was compiled. This caused two problems: (a) If there were more than 100
1.1708 + such subpatterns, the calculation of the memory needed for the whole
1.1709 + compiled pattern went wrong, leading to an overflow error. (b) Numerical
1.1710 + back references of the form \12, where the number was greater than 9, were
1.1711 + not recognized as back references, even though there were sufficient
1.1712 + previous subpatterns.
1.1713 +
1.1714 + 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
1.1715 + versions of gcc, e.g. 2.95.4.
1.1716 +
1.1717 +
1.1718 +Version 6.1 21-Jun-05
1.1719 +---------------------
1.1720 +
1.1721 + 1. There was one reference to the variable "posix" in pcretest.c that was not
1.1722 + surrounded by "#if !defined NOPOSIX".
1.1723 +
1.1724 + 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
1.1725 + the cross-check on the old pcre_info() function, for the benefit of the
1.1726 + cut-down version of PCRE that is currently imported into Exim.
1.1727 +
1.1728 + 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
1.1729 + allocation error. I've done the easy fix, which wastes 2 bytes for sensible
1.1730 + patterns that start (?i) but I don't think that matters. The use of (?i) is
1.1731 + just an example; this all applies to the other options as well.
1.1732 +
1.1733 + 4. Since libtool seems to echo the compile commands it is issuing, the output
1.1734 + from "make" can be reduced a bit by putting "@" in front of each libtool
1.1735 + compile command.
1.1736 +
1.1737 + 5. Patch from the folks at Google for configure.in to be a bit more thorough
1.1738 + in checking for a suitable C++ installation before trying to compile the
1.1739 + C++ stuff. This should fix a reported problem when a compiler was present,
1.1740 + but no suitable headers.
1.1741 +
1.1742 + 6. The man pages all had just "PCRE" as their title. I have changed them to
1.1743 + be the relevant file name. I have also arranged that these names are
1.1744 + retained in the file doc/pcre.txt, which is a concatenation in text format
1.1745 + of all the man pages except the little individual ones for each function.
1.1746 +
1.1747 + 7. The NON-UNIX-USE file had not been updated for the different set of source
1.1748 + files that come with release 6. I also added a few comments about the C++
1.1749 + wrapper.
1.1750 +
1.1751 +
1.1752 +Version 6.0 07-Jun-05
1.1753 +---------------------
1.1754 +
1.1755 + 1. Some minor internal re-organization to help with my DFA experiments.
1.1756 +
1.1757 + 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
1.1758 + didn't matter for the library itself when fully configured, but did matter
1.1759 + when compiling without UCP support, or within Exim, where the ucp files are
1.1760 + not imported.
1.1761 +
1.1762 + 3. Refactoring of the library code to split up the various functions into
1.1763 + different source modules. The addition of the new DFA matching code (see
1.1764 + below) to a single monolithic source would have made it really too
1.1765 + unwieldy, quite apart from causing all the code to be include in a
1.1766 + statically linked application, when only some functions are used. This is
1.1767 + relevant even without the DFA addition now that patterns can be compiled in
1.1768 + one application and matched in another.
1.1769 +
1.1770 + The downside of splitting up is that there have to be some external
1.1771 + functions and data tables that are used internally in different modules of
1.1772 + the library but which are not part of the API. These have all had their
1.1773 + names changed to start with "_pcre_" so that they are unlikely to clash
1.1774 + with other external names.
1.1775 +
1.1776 + 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
1.1777 + a different (DFA) algorithm. Although it is slower than the original
1.1778 + function, it does have some advantages for certain types of matching
1.1779 + problem.
1.1780 +
1.1781 + 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
1.1782 + including restarting after a partial match.
1.1783 +
1.1784 + 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
1.1785 + defined when compiling for Windows was sent to me. I have put it into the
1.1786 + code, though I have no means of testing or verifying it.
1.1787 +
1.1788 + 7. Added the pcre_refcount() auxiliary function.
1.1789 +
1.1790 + 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
1.1791 + match before or at the first newline in the subject string. In pcretest,
1.1792 + the /f option on a pattern can be used to set this.
1.1793 +
1.1794 + 9. A repeated \w when used in UTF-8 mode with characters greater than 256
1.1795 + would behave wrongly. This has been present in PCRE since release 4.0.
1.1796 +
1.1797 +10. A number of changes to the pcregrep command:
1.1798 +
1.1799 + (a) Refactored how -x works; insert ^(...)$ instead of setting
1.1800 + PCRE_ANCHORED and checking the length, in preparation for adding
1.1801 + something similar for -w.
1.1802 +
1.1803 + (b) Added the -w (match as a word) option.
1.1804 +
1.1805 + (c) Refactored the way lines are read and buffered so as to have more
1.1806 + than one at a time available.
1.1807 +
1.1808 + (d) Implemented a pcregrep test script.
1.1809 +
1.1810 + (e) Added the -M (multiline match) option. This allows patterns to match
1.1811 + over several lines of the subject. The buffering ensures that at least
1.1812 + 8K, or the rest of the document (whichever is the shorter) is available
1.1813 + for matching (and similarly the previous 8K for lookbehind assertions).
1.1814 +
1.1815 + (f) Changed the --help output so that it now says
1.1816 +
1.1817 + -w, --word-regex(p)
1.1818 +
1.1819 + instead of two lines, one with "regex" and the other with "regexp"
1.1820 + because that confused at least one person since the short forms are the
1.1821 + same. (This required a bit of code, as the output is generated
1.1822 + automatically from a table. It wasn't just a text change.)
1.1823 +
1.1824 + (g) -- can be used to terminate pcregrep options if the next thing isn't an
1.1825 + option but starts with a hyphen. Could be a pattern or a path name
1.1826 + starting with a hyphen, for instance.
1.1827 +
1.1828 + (h) "-" can be given as a file name to represent stdin.
1.1829 +
1.1830 + (i) When file names are being printed, "(standard input)" is used for
1.1831 + the standard input, for compatibility with GNU grep. Previously
1.1832 + "<stdin>" was used.
1.1833 +
1.1834 + (j) The option --label=xxx can be used to supply a name to be used for
1.1835 + stdin when file names are being printed. There is no short form.
1.1836 +
1.1837 + (k) Re-factored the options decoding logic because we are going to add
1.1838 + two more options that take data. Such options can now be given in four
1.1839 + different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
1.1840 +
1.1841 + (l) Added the -A, -B, and -C options for requesting that lines of context
1.1842 + around matches be printed.
1.1843 +
1.1844 + (m) Added the -L option to print the names of files that do not contain
1.1845 + any matching lines, that is, the complement of -l.
1.1846 +
1.1847 + (n) The return code is 2 if any file cannot be opened, but pcregrep does
1.1848 + continue to scan other files.
1.1849 +
1.1850 + (o) The -s option was incorrectly implemented. For compatibility with other
1.1851 + greps, it now suppresses the error message for a non-existent or non-
1.1852 + accessible file (but not the return code). There is a new option called
1.1853 + -q that suppresses the output of matching lines, which was what -s was
1.1854 + previously doing.
1.1855 +
1.1856 + (p) Added --include and --exclude options to specify files for inclusion
1.1857 + and exclusion when recursing.
1.1858 +
1.1859 +11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
1.1860 + Hopefully, it now does.
1.1861 +
1.1862 +12. Missing cast in pcre_study().
1.1863 +
1.1864 +13. Added an "uninstall" target to the makefile.
1.1865 +
1.1866 +14. Replaced "extern" in the function prototypes in Makefile.in with
1.1867 + "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
1.1868 + world, but is set differently for Windows.
1.1869 +
1.1870 +15. Added a second compiling function called pcre_compile2(). The only
1.1871 + difference is that it has an extra argument, which is a pointer to an
1.1872 + integer error code. When there is a compile-time failure, this is set
1.1873 + non-zero, in addition to the error test pointer being set to point to an
1.1874 + error message. The new argument may be NULL if no error number is required
1.1875 + (but then you may as well call pcre_compile(), which is now just a
1.1876 + wrapper). This facility is provided because some applications need a
1.1877 + numeric error indication, but it has also enabled me to tidy up the way
1.1878 + compile-time errors are handled in the POSIX wrapper.
1.1879 +
1.1880 +16. Added VPATH=.libs to the makefile; this should help when building with one
1.1881 + prefix path and installing with another. (Or so I'm told by someone who
1.1882 + knows more about this stuff than I do.)
1.1883 +
1.1884 +17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
1.1885 + passes PCRE_DOTALL to the pcre_compile() function, making the "." character
1.1886 + match everything, including newlines. This is not POSIX-compatible, but
1.1887 + somebody wanted the feature. From pcretest it can be activated by using
1.1888 + both the P and the s flags.
1.1889 +
1.1890 +18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
1.1891 +
1.1892 +19. libpcre.pc was being incorrectly installed as executable.
1.1893 +
1.1894 +20. A couple of places in pcretest check for end-of-line by looking for '\n';
1.1895 + it now also looks for '\r' so that it will work unmodified on Windows.
1.1896 +
1.1897 +21. Added Google's contributed C++ wrapper to the distribution.
1.1898 +
1.1899 +22. Added some untidy missing memory free() calls in pcretest, to keep
1.1900 + Electric Fence happy when testing.
1.1901 +
1.1902 +
1.1903 +
1.1904 +Version 5.0 13-Sep-04
1.1905 +---------------------
1.1906 +
1.1907 + 1. Internal change: literal characters are no longer packed up into items
1.1908 + containing multiple characters in a single byte-string. Each character
1.1909 + is now matched using a separate opcode. However, there may be more than one
1.1910 + byte in the character in UTF-8 mode.
1.1911 +
1.1912 + 2. The pcre_callout_block structure has two new fields: pattern_position and
1.1913 + next_item_length. These contain the offset in the pattern to the next match
1.1914 + item, and its length, respectively.
1.1915 +
1.1916 + 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
1.1917 + insertion of callouts before each pattern item. Added the /C option to
1.1918 + pcretest to make use of this.
1.1919 +
1.1920 + 4. On the advice of a Windows user, the lines
1.1921 +
1.1922 + #if defined(_WIN32) || defined(WIN32)
1.1923 + _setmode( _fileno( stdout ), 0x8000 );
1.1924 + #endif /* defined(_WIN32) || defined(WIN32) */
1.1925 +
1.1926 + have been added to the source of pcretest. This apparently does useful
1.1927 + magic in relation to line terminators.
1.1928 +
1.1929 + 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
1.1930 + for the benefit of those environments where the "b" makes a difference.
1.1931 +
1.1932 + 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
1.1933 + to know about it. I have put a hack into configure.in that adds in code
1.1934 + to set GCC=yes if CC=icc. This seems to end up at a point in the
1.1935 + generated configure script that is early enough to affect the setting of
1.1936 + compiler options, which is what is needed, but I have no means of testing
1.1937 + whether it really works. (The user who reported this had patched the
1.1938 + generated configure script, which of course I cannot do.)
1.1939 +
1.1940 + LATER: After change 22 below (new libtool files), the configure script
1.1941 + seems to know about icc (and also ecc). Therefore, I have commented out
1.1942 + this hack in configure.in.
1.1943 +
1.1944 + 7. Added support for pkg-config (2 patches were sent in).
1.1945 +
1.1946 + 8. Negated POSIX character classes that used a combination of internal tables
1.1947 + were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
1.1948 + [[:^ascii]]. Typically, they would match almost any characters. The other
1.1949 + POSIX classes were not broken in this way.
1.1950 +
1.1951 + 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
1.1952 + to find the match, as PCRE was deluded into thinking that the match had to
1.1953 + start at the start point or following a newline. The same bug applied to
1.1954 + patterns with negative forward assertions or any backward assertions
1.1955 + preceding ".*" at the start, unless the pattern required a fixed first
1.1956 + character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
1.1957 +
1.1958 +10. In UTF-8 mode, when moving forwards in the subject after a failed match
1.1959 + starting at the last subject character, bytes beyond the end of the subject
1.1960 + string were read.
1.1961 +
1.1962 +11. Renamed the variable "class" as "classbits" to make life easier for C++
1.1963 + users. (Previously there was a macro definition, but it apparently wasn't
1.1964 + enough.)
1.1965 +
1.1966 +12. Added the new field "tables" to the extra data so that tables can be passed
1.1967 + in at exec time, or the internal tables can be re-selected. This allows
1.1968 + a compiled regex to be saved and re-used at a later time by a different
1.1969 + program that might have everything at different addresses.
1.1970 +
1.1971 +13. Modified the pcre-config script so that, when run on Solaris, it shows a
1.1972 + -R library as well as a -L library.
1.1973 +
1.1974 +14. The debugging options of pcretest (-d on the command line or D on a
1.1975 + pattern) showed incorrect output for anything following an extended class
1.1976 + that contained multibyte characters and which was followed by a quantifier.
1.1977 +
1.1978 +15. Added optional support for general category Unicode character properties
1.1979 + via the \p, \P, and \X escapes. Unicode property support implies UTF-8
1.1980 + support. It adds about 90K to the size of the library. The meanings of the
1.1981 + inbuilt class escapes such as \d and \s have NOT been changed.
1.1982 +
1.1983 +16. Updated pcredemo.c to include calls to free() to release the memory for the
1.1984 + compiled pattern.
1.1985 +
1.1986 +17. The generated file chartables.c was being created in the source directory
1.1987 + instead of in the building directory. This caused the build to fail if the
1.1988 + source directory was different from the building directory, and was
1.1989 + read-only.
1.1990 +
1.1991 +18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
1.1992 + file. No doubt somebody will tell me if they don't make sense... Also added
1.1993 + Dan Mooney's comments about building on OpenVMS.
1.1994 +
1.1995 +19. Added support for partial matching via the PCRE_PARTIAL option for
1.1996 + pcre_exec() and the \P data escape in pcretest.
1.1997 +
1.1998 +20. Extended pcretest with 3 new pattern features:
1.1999 +
1.2000 + (i) A pattern option of the form ">rest-of-line" causes pcretest to
1.2001 + write the compiled pattern to the file whose name is "rest-of-line".
1.2002 + This is a straight binary dump of the data, with the saved pointer to
1.2003 + the character tables forced to be NULL. The study data, if any, is
1.2004 + written too. After writing, pcretest reads a new pattern.
1.2005 +
1.2006 + (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
1.2007 + compiled pattern from the given file. There must not be any
1.2008 + occurrences of "<" in the file name (pretty unlikely); if there are,
1.2009 + pcretest will instead treat the initial "<" as a pattern delimiter.
1.2010 + After reading in the pattern, pcretest goes on to read data lines as
1.2011 + usual.
1.2012 +
1.2013 + (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
1.2014 + and 16-bit fields in a compiled pattern, to simulate a pattern that
1.2015 + was compiled on a host of opposite endianness.
1.2016 +
1.2017 +21. The pcre-exec() function can now cope with patterns that were compiled on
1.2018 + hosts of opposite endianness, with this restriction:
1.2019 +
1.2020 + As for any compiled expression that is saved and used later, the tables
1.2021 + pointer field cannot be preserved; the extra_data field in the arguments
1.2022 + to pcre_exec() should be used to pass in a tables address if a value
1.2023 + other than the default internal tables were used at compile time.
1.2024 +
1.2025 +22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
1.2026 + now diagnosed as an error. Previously, most of the time, a negative number
1.2027 + would have been treated as zero, but if in addition "ovector" was passed as
1.2028 + NULL, a crash could occur.
1.2029 +
1.2030 +23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
1.2031 + new versions from the libtool 1.5 distribution (the last one is a copy of
1.2032 + a file called libtool.m4). This seems to have fixed the need to patch
1.2033 + "configure" to support Darwin 1.3 (which I used to do). However, I still
1.2034 + had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
1.2035 + workstation).
1.2036 +
1.2037 +24. Changed the PCRE licence to be the more standard "BSD" licence.
1.2038 +
1.2039 +
1.2040 +Version 4.5 01-Dec-03
1.2041 +---------------------
1.2042 +
1.2043 + 1. There has been some re-arrangement of the code for the match() function so
1.2044 + that it can be compiled in a version that does not call itself recursively.
1.2045 + Instead, it keeps those local variables that need separate instances for
1.2046 + each "recursion" in a frame on the heap, and gets/frees frames whenever it
1.2047 + needs to "recurse". Keeping track of where control must go is done by means
1.2048 + of setjmp/longjmp. The whole thing is implemented by a set of macros that
1.2049 + hide most of the details from the main code, and operates only if
1.2050 + NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
1.2051 + "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
1.2052 + operating.
1.2053 +
1.2054 + To make it easier for callers to provide specially tailored get/free
1.2055 + functions for this usage, two new functions, pcre_stack_malloc, and
1.2056 + pcre_stack_free, are used. They are always called in strict stacking order,
1.2057 + and the size of block requested is always the same.
1.2058 +
1.2059 + The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
1.2060 + PCRE has been compiled to use the stack or the heap for recursion. The
1.2061 + -C option of pcretest uses this to show which version is compiled.
1.2062 +
1.2063 + A new data escape \S, is added to pcretest; it causes the amounts of store
1.2064 + obtained and freed by both kinds of malloc/free at match time to be added
1.2065 + to the output.
1.2066 +
1.2067 + 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
1.2068 + what's available on my current Linux desktop machine.
1.2069 +
1.2070 + 3. When matching a UTF-8 string, the test for a valid string at the start has
1.2071 + been extended. If start_offset is not zero, PCRE now checks that it points
1.2072 + to a byte that is the start of a UTF-8 character. If not, it returns
1.2073 + PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
1.2074 + this is necessary because there may be backward assertions in the pattern.
1.2075 + When matching the same subject several times, it may save resources to use
1.2076 + PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
1.2077 +
1.2078 + 4. The code for checking the validity of UTF-8 strings has been tightened so
1.2079 + that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
1.2080 + containing "overlong sequences".
1.2081 +
1.2082 + 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
1.2083 + I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
1.2084 + should have been "&", but it just so happened that all the cases this let
1.2085 + through by mistake were picked up later in the function.
1.2086 +
1.2087 + 6. I had used a variable called "isblank" - this is a C99 function, causing
1.2088 + some compilers to warn. To avoid this, I renamed it (as "blankclass").
1.2089 +
1.2090 + 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
1.2091 + prompting; (b) run "./pcretest /dev/null" at the start of the test script
1.2092 + so the version is shown; (c) stop "make test" echoing "./RunTest".
1.2093 +
1.2094 + 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
1.2095 +
1.2096 + 9. The prototype for memmove() for systems that don't have it was using
1.2097 + size_t, but the inclusion of the header that defines size_t was later. I've
1.2098 + moved the #includes for the C headers earlier to avoid this.
1.2099 +
1.2100 +10. Added some adjustments to the code to make it easier to compiler on certain
1.2101 + special systems:
1.2102 +
1.2103 + (a) Some "const" qualifiers were missing.
1.2104 + (b) Added the macro EXPORT before all exported functions; by default this
1.2105 + is defined to be empty.
1.2106 + (c) Changed the dftables auxiliary program (that builds chartables.c) so
1.2107 + that it reads its output file name as an argument instead of writing
1.2108 + to the standard output and assuming this can be redirected.
1.2109 +
1.2110 +11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
1.2111 + class containing characters with values greater than 255, PCRE compilation
1.2112 + went into a loop.
1.2113 +
1.2114 +12. A recursive reference to a subpattern that was within another subpattern
1.2115 + that had a minimum quantifier of zero caused PCRE to crash. For example,
1.2116 + (x(y(?2))z)? provoked this bug with a subject that got as far as the
1.2117 + recursion. If the recursively-called subpattern itself had a zero repeat,
1.2118 + that was OK.
1.2119 +
1.2120 +13. In pcretest, the buffer for reading a data line was set at 30K, but the
1.2121 + buffer into which it was copied (for escape processing) was still set at
1.2122 + 1024, so long lines caused crashes.
1.2123 +
1.2124 +14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
1.2125 + "internal error: code overflow...". This applied to any character class
1.2126 + that was followed by a possessive quantifier.
1.2127 +
1.2128 +15. Modified the Makefile to add libpcre.la as a prerequisite for
1.2129 + libpcreposix.la because I was told this is needed for a parallel build to
1.2130 + work.
1.2131 +
1.2132 +16. If a pattern that contained .* following optional items at the start was
1.2133 + studied, the wrong optimizing data was generated, leading to matching
1.2134 + errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
1.2135 + matching string must start with a or b or c. The correct conclusion for
1.2136 + this pattern is that a match can start with any character.
1.2137 +
1.2138 +
1.2139 +Version 4.4 13-Aug-03
1.2140 +---------------------
1.2141 +
1.2142 + 1. In UTF-8 mode, a character class containing characters with values between
1.2143 + 127 and 255 was not handled correctly if the compiled pattern was studied.
1.2144 + In fixing this, I have also improved the studying algorithm for such
1.2145 + classes (slightly).
1.2146 +
1.2147 + 2. Three internal functions had redundant arguments passed to them. Removal
1.2148 + might give a very teeny performance improvement.
1.2149 +
1.2150 + 3. Documentation bug: the value of the capture_top field in a callout is *one
1.2151 + more than* the number of the hightest numbered captured substring.
1.2152 +
1.2153 + 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
1.2154 + in incorrectly linking with a previously installed version. They now link
1.2155 + explicitly with libpcre.la.
1.2156 +
1.2157 + 5. configure.in no longer needs to recognize Cygwin specially.
1.2158 +
1.2159 + 6. A problem in pcre.in for Windows platforms is fixed.
1.2160 +
1.2161 + 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
1.2162 + pcretest, it used to include the size of the study block as part of its
1.2163 + output. Unfortunately, the structure contains a field that has a different
1.2164 + size on different hardware architectures. This meant that the tests that
1.2165 + showed this size failed. As the block is currently always of a fixed size,
1.2166 + this information isn't actually particularly useful in pcretest output, so
1.2167 + I have just removed it.
1.2168 +
1.2169 + 8. Three pre-processor statements accidentally did not start in column 1.
1.2170 + Sadly, there are *still* compilers around that complain, even though
1.2171 + standard C has not required this for well over a decade. Sigh.
1.2172 +
1.2173 + 9. In pcretest, the code for checking callouts passed small integers in the
1.2174 + callout_data field, which is a void * field. However, some picky compilers
1.2175 + complained about the casts involved for this on 64-bit systems. Now
1.2176 + pcretest passes the address of the small integer instead, which should get
1.2177 + rid of the warnings.
1.2178 +
1.2179 +10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
1.2180 + both compile and run time, and gives an error if an invalid UTF-8 sequence
1.2181 + is found. There is a option for disabling this check in cases where the
1.2182 + string is known to be correct and/or the maximum performance is wanted.
1.2183 +
1.2184 +11. In response to a bug report, I changed one line in Makefile.in from
1.2185 +
1.2186 + -Wl,--out-implib,.libs/lib@[email protected] \
1.2187 + to
1.2188 + -Wl,--out-implib,.libs/@[email protected] \
1.2189 +
1.2190 + to look similar to other lines, but I have no way of telling whether this
1.2191 + is the right thing to do, as I do not use Windows. No doubt I'll get told
1.2192 + if it's wrong...
1.2193 +
1.2194 +
1.2195 +Version 4.3 21-May-03
1.2196 +---------------------
1.2197 +
1.2198 +1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
1.2199 + Makefile.
1.2200 +
1.2201 +2. Some refactoring to improve the quality of the code:
1.2202 +
1.2203 + (i) The utf8_table... variables are now declared "const".
1.2204 +
1.2205 + (ii) The code for \cx, which used the "case flipping" table to upper case
1.2206 + lower case letters, now just substracts 32. This is ASCII-specific,
1.2207 + but the whole concept of \cx is ASCII-specific, so it seems
1.2208 + reasonable.
1.2209 +
1.2210 + (iii) PCRE was using its character types table to recognize decimal and
1.2211 + hexadecimal digits in the pattern. This is silly, because it handles
1.2212 + only 0-9, a-f, and A-F, but the character types table is locale-
1.2213 + specific, which means strange things might happen. A private
1.2214 + table is now used for this - though it costs 256 bytes, a table is
1.2215 + much faster than multiple explicit tests. Of course, the standard
1.2216 + character types table is still used for matching digits in subject
1.2217 + strings against \d.
1.2218 +
1.2219 + (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
1.2220 + ending in _t are). So I've renamed it as ESC_tee.
1.2221 +
1.2222 +3. The first argument for regexec() in the POSIX wrapper should have been
1.2223 + defined as "const".
1.2224 +
1.2225 +4. Changed pcretest to use malloc() for its buffers so that they can be
1.2226 + Electric Fenced for debugging.
1.2227 +
1.2228 +5. There were several places in the code where, in UTF-8 mode, PCRE would try
1.2229 + to read one or more bytes before the start of the subject string. Often this
1.2230 + had no effect on PCRE's behaviour, but in some circumstances it could
1.2231 + provoke a segmentation fault.
1.2232 +
1.2233 +6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
1.2234 + to try to read one or more bytes before the start of the subject string.
1.2235 +
1.2236 +7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
1.2237 + UTF-8 support could misbehave in various ways if the subject string
1.2238 + contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
1.2239 + area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
1.2240 + back over UTF-8 characters.)
1.2241 +
1.2242 +
1.2243 +Version 4.2 14-Apr-03
1.2244 +---------------------
1.2245 +
1.2246 +1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
1.2247 +
1.2248 +2. Changes to the building process, supplied by Ronald Landheer-Cieslak
1.2249 + [ON_WINDOWS]: new variable, "#" on non-Windows platforms
1.2250 + [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
1.2251 + [WIN_PREFIX]: new variable, "cyg" for Cygwin
1.2252 + * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
1.2253 + and BUILD_EXEEXT
1.2254 + Note: automatic setting of the BUILD variables is not yet working
1.2255 + set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
1.2256 + compile-time but not at link-time
1.2257 + [LINK]: use for linking executables only
1.2258 + make different versions for Windows and non-Windows
1.2259 + [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
1.2260 + libraries
1.2261 + [LINK_FOR_BUILD]: new variable
1.2262 + [OBJEXT]: use throughout
1.2263 + [EXEEXT]: use throughout
1.2264 + <winshared>: new target
1.2265 + <wininstall>: new target
1.2266 + <dftables.o>: use native compiler
1.2267 + <dftables>: use native linker
1.2268 + <install>: handle Windows platform correctly
1.2269 + <clean>: ditto
1.2270 + <check>: ditto
1.2271 + copy DLL to top builddir before testing
1.2272 +
1.2273 + As part of these changes, -no-undefined was removed again. This was reported
1.2274 + to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
1.2275 + in any case.
1.2276 +
1.2277 +3. Some tidies to get rid of compiler warnings:
1.2278 +
1.2279 + . In the match_data structure, match_limit was an unsigned long int, whereas
1.2280 + match_call_count was an int. I've made them both unsigned long ints.
1.2281 +
1.2282 + . In pcretest the fact that a const uschar * doesn't automatically cast to
1.2283 + a void * provoked a warning.
1.2284 +
1.2285 + . Turning on some more compiler warnings threw up some "shadow" variables
1.2286 + and a few more missing casts.
1.2287 +
1.2288 +4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
1.2289 + option, a class that contained a single character with a value between 128
1.2290 + and 255 (e.g. /[\xFF]/) caused PCRE to crash.
1.2291 +
1.2292 +5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
1.2293 + option, a class that contained several characters, but with at least one
1.2294 + whose value was between 128 and 255 caused PCRE to crash.
1.2295 +
1.2296 +
1.2297 +Version 4.1 12-Mar-03
1.2298 +---------------------
1.2299 +
1.2300 +1. Compiling with gcc -pedantic found a couple of places where casts were
1.2301 +needed, and a string in dftables.c that was longer than standard compilers are
1.2302 +required to support.
1.2303 +
1.2304 +2. Compiling with Sun's compiler found a few more places where the code could
1.2305 +be tidied up in order to avoid warnings.
1.2306 +
1.2307 +3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
1.2308 +first of these names is deprecated in the latest Autoconf in favour of the name
1.2309 +CC_FOR_BUILD, because "host" is typically used to mean the system on which the
1.2310 +compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
1.2311 +analogy I have changed it to CFLAGS_FOR_BUILD.
1.2312 +
1.2313 +4. Added -no-undefined to the linking command in the Makefile, because this is
1.2314 +apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
1.2315 +linking step for the pcreposix library.
1.2316 +
1.2317 +5. PCRE was failing to diagnose the case of two named groups with the same
1.2318 +name.
1.2319 +
1.2320 +6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
1.2321 +literal character that is needed in the subject for a match, and scans along to
1.2322 +ensure that it is present before embarking on the full matching process. This
1.2323 +saves time in cases of nested unlimited repeats that are never going to match.
1.2324 +Problem: the scan can take a lot of time if the subject is very long (e.g.
1.2325 +megabytes), thus penalizing straightforward matches. It is now done only if the
1.2326 +amount of subject to be scanned is less than 1000 bytes.
1.2327 +
1.2328 +7. A lesser problem with the same optimization is that it was recording the
1.2329 +first character of an anchored pattern as "needed", thus provoking a search
1.2330 +right along the subject, even when the first match of the pattern was going to
1.2331 +fail. The "needed" character is now not set for anchored patterns, unless it
1.2332 +follows something in the pattern that is of non-fixed length. Thus, it still
1.2333 +fulfils its original purpose of finding quick non-matches in cases of nested
1.2334 +unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
1.2335 +
1.2336 +
1.2337 +Version 4.0 17-Feb-03
1.2338 +---------------------
1.2339 +
1.2340 +1. If a comment in an extended regex that started immediately after a meta-item
1.2341 +extended to the end of string, PCRE compiled incorrect data. This could lead to
1.2342 +all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
1.2343 +
1.2344 +2. Moved to autoconf 2.53 and libtool 1.4.2.
1.2345 +
1.2346 +3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
1.2347 +the special perltest8 script is no longer needed - all the tests can be run
1.2348 +from a single perltest script.
1.2349 +
1.2350 +4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
1.2351 +by \s. It has now been removed in PCRE. This means it isn't recognized as
1.2352 +whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
1.2353 +class [:space:] *does* include VT, thereby creating a mess.
1.2354 +
1.2355 +5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
1.2356 +space and tab.
1.2357 +
1.2358 +6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
1.2359 +its new features into the main test script, reducing the number of scripts.
1.2360 +
1.2361 +7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
1.2362 +were backward compatible, and made the (?i) apply to the whole pattern, as if
1.2363 +/i were given. Now it behaves more logically, and applies the option setting
1.2364 +only to what follows. PCRE has been changed to follow suit. However, if it
1.2365 +finds options settings right at the start of the pattern, it extracts them into
1.2366 +the global options, as before. Thus, they show up in the info data.
1.2367 +
1.2368 +8. Added support for the \Q...\E escape sequence. Characters in between are
1.2369 +treated as literals. This is slightly different from Perl in that $ and @ are
1.2370 +also handled as literals inside the quotes. In Perl, they will cause variable
1.2371 +interpolation. Note the following examples:
1.2372 +
1.2373 + Pattern PCRE matches Perl matches
1.2374 +
1.2375 + \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
1.2376 + \Qabc\$xyz\E abc\$xyz abc\$xyz
1.2377 + \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
1.2378 +
1.2379 +For compatibility with Perl, \Q...\E sequences are recognized inside character
1.2380 +classes as well as outside them.
1.2381 +
1.2382 +9. Re-organized 3 code statements in pcretest to avoid "overflow in
1.2383 +floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
1.2384 +(size_t) cast to one statement in pcretest and one in pcreposix to avoid
1.2385 +signed/unsigned warnings.
1.2386 +
1.2387 +10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
1.2388 +option for pcretest, so I've replaced it by a simple function that does just
1.2389 +that job.
1.2390 +
1.2391 +11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
1.2392 +"pcregrep -".
1.2393 +
1.2394 +12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
1.2395 +Java package. This provides some syntactic sugar for simple cases of what my
1.2396 +documentation calls "once-only subpatterns". A pattern such as x*+ is the same
1.2397 +as (?>x*). In other words, if what is inside (?>...) is just a single repeated
1.2398 +item, you can use this simplified notation. Note that only makes sense with
1.2399 +greedy quantifiers. Consequently, the use of the possessive quantifier forces
1.2400 +greediness, whatever the setting of the PCRE_UNGREEDY option.
1.2401 +
1.2402 +13. A change of greediness default within a pattern was not taking effect at
1.2403 +the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
1.2404 +subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
1.2405 +was abstracted outside.
1.2406 +
1.2407 +14. PCRE now supports the \G assertion. It is true when the current matching
1.2408 +position is at the start point of the match. This differs from \A when the
1.2409 +starting offset is non-zero. Used with the /g option of pcretest (or similar
1.2410 +code), it works in the same way as it does for Perl's /g option. If all
1.2411 +alternatives of a regex begin with \G, the expression is anchored to the start
1.2412 +match position, and the "anchored" flag is set in the compiled expression.
1.2413 +
1.2414 +15. Some bugs concerning the handling of certain option changes within patterns
1.2415 +have been fixed. These applied to options other than (?ims). For example,
1.2416 +"a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
1.2417 +been the other way round. Some of this was related to change 7 above.
1.2418 +
1.2419 +16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
1.2420 +features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
1.2421 +and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
1.2422 +POSIX classes only within a class (e.g. /[[:alpha:]]/).
1.2423 +
1.2424 +17. Added support for Perl's \C escape. This matches one byte, even in UTF8
1.2425 +mode. Unlike ".", it always matches newline, whatever the setting of
1.2426 +PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
1.2427 +assertions. Perl allows it, but it doesn't (in general) work because it can't
1.2428 +calculate the length of the lookbehind. At least, that's the case for Perl
1.2429 +5.8.0 - I've been told they are going to document that it doesn't work in
1.2430 +future.
1.2431 +
1.2432 +18. Added an error diagnosis for escapes that PCRE does not support: these are
1.2433 +\L, \l, \N, \P, \p, \U, \u, and \X.
1.2434 +
1.2435 +19. Although correctly diagnosing a missing ']' in a character class, PCRE was
1.2436 +reading past the end of the pattern in cases such as /[abcd/.
1.2437 +
1.2438 +20. PCRE was getting more memory than necessary for patterns with classes that
1.2439 +contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
1.2440 +
1.2441 +21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
1.2442 +compiling PCRE for use with Virtual Pascal.
1.2443 +
1.2444 +22. Small fix to the Makefile to make it work properly if the build is done
1.2445 +outside the source tree.
1.2446 +
1.2447 +23. Added a new extension: a condition to go with recursion. If a conditional
1.2448 +subpattern starts with (?(R) the "true" branch is used if recursion has
1.2449 +happened, whereas the "false" branch is used only at the top level.
1.2450 +
1.2451 +24. When there was a very long string of literal characters (over 255 bytes
1.2452 +without UTF support, over 250 bytes with UTF support), the computation of how
1.2453 +much memory was required could be incorrect, leading to segfaults or other
1.2454 +strange effects.
1.2455 +
1.2456 +25. PCRE was incorrectly assuming anchoring (either to start of subject or to
1.2457 +start of line for a non-DOTALL pattern) when a pattern started with (.*) and
1.2458 +there was a subsequent back reference to those brackets. This meant that, for
1.2459 +example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
1.2460 +possible to check for precisely this case. All we can do is abandon the
1.2461 +optimization if .* occurs inside capturing brackets when there are any back
1.2462 +references whatsoever. (See below for a better fix that came later.)
1.2463 +
1.2464 +26. The handling of the optimization for finding the first character of a
1.2465 +non-anchored pattern, and for finding a character that is required later in the
1.2466 +match were failing in some cases. This didn't break the matching; it just
1.2467 +failed to optimize when it could. The way this is done has been re-implemented.
1.2468 +
1.2469 +27. Fixed typo in error message for invalid (?R item (it said "(?p").
1.2470 +
1.2471 +28. Added a new feature that provides some of the functionality that Perl
1.2472 +provides with (?{...}). The facility is termed a "callout". The way it is done
1.2473 +in PCRE is for the caller to provide an optional function, by setting
1.2474 +pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
1.2475 +global variable. By default it is unset, which disables all calling out. To get
1.2476 +the function called, the regex must include (?C) at appropriate points. This
1.2477 +is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
1.2478 +This provides a means of identifying different callout points. When PCRE
1.2479 +reaches such a point in the regex, if pcre_callout has been set, the external
1.2480 +function is called. It is provided with data in a structure called
1.2481 +pcre_callout_block, which is defined in pcre.h. If the function returns 0,
1.2482 +matching continues; if it returns a non-zero value, the match at the current
1.2483 +point fails. However, backtracking will occur if possible. [This was changed
1.2484 +later and other features added - see item 49 below.]
1.2485 +
1.2486 +29. pcretest is upgraded to test the callout functionality. It provides a
1.2487 +callout function that displays information. By default, it shows the start of
1.2488 +the match and the current position in the text. There are some new data escapes
1.2489 +to vary what happens:
1.2490 +
1.2491 + \C+ in addition, show current contents of captured substrings
1.2492 + \C- do not supply a callout function
1.2493 + \C!n return 1 when callout number n is reached
1.2494 + \C!n!m return 1 when callout number n is reached for the mth time
1.2495 +
1.2496 +30. If pcregrep was called with the -l option and just a single file name, it
1.2497 +output "<stdin>" if a match was found, instead of the file name.
1.2498 +
1.2499 +31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
1.2500 +slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
1.2501 +pcre_exec(). This saves a malloc/free per call. The default value of
1.2502 +POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
1.2503 +when configuring.
1.2504 +
1.2505 +32. The default maximum size of a compiled pattern is 64K. There have been a
1.2506 +few cases of people hitting this limit. The code now uses macros to handle the
1.2507 +storing of links as offsets within the compiled pattern. It defaults to 2-byte
1.2508 +links, but this can be changed to 3 or 4 bytes by --with-link-size when
1.2509 +configuring. Tests 2 and 5 work only with 2-byte links because they output
1.2510 +debugging information about compiled patterns.
1.2511 +
1.2512 +33. Internal code re-arrangements:
1.2513 +
1.2514 +(a) Moved the debugging function for printing out a compiled regex into
1.2515 + its own source file (printint.c) and used #include to pull it into
1.2516 + pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
1.2517 + separate copies.
1.2518 +
1.2519 +(b) Defined the list of op-code names for debugging as a macro in
1.2520 + internal.h so that it is next to the definition of the opcodes.
1.2521 +
1.2522 +(c) Defined a table of op-code lengths for simpler skipping along compiled
1.2523 + code. This is again a macro in internal.h so that it is next to the
1.2524 + definition of the opcodes.
1.2525 +
1.2526 +34. Added support for recursive calls to individual subpatterns, along the
1.2527 +lines of Robin Houston's patch (but implemented somewhat differently).
1.2528 +
1.2529 +35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
1.2530 +allow it to read and process whole directories in Win32. This code was
1.2531 +contributed by Lionel Fourquaux; it has not been tested by me.
1.2532 +
1.2533 +36. Added support for named subpatterns. The Python syntax (?P<name>...) is
1.2534 +used to name a group. Names consist of alphanumerics and underscores, and must
1.2535 +be unique. Back references use the syntax (?P=name) and recursive calls use
1.2536 +(?P>name) which is a PCRE extension to the Python extension. Groups still have
1.2537 +numbers. The function pcre_fullinfo() can be used after compilation to extract
1.2538 +a name/number map. There are three relevant calls:
1.2539 +
1.2540 + PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
1.2541 + PCRE_INFO_NAMECOUNT yields the number of entries
1.2542 + PCRE_INFO_NAMETABLE yields a pointer to the map.
1.2543 +
1.2544 +The map is a vector of fixed-size entries. The size of each entry depends on
1.2545 +the length of the longest name used. The first two bytes of each entry are the
1.2546 +group number, most significant byte first. There follows the corresponding
1.2547 +name, zero terminated. The names are in alphabetical order.
1.2548 +
1.2549 +37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
1.2550 +case instead of 255. Making it the same both with and without UTF-8 support
1.2551 +means that the same test output works with both.
1.2552 +
1.2553 +38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
1.2554 +calling malloc() with a zero argument.
1.2555 +
1.2556 +39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
1.2557 +optimization. I've improved things by keeping a bitmap of backreferences with
1.2558 +numbers 1-31 so that if .* occurs inside capturing brackets that are not in
1.2559 +fact referenced, the optimization can be applied. It is unlikely that a
1.2560 +relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
1.2561 +the match to follow \n) will appear inside brackets with a number greater than
1.2562 +31, but if it does, any back reference > 31 suppresses the optimization.
1.2563 +
1.2564 +40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
1.2565 +of disabling numbered capturing parentheses. Any opening parenthesis that is
1.2566 +not followed by ? behaves as if it were followed by ?: but named parentheses
1.2567 +can still be used for capturing (and they will acquire numbers in the usual
1.2568 +way).
1.2569 +
1.2570 +41. Redesigned the return codes from the match() function into yes/no/error so
1.2571 +that errors can be passed back from deep inside the nested calls. A malloc
1.2572 +failure while inside a recursive subpattern call now causes the
1.2573 +PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
1.2574 +
1.2575 +42. It is now possible to set a limit on the number of times the match()
1.2576 +function is called in a call to pcre_exec(). This facility makes it possible to
1.2577 +limit the amount of recursion and backtracking, though not in a directly
1.2578 +obvious way, because the match() function is used in a number of different
1.2579 +circumstances. The count starts from zero for each position in the subject
1.2580 +string (for non-anchored patterns). The default limit is, for compatibility, a
1.2581 +large number, namely 10 000 000. You can change this in two ways:
1.2582 +
1.2583 +(a) When configuring PCRE before making, you can use --with-match-limit=n
1.2584 + to set a default value for the compiled library.
1.2585 +
1.2586 +(b) For each call to pcre_exec(), you can pass a pcre_extra block in which
1.2587 + a different value is set. See 45 below.
1.2588 +
1.2589 +If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
1.2590 +
1.2591 +43. Added a new function pcre_config(int, void *) to enable run-time extraction
1.2592 +of things that can be changed at compile time. The first argument specifies
1.2593 +what is wanted and the second points to where the information is to be placed.
1.2594 +The current list of available information is:
1.2595 +
1.2596 + PCRE_CONFIG_UTF8
1.2597 +
1.2598 +The output is an integer that is set to one if UTF-8 support is available;
1.2599 +otherwise it is set to zero.
1.2600 +
1.2601 + PCRE_CONFIG_NEWLINE
1.2602 +
1.2603 +The output is an integer that it set to the value of the code that is used for
1.2604 +newline. It is either LF (10) or CR (13).
1.2605 +
1.2606 + PCRE_CONFIG_LINK_SIZE
1.2607 +
1.2608 +The output is an integer that contains the number of bytes used for internal
1.2609 +linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
1.2610 +
1.2611 + PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
1.2612 +
1.2613 +The output is an integer that contains the threshold above which the POSIX
1.2614 +interface uses malloc() for output vectors. See item 31 above.
1.2615 +
1.2616 + PCRE_CONFIG_MATCH_LIMIT
1.2617 +
1.2618 +The output is an unsigned integer that contains the default limit of the number
1.2619 +of match() calls in a pcre_exec() execution. See 42 above.
1.2620 +
1.2621 +44. pcretest has been upgraded by the addition of the -C option. This causes it
1.2622 +to extract all the available output from the new pcre_config() function, and to
1.2623 +output it. The program then exits immediately.
1.2624 +
1.2625 +45. A need has arisen to pass over additional data with calls to pcre_exec() in
1.2626 +order to support additional features. One way would have been to define
1.2627 +pcre_exec2() (for example) with extra arguments, but this would not have been
1.2628 +extensible, and would also have required all calls to the original function to
1.2629 +be mapped to the new one. Instead, I have chosen to extend the mechanism that
1.2630 +is used for passing in "extra" data from pcre_study().
1.2631 +
1.2632 +The pcre_extra structure is now exposed and defined in pcre.h. It currently
1.2633 +contains the following fields:
1.2634 +
1.2635 + flags a bitmap indicating which of the following fields are set
1.2636 + study_data opaque data from pcre_study()
1.2637 + match_limit a way of specifying a limit on match() calls for a specific
1.2638 + call to pcre_exec()
1.2639 + callout_data data for callouts (see 49 below)
1.2640 +
1.2641 +The flag bits are also defined in pcre.h, and are
1.2642 +
1.2643 + PCRE_EXTRA_STUDY_DATA
1.2644 + PCRE_EXTRA_MATCH_LIMIT
1.2645 + PCRE_EXTRA_CALLOUT_DATA
1.2646 +
1.2647 +The pcre_study() function now returns one of these new pcre_extra blocks, with
1.2648 +the actual study data pointed to by the study_data field, and the
1.2649 +PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
1.2650 +before. That is, this change is entirely upwards-compatible and requires no
1.2651 +change to existing code.
1.2652 +
1.2653 +If you want to pass in additional data to pcre_exec(), you can either place it
1.2654 +in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
1.2655 +block.
1.2656 +
1.2657 +46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
1.2658 +data string contains the escape sequence \M, pcretest calls pcre_exec() several
1.2659 +times with different match limits, until it finds the minimum value needed for
1.2660 +pcre_exec() to complete. The value is then output. This can be instructive; for
1.2661 +most simple matches the number is quite small, but for pathological cases it
1.2662 +gets very large very quickly.
1.2663 +
1.2664 +47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
1.2665 +returns the size of the data block pointed to by the study_data field in a
1.2666 +pcre_extra block, that is, the value that was passed as the argument to
1.2667 +pcre_malloc() when PCRE was getting memory in which to place the information
1.2668 +created by pcre_study(). The fourth argument should point to a size_t variable.
1.2669 +pcretest has been extended so that this information is shown after a successful
1.2670 +pcre_study() call when information about the compiled regex is being displayed.
1.2671 +
1.2672 +48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
1.2673 +because what follows is always an absolute path. (Later: it turns out that this
1.2674 +is more than cosmetic for MinGW, because it doesn't like empty path
1.2675 +components.)
1.2676 +
1.2677 +49. Some changes have been made to the callout feature (see 28 above):
1.2678 +
1.2679 +(i) A callout function now has three choices for what it returns:
1.2680 +
1.2681 + 0 => success, carry on matching
1.2682 + > 0 => failure at this point, but backtrack if possible
1.2683 + < 0 => serious error, return this value from pcre_exec()
1.2684 +
1.2685 + Negative values should normally be chosen from the set of PCRE_ERROR_xxx
1.2686 + values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
1.2687 + "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
1.2688 + use by callout functions. It will never be used by PCRE itself.
1.2689 +
1.2690 +(ii) The pcre_extra structure (see 45 above) has a void * field called
1.2691 + callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
1.2692 + pcre_callout_block structure has a field of the same name. The contents of
1.2693 + the field passed in the pcre_extra structure are passed to the callout
1.2694 + function in the corresponding field in the callout block. This makes it
1.2695 + easier to use the same callout-containing regex from multiple threads. For
1.2696 + testing, the pcretest program has a new data escape
1.2697 +
1.2698 + \C*n pass the number n (may be negative) as callout_data
1.2699 +
1.2700 + If the callout function in pcretest receives a non-zero value as
1.2701 + callout_data, it returns that value.
1.2702 +
1.2703 +50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
1.2704 +there were some redundant $(CFLAGS) in commands that are now specified as
1.2705 +$(LINK), which already includes $(CFLAGS).
1.2706 +
1.2707 +51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
1.2708 +has been compiled with UTF-8 support *and* pcre_compile() has been compiled
1.2709 +with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
1.2710 +one-byte characters throughout. Note that case-insensitive matching applies
1.2711 +only to characters whose values are less than 256. PCRE doesn't support the
1.2712 +notion of cases for higher-valued characters.
1.2713 +
1.2714 +(i) A character class whose characters are all within 0-255 is handled as
1.2715 + a bit map, and the map is inverted for negative classes. Previously, a
1.2716 + character > 255 always failed to match such a class; however it should
1.2717 + match if the class was a negative one (e.g. [^ab]). This has been fixed.
1.2718 +
1.2719 +(ii) A negated character class with a single character < 255 is coded as
1.2720 + "not this character" (OP_NOT). This wasn't working properly when the test
1.2721 + character was multibyte, either singly or repeated.
1.2722 +
1.2723 +(iii) Repeats of multibyte characters are now handled correctly in UTF-8
1.2724 + mode, for example: \x{100}{2,3}.
1.2725 +
1.2726 +(iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
1.2727 + singly or repeated) now correctly test multibyte characters. However,
1.2728 + PCRE doesn't recognize any characters with values greater than 255 as
1.2729 + digits, spaces, or word characters. Such characters always match \D, \S,
1.2730 + and \W, and never match \d, \s, or \w.
1.2731 +
1.2732 +(v) Classes may now contain characters and character ranges with values
1.2733 + greater than 255. For example: [ab\x{100}-\x{400}].
1.2734 +
1.2735 +(vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
1.2736 + PCRE in UTF-8 mode.
1.2737 +
1.2738 +52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
1.2739 +PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
1.2740 +retained for backwards compatibility. (Note that LASTLITERAL is also a byte
1.2741 +value.)
1.2742 +
1.2743 +53. The single man page has become too large. I have therefore split it up into
1.2744 +a number of separate man pages. These also give rise to individual HTML pages;
1.2745 +these are now put in a separate directory, and there is an index.html page that
1.2746 +lists them all. Some hyperlinking between the pages has been installed.
1.2747 +
1.2748 +54. Added convenience functions for handling named capturing parentheses.
1.2749 +
1.2750 +55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
1.2751 +aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
1.2752 +true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
1.2753 +are faulted.
1.2754 +
1.2755 +56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
1.2756 +calling configure. These values are used when compiling the dftables.c program
1.2757 +which is run to generate the source of the default character tables. They
1.2758 +default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
1.2759 +you will need to set these values.
1.2760 +
1.2761 +57. Updated the building process for Windows DLL, as provided by Fred Cox.
1.2762 +
1.2763 +
1.2764 +Version 3.9 02-Jan-02
1.2765 +---------------------
1.2766 +
1.2767 +1. A bit of extraneous text had somehow crept into the pcregrep documentation.
1.2768 +
1.2769 +2. If --disable-static was given, the building process failed when trying to
1.2770 +build pcretest and pcregrep. (For some reason it was using libtool to compile
1.2771 +them, which is not right, as they aren't part of the library.)
1.2772 +
1.2773 +
1.2774 +Version 3.8 18-Dec-01
1.2775 +---------------------
1.2776 +
1.2777 +1. The experimental UTF-8 code was completely screwed up. It was packing the
1.2778 +bytes in the wrong order. How dumb can you get?
1.2779 +
1.2780 +
1.2781 +Version 3.7 29-Oct-01
1.2782 +---------------------
1.2783 +
1.2784 +1. In updating pcretest to check change 1 of version 3.6, I screwed up.
1.2785 +This caused pcretest, when used on the test data, to segfault. Unfortunately,
1.2786 +this didn't happen under Solaris 8, where I normally test things.
1.2787 +
1.2788 +2. The Makefile had to be changed to make it work on BSD systems, where 'make'
1.2789 +doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
1.2790 +isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
1.2791 +this fix an hour or so after the initial 3.7 release.)
1.2792 +
1.2793 +
1.2794 +Version 3.6 23-Oct-01
1.2795 +---------------------
1.2796 +
1.2797 +1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
1.2798 +offsets passed as NULL with zero offset count.
1.2799 +
1.2800 +2. The config.guess and config.sub files had not been updated when I moved to
1.2801 +the latest autoconf.
1.2802 +
1.2803 +
1.2804 +Version 3.5 15-Aug-01
1.2805 +---------------------
1.2806 +
1.2807 +1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
1.2808 +had been forgotten.
1.2809 +
1.2810 +2. By using declared but undefined structures, we can avoid using "void"
1.2811 +definitions in pcre.h while keeping the internal definitions of the structures
1.2812 +private.
1.2813 +
1.2814 +3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
1.2815 +user point of view, this means that both static and shared libraries are built
1.2816 +by default, but this can be individually controlled. More of the work of
1.2817 +handling this static/shared cases is now inside libtool instead of PCRE's make
1.2818 +file.
1.2819 +
1.2820 +4. The pcretest utility is now installed along with pcregrep because it is
1.2821 +useful for users (to test regexs) and by doing this, it automatically gets
1.2822 +relinked by libtool. The documentation has been turned into a man page, so
1.2823 +there are now .1, .txt, and .html versions in /doc.
1.2824 +
1.2825 +5. Upgrades to pcregrep:
1.2826 + (i) Added long-form option names like gnu grep.
1.2827 + (ii) Added --help to list all options with an explanatory phrase.
1.2828 + (iii) Added -r, --recursive to recurse into sub-directories.
1.2829 + (iv) Added -f, --file to read patterns from a file.
1.2830 +
1.2831 +6. pcre_exec() was referring to its "code" argument before testing that
1.2832 +argument for NULL (and giving an error if it was NULL).
1.2833 +
1.2834 +7. Upgraded Makefile.in to allow for compiling in a different directory from
1.2835 +the source directory.
1.2836 +
1.2837 +8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
1.2838 +options bits, the pointer it was passed was to an int instead of to an unsigned
1.2839 +long int. This mattered only on 64-bit systems.
1.2840 +
1.2841 +9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
1.2842 +generated) instead of pcre.in, which it its source. Also made the same change
1.2843 +in several of the .c files.
1.2844 +
1.2845 +10. A new release of gcc defines printf() as a macro, which broke pcretest
1.2846 +because it had an ifdef in the middle of a string argument for printf(). Fixed
1.2847 +by using separate calls to printf().
1.2848 +
1.2849 +11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
1.2850 +script, to force use of CR or LF instead of \n in the source. On non-Unix
1.2851 +systems, the value can be set in config.h.
1.2852 +
1.2853 +12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
1.2854 +absolute limit. Changed the text of the error message to make this clear, and
1.2855 +likewise updated the man page.
1.2856 +
1.2857 +13. The limit of 99 on the number of capturing subpatterns has been removed.
1.2858 +The new limit is 65535, which I hope will not be a "real" limit.
1.2859 +
1.2860 +
1.2861 +Version 3.4 22-Aug-00
1.2862 +---------------------
1.2863 +
1.2864 +1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
1.2865 +
1.2866 +2. Diagnose condition (?(0) as an error instead of crashing on matching.
1.2867 +
1.2868 +
1.2869 +Version 3.3 01-Aug-00
1.2870 +---------------------
1.2871 +
1.2872 +1. If an octal character was given, but the value was greater than \377, it
1.2873 +was not getting masked to the least significant bits, as documented. This could
1.2874 +lead to crashes in some systems.
1.2875 +
1.2876 +2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
1.2877 +the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
1.2878 +
1.2879 +3. Added the functions pcre_free_substring() and pcre_free_substring_list().
1.2880 +These just pass their arguments on to (pcre_free)(), but they are provided
1.2881 +because some uses of PCRE bind it to non-C systems that can call its functions,
1.2882 +but cannot call free() or pcre_free() directly.
1.2883 +
1.2884 +4. Add "make test" as a synonym for "make check". Corrected some comments in
1.2885 +the Makefile.
1.2886 +
1.2887 +5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
1.2888 +Makefile.
1.2889 +
1.2890 +6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
1.2891 +command called pgrep for grepping around the active processes.
1.2892 +
1.2893 +7. Added the beginnings of support for UTF-8 character strings.
1.2894 +
1.2895 +8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
1.2896 +RANLIB to ./ltconfig so that they are used by libtool. I think these are all
1.2897 +the relevant ones. (AR is not passed because ./ltconfig does its own figuring
1.2898 +out for the ar command.)
1.2899 +
1.2900 +
1.2901 +Version 3.2 12-May-00
1.2902 +---------------------
1.2903 +
1.2904 +This is purely a bug fixing release.
1.2905 +
1.2906 +1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
1.2907 +of ZA. This was just one example of several cases that could provoke this bug,
1.2908 +which was introduced by change 9 of version 2.00. The code for breaking
1.2909 +infinite loops after an iteration that matches an empty string was't working
1.2910 +correctly.
1.2911 +
1.2912 +2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
1.2913 +when matched against abbab (for example). After matching an empty string, it
1.2914 +wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
1.2915 +caused it to match further down the string than it should.
1.2916 +
1.2917 +3. The code contained an inclusion of sys/types.h. It isn't clear why this
1.2918 +was there because it doesn't seem to be needed, and it causes trouble on some
1.2919 +systems, as it is not a Standard C header. It has been removed.
1.2920 +
1.2921 +4. Made 4 silly changes to the source to avoid stupid compiler warnings that
1.2922 +were reported on the Macintosh. The changes were from
1.2923 +
1.2924 + while ((c = *(++ptr)) != 0 && c != '\n');
1.2925 +to
1.2926 + while ((c = *(++ptr)) != 0 && c != '\n') ;
1.2927 +
1.2928 +Totally extraordinary, but if that's what it takes...
1.2929 +
1.2930 +5. PCRE is being used in one environment where neither memmove() nor bcopy() is
1.2931 +available. Added HAVE_BCOPY and an autoconf test for it; if neither
1.2932 +HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
1.2933 +assumes the way PCRE uses memmove() (always moving upwards).
1.2934 +
1.2935 +6. PCRE is being used in one environment where strchr() is not available. There
1.2936 +was only one use in pcre.c, and writing it out to avoid strchr() probably gives
1.2937 +faster code anyway.
1.2938 +
1.2939 +
1.2940 +Version 3.1 09-Feb-00
1.2941 +---------------------
1.2942 +
1.2943 +The only change in this release is the fixing of some bugs in Makefile.in for
1.2944 +the "install" target:
1.2945 +
1.2946 +(1) It was failing to install pcreposix.h.
1.2947 +
1.2948 +(2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
1.2949 +
1.2950 +
1.2951 +Version 3.0 01-Feb-00
1.2952 +---------------------
1.2953 +
1.2954 +1. Add support for the /+ modifier to perltest (to output $` like it does in
1.2955 +pcretest).
1.2956 +
1.2957 +2. Add support for the /g modifier to perltest.
1.2958 +
1.2959 +3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
1.2960 +matches null strings.
1.2961 +
1.2962 +4. Fix perltest so that it doesn't do unwanted things when fed an empty
1.2963 +pattern. Perl treats empty patterns specially - it reuses the most recent
1.2964 +pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
1.2965 +effect.
1.2966 +
1.2967 +5. The POSIX interface was broken in that it was just handing over the POSIX
1.2968 +captured string vector to pcre_exec(), but (since release 2.00) PCRE has
1.2969 +required a bigger vector, with some working space on the end. This means that
1.2970 +the POSIX wrapper now has to get and free some memory, and copy the results.
1.2971 +
1.2972 +6. Added some simple autoconf support, placing the test data and the
1.2973 +documentation in separate directories, re-organizing some of the
1.2974 +information files, and making it build pcre-config (a GNU standard). Also added
1.2975 +libtool support for building PCRE as a shared library, which is now the
1.2976 +default.
1.2977 +
1.2978 +7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
1.2979 +09 are not valid octal constants. Single digits will be used for minor values
1.2980 +less than 10.
1.2981 +
1.2982 +8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
1.2983 +existing programs that set these in the POSIX interface can use PCRE without
1.2984 +modification.
1.2985 +
1.2986 +9. Added a new function, pcre_fullinfo() with an extensible interface. It can
1.2987 +return all that pcre_info() returns, plus additional data. The pcre_info()
1.2988 +function is retained for compatibility, but is considered to be obsolete.
1.2989 +
1.2990 +10. Added experimental recursion feature (?R) to handle one common case that
1.2991 +Perl 5.6 will be able to do with (?p{...}).
1.2992 +
1.2993 +11. Added support for POSIX character classes like [:alpha:], which Perl is
1.2994 +adopting.
1.2995 +
1.2996 +
1.2997 +Version 2.08 31-Aug-99
1.2998 +----------------------
1.2999 +
1.3000 +1. When startoffset was not zero and the pattern began with ".*", PCRE was not
1.3001 +trying to match at the startoffset position, but instead was moving forward to
1.3002 +the next newline as if a previous match had failed.
1.3003 +
1.3004 +2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
1.3005 +and could get into a loop if a null string was matched other than at the start
1.3006 +of the subject.
1.3007 +
1.3008 +3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
1.3009 +be distinguished at compile time, and for completeness also added PCRE_DATE.
1.3010 +
1.3011 +5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
1.3012 +in GnuWin32 environments.
1.3013 +
1.3014 +
1.3015 +Version 2.07 29-Jul-99
1.3016 +----------------------
1.3017 +
1.3018 +1. The documentation is now supplied in plain text form and HTML as well as in
1.3019 +the form of man page sources.
1.3020 +
1.3021 +2. C++ compilers don't like assigning (void *) values to other pointer types.
1.3022 +In particular this affects malloc(). Although there is no problem in Standard
1.3023 +C, I've put in casts to keep C++ compilers happy.
1.3024 +
1.3025 +3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
1.3026 +should be (const char *).
1.3027 +
1.3028 +4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
1.3029 +be useful for non-Unix systems who don't want to bother with the POSIX stuff.
1.3030 +However, I haven't made this a standard facility. The documentation doesn't
1.3031 +mention it, and the Makefile doesn't support it.
1.3032 +
1.3033 +5. The Makefile now contains an "install" target, with editable destinations at
1.3034 +the top of the file. The pcretest program is not installed.
1.3035 +
1.3036 +6. pgrep -V now gives the PCRE version number and date.
1.3037 +
1.3038 +7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
1.3039 +causing the entire string to be ignored, instead of just the last character.
1.3040 +
1.3041 +8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
1.3042 +non-matching string, it can take a very, very long time, even for strings of
1.3043 +quite modest length, because of the nested recursion. PCRE now does better in
1.3044 +some of these cases. It does this by remembering the last required literal
1.3045 +character in the pattern, and pre-searching the subject to ensure it is present
1.3046 +before running the real match. In other words, it applies a heuristic to detect
1.3047 +some types of certain failure quickly, and in the above example, if presented
1.3048 +with a string that has no trailing " it gives "no match" very quickly.
1.3049 +
1.3050 +9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
1.3051 +other alternatives are tried instead.
1.3052 +
1.3053 +
1.3054 +Version 2.06 09-Jun-99
1.3055 +----------------------
1.3056 +
1.3057 +1. Change pcretest's output for amount of store used to show just the code
1.3058 +space, because the remainder (the data block) varies in size between 32-bit and
1.3059 +64-bit systems.
1.3060 +
1.3061 +2. Added an extra argument to pcre_exec() to supply an offset in the subject to
1.3062 +start matching at. This allows lookbehinds to work when searching for multiple
1.3063 +occurrences in a string.
1.3064 +
1.3065 +3. Added additional options to pcretest for testing multiple occurrences:
1.3066 +
1.3067 + /+ outputs the rest of the string that follows a match
1.3068 + /g loops for multiple occurrences, using the new startoffset argument
1.3069 + /G loops for multiple occurrences by passing an incremented pointer
1.3070 +
1.3071 +4. PCRE wasn't doing the "first character" optimization for patterns starting
1.3072 +with \b or \B, though it was doing it for other lookbehind assertions. That is,
1.3073 +it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
1.3074 +the letter 'x'. On long subject strings, this gives a significant speed-up.
1.3075 +
1.3076 +
1.3077 +Version 2.05 21-Apr-99
1.3078 +----------------------
1.3079 +
1.3080 +1. Changed the type of magic_number from int to long int so that it works
1.3081 +properly on 16-bit systems.
1.3082 +
1.3083 +2. Fixed a bug which caused patterns starting with .* not to work correctly
1.3084 +when the subject string contained newline characters. PCRE was assuming
1.3085 +anchoring for such patterns in all cases, which is not correct because .* will
1.3086 +not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
1.3087 +DOTALL is set at top level; otherwise it knows that patterns starting with .*
1.3088 +must be retried after every newline in the subject.
1.3089 +
1.3090 +
1.3091 +Version 2.04 18-Feb-99
1.3092 +----------------------
1.3093 +
1.3094 +1. For parenthesized subpatterns with repeats whose minimum was zero, the
1.3095 +computation of the store needed to hold the pattern was incorrect (too large).
1.3096 +If such patterns were nested a few deep, this could multiply and become a real
1.3097 +problem.
1.3098 +
1.3099 +2. Added /M option to pcretest to show the memory requirement of a specific
1.3100 +pattern. Made -m a synonym of -s (which does this globally) for compatibility.
1.3101 +
1.3102 +3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
1.3103 +compiled in such a way that the backtracking after subsequent failure was
1.3104 +pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
1.3105 +((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
1.3106 +
1.3107 +
1.3108 +Version 2.03 02-Feb-99
1.3109 +----------------------
1.3110 +
1.3111 +1. Fixed typo and small mistake in man page.
1.3112 +
1.3113 +2. Added 4th condition (GPL supersedes if conflict) and created separate
1.3114 +LICENCE file containing the conditions.
1.3115 +
1.3116 +3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
1.3117 +Perl, that is the internal \ allows the delimiter to be included in the
1.3118 +pattern. Locked out the use of \ as a delimiter. If \ immediately follows
1.3119 +the final delimiter, add \ to the end of the pattern (to test the error).
1.3120 +
1.3121 +4. Added the convenience functions for extracting substrings after a successful
1.3122 +match. Updated pcretest to make it able to test these functions.
1.3123 +
1.3124 +
1.3125 +Version 2.02 14-Jan-99
1.3126 +----------------------
1.3127 +
1.3128 +1. Initialized the working variables associated with each extraction so that
1.3129 +their saving and restoring doesn't refer to uninitialized store.
1.3130 +
1.3131 +2. Put dummy code into study.c in order to trick the optimizer of the IBM C
1.3132 +compiler for OS/2 into generating correct code. Apparently IBM isn't going to
1.3133 +fix the problem.
1.3134 +
1.3135 +3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
1.3136 +calls, and wasn't printing the correct value for compiling calls. Increased the
1.3137 +default value of LOOPREPEAT, and the number of significant figures in the
1.3138 +times.
1.3139 +
1.3140 +4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
1.3141 +
1.3142 +5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
1.3143 +a building problem on Windows NT with a FAT file system.
1.3144 +
1.3145 +
1.3146 +Version 2.01 21-Oct-98
1.3147 +----------------------
1.3148 +
1.3149 +1. Changed the API for pcre_compile() to allow for the provision of a pointer
1.3150 +to character tables built by pcre_maketables() in the current locale. If NULL
1.3151 +is passed, the default tables are used.
1.3152 +
1.3153 +
1.3154 +Version 2.00 24-Sep-98
1.3155 +----------------------
1.3156 +
1.3157 +1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
1.3158 +it any more.
1.3159 +
1.3160 +2. Allow quantification of (?>) groups, and make it work correctly.
1.3161 +
1.3162 +3. The first character computation wasn't working for (?>) groups.
1.3163 +
1.3164 +4. Correct the implementation of \Z (it is permitted to match on the \n at the
1.3165 +end of the subject) and add 5.005's \z, which really does match only at the
1.3166 +very end of the subject.
1.3167 +
1.3168 +5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
1.3169 +
1.3170 +6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
1.3171 +DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
1.3172 +localized options. All options to pcre_study() were also removed.
1.3173 +
1.3174 +7. Add other new features from 5.005:
1.3175 +
1.3176 + $(?<= positive lookbehind
1.3177 + $(?<! negative lookbehind
1.3178 + (?imsx-imsx) added the unsetting capability
1.3179 + such a setting is global if at outer level; local otherwise
1.3180 + (?imsx-imsx:) non-capturing groups with option setting
1.3181 + (?(cond)re|re) conditional pattern matching
1.3182 +
1.3183 + A backreference to itself in a repeated group matches the previous
1.3184 + captured string.
1.3185 +
1.3186 +8. General tidying up of studying (both automatic and via "study")
1.3187 +consequential on the addition of new assertions.
1.3188 +
1.3189 +9. As in 5.005, unlimited repeated groups that could match an empty substring
1.3190 +are no longer faulted at compile time. Instead, the loop is forcibly broken at
1.3191 +runtime if any iteration does actually match an empty substring.
1.3192 +
1.3193 +10. Include the RunTest script in the distribution.
1.3194 +
1.3195 +11. Added tests from the Perl 5.005_02 distribution. This showed up a few
1.3196 +discrepancies, some of which were old and were also with respect to 5.004. They
1.3197 +have now been fixed.
1.3198 +
1.3199 +
1.3200 +Version 1.09 28-Apr-98
1.3201 +----------------------
1.3202 +
1.3203 +1. A negated single character class followed by a quantifier with a minimum
1.3204 +value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
1.3205 +program crashes, or just wrong answers. This did not apply to negated classes
1.3206 +containing more than one character, or to minima other than one.
1.3207 +
1.3208 +
1.3209 +Version 1.08 27-Mar-98
1.3210 +----------------------
1.3211 +
1.3212 +1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
1.3213 +
1.3214 +2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
1.3215 +latter must appear before anything that relies on it in the pattern.
1.3216 +
1.3217 +
1.3218 +Version 1.07 16-Feb-98
1.3219 +----------------------
1.3220 +
1.3221 +1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
1.3222 +repeat of a potentially empty string).
1.3223 +
1.3224 +
1.3225 +Version 1.06 23-Jan-98
1.3226 +----------------------
1.3227 +
1.3228 +1. Added Markus Oberhumer's little patches for C++.
1.3229 +
1.3230 +2. Literal strings longer than 255 characters were broken.
1.3231 +
1.3232 +
1.3233 +Version 1.05 23-Dec-97
1.3234 +----------------------
1.3235 +
1.3236 +1. Negated character classes containing more than one character were failing if
1.3237 +PCRE_CASELESS was set at run time.
1.3238 +
1.3239 +
1.3240 +Version 1.04 19-Dec-97
1.3241 +----------------------
1.3242 +
1.3243 +1. Corrected the man page, where some "const" qualifiers had been omitted.
1.3244 +
1.3245 +2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
1.3246 +input syntax.
1.3247 +
1.3248 +3. Fixed memory leak which occurred when a regex with back references was
1.3249 +matched with an offsets vector that wasn't big enough. The temporary memory
1.3250 +that is used in this case wasn't being freed if the match failed.
1.3251 +
1.3252 +4. Tidied pcretest to ensure it frees memory that it gets.
1.3253 +
1.3254 +5. Temporary memory was being obtained in the case where the passed offsets
1.3255 +vector was exactly big enough.
1.3256 +
1.3257 +6. Corrected definition of offsetof() from change 5 below.
1.3258 +
1.3259 +7. I had screwed up change 6 below and broken the rules for the use of
1.3260 +setjmp(). Now fixed.
1.3261 +
1.3262 +
1.3263 +Version 1.03 18-Dec-97
1.3264 +----------------------
1.3265 +
1.3266 +1. A erroneous regex with a missing opening parenthesis was correctly
1.3267 +diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
1.3268 +on some systems.
1.3269 +
1.3270 +2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
1.3271 +it was reported that one broken compiler failed on the former because "code" is
1.3272 +also an independent variable.
1.3273 +
1.3274 +3. The erroneous regex a[]b caused an array overrun reference.
1.3275 +
1.3276 +4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
1.3277 +fail on data ending with that character. (It was going on too far, and checking
1.3278 +the next character, typically a binary zero.) This was specific to the
1.3279 +optimized code for single-character negative classes.
1.3280 +
1.3281 +5. Added a contributed patch from the TIN world which does the following:
1.3282 +
1.3283 + + Add an undef for memmove, in case the the system defines a macro for it.
1.3284 +
1.3285 + + Add a definition of offsetof(), in case there isn't one. (I don't know
1.3286 + the reason behind this - offsetof() is part of the ANSI standard - but
1.3287 + it does no harm).
1.3288 +
1.3289 + + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
1.3290 + most of the places where whitespace preceded '#'. I have given up and
1.3291 + allowed the remaining 2 cases to be at the margin.
1.3292 +
1.3293 + + Rename some variables in pcre to eliminate shadowing. This seems very
1.3294 + pedantic, but does no harm, of course.
1.3295 +
1.3296 +6. Moved the call to setjmp() into its own function, to get rid of warnings
1.3297 +from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
1.3298 +
1.3299 +7. Constructs such as \d{8,} were compiling into the equivalent of
1.3300 +\d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
1.3301 +outcome, but in this particular case used more store than had been allocated,
1.3302 +which caused the bug to be discovered because it threw up an internal error.
1.3303 +
1.3304 +8. The debugging code in both pcre and pcretest for outputting the compiled
1.3305 +form of a regex was going wrong in the case of back references followed by
1.3306 +curly-bracketed repeats.
1.3307 +
1.3308 +
1.3309 +Version 1.02 12-Dec-97
1.3310 +----------------------
1.3311 +
1.3312 +1. Typos in pcre.3 and comments in the source fixed.
1.3313 +
1.3314 +2. Applied a contributed patch to get rid of places where it used to remove
1.3315 +'const' from variables, and fixed some signed/unsigned and uninitialized
1.3316 +variable warnings.
1.3317 +
1.3318 +3. Added the "runtest" target to Makefile.
1.3319 +
1.3320 +4. Set default compiler flag to -O2 rather than just -O.
1.3321 +
1.3322 +
1.3323 +Version 1.01 19-Nov-97
1.3324 +----------------------
1.3325 +
1.3326 +1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
1.3327 +like /([ab]*)*/, that is, for classes with more than one character in them.
1.3328 +
1.3329 +2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
1.3330 +as /((?>a*))*/ (a PCRE_EXTRA facility).
1.3331 +
1.3332 +
1.3333 +Version 1.00 18-Nov-97
1.3334 +----------------------
1.3335 +
1.3336 +1. Added compile-time macros to support systems such as SunOS4 which don't have
1.3337 +memmove() or strerror() but have other things that can be used instead.
1.3338 +
1.3339 +2. Arranged that "make clean" removes the executables.
1.3340 +
1.3341 +
1.3342 +Version 0.99 27-Oct-97
1.3343 +----------------------
1.3344 +
1.3345 +1. Fixed bug in code for optimizing classes with only one character. It was
1.3346 +initializing a 32-byte map regardless, which could cause it to run off the end
1.3347 +of the memory it had got.
1.3348 +
1.3349 +2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
1.3350 +
1.3351 +
1.3352 +Version 0.98 22-Oct-97
1.3353 +----------------------
1.3354 +
1.3355 +1. Fixed bug in code for handling temporary memory usage when there are more
1.3356 +back references than supplied space in the ovector. This could cause segfaults.
1.3357 +
1.3358 +
1.3359 +Version 0.97 21-Oct-97
1.3360 +----------------------
1.3361 +
1.3362 +1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
1.3363 +
1.3364 +2. Optimized negated single characters not to use a bit map.
1.3365 +
1.3366 +3. Brought error texts together as macro definitions; clarified some of them;
1.3367 +fixed one that was wrong - it said "range out of order" when it meant "invalid
1.3368 +escape sequence".
1.3369 +
1.3370 +4. Changed some char * arguments to const char *.
1.3371 +
1.3372 +5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
1.3373 +
1.3374 +6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
1.3375 +pcretest.
1.3376 +
1.3377 +
1.3378 +Version 0.96 16-Oct-97
1.3379 +----------------------
1.3380 +
1.3381 +1. Added a simple "pgrep" utility to the distribution.
1.3382 +
1.3383 +2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
1.3384 +unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
1.3385 +where "ddd" means "one or more decimal digits".
1.3386 +
1.3387 +3. Fixed serious bug. If a pattern had a back reference, but the call to
1.3388 +pcre_exec() didn't supply a large enough ovector to record the related
1.3389 +identifying subpattern, the match always failed. PCRE now remembers the number
1.3390 +of the largest back reference, and gets some temporary memory in which to save
1.3391 +the offsets during matching if necessary, in order to ensure that
1.3392 +backreferences always work.
1.3393 +
1.3394 +4. Increased the compatibility with Perl in a number of ways:
1.3395 +
1.3396 + (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
1.3397 + to request this handling. The option can be set at compile or exec time.
1.3398 +
1.3399 + (b) $ matches before a terminating newline by default; an option
1.3400 + PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
1.3401 + mode). The option can be set at compile or exec time.
1.3402 +
1.3403 + (c) The handling of \ followed by a digit other than 0 is now supposed to be
1.3404 + the same as Perl's. If the decimal number it represents is less than 10
1.3405 + or there aren't that many previous left capturing parentheses, an octal
1.3406 + escape is read. Inside a character class, it's always an octal escape,
1.3407 + even if it is a single digit.
1.3408 +
1.3409 + (d) An escaped but undefined alphabetic character is taken as a literal,
1.3410 + unless PCRE_EXTRA is set. Currently this just reserves the remaining
1.3411 + escapes.
1.3412 +
1.3413 + (e) {0} is now permitted. (The previous item is removed from the compiled
1.3414 + pattern).
1.3415 +
1.3416 +5. Changed all the names of code files so that the basic parts are no longer
1.3417 +than 10 characters, and abolished the teeny "globals.c" file.
1.3418 +
1.3419 +6. Changed the handling of character classes; they are now done with a 32-byte
1.3420 +bit map always.
1.3421 +
1.3422 +7. Added the -d and /D options to pcretest to make it possible to look at the
1.3423 +internals of compilation without having to recompile pcre.
1.3424 +
1.3425 +
1.3426 +Version 0.95 23-Sep-97
1.3427 +----------------------
1.3428 +
1.3429 +1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
1.3430 +\x20 at the start of a run of normal characters. These were being treated as
1.3431 +real characters, instead of the source characters being re-checked.
1.3432 +
1.3433 +
1.3434 +Version 0.94 18-Sep-97
1.3435 +----------------------
1.3436 +
1.3437 +1. The functions are now thread-safe, with the caveat that the global variables
1.3438 +containing pointers to malloc() and free() or alternative functions are the
1.3439 +same for all threads.
1.3440 +
1.3441 +2. Get pcre_study() to generate a bitmap of initial characters for non-
1.3442 +anchored patterns when this is possible, and use it if passed to pcre_exec().
1.3443 +
1.3444 +
1.3445 +Version 0.93 15-Sep-97
1.3446 +----------------------
1.3447 +
1.3448 +1. /(b)|(:+)/ was computing an incorrect first character.
1.3449 +
1.3450 +2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
1.3451 +but not actually doing anything yet.
1.3452 +
1.3453 +3. Treat "-" characters in classes that cannot be part of ranges as literals,
1.3454 +as Perl does (e.g. [-az] or [az-]).
1.3455 +
1.3456 +4. Set the anchored flag if a branch starts with .* or .*? because that tests
1.3457 +all possible positions.
1.3458 +
1.3459 +5. Split up into different modules to avoid including unneeded functions in a
1.3460 +compiled binary. However, compile and exec are still in one module. The "study"
1.3461 +function is split off.
1.3462 +
1.3463 +6. The character tables are now in a separate module whose source is generated
1.3464 +by an auxiliary program - but can then be edited by hand if required. There are
1.3465 +now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
1.3466 +toupper() in the code.
1.3467 +
1.3468 +7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
1.3469 +make them global. Abolish the function for setting them, as the caller can now
1.3470 +set them directly.
1.3471 +
1.3472 +
1.3473 +Version 0.92 11-Sep-97
1.3474 +----------------------
1.3475 +
1.3476 +1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
1.3477 +(e.g. /a{1,3}/) was broken (I mis-optimized it).
1.3478 +
1.3479 +2. Caseless matching was not working in character classes if the characters in
1.3480 +the pattern were in upper case.
1.3481 +
1.3482 +3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
1.3483 +
1.3484 +4. Make PCRE_ANCHORED public and accept as a compile option.
1.3485 +
1.3486 +5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
1.3487 +PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
1.3488 +pass them.
1.3489 +
1.3490 +6. Give an error if bad option bits passed at compile or run time.
1.3491 +
1.3492 +7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
1.3493 +pcretest to cause it to pass that flag.
1.3494 +
1.3495 +8. Add pcre_info(), to get the number of identifying subpatterns, the stored
1.3496 +options, and the first character, if set.
1.3497 +
1.3498 +9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
1.3499 +
1.3500 +
1.3501 +Version 0.91 10-Sep-97
1.3502 +----------------------
1.3503 +
1.3504 +1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
1.3505 +match the empty string as in /(a*)*/. It was looping and ultimately crashing.
1.3506 +
1.3507 +2. PCRE was looping on encountering an indefinitely repeated back reference to
1.3508 +a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
1.3509 +Perl does - treats the match as successful.
1.3510 +
1.3511 +****
