# # # add_dir "pcre" # # add_file "m4/pcre.m4" # content [806d70df6bad48195a10ab334738542d17eb456c] # # add_file "pcre/README" # content [d83b21c4e1e1332190090230099bccd633988985] # # add_file "pcre/pcre.h" # content [845646297e364447eb985e62d9b838c4e26fdbc2] # # add_file "pcre/pcre_chartables.c" # content [aa6b49b4d832ffdad9d5ca3fd8013ec371fc417a] # # add_file "pcre/pcre_compile.c" # content [2a7d0d4b9860bb85f50197170d47926a8526269b] # # add_file "pcre/pcre_config.c" # content [3197ae2ed4c54cc20c770044d842038ba7e9f2b5] # # add_file "pcre/pcre_config.h" # content [cdd4d95dc55d8f805da3888d61eb630148b39eaf] # # add_file "pcre/pcre_dfa_exec.c" # content [e47bdb1351432a867879ff71b5439285458202b6] # # add_file "pcre/pcre_exec.c" # content [67b6e933c1bf48822748163ca104e433a79208b5] # # add_file "pcre/pcre_fullinfo.c" # content [65c02930182c0ca0a12b7a2cd46cce76a0f89457] # # add_file "pcre/pcre_get.c" # content [4dd0a6df9fa0c3a4fb9a3356ad2b9623d9febf57] # # add_file "pcre/pcre_globals.c" # content [4a27067a587c5cd92ad9671107a3541f6429082c] # # add_file "pcre/pcre_info.c" # content [14399144255ceaa0d09a0d16ef030e3dc91cff5b] # # add_file "pcre/pcre_internal.h" # content [d6b78d25230e9f7d3cdf183f2fb45b6de45dc062] # # add_file "pcre/pcre_maketables.c" # content [03f5e59dd47ef2874398251712ff6ee454d76d13] # # add_file "pcre/pcre_refcount.c" # content [ebd0c3dfce2e8dc81a179e2b033364fcec82f55b] # # add_file "pcre/pcre_study.c" # content [9a88f30ab85c1a4df719efb21352da8ec0ff0a60] # # add_file "pcre/pcre_tables.c" # content [f504519f4ad83001af74dca4ada66d1fe68f64aa] # # add_file "pcre/pcre_try_flipped.c" # content [4d22ad726b0e1decb6c0ebaecd607b7def17d642] # # add_file "pcre/pcre_version.c" # content [0e4d6ac360d5acfba16e4d071b69d954488e7a5f] # # add_file "pcre/pcre_xclass.c" # content [d672ac5b6a31841662bca6917df7d31537ae7f22] # # patch ".mtn-ignore" # from [53ba0a32a8d3d257787d12f2e37b8c1329c664b4] # to [b587a7b3db89e9f0549ddcae609f5ccf3853f1eb] # # patch "ChangeLog" # from [a7d6d3a553ba406d13b20f6e11d251ca4158a339] # to [49c9aa7b2d19c2e0d8a55800e9f9082df054ec9d] # # patch "Makefile.am" # from [cdd7815c1e675ef9cb034b2a0fd9d2bdb36a1f55] # to [2cdac9ee2ebeba3ebcf3ff0855c993416e227207] # # patch "configure.ac" # from [661fc6adca568307e46e8d53d8dbd51c6ead71f0] # to [741803702d24a7cc1ebb7889cd061bfa35b0aaed] # # patch "debian/control" # from [72d4ce11e07e8653805b1872da91c6b52cc4acb3] # to [a04bbd93b10b7bc711c39b63bad977f1a20cb081] # # patch "m4/boost.m4" # from [005410a3f9afe57b52fd5bfc18330155d400ac6f] # to [d98c8d3dc1f0acf7d6ca96acf4418eaa259d04cf] # ============================================================ --- m4/pcre.m4 806d70df6bad48195a10ab334738542d17eb456c +++ m4/pcre.m4 806d70df6bad48195a10ab334738542d17eb456c @@ -0,0 +1,81 @@ +# Detect libpcre or fall back to our bundled version. +AC_DEFUN([MTN_LIB_PCRE], +[AC_ARG_WITH([included-pcre], + AC_HELP_STRING([--with-included-pcre], + [force use of the included copy of libpcre. (To use a specific + installed version, use the environment variables PCRE_CFLAGS and/or + PCRE_LIBS.)]), + [case "$withval" in + (""|yes) with_included_pcre=yes ;; + (no) with_included_pcre=no ;; + (*) AC_MSG_ERROR([--with(out)-included-pcre takes no argument]) ;; + esac], + [with_included_pcre=no]) + if test "$with_included_pcre" = no; then + # We manually test the variables here because we want them to work + # even if pkg-config isn't installed. The use of + instead of :+ is + # deliberate; the user should be able to tell us that the empty string + # is the correct set of flags. (PKG_CHECK_MODULES gets this wrong!) + if test -n "${PCRE_CFLAGS+set}" || test -n "${PCRE_LIBS+set}"; then + found_libpcre=yes + else + PKG_CHECK_MODULES([PCRE], [libpcre], + [found_libpcre=yes], [found_libpcre=no]) + fi + if test $found_libpcre = no; then + # try pcre-config, in case we're on a system with no pkg-config + AC_MSG_CHECKING([for PCRE using pcre-config]) + if test -n "`type -p pcre-config`"; then + PCRE_CFLAGS="`pcre-config --cflags`" + # pcre-config has the annoying habit of telling us to use + # -L switches for directories that the compiler will search + # automatically. + PCRE_LIBS="`pcre-config --libs | \ + sed -e 's:-L */usr/lib/*::' -e 's:-R */usr/lib/*::' \ + -e 's:-L */lib/*:: -e 's:-R */lib/*::'`" + found_libpcre=yes + AC_MSG_RESULT([yes]) + fi + fi + if test $found_libpcre = no; then + AC_MSG_RESULT([no; guessing]) + PCRE_CFLAGS= + PCRE_LIBS=-lpcre + fi + + # Wherever we got the settings from, make sure they work. + PCRE_CFLAGS="`echo :$PCRE_CFLAGS | sed -e 's/^:@<:@ @:>@*//; s/@<:@ @:>@*$//'`" + PCRE_LIBS="`echo :$PCRE_LIBS | sed -e 's/^:@<:@ @:>@*//; s/@<:@ @:>@*$//'`" + if test x"$PCRE_CFLAGS" != x""; then + AC_MSG_NOTICE([using PCRE compile flags: "$PCRE_CFLAGS"]) + fi + if test x"$PCRE_LIBS" != x"-lpcre"; then + AC_MSG_NOTICE([using PCRE link flags: "$PCRE_LIBS"]) + fi + AC_CACHE_CHECK([whether the system libpcre works], ac_cv_lib_pcre_works, + [save_LIBS="$LIBS" + save_CFLAGS="$CFLAGS" + LIBS="$LIBS $PCRE_LIBS" + CFLAGS="$CFLAGS $PCRE_CFLAGS" + AC_LINK_IFELSE([AC_LANG_PROGRAM( + [#include ], + [const char *e; + int o; + pcre *re = pcre_compile("foo", 0, &e, &o, 0);])], + [ac_cv_lib_pcre_works=yes], [ac_cv_lib_pcre_works=no]) + LIBS="$save_LIBS" + CFLAGS="$save_CFLAGS"]) + if test $ac_cv_lib_pcre_works = no; then + with_included_pcre=yes + fi + fi + if test $with_included_pcre = yes; then + AC_MSG_NOTICE([using the PCRE library included with the distribution]) + fi + AM_CONDITIONAL([INCLUDED_PCRE], [test $with_included_pcre = yes]) + AC_SUBST([PCRE_CFLAGS]) + AC_SUBST([PCRE_LIBS]) +]) + + + ============================================================ --- pcre/README d83b21c4e1e1332190090230099bccd633988985 +++ pcre/README d83b21c4e1e1332190090230099bccd633988985 @@ -0,0 +1,8 @@ +This is most of pcre 6.7. Configuration settings have been HARDWIRED. +(pcre does not have any portability constraints that we need to worry +about.) Unicode support has been removed, and the "dftables.c" +program has been run and its result checked in. pcre_internal.h has +been modified to include pcre_config.h instead of bare config.h. Some +#ifdefs have been added to pcre_tables.c, pcre_internal.h, and +pcre_dfa_exec.c to dike out Unicode more thoroughly. + ============================================================ --- pcre/pcre.h 845646297e364447eb985e62d9b838c4e26fdbc2 +++ pcre/pcre.h 845646297e364447eb985e62d9b838c4e26fdbc2 @@ -0,0 +1,290 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + +/* This is the public header file for the PCRE library, to be #included by +applications that call the PCRE functions. + + Copyright (c) 1997-2005 University of Cambridge + +----------------------------------------------------------------------------- +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + + * Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimer. + + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + * Neither the name of the University of Cambridge nor the names of its + contributors may be used to endorse or promote products derived from + this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE +LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR +CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +----------------------------------------------------------------------------- +*/ + +#ifndef _PCRE_H +#define _PCRE_H + +/* The current PCRE version information. */ + +/* NOTES FOR FUTURE MAINTAINERS: Do not use numbers with leading zeros, because +they may be treated as octal constants. The PCRE_PRERELEASE feature is for +identifying release candidates. It might be defined as -RC2, for example. In +real releases, it should be defined empty. Do not change the alignment of these +statments. The code in ./configure greps out the version numbers by using "cut" +to get values from column 29 onwards. These are substituted into pcre-config +and libpcre.pc. The values are not put into configure.ac and substituted here +(which would simplify this issue) because that makes life harder for those who +cannot run ./configure. As it now stands, this file need not be edited in that +circumstance. */ + +#define PCRE_MAJOR 6 +#define PCRE_MINOR 7 +#define PCRE_PRERELEASE +#define PCRE_DATE 04-Jul-2006 + +/* Win32 uses DLL by default; it needs special stuff for exported functions +when building PCRE. */ + +#ifdef _WIN32 +# ifdef PCRE_DEFINITION +# ifdef DLL_EXPORT +# define PCRE_DATA_SCOPE __declspec(dllexport) +# endif +# else +# ifndef PCRE_STATIC +# define PCRE_DATA_SCOPE extern __declspec(dllimport) +# endif +# endif +#endif + +/* Otherwise, we use the standard "extern". */ + +#ifndef PCRE_DATA_SCOPE +# ifdef __cplusplus +# define PCRE_DATA_SCOPE extern "C" +# else +# define PCRE_DATA_SCOPE extern +# endif +#endif + +/* Have to include stdlib.h in order to ensure that size_t is defined; +it is needed here for malloc. */ + +#include + +/* Allow for C++ users */ + +#ifdef __cplusplus +extern "C" { +#endif + +/* Options */ + +#define PCRE_CASELESS 0x00000001 +#define PCRE_MULTILINE 0x00000002 +#define PCRE_DOTALL 0x00000004 +#define PCRE_EXTENDED 0x00000008 +#define PCRE_ANCHORED 0x00000010 +#define PCRE_DOLLAR_ENDONLY 0x00000020 +#define PCRE_EXTRA 0x00000040 +#define PCRE_NOTBOL 0x00000080 +#define PCRE_NOTEOL 0x00000100 +#define PCRE_UNGREEDY 0x00000200 +#define PCRE_NOTEMPTY 0x00000400 +#define PCRE_UTF8 0x00000800 +#define PCRE_NO_AUTO_CAPTURE 0x00001000 +#define PCRE_NO_UTF8_CHECK 0x00002000 +#define PCRE_AUTO_CALLOUT 0x00004000 +#define PCRE_PARTIAL 0x00008000 +#define PCRE_DFA_SHORTEST 0x00010000 +#define PCRE_DFA_RESTART 0x00020000 +#define PCRE_FIRSTLINE 0x00040000 +#define PCRE_DUPNAMES 0x00080000 +#define PCRE_NEWLINE_CR 0x00100000 +#define PCRE_NEWLINE_LF 0x00200000 +#define PCRE_NEWLINE_CRLF 0x00300000 + +/* Exec-time and get/set-time error codes */ + +#define PCRE_ERROR_NOMATCH (-1) +#define PCRE_ERROR_NULL (-2) +#define PCRE_ERROR_BADOPTION (-3) +#define PCRE_ERROR_BADMAGIC (-4) +#define PCRE_ERROR_UNKNOWN_NODE (-5) +#define PCRE_ERROR_NOMEMORY (-6) +#define PCRE_ERROR_NOSUBSTRING (-7) +#define PCRE_ERROR_MATCHLIMIT (-8) +#define PCRE_ERROR_CALLOUT (-9) /* Never used by PCRE itself */ +#define PCRE_ERROR_BADUTF8 (-10) +#define PCRE_ERROR_BADUTF8_OFFSET (-11) +#define PCRE_ERROR_PARTIAL (-12) +#define PCRE_ERROR_BADPARTIAL (-13) +#define PCRE_ERROR_INTERNAL (-14) +#define PCRE_ERROR_BADCOUNT (-15) +#define PCRE_ERROR_DFA_UITEM (-16) +#define PCRE_ERROR_DFA_UCOND (-17) +#define PCRE_ERROR_DFA_UMLIMIT (-18) +#define PCRE_ERROR_DFA_WSSIZE (-19) +#define PCRE_ERROR_DFA_RECURSE (-20) +#define PCRE_ERROR_RECURSIONLIMIT (-21) + +/* Request types for pcre_fullinfo() */ + +#define PCRE_INFO_OPTIONS 0 +#define PCRE_INFO_SIZE 1 +#define PCRE_INFO_CAPTURECOUNT 2 +#define PCRE_INFO_BACKREFMAX 3 +#define PCRE_INFO_FIRSTBYTE 4 +#define PCRE_INFO_FIRSTCHAR 4 /* For backwards compatibility */ +#define PCRE_INFO_FIRSTTABLE 5 +#define PCRE_INFO_LASTLITERAL 6 +#define PCRE_INFO_NAMEENTRYSIZE 7 +#define PCRE_INFO_NAMECOUNT 8 +#define PCRE_INFO_NAMETABLE 9 +#define PCRE_INFO_STUDYSIZE 10 +#define PCRE_INFO_DEFAULT_TABLES 11 + +/* Request types for pcre_config(). Do not re-arrange, in order to remain +compatible. */ + +#define PCRE_CONFIG_UTF8 0 +#define PCRE_CONFIG_NEWLINE 1 +#define PCRE_CONFIG_LINK_SIZE 2 +#define PCRE_CONFIG_POSIX_MALLOC_THRESHOLD 3 +#define PCRE_CONFIG_MATCH_LIMIT 4 +#define PCRE_CONFIG_STACKRECURSE 5 +#define PCRE_CONFIG_UNICODE_PROPERTIES 6 +#define PCRE_CONFIG_MATCH_LIMIT_RECURSION 7 + +/* Bit flags for the pcre_extra structure. Do not re-arrange or redefine +these bits, just add new ones on the end, in order to remain compatible. */ + +#define PCRE_EXTRA_STUDY_DATA 0x0001 +#define PCRE_EXTRA_MATCH_LIMIT 0x0002 +#define PCRE_EXTRA_CALLOUT_DATA 0x0004 +#define PCRE_EXTRA_TABLES 0x0008 +#define PCRE_EXTRA_MATCH_LIMIT_RECURSION 0x0010 + +/* Types */ + +struct real_pcre; /* declaration; the definition is private */ +typedef struct real_pcre pcre; + +/* When PCRE is compiled as a C++ library, the subject pointer type can be +replaced with a custom type. For conventional use, the public interface is a +const char *. */ + +#ifndef PCRE_SPTR +#define PCRE_SPTR const char * +#endif + +/* The structure for passing additional data to pcre_exec(). This is defined in +such as way as to be extensible. Always add new fields at the end, in order to +remain compatible. */ + +typedef struct pcre_extra { + unsigned long int flags; /* Bits for which fields are set */ + void *study_data; /* Opaque data from pcre_study() */ + unsigned long int match_limit; /* Maximum number of calls to match() */ + void *callout_data; /* Data passed back in callouts */ + const unsigned char *tables; /* Pointer to character tables */ + unsigned long int match_limit_recursion; /* Max recursive calls to match() */ +} pcre_extra; + +/* The structure for passing out data via the pcre_callout_function. We use a +structure so that new fields can be added on the end in future versions, +without changing the API of the function, thereby allowing old clients to work +without modification. */ + +typedef struct pcre_callout_block { + int version; /* Identifies version of block */ + /* ------------------------ Version 0 ------------------------------- */ + int callout_number; /* Number compiled into pattern */ + int *offset_vector; /* The offset vector */ + PCRE_SPTR subject; /* The subject being matched */ + int subject_length; /* The length of the subject */ + int start_match; /* Offset to start of this match attempt */ + int current_position; /* Where we currently are in the subject */ + int capture_top; /* Max current capture */ + int capture_last; /* Most recently closed capture */ + void *callout_data; /* Data passed in with the call */ + /* ------------------- Added for Version 1 -------------------------- */ + int pattern_position; /* Offset to next item in the pattern */ + int next_item_length; /* Length of next item in the pattern */ + /* ------------------------------------------------------------------ */ +} pcre_callout_block; + +/* Indirection for store get and free functions. These can be set to +alternative malloc/free functions if required. Special ones are used in the +non-recursive case for "frames". There is also an optional callout function +that is triggered by the (?) regex item. For Virtual Pascal, these definitions +have to take another form. */ + +#ifndef VPCOMPAT +PCRE_DATA_SCOPE void *(*pcre_malloc)(size_t); +PCRE_DATA_SCOPE void (*pcre_free)(void *); +PCRE_DATA_SCOPE void *(*pcre_stack_malloc)(size_t); +PCRE_DATA_SCOPE void (*pcre_stack_free)(void *); +PCRE_DATA_SCOPE int (*pcre_callout)(pcre_callout_block *); +#else /* VPCOMPAT */ +PCRE_DATA_SCOPE void *pcre_malloc(size_t); +PCRE_DATA_SCOPE void pcre_free(void *); +PCRE_DATA_SCOPE void *pcre_stack_malloc(size_t); +PCRE_DATA_SCOPE void pcre_stack_free(void *); +PCRE_DATA_SCOPE int pcre_callout(pcre_callout_block *); +#endif /* VPCOMPAT */ + +/* Exported PCRE functions */ + +PCRE_DATA_SCOPE pcre *pcre_compile(const char *, int, const char **, int *, + const unsigned char *); +PCRE_DATA_SCOPE pcre *pcre_compile2(const char *, int, int *, const char **, + int *, const unsigned char *); +PCRE_DATA_SCOPE int pcre_config(int, void *); +PCRE_DATA_SCOPE int pcre_copy_named_substring(const pcre *, const char *, + int *, int, const char *, char *, int); +PCRE_DATA_SCOPE int pcre_copy_substring(const char *, int *, int, int, char *, + int); +PCRE_DATA_SCOPE int pcre_dfa_exec(const pcre *, const pcre_extra *, + const char *, int, int, int, int *, int , int *, int); +PCRE_DATA_SCOPE int pcre_exec(const pcre *, const pcre_extra *, PCRE_SPTR, + int, int, int, int *, int); +PCRE_DATA_SCOPE void pcre_free_substring(const char *); +PCRE_DATA_SCOPE void pcre_free_substring_list(const char **); +PCRE_DATA_SCOPE int pcre_fullinfo(const pcre *, const pcre_extra *, int, + void *); +PCRE_DATA_SCOPE int pcre_get_named_substring(const pcre *, const char *, + int *, int, const char *, const char **); +PCRE_DATA_SCOPE int pcre_get_stringnumber(const pcre *, const char *); +PCRE_DATA_SCOPE int pcre_get_stringtable_entries(const pcre *, const char *, + char **, char **); +PCRE_DATA_SCOPE int pcre_get_substring(const char *, int *, int, int, + const char **); +PCRE_DATA_SCOPE int pcre_get_substring_list(const char *, int *, int, + const char ***); +PCRE_DATA_SCOPE int pcre_info(const pcre *, int *, int *); +PCRE_DATA_SCOPE const unsigned char *pcre_maketables(void); +PCRE_DATA_SCOPE int pcre_refcount(pcre *, int); +PCRE_DATA_SCOPE pcre_extra *pcre_study(const pcre *, int, const char **); +PCRE_DATA_SCOPE const char *pcre_version(void); + +#ifdef __cplusplus +} /* extern "C" */ +#endif + +#endif /* End of pcre.h */ ============================================================ --- pcre/pcre_chartables.c aa6b49b4d832ffdad9d5ca3fd8013ec371fc417a +++ pcre/pcre_chartables.c aa6b49b4d832ffdad9d5ca3fd8013ec371fc417a @@ -0,0 +1,183 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + +/* This file is automatically written by the dftables auxiliary +program. If you edit it by hand, you might like to edit the Makefile to +prevent its ever being regenerated. + +This file contains the default tables for characters with codes less than +128 (ASCII characters). These tables are used when no external tables are +passed to PCRE. */ + +const unsigned char _pcre_default_tables[] = { + +/* This table is a lower casing table. */ + + 0, 1, 2, 3, 4, 5, 6, 7, + 8, 9, 10, 11, 12, 13, 14, 15, + 16, 17, 18, 19, 20, 21, 22, 23, + 24, 25, 26, 27, 28, 29, 30, 31, + 32, 33, 34, 35, 36, 37, 38, 39, + 40, 41, 42, 43, 44, 45, 46, 47, + 48, 49, 50, 51, 52, 53, 54, 55, + 56, 57, 58, 59, 60, 61, 62, 63, + 64, 97, 98, 99,100,101,102,103, + 104,105,106,107,108,109,110,111, + 112,113,114,115,116,117,118,119, + 120,121,122, 91, 92, 93, 94, 95, + 96, 97, 98, 99,100,101,102,103, + 104,105,106,107,108,109,110,111, + 112,113,114,115,116,117,118,119, + 120,121,122,123,124,125,126,127, + 128,129,130,131,132,133,134,135, + 136,137,138,139,140,141,142,143, + 144,145,146,147,148,149,150,151, + 152,153,154,155,156,157,158,159, + 160,161,162,163,164,165,166,167, + 168,169,170,171,172,173,174,175, + 176,177,178,179,180,181,182,183, + 184,185,186,187,188,189,190,191, + 192,193,194,195,196,197,198,199, + 200,201,202,203,204,205,206,207, + 208,209,210,211,212,213,214,215, + 216,217,218,219,220,221,222,223, + 224,225,226,227,228,229,230,231, + 232,233,234,235,236,237,238,239, + 240,241,242,243,244,245,246,247, + 248,249,250,251,252,253,254,255, + +/* This table is a case flipping table. */ + + 0, 1, 2, 3, 4, 5, 6, 7, + 8, 9, 10, 11, 12, 13, 14, 15, + 16, 17, 18, 19, 20, 21, 22, 23, + 24, 25, 26, 27, 28, 29, 30, 31, + 32, 33, 34, 35, 36, 37, 38, 39, + 40, 41, 42, 43, 44, 45, 46, 47, + 48, 49, 50, 51, 52, 53, 54, 55, + 56, 57, 58, 59, 60, 61, 62, 63, + 64, 97, 98, 99,100,101,102,103, + 104,105,106,107,108,109,110,111, + 112,113,114,115,116,117,118,119, + 120,121,122, 91, 92, 93, 94, 95, + 96, 65, 66, 67, 68, 69, 70, 71, + 72, 73, 74, 75, 76, 77, 78, 79, + 80, 81, 82, 83, 84, 85, 86, 87, + 88, 89, 90,123,124,125,126,127, + 128,129,130,131,132,133,134,135, + 136,137,138,139,140,141,142,143, + 144,145,146,147,148,149,150,151, + 152,153,154,155,156,157,158,159, + 160,161,162,163,164,165,166,167, + 168,169,170,171,172,173,174,175, + 176,177,178,179,180,181,182,183, + 184,185,186,187,188,189,190,191, + 192,193,194,195,196,197,198,199, + 200,201,202,203,204,205,206,207, + 208,209,210,211,212,213,214,215, + 216,217,218,219,220,221,222,223, + 224,225,226,227,228,229,230,231, + 232,233,234,235,236,237,238,239, + 240,241,242,243,244,245,246,247, + 248,249,250,251,252,253,254,255, + +/* This table contains bit maps for various character classes. +Each map is 32 bytes long and the bits run from the least +significant end of each byte. The classes that have their own +maps are: space, xdigit, digit, upper, lower, word, graph +print, punct, and cntrl. Other classes are built from combinations. */ + + 0x00,0x3e,0x00,0x00,0x01,0x00,0x00,0x00, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + + 0x00,0x00,0x00,0x00,0x00,0x00,0xff,0x03, + 0x7e,0x00,0x00,0x00,0x7e,0x00,0x00,0x00, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + + 0x00,0x00,0x00,0x00,0x00,0x00,0xff,0x03, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + 0xfe,0xff,0xff,0x07,0x00,0x00,0x00,0x00, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + 0x00,0x00,0x00,0x00,0xfe,0xff,0xff,0x07, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + + 0x00,0x00,0x00,0x00,0x00,0x00,0xff,0x03, + 0xfe,0xff,0xff,0x87,0xfe,0xff,0xff,0x07, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + + 0x00,0x00,0x00,0x00,0xfe,0xff,0xff,0xff, + 0xff,0xff,0xff,0xff,0xff,0xff,0xff,0x7f, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + + 0x00,0x00,0x00,0x00,0xff,0xff,0xff,0xff, + 0xff,0xff,0xff,0xff,0xff,0xff,0xff,0x7f, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + + 0x00,0x00,0x00,0x00,0xfe,0xff,0x00,0xfc, + 0x01,0x00,0x00,0xf8,0x01,0x00,0x00,0x78, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + + 0xff,0xff,0xff,0xff,0x00,0x00,0x00,0x00, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x80, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, + +/* This table identifies various classes of character by individual bits: + 0x01 white space character + 0x02 letter + 0x04 decimal digit + 0x08 hexadecimal digit + 0x10 alphanumeric or '_' + 0x80 regular expression metacharacter or binary zero +*/ + + 0x80,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 0- 7 */ + 0x00,0x01,0x01,0x00,0x01,0x01,0x00,0x00, /* 8- 15 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 16- 23 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 24- 31 */ + 0x01,0x00,0x00,0x00,0x80,0x00,0x00,0x00, /* - ' */ + 0x80,0x80,0x80,0x80,0x00,0x00,0x80,0x00, /* ( - / */ + 0x1c,0x1c,0x1c,0x1c,0x1c,0x1c,0x1c,0x1c, /* 0 - 7 */ + 0x1c,0x1c,0x00,0x00,0x00,0x00,0x00,0x80, /* 8 - ? */ + 0x00,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x12, /* @ - G */ + 0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12, /* H - O */ + 0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12, /* P - W */ + 0x12,0x12,0x12,0x80,0x00,0x00,0x80,0x10, /* X - _ */ + 0x00,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x12, /* ` - g */ + 0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12, /* h - o */ + 0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12, /* p - w */ + 0x12,0x12,0x12,0x80,0x80,0x00,0x00,0x00, /* x -127 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 128-135 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 136-143 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 144-151 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 152-159 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 160-167 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 168-175 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 176-183 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 184-191 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 192-199 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 200-207 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 208-215 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 216-223 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 224-231 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 232-239 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 240-247 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00};/* 248-255 */ + +/* End of chartables.c */ ============================================================ --- pcre/pcre_compile.c 2a7d0d4b9860bb85f50197170d47926a8526269b +++ pcre/pcre_compile.c 2a7d0d4b9860bb85f50197170d47926a8526269b @@ -0,0 +1,5281 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + +/* PCRE is a library of functions to support regular expressions whose syntax +and semantics are as close as possible to those of the Perl 5 language. + + Written by Philip Hazel + Copyright (c) 1997-2006 University of Cambridge + +----------------------------------------------------------------------------- +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + + * Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimer. + + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + * Neither the name of the University of Cambridge nor the names of its + contributors may be used to endorse or promote products derived from + this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE +LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR +CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +----------------------------------------------------------------------------- +*/ + + +/* This module contains the external function pcre_compile(), along with +supporting internal functions that are not used by other modules. */ + + +#define NLBLOCK cd /* The block containing newline information */ +#include "pcre_internal.h" + + +/* When DEBUG is defined, we need the pcre_printint() function, which is also +used by pcretest. DEBUG is not defined when building a production library. */ + +#ifdef DEBUG +#include "pcre_printint.src" +#endif + + + +/************************************************* +* Code parameters and static tables * +*************************************************/ + +/* Maximum number of items on the nested bracket stacks at compile time. This +applies to the nesting of all kinds of parentheses. It does not limit +un-nested, non-capturing parentheses. This number can be made bigger if +necessary - it is used to dimension one int and one unsigned char vector at +compile time. */ + +#define BRASTACK_SIZE 200 + + +/* Table for handling escaped characters in the range '0'-'z'. Positive returns +are simple data values; negative values are for special things like \d and so +on. Zero means further processing is needed (for things like \x), or the escape +is invalid. */ + +#if !EBCDIC /* This is the "normal" table for ASCII systems */ +static const short int escapes[] = { + 0, 0, 0, 0, 0, 0, 0, 0, /* 0 - 7 */ + 0, 0, ':', ';', '<', '=', '>', '?', /* 8 - ? */ + '@', -ESC_A, -ESC_B, -ESC_C, -ESC_D, -ESC_E, 0, -ESC_G, /* @ - G */ + 0, 0, 0, 0, 0, 0, 0, 0, /* H - O */ +-ESC_P, -ESC_Q, 0, -ESC_S, 0, 0, 0, -ESC_W, /* P - W */ +-ESC_X, 0, -ESC_Z, '[', '\\', ']', '^', '_', /* X - _ */ + '`', 7, -ESC_b, 0, -ESC_d, ESC_e, ESC_f, 0, /* ` - g */ + 0, 0, 0, 0, 0, 0, ESC_n, 0, /* h - o */ +-ESC_p, 0, ESC_r, -ESC_s, ESC_tee, 0, 0, -ESC_w, /* p - w */ + 0, 0, -ESC_z /* x - z */ +}; + +#else /* This is the "abnormal" table for EBCDIC systems */ +static const short int escapes[] = { +/* 48 */ 0, 0, 0, '.', '<', '(', '+', '|', +/* 50 */ '&', 0, 0, 0, 0, 0, 0, 0, +/* 58 */ 0, 0, '!', '$', '*', ')', ';', '~', +/* 60 */ '-', '/', 0, 0, 0, 0, 0, 0, +/* 68 */ 0, 0, '|', ',', '%', '_', '>', '?', +/* 70 */ 0, 0, 0, 0, 0, 0, 0, 0, +/* 78 */ 0, '`', ':', '#', '@', '\'', '=', '"', +/* 80 */ 0, 7, -ESC_b, 0, -ESC_d, ESC_e, ESC_f, 0, +/* 88 */ 0, 0, 0, '{', 0, 0, 0, 0, +/* 90 */ 0, 0, 0, 'l', 0, ESC_n, 0, -ESC_p, +/* 98 */ 0, ESC_r, 0, '}', 0, 0, 0, 0, +/* A0 */ 0, '~', -ESC_s, ESC_tee, 0, 0, -ESC_w, 0, +/* A8 */ 0,-ESC_z, 0, 0, 0, '[', 0, 0, +/* B0 */ 0, 0, 0, 0, 0, 0, 0, 0, +/* B8 */ 0, 0, 0, 0, 0, ']', '=', '-', +/* C0 */ '{',-ESC_A, -ESC_B, -ESC_C, -ESC_D,-ESC_E, 0, -ESC_G, +/* C8 */ 0, 0, 0, 0, 0, 0, 0, 0, +/* D0 */ '}', 0, 0, 0, 0, 0, 0, -ESC_P, +/* D8 */-ESC_Q, 0, 0, 0, 0, 0, 0, 0, +/* E0 */ '\\', 0, -ESC_S, 0, 0, 0, -ESC_W, -ESC_X, +/* E8 */ 0,-ESC_Z, 0, 0, 0, 0, 0, 0, +/* F0 */ 0, 0, 0, 0, 0, 0, 0, 0, +/* F8 */ 0, 0, 0, 0, 0, 0, 0, 0 +}; +#endif + + +/* Tables of names of POSIX character classes and their lengths. The list is +terminated by a zero length entry. The first three must be alpha, lower, upper, +as this is assumed for handling case independence. */ + +static const char *const posix_names[] = { + "alpha", "lower", "upper", + "alnum", "ascii", "blank", "cntrl", "digit", "graph", + "print", "punct", "space", "word", "xdigit" }; + +static const uschar posix_name_lengths[] = { + 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 }; + +/* Table of class bit maps for each POSIX class. Each class is formed from a +base map, with an optional addition or removal of another map. Then, for some +classes, there is some additional tweaking: for [:blank:] the vertical space +characters are removed, and for [:alpha:] and [:alnum:] the underscore +character is removed. The triples in the table consist of the base map offset, +second map offset or -1 if no second map, and a non-negative value for map +addition or a negative value for map subtraction (if there are two maps). The +absolute value of the third field has these meanings: 0 => no tweaking, 1 => +remove vertical space characters, 2 => remove underscore. */ + +static const int posix_class_maps[] = { + cbit_word, cbit_digit, -2, /* alpha */ + cbit_lower, -1, 0, /* lower */ + cbit_upper, -1, 0, /* upper */ + cbit_word, -1, 2, /* alnum - word without underscore */ + cbit_print, cbit_cntrl, 0, /* ascii */ + cbit_space, -1, 1, /* blank - a GNU extension */ + cbit_cntrl, -1, 0, /* cntrl */ + cbit_digit, -1, 0, /* digit */ + cbit_graph, -1, 0, /* graph */ + cbit_print, -1, 0, /* print */ + cbit_punct, -1, 0, /* punct */ + cbit_space, -1, 0, /* space */ + cbit_word, -1, 0, /* word - a Perl extension */ + cbit_xdigit,-1, 0 /* xdigit */ +}; + + +/* The texts of compile-time error messages. These are "char *" because they +are passed to the outside world. */ + +static const char *error_texts[] = { + "no error", + "\\ at end of pattern", + "\\c at end of pattern", + "unrecognized character follows \\", + "numbers out of order in {} quantifier", + /* 5 */ + "number too big in {} quantifier", + "missing terminating ] for character class", + "invalid escape sequence in character class", + "range out of order in character class", + "nothing to repeat", + /* 10 */ + "operand of unlimited repeat could match the empty string", + "internal error: unexpected repeat", + "unrecognized character after (?", + "POSIX named classes are supported only within a class", + "missing )", + /* 15 */ + "reference to non-existent subpattern", + "erroffset passed as NULL", + "unknown option bit(s) set", + "missing ) after comment", + "parentheses nested too deeply", + /* 20 */ + "regular expression too large", + "failed to get memory", + "unmatched parentheses", + "internal error: code overflow", + "unrecognized character after (?<", + /* 25 */ + "lookbehind assertion is not fixed length", + "malformed number or name after (?(", + "conditional group contains more than two branches", + "assertion expected after (?(", + "(?R or (?digits must be followed by )", + /* 30 */ + "unknown POSIX class name", + "POSIX collating elements are not supported", + "this version of PCRE is not compiled with PCRE_UTF8 support", + "spare error", + "character value in \\x{...} sequence is too large", + /* 35 */ + "invalid condition (?(0)", + "\\C not allowed in lookbehind assertion", + "PCRE does not support \\L, \\l, \\N, \\U, or \\u", + "number after (?C is > 255", + "closing ) for (?C expected", + /* 40 */ + "recursive call could loop indefinitely", + "unrecognized character after (?P", + "syntax error after (?P", + "two named subpatterns have the same name", + "invalid UTF-8 string", + /* 45 */ + "support for \\P, \\p, and \\X has not been compiled", + "malformed \\P or \\p sequence", + "unknown property name after \\P or \\p", + "subpattern name is too long (maximum 32 characters)", + "too many named subpatterns (maximum 10,000)", + /* 50 */ + "repeated subpattern is too long", + "octal value is greater than \\377 (not in UTF-8 mode)" +}; + + +/* Table to identify digits and hex digits. This is used when compiling +patterns. Note that the tables in chartables are dependent on the locale, and +may mark arbitrary characters as digits - but the PCRE compiling code expects +to handle only 0-9, a-z, and A-Z as digits when compiling. That is why we have +a private table here. It costs 256 bytes, but it is a lot faster than doing +character value tests (at least in some simple cases I timed), and in some +applications one wants PCRE to compile efficiently as well as match +efficiently. + +For convenience, we use the same bit definitions as in chartables: + + 0x04 decimal digit + 0x08 hexadecimal digit + +Then we can use ctype_digit and ctype_xdigit in the code. */ + +#if !EBCDIC /* This is the "normal" case, for ASCII systems */ +static const unsigned char digitab[] = + { + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 0- 7 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 8- 15 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 16- 23 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 24- 31 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* - ' */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* ( - / */ + 0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /* 0 - 7 */ + 0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00, /* 8 - ? */ + 0x00,0x08,0x08,0x08,0x08,0x08,0x08,0x00, /* @ - G */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* H - O */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* P - W */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* X - _ */ + 0x00,0x08,0x08,0x08,0x08,0x08,0x08,0x00, /* ` - g */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* h - o */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* p - w */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* x -127 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 128-135 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 136-143 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 144-151 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 152-159 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 160-167 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 168-175 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 176-183 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 184-191 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 192-199 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 200-207 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 208-215 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 216-223 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 224-231 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 232-239 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 240-247 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00};/* 248-255 */ + +#else /* This is the "abnormal" case, for EBCDIC systems */ +static const unsigned char digitab[] = + { + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 0- 7 0 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 8- 15 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 16- 23 10 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 24- 31 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 32- 39 20 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 40- 47 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 48- 55 30 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 56- 63 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* - 71 40 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 72- | */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* & - 87 50 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 88- ¬ */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* - -103 60 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 104- ? */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 112-119 70 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 120- " */ + 0x00,0x08,0x08,0x08,0x08,0x08,0x08,0x00, /* 128- g 80 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* h -143 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 144- p 90 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* q -159 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 160- x A0 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* y -175 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* ^ -183 B0 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 184-191 */ + 0x00,0x08,0x08,0x08,0x08,0x08,0x08,0x00, /* { - G C0 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* H -207 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* } - P D0 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* Q -223 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* \ - X E0 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* Y -239 */ + 0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /* 0 - 7 F0 */ + 0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/* 8 -255 */ + +static const unsigned char ebcdic_chartab[] = { /* chartable partial dup */ + 0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /* 0- 7 */ + 0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /* 8- 15 */ + 0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /* 16- 23 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 24- 31 */ + 0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /* 32- 39 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 40- 47 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 48- 55 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 56- 63 */ + 0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* - 71 */ + 0x00,0x00,0x00,0x80,0x00,0x80,0x80,0x80, /* 72- | */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* & - 87 */ + 0x00,0x00,0x00,0x80,0x80,0x80,0x00,0x00, /* 88- ¬ */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* - -103 */ + 0x00,0x00,0x00,0x00,0x00,0x10,0x00,0x80, /* 104- ? */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 112-119 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 120- " */ + 0x00,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x12, /* 128- g */ + 0x12,0x12,0x00,0x00,0x00,0x00,0x00,0x00, /* h -143 */ + 0x00,0x12,0x12,0x12,0x12,0x12,0x12,0x12, /* 144- p */ + 0x12,0x12,0x00,0x00,0x00,0x00,0x00,0x00, /* q -159 */ + 0x00,0x00,0x12,0x12,0x12,0x12,0x12,0x12, /* 160- x */ + 0x12,0x12,0x00,0x00,0x00,0x00,0x00,0x00, /* y -175 */ + 0x80,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* ^ -183 */ + 0x00,0x00,0x80,0x00,0x00,0x00,0x00,0x00, /* 184-191 */ + 0x80,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x12, /* { - G */ + 0x12,0x12,0x00,0x00,0x00,0x00,0x00,0x00, /* H -207 */ + 0x00,0x12,0x12,0x12,0x12,0x12,0x12,0x12, /* } - P */ + 0x12,0x12,0x00,0x00,0x00,0x00,0x00,0x00, /* Q -223 */ + 0x00,0x00,0x12,0x12,0x12,0x12,0x12,0x12, /* \ - X */ + 0x12,0x12,0x00,0x00,0x00,0x00,0x00,0x00, /* Y -239 */ + 0x1c,0x1c,0x1c,0x1c,0x1c,0x1c,0x1c,0x1c, /* 0 - 7 */ + 0x1c,0x1c,0x00,0x00,0x00,0x00,0x00,0x00};/* 8 -255 */ +#endif + + +/* Definition to allow mutual recursion */ + +static BOOL + compile_regex(int, int, int *, uschar **, const uschar **, int *, BOOL, int, + int *, int *, branch_chain *, compile_data *); + + + +/************************************************* +* Handle escapes * +*************************************************/ + +/* This function is called when a \ has been encountered. It either returns a +positive value for a simple escape such as \n, or a negative value which +encodes one of the more complicated things such as \d. When UTF-8 is enabled, +a positive value greater than 255 may be returned. On entry, ptr is pointing at +the \. On exit, it is on the final character of the escape sequence. + +Arguments: + ptrptr points to the pattern position pointer + errorcodeptr points to the errorcode variable + bracount number of previous extracting brackets + options the options bits + isclass TRUE if inside a character class + +Returns: zero or positive => a data character + negative => a special escape sequence + on error, errorptr is set +*/ + +static int +check_escape(const uschar **ptrptr, int *errorcodeptr, int bracount, + int options, BOOL isclass) +{ +BOOL utf8 = (options & PCRE_UTF8) != 0; +const uschar *ptr = *ptrptr + 1; +int c, i; + +GETCHARINCTEST(c, ptr); /* Get character value, increment pointer */ +ptr--; /* Set pointer back to the last byte */ + +/* If backslash is at the end of the pattern, it's an error. */ + +if (c == 0) *errorcodeptr = ERR1; + +/* Non-alphamerics are literals. For digits or letters, do an initial lookup in +a table. A non-zero result is something that can be returned immediately. +Otherwise further processing may be required. */ + +#if !EBCDIC /* ASCII coding */ +else if (c < '0' || c > 'z') {} /* Not alphameric */ +else if ((i = escapes[c - '0']) != 0) c = i; + +#else /* EBCDIC coding */ +else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {} /* Not alphameric */ +else if ((i = escapes[c - 0x48]) != 0) c = i; +#endif + +/* Escapes that need further processing, or are illegal. */ + +else + { + const uschar *oldptr; + switch (c) + { + /* A number of Perl escapes are not handled by PCRE. We give an explicit + error. */ + + case 'l': + case 'L': + case 'N': + case 'u': + case 'U': + *errorcodeptr = ERR37; + break; + + /* The handling of escape sequences consisting of a string of digits + starting with one that is not zero is not straightforward. By experiment, + the way Perl works seems to be as follows: + + Outside a character class, the digits are read as a decimal number. If the + number is less than 10, or if there are that many previous extracting + left brackets, then it is a back reference. Otherwise, up to three octal + digits are read to form an escaped byte. Thus \123 is likely to be octal + 123 (cf \0123, which is octal 012 followed by the literal 3). If the octal + value is greater than 377, the least significant 8 bits are taken. Inside a + character class, \ followed by a digit is always an octal number. */ + + case '1': case '2': case '3': case '4': case '5': + case '6': case '7': case '8': case '9': + + if (!isclass) + { + oldptr = ptr; + c -= '0'; + while ((digitab[ptr[1]] & ctype_digit) != 0) + c = c * 10 + *(++ptr) - '0'; + if (c < 10 || c <= bracount) + { + c = -(ESC_REF + c); + break; + } + ptr = oldptr; /* Put the pointer back and fall through */ + } + + /* Handle an octal number following \. If the first digit is 8 or 9, Perl + generates a binary zero byte and treats the digit as a following literal. + Thus we have to pull back the pointer by one. */ + + if ((c = *ptr) >= '8') + { + ptr--; + c = 0; + break; + } + + /* \0 always starts an octal number, but we may drop through to here with a + larger first octal digit. The original code used just to take the least + significant 8 bits of octal numbers (I think this is what early Perls used + to do). Nowadays we allow for larger numbers in UTF-8 mode, but no more + than 3 octal digits. */ + + case '0': + c -= '0'; + while(i++ < 2 && ptr[1] >= '0' && ptr[1] <= '7') + c = c * 8 + *(++ptr) - '0'; + if (!utf8 && c > 255) *errorcodeptr = ERR51; + break; + + /* \x is complicated. \x{ddd} is a character number which can be greater + than 0xff in utf8 mode, but only if the ddd are hex digits. If not, { is + treated as a data character. */ + + case 'x': + if (ptr[1] == '{') + { + const uschar *pt = ptr + 2; + int count = 0; + + c = 0; + while ((digitab[*pt] & ctype_xdigit) != 0) + { + register int cc = *pt++; + if (c == 0 && cc == '0') continue; /* Leading zeroes */ + count++; + +#if !EBCDIC /* ASCII coding */ + if (cc >= 'a') cc -= 32; /* Convert to upper case */ + c = (c << 4) + cc - ((cc < 'A')? '0' : ('A' - 10)); +#else /* EBCDIC coding */ + if (cc >= 'a' && cc <= 'z') cc += 64; /* Convert to upper case */ + c = (c << 4) + cc - ((cc >= '0')? '0' : ('A' - 10)); +#endif + } + + if (*pt == '}') + { + if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34; + ptr = pt; + break; + } + + /* If the sequence of hex digits does not end with '}', then we don't + recognize this construct; fall through to the normal \x handling. */ + } + + /* Read just a single-byte hex-defined char */ + + c = 0; + while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0) + { + int cc; /* Some compilers don't like ++ */ + cc = *(++ptr); /* in initializers */ +#if !EBCDIC /* ASCII coding */ + if (cc >= 'a') cc -= 32; /* Convert to upper case */ + c = c * 16 + cc - ((cc < 'A')? '0' : ('A' - 10)); +#else /* EBCDIC coding */ + if (cc <= 'z') cc += 64; /* Convert to upper case */ + c = c * 16 + cc - ((cc >= '0')? '0' : ('A' - 10)); +#endif + } + break; + + /* Other special escapes not starting with a digit are straightforward */ + + case 'c': + c = *(++ptr); + if (c == 0) + { + *errorcodeptr = ERR2; + return 0; + } + + /* A letter is upper-cased; then the 0x40 bit is flipped. This coding + is ASCII-specific, but then the whole concept of \cx is ASCII-specific. + (However, an EBCDIC equivalent has now been added.) */ + +#if !EBCDIC /* ASCII coding */ + if (c >= 'a' && c <= 'z') c -= 32; + c ^= 0x40; +#else /* EBCDIC coding */ + if (c >= 'a' && c <= 'z') c += 64; + c ^= 0xC0; +#endif + break; + + /* PCRE_EXTRA enables extensions to Perl in the matter of escapes. Any + other alphameric following \ is an error if PCRE_EXTRA was set; otherwise, + for Perl compatibility, it is a literal. This code looks a bit odd, but + there used to be some cases other than the default, and there may be again + in future, so I haven't "optimized" it. */ + + default: + if ((options & PCRE_EXTRA) != 0) switch(c) + { + default: + *errorcodeptr = ERR3; + break; + } + break; + } + } + +*ptrptr = ptr; +return c; +} + + + +#ifdef SUPPORT_UCP +/************************************************* +* Handle \P and \p * +*************************************************/ + +/* This function is called after \P or \p has been encountered, provided that +PCRE is compiled with support for Unicode properties. On entry, ptrptr is +pointing at the P or p. On exit, it is pointing at the final character of the +escape sequence. + +Argument: + ptrptr points to the pattern position pointer + negptr points to a boolean that is set TRUE for negation else FALSE + dptr points to an int that is set to the detailed property value + errorcodeptr points to the error code variable + +Returns: type value from ucp_type_table, or -1 for an invalid type +*/ + +static int +get_ucp(const uschar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr) +{ +int c, i, bot, top; +const uschar *ptr = *ptrptr; +char name[32]; + +c = *(++ptr); +if (c == 0) goto ERROR_RETURN; + +*negptr = FALSE; + +/* \P or \p can be followed by a name in {}, optionally preceded by ^ for +negation. */ + +if (c == '{') + { + if (ptr[1] == '^') + { + *negptr = TRUE; + ptr++; + } + for (i = 0; i < sizeof(name) - 1; i++) + { + c = *(++ptr); + if (c == 0) goto ERROR_RETURN; + if (c == '}') break; + name[i] = c; + } + if (c !='}') goto ERROR_RETURN; + name[i] = 0; + } + +/* Otherwise there is just one following character */ + +else + { + name[0] = c; + name[1] = 0; + } + +*ptrptr = ptr; + +/* Search for a recognized property name using binary chop */ + +bot = 0; +top = _pcre_utt_size; + +while (bot < top) + { + i = (bot + top) >> 1; + c = strcmp(name, _pcre_utt[i].name); + if (c == 0) + { + *dptr = _pcre_utt[i].value; + return _pcre_utt[i].type; + } + if (c > 0) bot = i + 1; else top = i; + } + +*errorcodeptr = ERR47; +*ptrptr = ptr; +return -1; + +ERROR_RETURN: +*errorcodeptr = ERR46; +*ptrptr = ptr; +return -1; +} +#endif + + + + +/************************************************* +* Check for counted repeat * +*************************************************/ + +/* This function is called when a '{' is encountered in a place where it might +start a quantifier. It looks ahead to see if it really is a quantifier or not. +It is only a quantifier if it is one of the forms {ddd} {ddd,} or {ddd,ddd} +where the ddds are digits. + +Arguments: + p pointer to the first char after '{' + +Returns: TRUE or FALSE +*/ + +static BOOL +is_counted_repeat(const uschar *p) +{ +if ((digitab[*p++] & ctype_digit) == 0) return FALSE; +while ((digitab[*p] & ctype_digit) != 0) p++; +if (*p == '}') return TRUE; + +if (*p++ != ',') return FALSE; +if (*p == '}') return TRUE; + +if ((digitab[*p++] & ctype_digit) == 0) return FALSE; +while ((digitab[*p] & ctype_digit) != 0) p++; + +return (*p == '}'); +} + + + +/************************************************* +* Read repeat counts * +*************************************************/ + +/* Read an item of the form {n,m} and return the values. This is called only +after is_counted_repeat() has confirmed that a repeat-count quantifier exists, +so the syntax is guaranteed to be correct, but we need to check the values. + +Arguments: + p pointer to first char after '{' + minp pointer to int for min + maxp pointer to int for max + returned as -1 if no max + errorcodeptr points to error code variable + +Returns: pointer to '}' on success; + current ptr on error, with errorcodeptr set non-zero +*/ + +static const uschar * +read_repeat_counts(const uschar *p, int *minp, int *maxp, int *errorcodeptr) +{ +int min = 0; +int max = -1; + +/* Read the minimum value and do a paranoid check: a negative value indicates +an integer overflow. */ + +while ((digitab[*p] & ctype_digit) != 0) min = min * 10 + *p++ - '0'; +if (min < 0 || min > 65535) + { + *errorcodeptr = ERR5; + return p; + } + +/* Read the maximum value if there is one, and again do a paranoid on its size. +Also, max must not be less than min. */ + +if (*p == '}') max = min; else + { + if (*(++p) != '}') + { + max = 0; + while((digitab[*p] & ctype_digit) != 0) max = max * 10 + *p++ - '0'; + if (max < 0 || max > 65535) + { + *errorcodeptr = ERR5; + return p; + } + if (max < min) + { + *errorcodeptr = ERR4; + return p; + } + } + } + +/* Fill in the required variables, and pass back the pointer to the terminating +'}'. */ + +*minp = min; +*maxp = max; +return p; +} + + + +/************************************************* +* Find forward referenced named subpattern * +*************************************************/ + +/* This function scans along a pattern looking for capturing subpatterns, and +counting them. If it finds a named pattern that matches the name it is given, +it returns its number. This is used for forward references to named +subpatterns. We know that if (?P< is encountered, the name will be terminated +by '>' because that is checked in the first pass. + +Arguments: + pointer current position in the pattern + count current count of capturing parens + name name to seek + namelen name length + +Returns: the number of the named subpattern, or -1 if not found +*/ + +static int +find_named_parens(const uschar *ptr, int count, const uschar *name, int namelen) +{ +const uschar *thisname; +for (; *ptr != 0; ptr++) + { + if (*ptr == '\\' && ptr[1] != 0) { ptr++; continue; } + if (*ptr != '(') continue; + if (ptr[1] != '?') { count++; continue; } + if (ptr[2] == '(') { ptr += 2; continue; } + if (ptr[2] != 'P' || ptr[3] != '<') continue; + count++; + ptr += 4; + thisname = ptr; + while (*ptr != '>') ptr++; + if (namelen == ptr - thisname && strncmp(name, thisname, namelen) == 0) + return count; + } +return -1; +} + + + +/************************************************* +* Find first significant op code * +*************************************************/ + +/* This is called by several functions that scan a compiled expression looking +for a fixed first character, or an anchoring op code etc. It skips over things +that do not influence this. For some calls, a change of option is important. +For some calls, it makes sense to skip negative forward and all backward +assertions, and also the \b assertion; for others it does not. + +Arguments: + code pointer to the start of the group + options pointer to external options + optbit the option bit whose changing is significant, or + zero if none are + skipassert TRUE if certain assertions are to be skipped + +Returns: pointer to the first significant opcode +*/ + +static const uschar* +first_significant_code(const uschar *code, int *options, int optbit, + BOOL skipassert) +{ +for (;;) + { + switch ((int)*code) + { + case OP_OPT: + if (optbit > 0 && ((int)code[1] & optbit) != (*options & optbit)) + *options = (int)code[1]; + code += 2; + break; + + case OP_ASSERT_NOT: + case OP_ASSERTBACK: + case OP_ASSERTBACK_NOT: + if (!skipassert) return code; + do code += GET(code, 1); while (*code == OP_ALT); + code += _pcre_OP_lengths[*code]; + break; + + case OP_WORD_BOUNDARY: + case OP_NOT_WORD_BOUNDARY: + if (!skipassert) return code; + /* Fall through */ + + case OP_CALLOUT: + case OP_CREF: + case OP_BRANUMBER: + code += _pcre_OP_lengths[*code]; + break; + + default: + return code; + } + } +/* Control never reaches here */ +} + + + + +/************************************************* +* Find the fixed length of a pattern * +*************************************************/ + +/* Scan a pattern and compute the fixed length of subject that will match it, +if the length is fixed. This is needed for dealing with backward assertions. +In UTF8 mode, the result is in characters rather than bytes. + +Arguments: + code points to the start of the pattern (the bracket) + options the compiling options + +Returns: the fixed length, or -1 if there is no fixed length, + or -2 if \C was encountered +*/ + +static int +find_fixedlength(uschar *code, int options) +{ +int length = -1; + +register int branchlength = 0; +register uschar *cc = code + 1 + LINK_SIZE; + +/* Scan along the opcodes for this branch. If we get to the end of the +branch, check the length against that of the other branches. */ + +for (;;) + { + int d; + register int op = *cc; + if (op >= OP_BRA) op = OP_BRA; + + switch (op) + { + case OP_BRA: + case OP_ONCE: + case OP_COND: + d = find_fixedlength(cc, options); + if (d < 0) return d; + branchlength += d; + do cc += GET(cc, 1); while (*cc == OP_ALT); + cc += 1 + LINK_SIZE; + break; + + /* Reached end of a branch; if it's a ket it is the end of a nested + call. If it's ALT it is an alternation in a nested call. If it is + END it's the end of the outer call. All can be handled by the same code. */ + + case OP_ALT: + case OP_KET: + case OP_KETRMAX: + case OP_KETRMIN: + case OP_END: + if (length < 0) length = branchlength; + else if (length != branchlength) return -1; + if (*cc != OP_ALT) return length; + cc += 1 + LINK_SIZE; + branchlength = 0; + break; + + /* Skip over assertive subpatterns */ + + case OP_ASSERT: + case OP_ASSERT_NOT: + case OP_ASSERTBACK: + case OP_ASSERTBACK_NOT: + do cc += GET(cc, 1); while (*cc == OP_ALT); + /* Fall through */ + + /* Skip over things that don't match chars */ + + case OP_REVERSE: + case OP_BRANUMBER: + case OP_CREF: + case OP_OPT: + case OP_CALLOUT: + case OP_SOD: + case OP_SOM: + case OP_EOD: + case OP_EODN: + case OP_CIRC: + case OP_DOLL: + case OP_NOT_WORD_BOUNDARY: + case OP_WORD_BOUNDARY: + cc += _pcre_OP_lengths[*cc]; + break; + + /* Handle literal characters */ + + case OP_CHAR: + case OP_CHARNC: + case OP_NOT: + branchlength++; + cc += 2; +#ifdef SUPPORT_UTF8 + if ((options & PCRE_UTF8) != 0) + { + while ((*cc & 0xc0) == 0x80) cc++; + } +#endif + break; + + /* Handle exact repetitions. The count is already in characters, but we + need to skip over a multibyte character in UTF8 mode. */ + + case OP_EXACT: + branchlength += GET2(cc,1); + cc += 4; +#ifdef SUPPORT_UTF8 + if ((options & PCRE_UTF8) != 0) + { + while((*cc & 0x80) == 0x80) cc++; + } +#endif + break; + + case OP_TYPEEXACT: + branchlength += GET2(cc,1); + cc += 4; + break; + + /* Handle single-char matchers */ + + case OP_PROP: + case OP_NOTPROP: + cc += 2; + /* Fall through */ + + case OP_NOT_DIGIT: + case OP_DIGIT: + case OP_NOT_WHITESPACE: + case OP_WHITESPACE: + case OP_NOT_WORDCHAR: + case OP_WORDCHAR: + case OP_ANY: + branchlength++; + cc++; + break; + + /* The single-byte matcher isn't allowed */ + + case OP_ANYBYTE: + return -2; + + /* Check a class for variable quantification */ + +#ifdef SUPPORT_UTF8 + case OP_XCLASS: + cc += GET(cc, 1) - 33; + /* Fall through */ +#endif + + case OP_CLASS: + case OP_NCLASS: + cc += 33; + + switch (*cc) + { + case OP_CRSTAR: + case OP_CRMINSTAR: + case OP_CRQUERY: + case OP_CRMINQUERY: + return -1; + + case OP_CRRANGE: + case OP_CRMINRANGE: + if (GET2(cc,1) != GET2(cc,3)) return -1; + branchlength += GET2(cc,1); + cc += 5; + break; + + default: + branchlength++; + } + break; + + /* Anything else is variable length */ + + default: + return -1; + } + } +/* Control never gets here */ +} + + + + +/************************************************* +* Scan compiled regex for numbered bracket * +*************************************************/ + +/* This little function scans through a compiled pattern until it finds a +capturing bracket with the given number. + +Arguments: + code points to start of expression + utf8 TRUE in UTF-8 mode + number the required bracket number + +Returns: pointer to the opcode for the bracket, or NULL if not found +*/ + +static const uschar * +find_bracket(const uschar *code, BOOL utf8, int number) +{ +for (;;) + { + register int c = *code; + if (c == OP_END) return NULL; + + /* XCLASS is used for classes that cannot be represented just by a bit + map. This includes negated single high-valued characters. The length in + the table is zero; the actual length is stored in the compiled code. */ + + if (c == OP_XCLASS) code += GET(code, 1); + + /* Handle bracketed group */ + + else if (c > OP_BRA) + { + int n = c - OP_BRA; + if (n > EXTRACT_BASIC_MAX) n = GET2(code, 2+LINK_SIZE); + if (n == number) return (uschar *)code; + code += _pcre_OP_lengths[OP_BRA]; + } + + /* Otherwise, we get the item's length from the table. In UTF-8 mode, opcodes + that are followed by a character may be followed by a multi-byte character. + The length in the table is a minimum, so we have to scan along to skip the + extra bytes. All opcodes are less than 128, so we can use relatively + efficient code. */ + + else + { + code += _pcre_OP_lengths[c]; + if (utf8) switch(c) + { + case OP_CHAR: + case OP_CHARNC: + case OP_EXACT: + case OP_UPTO: + case OP_MINUPTO: + case OP_STAR: + case OP_MINSTAR: + case OP_PLUS: + case OP_MINPLUS: + case OP_QUERY: + case OP_MINQUERY: + while ((*code & 0xc0) == 0x80) code++; + break; + } + } + } +} + + + +/************************************************* +* Scan compiled regex for recursion reference * +*************************************************/ + +/* This little function scans through a compiled pattern until it finds an +instance of OP_RECURSE. + +Arguments: + code points to start of expression + utf8 TRUE in UTF-8 mode + +Returns: pointer to the opcode for OP_RECURSE, or NULL if not found +*/ + +static const uschar * +find_recurse(const uschar *code, BOOL utf8) +{ +for (;;) + { + register int c = *code; + if (c == OP_END) return NULL; + if (c == OP_RECURSE) return code; + + /* XCLASS is used for classes that cannot be represented just by a bit + map. This includes negated single high-valued characters. The length in + the table is zero; the actual length is stored in the compiled code. */ + + if (c == OP_XCLASS) code += GET(code, 1); + + /* All bracketed groups have the same length. */ + + else if (c > OP_BRA) + { + code += _pcre_OP_lengths[OP_BRA]; + } + + /* Otherwise, we get the item's length from the table. In UTF-8 mode, opcodes + that are followed by a character may be followed by a multi-byte character. + The length in the table is a minimum, so we have to scan along to skip the + extra bytes. All opcodes are less than 128, so we can use relatively + efficient code. */ + + else + { + code += _pcre_OP_lengths[c]; + if (utf8) switch(c) + { + case OP_CHAR: + case OP_CHARNC: + case OP_EXACT: + case OP_UPTO: + case OP_MINUPTO: + case OP_STAR: + case OP_MINSTAR: + case OP_PLUS: + case OP_MINPLUS: + case OP_QUERY: + case OP_MINQUERY: + while ((*code & 0xc0) == 0x80) code++; + break; + } + } + } +} + + + +/************************************************* +* Scan compiled branch for non-emptiness * +*************************************************/ + +/* This function scans through a branch of a compiled pattern to see whether it +can match the empty string or not. It is called only from could_be_empty() +below. Note that first_significant_code() skips over assertions. If we hit an +unclosed bracket, we return "empty" - this means we've struck an inner bracket +whose current branch will already have been scanned. + +Arguments: + code points to start of search + endcode points to where to stop + utf8 TRUE if in UTF8 mode + +Returns: TRUE if what is matched could be empty +*/ + +static BOOL +could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8) +{ +register int c; +for (code = first_significant_code(code + 1 + LINK_SIZE, NULL, 0, TRUE); + code < endcode; + code = first_significant_code(code + _pcre_OP_lengths[c], NULL, 0, TRUE)) + { + const uschar *ccode; + + c = *code; + + if (c >= OP_BRA) + { + BOOL empty_branch; + if (GET(code, 1) == 0) return TRUE; /* Hit unclosed bracket */ + + /* Scan a closed bracket */ + + empty_branch = FALSE; + do + { + if (!empty_branch && could_be_empty_branch(code, endcode, utf8)) + empty_branch = TRUE; + code += GET(code, 1); + } + while (*code == OP_ALT); + if (!empty_branch) return FALSE; /* All branches are non-empty */ + code += 1 + LINK_SIZE; + c = *code; + } + + else switch (c) + { + /* Check for quantifiers after a class */ + +#ifdef SUPPORT_UTF8 + case OP_XCLASS: + ccode = code + GET(code, 1); + goto CHECK_CLASS_REPEAT; +#endif + + case OP_CLASS: + case OP_NCLASS: + ccode = code + 33; + +#ifdef SUPPORT_UTF8 + CHECK_CLASS_REPEAT: +#endif + + switch (*ccode) + { + case OP_CRSTAR: /* These could be empty; continue */ + case OP_CRMINSTAR: + case OP_CRQUERY: + case OP_CRMINQUERY: + break; + + default: /* Non-repeat => class must match */ + case OP_CRPLUS: /* These repeats aren't empty */ + case OP_CRMINPLUS: + return FALSE; + + case OP_CRRANGE: + case OP_CRMINRANGE: + if (GET2(ccode, 1) > 0) return FALSE; /* Minimum > 0 */ + break; + } + break; + + /* Opcodes that must match a character */ + + case OP_PROP: + case OP_NOTPROP: + case OP_EXTUNI: + case OP_NOT_DIGIT: + case OP_DIGIT: + case OP_NOT_WHITESPACE: + case OP_WHITESPACE: + case OP_NOT_WORDCHAR: + case OP_WORDCHAR: + case OP_ANY: + case OP_ANYBYTE: + case OP_CHAR: + case OP_CHARNC: + case OP_NOT: + case OP_PLUS: + case OP_MINPLUS: + case OP_EXACT: + case OP_NOTPLUS: + case OP_NOTMINPLUS: + case OP_NOTEXACT: + case OP_TYPEPLUS: + case OP_TYPEMINPLUS: + case OP_TYPEEXACT: + return FALSE; + + /* End of branch */ + + case OP_KET: + case OP_KETRMAX: + case OP_KETRMIN: + case OP_ALT: + return TRUE; + + /* In UTF-8 mode, STAR, MINSTAR, QUERY, MINQUERY, UPTO, and MINUPTO may be + followed by a multibyte character */ + +#ifdef SUPPORT_UTF8 + case OP_STAR: + case OP_MINSTAR: + case OP_QUERY: + case OP_MINQUERY: + case OP_UPTO: + case OP_MINUPTO: + if (utf8) while ((code[2] & 0xc0) == 0x80) code++; + break; +#endif + } + } + +return TRUE; +} + + + +/************************************************* +* Scan compiled regex for non-emptiness * +*************************************************/ + +/* This function is called to check for left recursive calls. We want to check +the current branch of the current pattern to see if it could match the empty +string. If it could, we must look outwards for branches at other levels, +stopping when we pass beyond the bracket which is the subject of the recursion. + +Arguments: + code points to start of the recursion + endcode points to where to stop (current RECURSE item) + bcptr points to the chain of current (unclosed) branch starts + utf8 TRUE if in UTF-8 mode + +Returns: TRUE if what is matched could be empty +*/ + +static BOOL +could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr, + BOOL utf8) +{ +while (bcptr != NULL && bcptr->current >= code) + { + if (!could_be_empty_branch(bcptr->current, endcode, utf8)) return FALSE; + bcptr = bcptr->outer; + } +return TRUE; +} + + + +/************************************************* +* Check for POSIX class syntax * +*************************************************/ + +/* This function is called when the sequence "[:" or "[." or "[=" is +encountered in a character class. It checks whether this is followed by an +optional ^ and then a sequence of letters, terminated by a matching ":]" or +".]" or "=]". + +Argument: + ptr pointer to the initial [ + endptr where to return the end pointer + cd pointer to compile data + +Returns: TRUE or FALSE +*/ + +static BOOL +check_posix_syntax(const uschar *ptr, const uschar **endptr, compile_data *cd) +{ +int terminator; /* Don't combine these lines; the Solaris cc */ +terminator = *(++ptr); /* compiler warns about "non-constant" initializer. */ +if (*(++ptr) == '^') ptr++; +while ((cd->ctypes[*ptr] & ctype_letter) != 0) ptr++; +if (*ptr == terminator && ptr[1] == ']') + { + *endptr = ptr; + return TRUE; + } +return FALSE; +} + + + + +/************************************************* +* Check POSIX class name * +*************************************************/ + +/* This function is called to check the name given in a POSIX-style class entry +such as [:alnum:]. + +Arguments: + ptr points to the first letter + len the length of the name + +Returns: a value representing the name, or -1 if unknown +*/ + +static int +check_posix_name(const uschar *ptr, int len) +{ +register int yield = 0; +while (posix_name_lengths[yield] != 0) + { + if (len == posix_name_lengths[yield] && + strncmp((const char *)ptr, posix_names[yield], len) == 0) return yield; + yield++; + } +return -1; +} + + +/************************************************* +* Adjust OP_RECURSE items in repeated group * +*************************************************/ + +/* OP_RECURSE items contain an offset from the start of the regex to the group +that is referenced. This means that groups can be replicated for fixed +repetition simply by copying (because the recursion is allowed to refer to +earlier groups that are outside the current group). However, when a group is +optional (i.e. the minimum quantifier is zero), OP_BRAZERO is inserted before +it, after it has been compiled. This means that any OP_RECURSE items within it +that refer to the group itself or any contained groups have to have their +offsets adjusted. That is the job of this function. Before it is called, the +partially compiled regex must be temporarily terminated with OP_END. + +Arguments: + group points to the start of the group + adjust the amount by which the group is to be moved + utf8 TRUE in UTF-8 mode + cd contains pointers to tables etc. + +Returns: nothing +*/ + +static void +adjust_recurse(uschar *group, int adjust, BOOL utf8, compile_data *cd) +{ +uschar *ptr = group; +while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL) + { + int offset = GET(ptr, 1); + if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust); + ptr += 1 + LINK_SIZE; + } +} + + + +/************************************************* +* Insert an automatic callout point * +*************************************************/ + +/* This function is called when the PCRE_AUTO_CALLOUT option is set, to insert +callout points before each pattern item. + +Arguments: + code current code pointer + ptr current pattern pointer + cd pointers to tables etc + +Returns: new code pointer +*/ + +static uschar * +auto_callout(uschar *code, const uschar *ptr, compile_data *cd) +{ +*code++ = OP_CALLOUT; +*code++ = 255; +PUT(code, 0, ptr - cd->start_pattern); /* Pattern offset */ +PUT(code, LINK_SIZE, 0); /* Default length */ +return code + 2*LINK_SIZE; +} + + + +/************************************************* +* Complete a callout item * +*************************************************/ + +/* A callout item contains the length of the next item in the pattern, which +we can't fill in till after we have reached the relevant point. This is used +for both automatic and manual callouts. + +Arguments: + previous_callout points to previous callout item + ptr current pattern pointer + cd pointers to tables etc + +Returns: nothing +*/ + +static void +complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd) +{ +int length = ptr - cd->start_pattern - GET(previous_callout, 2); +PUT(previous_callout, 2 + LINK_SIZE, length); +} + + + +#ifdef SUPPORT_UCP +/************************************************* +* Get othercase range * +*************************************************/ + +/* This function is passed the start and end of a class range, in UTF-8 mode +with UCP support. It searches up the characters, looking for internal ranges of +characters in the "other" case. Each call returns the next one, updating the +start address. + +Arguments: + cptr points to starting character value; updated + d end value + ocptr where to put start of othercase range + odptr where to put end of othercase range + +Yield: TRUE when range returned; FALSE when no more +*/ + +static BOOL +get_othercase_range(int *cptr, int d, int *ocptr, int *odptr) +{ +int c, othercase, next; + +for (c = *cptr; c <= d; c++) + { if ((othercase = _pcre_ucp_othercase(c)) >= 0) break; } + +if (c > d) return FALSE; + +*ocptr = othercase; +next = othercase + 1; + +for (++c; c <= d; c++) + { + if (_pcre_ucp_othercase(c) != next) break; + next++; + } + +*odptr = next - 1; +*cptr = c; + +return TRUE; +} +#endif /* SUPPORT_UCP */ + + +/************************************************* +* Compile one branch * +*************************************************/ + +/* Scan the pattern, compiling it into the code vector. If the options are +changed during the branch, the pointer is used to change the external options +bits. + +Arguments: + optionsptr pointer to the option bits + brackets points to number of extracting brackets used + codeptr points to the pointer to the current code point + ptrptr points to the current pattern pointer + errorcodeptr points to error code variable + firstbyteptr set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE) + reqbyteptr set to the last literal character required, else < 0 + bcptr points to current branch chain + cd contains pointers to tables etc. + +Returns: TRUE on success + FALSE, with *errorcodeptr set non-zero on error +*/ + +static BOOL +compile_branch(int *optionsptr, int *brackets, uschar **codeptr, + const uschar **ptrptr, int *errorcodeptr, int *firstbyteptr, + int *reqbyteptr, branch_chain *bcptr, compile_data *cd) +{ +int repeat_type, op_type; +int repeat_min = 0, repeat_max = 0; /* To please picky compilers */ +int bravalue = 0; +int greedy_default, greedy_non_default; +int firstbyte, reqbyte; +int zeroreqbyte, zerofirstbyte; +int req_caseopt, reqvary, tempreqvary; +int options = *optionsptr; +int after_manual_callout = 0; +register int c; +register uschar *code = *codeptr; +uschar *tempcode; +BOOL inescq = FALSE; +BOOL groupsetfirstbyte = FALSE; +const uschar *ptr = *ptrptr; +const uschar *tempptr; +uschar *previous = NULL; +uschar *previous_callout = NULL; +uschar classbits[32]; + +#ifdef SUPPORT_UTF8 +BOOL class_utf8; +BOOL utf8 = (options & PCRE_UTF8) != 0; +uschar *class_utf8data; +uschar utf8_char[6]; +#else +BOOL utf8 = FALSE; +#endif + +/* Set up the default and non-default settings for greediness */ + +greedy_default = ((options & PCRE_UNGREEDY) != 0); +greedy_non_default = greedy_default ^ 1; + +/* Initialize no first byte, no required byte. REQ_UNSET means "no char +matching encountered yet". It gets changed to REQ_NONE if we hit something that +matches a non-fixed char first char; reqbyte just remains unset if we never +find one. + +When we hit a repeat whose minimum is zero, we may have to adjust these values +to take the zero repeat into account. This is implemented by setting them to +zerofirstbyte and zeroreqbyte when such a repeat is encountered. The individual +item types that can be repeated set these backoff variables appropriately. */ + +firstbyte = reqbyte = zerofirstbyte = zeroreqbyte = REQ_UNSET; + +/* The variable req_caseopt contains either the REQ_CASELESS value or zero, +according to the current setting of the caseless flag. REQ_CASELESS is a bit +value > 255. It is added into the firstbyte or reqbyte variables to record the +case status of the value. This is used only for ASCII characters. */ + +req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS : 0; + +/* Switch on next character until the end of the branch */ + +for (;; ptr++) + { + BOOL negate_class; + BOOL possessive_quantifier; + BOOL is_quantifier; + int class_charcount; + int class_lastchar; + int newoptions; + int recno; + int skipbytes; + int subreqbyte; + int subfirstbyte; + int mclength; + uschar mcbuffer[8]; + + /* Next byte in the pattern */ + + c = *ptr; + + /* If in \Q...\E, check for the end; if not, we have a literal */ + + if (inescq && c != 0) + { + if (c == '\\' && ptr[1] == 'E') + { + inescq = FALSE; + ptr++; + continue; + } + else + { + if (previous_callout != NULL) + { + complete_callout(previous_callout, ptr, cd); + previous_callout = NULL; + } + if ((options & PCRE_AUTO_CALLOUT) != 0) + { + previous_callout = code; + code = auto_callout(code, ptr, cd); + } + goto NORMAL_CHAR; + } + } + + /* Fill in length of a previous callout, except when the next thing is + a quantifier. */ + + is_quantifier = c == '*' || c == '+' || c == '?' || + (c == '{' && is_counted_repeat(ptr+1)); + + if (!is_quantifier && previous_callout != NULL && + after_manual_callout-- <= 0) + { + complete_callout(previous_callout, ptr, cd); + previous_callout = NULL; + } + + /* In extended mode, skip white space and comments */ + + if ((options & PCRE_EXTENDED) != 0) + { + if ((cd->ctypes[c] & ctype_space) != 0) continue; + if (c == '#') + { + while (*(++ptr) != 0) if (IS_NEWLINE(ptr)) break; + if (*ptr != 0) + { + ptr += cd->nllen - 1; + continue; + } + /* Else fall through to handle end of string */ + c = 0; + } + } + + /* No auto callout for quantifiers. */ + + if ((options & PCRE_AUTO_CALLOUT) != 0 && !is_quantifier) + { + previous_callout = code; + code = auto_callout(code, ptr, cd); + } + + switch(c) + { + /* The branch terminates at end of string, |, or ). */ + + case 0: + case '|': + case ')': + *firstbyteptr = firstbyte; + *reqbyteptr = reqbyte; + *codeptr = code; + *ptrptr = ptr; + return TRUE; + + /* Handle single-character metacharacters. In multiline mode, ^ disables + the setting of any following char as a first character. */ + + case '^': + if ((options & PCRE_MULTILINE) != 0) + { + if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE; + } + previous = NULL; + *code++ = OP_CIRC; + break; + + case '$': + previous = NULL; + *code++ = OP_DOLL; + break; + + /* There can never be a first char if '.' is first, whatever happens about + repeats. The value of reqbyte doesn't change either. */ + + case '.': + if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE; + zerofirstbyte = firstbyte; + zeroreqbyte = reqbyte; + previous = code; + *code++ = OP_ANY; + break; + + /* Character classes. If the included characters are all < 256, we build a + 32-byte bitmap of the permitted characters, except in the special case + where there is only one such character. For negated classes, we build the + map as usual, then invert it at the end. However, we use a different opcode + so that data characters > 255 can be handled correctly. + + If the class contains characters outside the 0-255 range, a different + opcode is compiled. It may optionally have a bit map for characters < 256, + but those above are are explicitly listed afterwards. A flag byte tells + whether the bitmap is present, and whether this is a negated class or not. + */ + + case '[': + previous = code; + + /* PCRE supports POSIX class stuff inside a class. Perl gives an error if + they are encountered at the top level, so we'll do that too. */ + + if ((ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') && + check_posix_syntax(ptr, &tempptr, cd)) + { + *errorcodeptr = (ptr[1] == ':')? ERR13 : ERR31; + goto FAILED; + } + + /* If the first character is '^', set the negation flag and skip it. */ + + if ((c = *(++ptr)) == '^') + { + negate_class = TRUE; + c = *(++ptr); + } + else + { + negate_class = FALSE; + } + + /* Keep a count of chars with values < 256 so that we can optimize the case + of just a single character (as long as it's < 256). For higher valued UTF-8 + characters, we don't yet do any optimization. */ + + class_charcount = 0; + class_lastchar = -1; + +#ifdef SUPPORT_UTF8 + class_utf8 = FALSE; /* No chars >= 256 */ + class_utf8data = code + LINK_SIZE + 34; /* For UTF-8 items */ +#endif + + /* Initialize the 32-char bit map to all zeros. We have to build the + map in a temporary bit of store, in case the class contains only 1 + character (< 256), because in that case the compiled code doesn't use the + bit map. */ + + memset(classbits, 0, 32 * sizeof(uschar)); + + /* Process characters until ] is reached. By writing this as a "do" it + means that an initial ] is taken as a data character. The first pass + through the regex checked the overall syntax, so we don't need to be very + strict here. At the start of the loop, c contains the first byte of the + character. */ + + do + { +#ifdef SUPPORT_UTF8 + if (utf8 && c > 127) + { /* Braces are required because the */ + GETCHARLEN(c, ptr, ptr); /* macro generates multiple statements */ + } +#endif + + /* Inside \Q...\E everything is literal except \E */ + + if (inescq) + { + if (c == '\\' && ptr[1] == 'E') + { + inescq = FALSE; + ptr++; + continue; + } + else goto LONE_SINGLE_CHARACTER; + } + + /* Handle POSIX class names. Perl allows a negation extension of the + form [:^name:]. A square bracket that doesn't match the syntax is + treated as a literal. We also recognize the POSIX constructions + [.ch.] and [=ch=] ("collating elements") and fault them, as Perl + 5.6 and 5.8 do. */ + + if (c == '[' && + (ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') && + check_posix_syntax(ptr, &tempptr, cd)) + { + BOOL local_negate = FALSE; + int posix_class, taboffset, tabopt; + register const uschar *cbits = cd->cbits; + uschar pbits[32]; + + if (ptr[1] != ':') + { + *errorcodeptr = ERR31; + goto FAILED; + } + + ptr += 2; + if (*ptr == '^') + { + local_negate = TRUE; + ptr++; + } + + posix_class = check_posix_name(ptr, tempptr - ptr); + if (posix_class < 0) + { + *errorcodeptr = ERR30; + goto FAILED; + } + + /* If matching is caseless, upper and lower are converted to + alpha. This relies on the fact that the class table starts with + alpha, lower, upper as the first 3 entries. */ + + if ((options & PCRE_CASELESS) != 0 && posix_class <= 2) + posix_class = 0; + + /* We build the bit map for the POSIX class in a chunk of local store + because we may be adding and subtracting from it, and we don't want to + subtract bits that may be in the main map already. At the end we or the + result into the bit map that is being built. */ + + posix_class *= 3; + + /* Copy in the first table (always present) */ + + memcpy(pbits, cbits + posix_class_maps[posix_class], + 32 * sizeof(uschar)); + + /* If there is a second table, add or remove it as required. */ + + taboffset = posix_class_maps[posix_class + 1]; + tabopt = posix_class_maps[posix_class + 2]; + + if (taboffset >= 0) + { + if (tabopt >= 0) + for (c = 0; c < 32; c++) pbits[c] |= cbits[c + taboffset]; + else + for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset]; + } + + /* Not see if we need to remove any special characters. An option + value of 1 removes vertical space and 2 removes underscore. */ + + if (tabopt < 0) tabopt = -tabopt; + if (tabopt == 1) pbits[1] &= ~0x3c; + else if (tabopt == 2) pbits[11] &= 0x7f; + + /* Add the POSIX table or its complement into the main table that is + being built and we are done. */ + + if (local_negate) + for (c = 0; c < 32; c++) classbits[c] |= ~pbits[c]; + else + for (c = 0; c < 32; c++) classbits[c] |= pbits[c]; + + ptr = tempptr + 1; + class_charcount = 10; /* Set > 1; assumes more than 1 per class */ + continue; /* End of POSIX syntax handling */ + } + + /* Backslash may introduce a single character, or it may introduce one + of the specials, which just set a flag. Escaped items are checked for + validity in the pre-compiling pass. The sequence \b is a special case. + Inside a class (and only there) it is treated as backspace. Elsewhere + it marks a word boundary. Other escapes have preset maps ready to + or into the one we are building. We assume they have more than one + character in them, so set class_charcount bigger than one. */ + + if (c == '\\') + { + c = check_escape(&ptr, errorcodeptr, *brackets, options, TRUE); + + if (-c == ESC_b) c = '\b'; /* \b is backslash in a class */ + else if (-c == ESC_X) c = 'X'; /* \X is literal X in a class */ + else if (-c == ESC_Q) /* Handle start of quoted string */ + { + if (ptr[1] == '\\' && ptr[2] == 'E') + { + ptr += 2; /* avoid empty string */ + } + else inescq = TRUE; + continue; + } + + if (c < 0) + { + register const uschar *cbits = cd->cbits; + class_charcount += 2; /* Greater than 1 is what matters */ + switch (-c) + { + case ESC_d: + for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_digit]; + continue; + + case ESC_D: + for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit]; + continue; + + case ESC_w: + for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_word]; + continue; + + case ESC_W: + for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word]; + continue; + + case ESC_s: + for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_space]; + classbits[1] &= ~0x08; /* Perl 5.004 onwards omits VT from \s */ + continue; + + case ESC_S: + for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space]; + classbits[1] |= 0x08; /* Perl 5.004 onwards omits VT from \s */ + continue; + +#ifdef SUPPORT_UCP + case ESC_p: + case ESC_P: + { + BOOL negated; + int pdata; + int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr); + if (ptype < 0) goto FAILED; + class_utf8 = TRUE; + *class_utf8data++ = ((-c == ESC_p) != negated)? + XCL_PROP : XCL_NOTPROP; + *class_utf8data++ = ptype; + *class_utf8data++ = pdata; + class_charcount -= 2; /* Not a < 256 character */ + } + continue; +#endif + + /* Unrecognized escapes are faulted if PCRE is running in its + strict mode. By default, for compatibility with Perl, they are + treated as literals. */ + + default: + if ((options & PCRE_EXTRA) != 0) + { + *errorcodeptr = ERR7; + goto FAILED; + } + c = *ptr; /* The final character */ + class_charcount -= 2; /* Undo the default count from above */ + } + } + + /* Fall through if we have a single character (c >= 0). This may be + > 256 in UTF-8 mode. */ + + } /* End of backslash handling */ + + /* A single character may be followed by '-' to form a range. However, + Perl does not permit ']' to be the end of the range. A '-' character + here is treated as a literal. */ + + if (ptr[1] == '-' && ptr[2] != ']') + { + int d; + ptr += 2; + +#ifdef SUPPORT_UTF8 + if (utf8) + { /* Braces are required because the */ + GETCHARLEN(d, ptr, ptr); /* macro generates multiple statements */ + } + else +#endif + d = *ptr; /* Not UTF-8 mode */ + + /* The second part of a range can be a single-character escape, but + not any of the other escapes. Perl 5.6 treats a hyphen as a literal + in such circumstances. */ + + if (d == '\\') + { + const uschar *oldptr = ptr; + d = check_escape(&ptr, errorcodeptr, *brackets, options, TRUE); + + /* \b is backslash; \X is literal X; any other special means the '-' + was literal */ + + if (d < 0) + { + if (d == -ESC_b) d = '\b'; + else if (d == -ESC_X) d = 'X'; else + { + ptr = oldptr - 2; + goto LONE_SINGLE_CHARACTER; /* A few lines below */ + } + } + } + + /* The check that the two values are in the correct order happens in + the pre-pass. Optimize one-character ranges */ + + if (d == c) goto LONE_SINGLE_CHARACTER; /* A few lines below */ + + /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless + matching, we have to use an XCLASS with extra data items. Caseless + matching for characters > 127 is available only if UCP support is + available. */ + +#ifdef SUPPORT_UTF8 + if (utf8 && (d > 255 || ((options & PCRE_CASELESS) != 0 && d > 127))) + { + class_utf8 = TRUE; + + /* With UCP support, we can find the other case equivalents of + the relevant characters. There may be several ranges. Optimize how + they fit with the basic range. */ + +#ifdef SUPPORT_UCP + if ((options & PCRE_CASELESS) != 0) + { + int occ, ocd; + int cc = c; + int origd = d; + while (get_othercase_range(&cc, origd, &occ, &ocd)) + { + if (occ >= c && ocd <= d) continue; /* Skip embedded ranges */ + + if (occ < c && ocd >= c - 1) /* Extend the basic range */ + { /* if there is overlap, */ + c = occ; /* noting that if occ < c */ + continue; /* we can't have ocd > d */ + } /* because a subrange is */ + if (ocd > d && occ <= d + 1) /* always shorter than */ + { /* the basic range. */ + d = ocd; + continue; + } + + if (occ == ocd) + { + *class_utf8data++ = XCL_SINGLE; + } + else + { + *class_utf8data++ = XCL_RANGE; + class_utf8data += _pcre_ord2utf8(occ, class_utf8data); + } + class_utf8data += _pcre_ord2utf8(ocd, class_utf8data); + } + } +#endif /* SUPPORT_UCP */ + + /* Now record the original range, possibly modified for UCP caseless + overlapping ranges. */ + + *class_utf8data++ = XCL_RANGE; + class_utf8data += _pcre_ord2utf8(c, class_utf8data); + class_utf8data += _pcre_ord2utf8(d, class_utf8data); + + /* With UCP support, we are done. Without UCP support, there is no + caseless matching for UTF-8 characters > 127; we can use the bit map + for the smaller ones. */ + +#ifdef SUPPORT_UCP + continue; /* With next character in the class */ +#else + if ((options & PCRE_CASELESS) == 0 || c > 127) continue; + + /* Adjust upper limit and fall through to set up the map */ + + d = 127; + +#endif /* SUPPORT_UCP */ + } +#endif /* SUPPORT_UTF8 */ + + /* We use the bit map for all cases when not in UTF-8 mode; else + ranges that lie entirely within 0-127 when there is UCP support; else + for partial ranges without UCP support. */ + + for (; c <= d; c++) + { + classbits[c/8] |= (1 << (c&7)); + if ((options & PCRE_CASELESS) != 0) + { + int uc = cd->fcc[c]; /* flip case */ + classbits[uc/8] |= (1 << (uc&7)); + } + class_charcount++; /* in case a one-char range */ + class_lastchar = c; + } + + continue; /* Go get the next char in the class */ + } + + /* Handle a lone single character - we can get here for a normal + non-escape char, or after \ that introduces a single character or for an + apparent range that isn't. */ + + LONE_SINGLE_CHARACTER: + + /* Handle a character that cannot go in the bit map */ + +#ifdef SUPPORT_UTF8 + if (utf8 && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127))) + { + class_utf8 = TRUE; + *class_utf8data++ = XCL_SINGLE; + class_utf8data += _pcre_ord2utf8(c, class_utf8data); + +#ifdef SUPPORT_UCP + if ((options & PCRE_CASELESS) != 0) + { + int othercase; + if ((othercase = _pcre_ucp_othercase(c)) >= 0) + { + *class_utf8data++ = XCL_SINGLE; + class_utf8data += _pcre_ord2utf8(othercase, class_utf8data); + } + } +#endif /* SUPPORT_UCP */ + + } + else +#endif /* SUPPORT_UTF8 */ + + /* Handle a single-byte character */ + { + classbits[c/8] |= (1 << (c&7)); + if ((options & PCRE_CASELESS) != 0) + { + c = cd->fcc[c]; /* flip case */ + classbits[c/8] |= (1 << (c&7)); + } + class_charcount++; + class_lastchar = c; + } + } + + /* Loop until ']' reached; the check for end of string happens inside the + loop. This "while" is the end of the "do" above. */ + + while ((c = *(++ptr)) != ']' || inescq); + + /* If class_charcount is 1, we saw precisely one character whose value is + less than 256. In non-UTF-8 mode we can always optimize. In UTF-8 mode, we + can optimize the negative case only if there were no characters >= 128 + because OP_NOT and the related opcodes like OP_NOTSTAR operate on + single-bytes only. This is an historical hangover. Maybe one day we can + tidy these opcodes to handle multi-byte characters. + + The optimization throws away the bit map. We turn the item into a + 1-character OP_CHAR[NC] if it's positive, or OP_NOT if it's negative. Note + that OP_NOT does not support multibyte characters. In the positive case, it + can cause firstbyte to be set. Otherwise, there can be no first char if + this item is first, whatever repeat count may follow. In the case of + reqbyte, save the previous value for reinstating. */ + +#ifdef SUPPORT_UTF8 + if (class_charcount == 1 && + (!utf8 || + (!class_utf8 && (!negate_class || class_lastchar < 128)))) + +#else + if (class_charcount == 1) +#endif + { + zeroreqbyte = reqbyte; + + /* The OP_NOT opcode works on one-byte characters only. */ + + if (negate_class) + { + if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE; + zerofirstbyte = firstbyte; + *code++ = OP_NOT; + *code++ = class_lastchar; + break; + } + + /* For a single, positive character, get the value into mcbuffer, and + then we can handle this with the normal one-character code. */ + +#ifdef SUPPORT_UTF8 + if (utf8 && class_lastchar > 127) + mclength = _pcre_ord2utf8(class_lastchar, mcbuffer); + else +#endif + { + mcbuffer[0] = class_lastchar; + mclength = 1; + } + goto ONE_CHAR; + } /* End of 1-char optimization */ + + /* The general case - not the one-char optimization. If this is the first + thing in the branch, there can be no first char setting, whatever the + repeat count. Any reqbyte setting must remain unchanged after any kind of + repeat. */ + + if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE; + zerofirstbyte = firstbyte; + zeroreqbyte = reqbyte; + + /* If there are characters with values > 255, we have to compile an + extended class, with its own opcode. If there are no characters < 256, + we can omit the bitmap. */ + +#ifdef SUPPORT_UTF8 + if (class_utf8) + { + *class_utf8data++ = XCL_END; /* Marks the end of extra data */ + *code++ = OP_XCLASS; + code += LINK_SIZE; + *code = negate_class? XCL_NOT : 0; + + /* If the map is required, install it, and move on to the end of + the extra data */ + + if (class_charcount > 0) + { + *code++ |= XCL_MAP; + memcpy(code, classbits, 32); + code = class_utf8data; + } + + /* If the map is not required, slide down the extra data. */ + + else + { + int len = class_utf8data - (code + 33); + memmove(code + 1, code + 33, len); + code += len + 1; + } + + /* Now fill in the complete length of the item */ + + PUT(previous, 1, code - previous); + break; /* End of class handling */ + } +#endif + + /* If there are no characters > 255, negate the 32-byte map if necessary, + and copy it into the code vector. If this is the first thing in the branch, + there can be no first char setting, whatever the repeat count. Any reqbyte + setting must remain unchanged after any kind of repeat. */ + + if (negate_class) + { + *code++ = OP_NCLASS; + for (c = 0; c < 32; c++) code[c] = ~classbits[c]; + } + else + { + *code++ = OP_CLASS; + memcpy(code, classbits, 32); + } + code += 32; + break; + + /* Various kinds of repeat; '{' is not necessarily a quantifier, but this + has been tested above. */ + + case '{': + if (!is_quantifier) goto NORMAL_CHAR; + ptr = read_repeat_counts(ptr+1, &repeat_min, &repeat_max, errorcodeptr); + if (*errorcodeptr != 0) goto FAILED; + goto REPEAT; + + case '*': + repeat_min = 0; + repeat_max = -1; + goto REPEAT; + + case '+': + repeat_min = 1; + repeat_max = -1; + goto REPEAT; + + case '?': + repeat_min = 0; + repeat_max = 1; + + REPEAT: + if (previous == NULL) + { + *errorcodeptr = ERR9; + goto FAILED; + } + + if (repeat_min == 0) + { + firstbyte = zerofirstbyte; /* Adjust for zero repeat */ + reqbyte = zeroreqbyte; /* Ditto */ + } + + /* Remember whether this is a variable length repeat */ + + reqvary = (repeat_min == repeat_max)? 0 : REQ_VARY; + + op_type = 0; /* Default single-char op codes */ + possessive_quantifier = FALSE; /* Default not possessive quantifier */ + + /* Save start of previous item, in case we have to move it up to make space + for an inserted OP_ONCE for the additional '+' extension. */ + + tempcode = previous; + + /* If the next character is '+', we have a possessive quantifier. This + implies greediness, whatever the setting of the PCRE_UNGREEDY option. + If the next character is '?' this is a minimizing repeat, by default, + but if PCRE_UNGREEDY is set, it works the other way round. We change the + repeat type to the non-default. */ + + if (ptr[1] == '+') + { + repeat_type = 0; /* Force greedy */ + possessive_quantifier = TRUE; + ptr++; + } + else if (ptr[1] == '?') + { + repeat_type = greedy_non_default; + ptr++; + } + else repeat_type = greedy_default; + + /* If previous was a recursion, we need to wrap it inside brackets so that + it can be replicated if necessary. */ + + if (*previous == OP_RECURSE) + { + memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE); + code += 1 + LINK_SIZE; + *previous = OP_BRA; + PUT(previous, 1, code - previous); + *code = OP_KET; + PUT(code, 1, code - previous); + code += 1 + LINK_SIZE; + } + + /* If previous was a character match, abolish the item and generate a + repeat item instead. If a char item has a minumum of more than one, ensure + that it is set in reqbyte - it might not be if a sequence such as x{3} is + the first thing in a branch because the x will have gone into firstbyte + instead. */ + + if (*previous == OP_CHAR || *previous == OP_CHARNC) + { + /* Deal with UTF-8 characters that take up more than one byte. It's + easier to write this out separately than try to macrify it. Use c to + hold the length of the character in bytes, plus 0x80 to flag that it's a + length rather than a small character. */ + +#ifdef SUPPORT_UTF8 + if (utf8 && (code[-1] & 0x80) != 0) + { + uschar *lastchar = code - 1; + while((*lastchar & 0xc0) == 0x80) lastchar--; + c = code - lastchar; /* Length of UTF-8 character */ + memcpy(utf8_char, lastchar, c); /* Save the char */ + c |= 0x80; /* Flag c as a length */ + } + else +#endif + + /* Handle the case of a single byte - either with no UTF8 support, or + with UTF-8 disabled, or for a UTF-8 character < 128. */ + + { + c = code[-1]; + if (repeat_min > 1) reqbyte = c | req_caseopt | cd->req_varyopt; + } + + goto OUTPUT_SINGLE_REPEAT; /* Code shared with single character types */ + } + + /* If previous was a single negated character ([^a] or similar), we use + one of the special opcodes, replacing it. The code is shared with single- + character repeats by setting opt_type to add a suitable offset into + repeat_type. OP_NOT is currently used only for single-byte chars. */ + + else if (*previous == OP_NOT) + { + op_type = OP_NOTSTAR - OP_STAR; /* Use "not" opcodes */ + c = previous[1]; + goto OUTPUT_SINGLE_REPEAT; + } + + /* If previous was a character type match (\d or similar), abolish it and + create a suitable repeat item. The code is shared with single-character + repeats by setting op_type to add a suitable offset into repeat_type. Note + the the Unicode property types will be present only when SUPPORT_UCP is + defined, but we don't wrap the little bits of code here because it just + makes it horribly messy. */ + + else if (*previous < OP_EODN) + { + uschar *oldcode; + int prop_type, prop_value; + op_type = OP_TYPESTAR - OP_STAR; /* Use type opcodes */ + c = *previous; + + OUTPUT_SINGLE_REPEAT: + if (*previous == OP_PROP || *previous == OP_NOTPROP) + { + prop_type = previous[1]; + prop_value = previous[2]; + } + else prop_type = prop_value = -1; + + oldcode = code; + code = previous; /* Usually overwrite previous item */ + + /* If the maximum is zero then the minimum must also be zero; Perl allows + this case, so we do too - by simply omitting the item altogether. */ + + if (repeat_max == 0) goto END_REPEAT; + + /* All real repeats make it impossible to handle partial matching (maybe + one day we will be able to remove this restriction). */ + + if (repeat_max != 1) cd->nopartial = TRUE; + + /* Combine the op_type with the repeat_type */ + + repeat_type += op_type; + + /* A minimum of zero is handled either as the special case * or ?, or as + an UPTO, with the maximum given. */ + + if (repeat_min == 0) + { + if (repeat_max == -1) *code++ = OP_STAR + repeat_type; + else if (repeat_max == 1) *code++ = OP_QUERY + repeat_type; + else + { + *code++ = OP_UPTO + repeat_type; + PUT2INC(code, 0, repeat_max); + } + } + + /* A repeat minimum of 1 is optimized into some special cases. If the + maximum is unlimited, we use OP_PLUS. Otherwise, the original item it + left in place and, if the maximum is greater than 1, we use OP_UPTO with + one less than the maximum. */ + + else if (repeat_min == 1) + { + if (repeat_max == -1) + *code++ = OP_PLUS + repeat_type; + else + { + code = oldcode; /* leave previous item in place */ + if (repeat_max == 1) goto END_REPEAT; + *code++ = OP_UPTO + repeat_type; + PUT2INC(code, 0, repeat_max - 1); + } + } + + /* The case {n,n} is just an EXACT, while the general case {n,m} is + handled as an EXACT followed by an UPTO. */ + + else + { + *code++ = OP_EXACT + op_type; /* NB EXACT doesn't have repeat_type */ + PUT2INC(code, 0, repeat_min); + + /* If the maximum is unlimited, insert an OP_STAR. Before doing so, + we have to insert the character for the previous code. For a repeated + Unicode property match, there are two extra bytes that define the + required property. In UTF-8 mode, long characters have their length in + c, with the 0x80 bit as a flag. */ + + if (repeat_max < 0) + { +#ifdef SUPPORT_UTF8 + if (utf8 && c >= 128) + { + memcpy(code, utf8_char, c & 7); + code += c & 7; + } + else +#endif + { + *code++ = c; + if (prop_type >= 0) + { + *code++ = prop_type; + *code++ = prop_value; + } + } + *code++ = OP_STAR + repeat_type; + } + + /* Else insert an UPTO if the max is greater than the min, again + preceded by the character, for the previously inserted code. */ + + else if (repeat_max != repeat_min) + { +#ifdef SUPPORT_UTF8 + if (utf8 && c >= 128) + { + memcpy(code, utf8_char, c & 7); + code += c & 7; + } + else +#endif + *code++ = c; + if (prop_type >= 0) + { + *code++ = prop_type; + *code++ = prop_value; + } + repeat_max -= repeat_min; + *code++ = OP_UPTO + repeat_type; + PUT2INC(code, 0, repeat_max); + } + } + + /* The character or character type itself comes last in all cases. */ + +#ifdef SUPPORT_UTF8 + if (utf8 && c >= 128) + { + memcpy(code, utf8_char, c & 7); + code += c & 7; + } + else +#endif + *code++ = c; + + /* For a repeated Unicode property match, there are two extra bytes that + define the required property. */ + +#ifdef SUPPORT_UCP + if (prop_type >= 0) + { + *code++ = prop_type; + *code++ = prop_value; + } +#endif + } + + /* If previous was a character class or a back reference, we put the repeat + stuff after it, but just skip the item if the repeat was {0,0}. */ + + else if (*previous == OP_CLASS || + *previous == OP_NCLASS || +#ifdef SUPPORT_UTF8 + *previous == OP_XCLASS || +#endif + *previous == OP_REF) + { + if (repeat_max == 0) + { + code = previous; + goto END_REPEAT; + } + + /* All real repeats make it impossible to handle partial matching (maybe + one day we will be able to remove this restriction). */ + + if (repeat_max != 1) cd->nopartial = TRUE; + + if (repeat_min == 0 && repeat_max == -1) + *code++ = OP_CRSTAR + repeat_type; + else if (repeat_min == 1 && repeat_max == -1) + *code++ = OP_CRPLUS + repeat_type; + else if (repeat_min == 0 && repeat_max == 1) + *code++ = OP_CRQUERY + repeat_type; + else + { + *code++ = OP_CRRANGE + repeat_type; + PUT2INC(code, 0, repeat_min); + if (repeat_max == -1) repeat_max = 0; /* 2-byte encoding for max */ + PUT2INC(code, 0, repeat_max); + } + } + + /* If previous was a bracket group, we may have to replicate it in certain + cases. */ + + else if (*previous >= OP_BRA || *previous == OP_ONCE || + *previous == OP_COND) + { + register int i; + int ketoffset = 0; + int len = code - previous; + uschar *bralink = NULL; + + /* If the maximum repeat count is unlimited, find the end of the bracket + by scanning through from the start, and compute the offset back to it + from the current code pointer. There may be an OP_OPT setting following + the final KET, so we can't find the end just by going back from the code + pointer. */ + + if (repeat_max == -1) + { + register uschar *ket = previous; + do ket += GET(ket, 1); while (*ket != OP_KET); + ketoffset = code - ket; + } + + /* The case of a zero minimum is special because of the need to stick + OP_BRAZERO in front of it, and because the group appears once in the + data, whereas in other cases it appears the minimum number of times. For + this reason, it is simplest to treat this case separately, as otherwise + the code gets far too messy. There are several special subcases when the + minimum is zero. */ + + if (repeat_min == 0) + { + /* If the maximum is also zero, we just omit the group from the output + altogether. */ + + if (repeat_max == 0) + { + code = previous; + goto END_REPEAT; + } + + /* If the maximum is 1 or unlimited, we just have to stick in the + BRAZERO and do no more at this point. However, we do need to adjust + any OP_RECURSE calls inside the group that refer to the group itself or + any internal group, because the offset is from the start of the whole + regex. Temporarily terminate the pattern while doing this. */ + + if (repeat_max <= 1) + { + *code = OP_END; + adjust_recurse(previous, 1, utf8, cd); + memmove(previous+1, previous, len); + code++; + *previous++ = OP_BRAZERO + repeat_type; + } + + /* If the maximum is greater than 1 and limited, we have to replicate + in a nested fashion, sticking OP_BRAZERO before each set of brackets. + The first one has to be handled carefully because it's the original + copy, which has to be moved up. The remainder can be handled by code + that is common with the non-zero minimum case below. We have to + adjust the value or repeat_max, since one less copy is required. Once + again, we may have to adjust any OP_RECURSE calls inside the group. */ + + else + { + int offset; + *code = OP_END; + adjust_recurse(previous, 2 + LINK_SIZE, utf8, cd); + memmove(previous + 2 + LINK_SIZE, previous, len); + code += 2 + LINK_SIZE; + *previous++ = OP_BRAZERO + repeat_type; + *previous++ = OP_BRA; + + /* We chain together the bracket offset fields that have to be + filled in later when the ends of the brackets are reached. */ + + offset = (bralink == NULL)? 0 : previous - bralink; + bralink = previous; + PUTINC(previous, 0, offset); + } + + repeat_max--; + } + + /* If the minimum is greater than zero, replicate the group as many + times as necessary, and adjust the maximum to the number of subsequent + copies that we need. If we set a first char from the group, and didn't + set a required char, copy the latter from the former. */ + + else + { + if (repeat_min > 1) + { + if (groupsetfirstbyte && reqbyte < 0) reqbyte = firstbyte; + for (i = 1; i < repeat_min; i++) + { + memcpy(code, previous, len); + code += len; + } + } + if (repeat_max > 0) repeat_max -= repeat_min; + } + + /* This code is common to both the zero and non-zero minimum cases. If + the maximum is limited, it replicates the group in a nested fashion, + remembering the bracket starts on a stack. In the case of a zero minimum, + the first one was set up above. In all cases the repeat_max now specifies + the number of additional copies needed. */ + + if (repeat_max >= 0) + { + for (i = repeat_max - 1; i >= 0; i--) + { + *code++ = OP_BRAZERO + repeat_type; + + /* All but the final copy start a new nesting, maintaining the + chain of brackets outstanding. */ + + if (i != 0) + { + int offset; + *code++ = OP_BRA; + offset = (bralink == NULL)? 0 : code - bralink; + bralink = code; + PUTINC(code, 0, offset); + } + + memcpy(code, previous, len); + code += len; + } + + /* Now chain through the pending brackets, and fill in their length + fields (which are holding the chain links pro tem). */ + + while (bralink != NULL) + { + int oldlinkoffset; + int offset = code - bralink + 1; + uschar *bra = code - offset; + oldlinkoffset = GET(bra, 1); + bralink = (oldlinkoffset == 0)? NULL : bralink - oldlinkoffset; + *code++ = OP_KET; + PUTINC(code, 0, offset); + PUT(bra, 1, offset); + } + } + + /* If the maximum is unlimited, set a repeater in the final copy. We + can't just offset backwards from the current code point, because we + don't know if there's been an options resetting after the ket. The + correct offset was computed above. */ + + else code[-ketoffset] = OP_KETRMAX + repeat_type; + } + + /* Else there's some kind of shambles */ + + else + { + *errorcodeptr = ERR11; + goto FAILED; + } + + /* If the character following a repeat is '+', we wrap the entire repeated + item inside OP_ONCE brackets. This is just syntactic sugar, taken from + Sun's Java package. The repeated item starts at tempcode, not at previous, + which might be the first part of a string whose (former) last char we + repeated. However, we don't support '+' after a greediness '?'. */ + + if (possessive_quantifier) + { + int len = code - tempcode; + memmove(tempcode + 1+LINK_SIZE, tempcode, len); + code += 1 + LINK_SIZE; + len += 1 + LINK_SIZE; + tempcode[0] = OP_ONCE; + *code++ = OP_KET; + PUTINC(code, 0, len); + PUT(tempcode, 1, len); + } + + /* In all case we no longer have a previous item. We also set the + "follows varying string" flag for subsequently encountered reqbytes if + it isn't already set and we have just passed a varying length item. */ + + END_REPEAT: + previous = NULL; + cd->req_varyopt |= reqvary; + break; + + + /* Start of nested bracket sub-expression, or comment or lookahead or + lookbehind or option setting or condition. First deal with special things + that can come after a bracket; all are introduced by ?, and the appearance + of any of them means that this is not a referencing group. They were + checked for validity in the first pass over the string, so we don't have to + check for syntax errors here. */ + + case '(': + newoptions = options; + skipbytes = 0; + + if (*(++ptr) == '?') + { + int set, unset; + int *optset; + + switch (*(++ptr)) + { + case '#': /* Comment; skip to ket */ + ptr++; + while (*ptr != ')') ptr++; + continue; + + case ':': /* Non-extracting bracket */ + bravalue = OP_BRA; + ptr++; + break; + + case '(': + bravalue = OP_COND; /* Conditional group */ + + /* A condition can be a number, referring to a numbered group, a name, + referring to a named group, 'R', referring to recursion, or an + assertion. There are two unfortunate ambiguities, caused by history. + (a) 'R' can be the recursive thing or the name 'R', and (b) a number + could be a name that consists of digits. In both cases, we look for a + name first; if not found, we try the other cases. If the first + character after (?( is a word character, we know the rest up to ) will + also be word characters because the syntax was checked in the first + pass. */ + + if ((cd->ctypes[ptr[1]] & ctype_word) != 0) + { + int i, namelen; + int condref = 0; + const uschar *name; + uschar *slot = cd->name_table; + + /* This is needed for all successful cases. */ + + skipbytes = 3; + + /* Read the name, but also get it as a number if it's all digits */ + + name = ++ptr; + while (*ptr != ')') + { + if (condref >= 0) + condref = ((digitab[*ptr] & ctype_digit) != 0)? + condref * 10 + *ptr - '0' : -1; + ptr++; + } + namelen = ptr - name; + ptr++; + + for (i = 0; i < cd->names_found; i++) + { + if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break; + slot += cd->name_entry_size; + } + + /* Found a previous named subpattern */ + + if (i < cd->names_found) + { + condref = GET2(slot, 0); + code[1+LINK_SIZE] = OP_CREF; + PUT2(code, 2+LINK_SIZE, condref); + } + + /* Search the pattern for a forward reference */ + + else if ((i = find_named_parens(ptr, *brackets, name, namelen)) > 0) + { + code[1+LINK_SIZE] = OP_CREF; + PUT2(code, 2+LINK_SIZE, i); + } + + /* Check for 'R' for recursion */ + + else if (namelen == 1 && *name == 'R') + { + code[1+LINK_SIZE] = OP_CREF; + PUT2(code, 2+LINK_SIZE, CREF_RECURSE); + } + + /* Check for a subpattern number */ + + else if (condref > 0) + { + code[1+LINK_SIZE] = OP_CREF; + PUT2(code, 2+LINK_SIZE, condref); + } + + /* Either an unidentified subpattern, or a reference to (?(0) */ + + else + { + *errorcodeptr = (condref == 0)? ERR35: ERR15; + goto FAILED; + } + } + + /* For conditions that are assertions, we just fall through, having + set bravalue above. */ + + break; + + case '=': /* Positive lookahead */ + bravalue = OP_ASSERT; + ptr++; + break; + + case '!': /* Negative lookahead */ + bravalue = OP_ASSERT_NOT; + ptr++; + break; + + case '<': /* Lookbehinds */ + switch (*(++ptr)) + { + case '=': /* Positive lookbehind */ + bravalue = OP_ASSERTBACK; + ptr++; + break; + + case '!': /* Negative lookbehind */ + bravalue = OP_ASSERTBACK_NOT; + ptr++; + break; + } + break; + + case '>': /* One-time brackets */ + bravalue = OP_ONCE; + ptr++; + break; + + case 'C': /* Callout - may be followed by digits; */ + previous_callout = code; /* Save for later completion */ + after_manual_callout = 1; /* Skip one item before completing */ + *code++ = OP_CALLOUT; /* Already checked that the terminating */ + { /* closing parenthesis is present. */ + int n = 0; + while ((digitab[*(++ptr)] & ctype_digit) != 0) + n = n * 10 + *ptr - '0'; + if (n > 255) + { + *errorcodeptr = ERR38; + goto FAILED; + } + *code++ = n; + PUT(code, 0, ptr - cd->start_pattern + 1); /* Pattern offset */ + PUT(code, LINK_SIZE, 0); /* Default length */ + code += 2 * LINK_SIZE; + } + previous = NULL; + continue; + + case 'P': /* Named subpattern handling */ + if (*(++ptr) == '<') /* Definition */ + { + int i, namelen; + uschar *slot = cd->name_table; + const uschar *name; /* Don't amalgamate; some compilers */ + name = ++ptr; /* grumble at autoincrement in declaration */ + + while (*ptr++ != '>'); + namelen = ptr - name - 1; + + for (i = 0; i < cd->names_found; i++) + { + int crc = memcmp(name, slot+2, namelen); + if (crc == 0) + { + if (slot[2+namelen] == 0) + { + if ((options & PCRE_DUPNAMES) == 0) + { + *errorcodeptr = ERR43; + goto FAILED; + } + } + else crc = -1; /* Current name is substring */ + } + if (crc < 0) + { + memmove(slot + cd->name_entry_size, slot, + (cd->names_found - i) * cd->name_entry_size); + break; + } + slot += cd->name_entry_size; + } + + PUT2(slot, 0, *brackets + 1); + memcpy(slot + 2, name, namelen); + slot[2+namelen] = 0; + cd->names_found++; + goto NUMBERED_GROUP; + } + + if (*ptr == '=' || *ptr == '>') /* Reference or recursion */ + { + int i, namelen; + int type = *ptr++; + const uschar *name = ptr; + uschar *slot = cd->name_table; + + while (*ptr != ')') ptr++; + namelen = ptr - name; + + for (i = 0; i < cd->names_found; i++) + { + if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break; + slot += cd->name_entry_size; + } + + if (i < cd->names_found) /* Back reference */ + { + recno = GET2(slot, 0); + } + else if ((recno = /* Forward back reference */ + find_named_parens(ptr, *brackets, name, namelen)) <= 0) + { + *errorcodeptr = ERR15; + goto FAILED; + } + + if (type == '>') goto HANDLE_RECURSION; /* A few lines below */ + + /* Back reference */ + + previous = code; + *code++ = OP_REF; + PUT2INC(code, 0, recno); + cd->backref_map |= (recno < 32)? (1 << recno) : 1; + if (recno > cd->top_backref) cd->top_backref = recno; + continue; + } + + /* Should never happen */ + break; + + case 'R': /* Pattern recursion */ + ptr++; /* Same as (?0) */ + /* Fall through */ + + /* Recursion or "subroutine" call */ + + case '0': case '1': case '2': case '3': case '4': + case '5': case '6': case '7': case '8': case '9': + { + const uschar *called; + recno = 0; + while((digitab[*ptr] & ctype_digit) != 0) + recno = recno * 10 + *ptr++ - '0'; + + /* Come here from code above that handles a named recursion */ + + HANDLE_RECURSION: + + previous = code; + + /* Find the bracket that is being referenced. Temporarily end the + regex in case it doesn't exist. */ + + *code = OP_END; + called = (recno == 0)? cd->start_code : + find_bracket(cd->start_code, utf8, recno); + if (called == NULL) + { + *errorcodeptr = ERR15; + goto FAILED; + } + + /* If the subpattern is still open, this is a recursive call. We + check to see if this is a left recursion that could loop for ever, + and diagnose that case. */ + + if (GET(called, 1) == 0 && could_be_empty(called, code, bcptr, utf8)) + { + *errorcodeptr = ERR40; + goto FAILED; + } + + /* Insert the recursion/subroutine item, automatically wrapped inside + "once" brackets. */ + + *code = OP_ONCE; + PUT(code, 1, 2 + 2*LINK_SIZE); + code += 1 + LINK_SIZE; + + *code = OP_RECURSE; + PUT(code, 1, called - cd->start_code); + code += 1 + LINK_SIZE; + + *code = OP_KET; + PUT(code, 1, 2 + 2*LINK_SIZE); + code += 1 + LINK_SIZE; + } + continue; + + /* Character after (? not specially recognized */ + + default: /* Option setting */ + set = unset = 0; + optset = &set; + + while (*ptr != ')' && *ptr != ':') + { + switch (*ptr++) + { + case '-': optset = &unset; break; + + case 'i': *optset |= PCRE_CASELESS; break; + case 'J': *optset |= PCRE_DUPNAMES; break; + case 'm': *optset |= PCRE_MULTILINE; break; + case 's': *optset |= PCRE_DOTALL; break; + case 'x': *optset |= PCRE_EXTENDED; break; + case 'U': *optset |= PCRE_UNGREEDY; break; + case 'X': *optset |= PCRE_EXTRA; break; + } + } + + /* Set up the changed option bits, but don't change anything yet. */ + + newoptions = (options | set) & (~unset); + + /* If the options ended with ')' this is not the start of a nested + group with option changes, so the options change at this level. Compile + code to change the ims options if this setting actually changes any of + them. We also pass the new setting back so that it can be put at the + start of any following branches, and when this group ends (if we are in + a group), a resetting item can be compiled. + + Note that if this item is right at the start of the pattern, the + options will have been abstracted and made global, so there will be no + change to compile. */ + + if (*ptr == ')') + { + if ((options & PCRE_IMS) != (newoptions & PCRE_IMS)) + { + *code++ = OP_OPT; + *code++ = newoptions & PCRE_IMS; + } + + /* Change options at this level, and pass them back for use + in subsequent branches. Reset the greedy defaults and the case + value for firstbyte and reqbyte. */ + + *optionsptr = options = newoptions; + greedy_default = ((newoptions & PCRE_UNGREEDY) != 0); + greedy_non_default = greedy_default ^ 1; + req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS : 0; + + previous = NULL; /* This item can't be repeated */ + continue; /* It is complete */ + } + + /* If the options ended with ':' we are heading into a nested group + with possible change of options. Such groups are non-capturing and are + not assertions of any kind. All we need to do is skip over the ':'; + the newoptions value is handled below. */ + + bravalue = OP_BRA; + ptr++; + } + } + + /* If PCRE_NO_AUTO_CAPTURE is set, all unadorned brackets become + non-capturing and behave like (?:...) brackets */ + + else if ((options & PCRE_NO_AUTO_CAPTURE) != 0) + { + bravalue = OP_BRA; + } + + /* Else we have a referencing group; adjust the opcode. If the bracket + number is greater than EXTRACT_BASIC_MAX, we set the opcode one higher, and + arrange for the true number to follow later, in an OP_BRANUMBER item. */ + + else + { + NUMBERED_GROUP: + if (++(*brackets) > EXTRACT_BASIC_MAX) + { + bravalue = OP_BRA + EXTRACT_BASIC_MAX + 1; + code[1+LINK_SIZE] = OP_BRANUMBER; + PUT2(code, 2+LINK_SIZE, *brackets); + skipbytes = 3; + } + else bravalue = OP_BRA + *brackets; + } + + /* Process nested bracketed re. Assertions may not be repeated, but other + kinds can be. We copy code into a non-register variable in order to be able + to pass its address because some compilers complain otherwise. Pass in a + new setting for the ims options if they have changed. */ + + previous = (bravalue >= OP_ONCE)? code : NULL; + *code = bravalue; + tempcode = code; + tempreqvary = cd->req_varyopt; /* Save value before bracket */ + + if (!compile_regex( + newoptions, /* The complete new option state */ + options & PCRE_IMS, /* The previous ims option state */ + brackets, /* Extracting bracket count */ + &tempcode, /* Where to put code (updated) */ + &ptr, /* Input pointer (updated) */ + errorcodeptr, /* Where to put an error message */ + (bravalue == OP_ASSERTBACK || + bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */ + skipbytes, /* Skip over OP_COND/OP_BRANUMBER */ + &subfirstbyte, /* For possible first char */ + &subreqbyte, /* For possible last char */ + bcptr, /* Current branch chain */ + cd)) /* Tables block */ + goto FAILED; + + /* At the end of compiling, code is still pointing to the start of the + group, while tempcode has been updated to point past the end of the group + and any option resetting that may follow it. The pattern pointer (ptr) + is on the bracket. */ + + /* If this is a conditional bracket, check that there are no more than + two branches in the group. */ + + else if (bravalue == OP_COND) + { + uschar *tc = code; + int condcount = 0; + + do { + condcount++; + tc += GET(tc,1); + } + while (*tc != OP_KET); + + if (condcount > 2) + { + *errorcodeptr = ERR27; + goto FAILED; + } + + /* If there is just one branch, we must not make use of its firstbyte or + reqbyte, because this is equivalent to an empty second branch. */ + + if (condcount == 1) subfirstbyte = subreqbyte = REQ_NONE; + } + + /* Handle updating of the required and first characters. Update for normal + brackets of all kinds, and conditions with two branches (see code above). + If the bracket is followed by a quantifier with zero repeat, we have to + back off. Hence the definition of zeroreqbyte and zerofirstbyte outside the + main loop so that they can be accessed for the back off. */ + + zeroreqbyte = reqbyte; + zerofirstbyte = firstbyte; + groupsetfirstbyte = FALSE; + + if (bravalue >= OP_BRA || bravalue == OP_ONCE || bravalue == OP_COND) + { + /* If we have not yet set a firstbyte in this branch, take it from the + subpattern, remembering that it was set here so that a repeat of more + than one can replicate it as reqbyte if necessary. If the subpattern has + no firstbyte, set "none" for the whole branch. In both cases, a zero + repeat forces firstbyte to "none". */ + + if (firstbyte == REQ_UNSET) + { + if (subfirstbyte >= 0) + { + firstbyte = subfirstbyte; + groupsetfirstbyte = TRUE; + } + else firstbyte = REQ_NONE; + zerofirstbyte = REQ_NONE; + } + + /* If firstbyte was previously set, convert the subpattern's firstbyte + into reqbyte if there wasn't one, using the vary flag that was in + existence beforehand. */ + + else if (subfirstbyte >= 0 && subreqbyte < 0) + subreqbyte = subfirstbyte | tempreqvary; + + /* If the subpattern set a required byte (or set a first byte that isn't + really the first byte - see above), set it. */ + + if (subreqbyte >= 0) reqbyte = subreqbyte; + } + + /* For a forward assertion, we take the reqbyte, if set. This can be + helpful if the pattern that follows the assertion doesn't set a different + char. For example, it's useful for /(?=abcde).+/. We can't set firstbyte + for an assertion, however because it leads to incorrect effect for patterns + such as /(?=a)a.+/ when the "real" "a" would then become a reqbyte instead + of a firstbyte. This is overcome by a scan at the end if there's no + firstbyte, looking for an asserted first char. */ + + else if (bravalue == OP_ASSERT && subreqbyte >= 0) reqbyte = subreqbyte; + + /* Now update the main code pointer to the end of the group. */ + + code = tempcode; + + /* Error if hit end of pattern */ + + if (*ptr != ')') + { + *errorcodeptr = ERR14; + goto FAILED; + } + break; + + /* Check \ for being a real metacharacter; if not, fall through and handle + it as a data character at the start of a string. Escape items are checked + for validity in the pre-compiling pass. */ + + case '\\': + tempptr = ptr; + c = check_escape(&ptr, errorcodeptr, *brackets, options, FALSE); + + /* Handle metacharacters introduced by \. For ones like \d, the ESC_ values + are arranged to be the negation of the corresponding OP_values. For the + back references, the values are ESC_REF plus the reference number. Only + back references and those types that consume a character may be repeated. + We can test for values between ESC_b and ESC_Z for the latter; this may + have to change if any new ones are ever created. */ + + if (c < 0) + { + if (-c == ESC_Q) /* Handle start of quoted string */ + { + if (ptr[1] == '\\' && ptr[2] == 'E') ptr += 2; /* avoid empty string */ + else inescq = TRUE; + continue; + } + + /* For metasequences that actually match a character, we disable the + setting of a first character if it hasn't already been set. */ + + if (firstbyte == REQ_UNSET && -c > ESC_b && -c < ESC_Z) + firstbyte = REQ_NONE; + + /* Set values to reset to if this is followed by a zero repeat. */ + + zerofirstbyte = firstbyte; + zeroreqbyte = reqbyte; + + /* Back references are handled specially */ + + if (-c >= ESC_REF) + { + int number = -c - ESC_REF; + previous = code; + *code++ = OP_REF; + PUT2INC(code, 0, number); + } + + /* So are Unicode property matches, if supported. We know that get_ucp + won't fail because it was tested in the pre-pass. */ + +#ifdef SUPPORT_UCP + else if (-c == ESC_P || -c == ESC_p) + { + BOOL negated; + int pdata; + int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr); + previous = code; + *code++ = ((-c == ESC_p) != negated)? OP_PROP : OP_NOTPROP; + *code++ = ptype; + *code++ = pdata; + } +#endif + + /* For the rest, we can obtain the OP value by negating the escape + value */ + + else + { + previous = (-c > ESC_b && -c < ESC_Z)? code : NULL; + *code++ = -c; + } + continue; + } + + /* We have a data character whose value is in c. In UTF-8 mode it may have + a value > 127. We set its representation in the length/buffer, and then + handle it as a data character. */ + +#ifdef SUPPORT_UTF8 + if (utf8 && c > 127) + mclength = _pcre_ord2utf8(c, mcbuffer); + else +#endif + + { + mcbuffer[0] = c; + mclength = 1; + } + + goto ONE_CHAR; + + /* Handle a literal character. It is guaranteed not to be whitespace or # + when the extended flag is set. If we are in UTF-8 mode, it may be a + multi-byte literal character. */ + + default: + NORMAL_CHAR: + mclength = 1; + mcbuffer[0] = c; + +#ifdef SUPPORT_UTF8 + if (utf8 && (c & 0xc0) == 0xc0) + { + while ((ptr[1] & 0xc0) == 0x80) + mcbuffer[mclength++] = *(++ptr); + } +#endif + + /* At this point we have the character's bytes in mcbuffer, and the length + in mclength. When not in UTF-8 mode, the length is always 1. */ + + ONE_CHAR: + previous = code; + *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR; + for (c = 0; c < mclength; c++) *code++ = mcbuffer[c]; + + /* Set the first and required bytes appropriately. If no previous first + byte, set it from this character, but revert to none on a zero repeat. + Otherwise, leave the firstbyte value alone, and don't change it on a zero + repeat. */ + + if (firstbyte == REQ_UNSET) + { + zerofirstbyte = REQ_NONE; + zeroreqbyte = reqbyte; + + /* If the character is more than one byte long, we can set firstbyte + only if it is not to be matched caselessly. */ + + if (mclength == 1 || req_caseopt == 0) + { + firstbyte = mcbuffer[0] | req_caseopt; + if (mclength != 1) reqbyte = code[-1] | cd->req_varyopt; + } + else firstbyte = reqbyte = REQ_NONE; + } + + /* firstbyte was previously set; we can set reqbyte only the length is + 1 or the matching is caseful. */ + + else + { + zerofirstbyte = firstbyte; + zeroreqbyte = reqbyte; + if (mclength == 1 || req_caseopt == 0) + reqbyte = code[-1] | req_caseopt | cd->req_varyopt; + } + + break; /* End of literal character handling */ + } + } /* end of big loop */ + +/* Control never reaches here by falling through, only by a goto for all the +error states. Pass back the position in the pattern so that it can be displayed +to the user for diagnosing the error. */ + +FAILED: +*ptrptr = ptr; +return FALSE; +} + + + + +/************************************************* +* Compile sequence of alternatives * +*************************************************/ + +/* On entry, ptr is pointing past the bracket character, but on return +it points to the closing bracket, or vertical bar, or end of string. +The code variable is pointing at the byte into which the BRA operator has been +stored. If the ims options are changed at the start (for a (?ims: group) or +during any branch, we need to insert an OP_OPT item at the start of every +following branch to ensure they get set correctly at run time, and also pass +the new options into every subsequent branch compile. + +Argument: + options option bits, including any changes for this subpattern + oldims previous settings of ims option bits + brackets -> int containing the number of extracting brackets used + codeptr -> the address of the current code pointer + ptrptr -> the address of the current pattern pointer + errorcodeptr -> pointer to error code variable + lookbehind TRUE if this is a lookbehind assertion + skipbytes skip this many bytes at start (for OP_COND, OP_BRANUMBER) + firstbyteptr place to put the first required character, or a negative number + reqbyteptr place to put the last required character, or a negative number + bcptr pointer to the chain of currently open branches + cd points to the data block with tables pointers etc. + +Returns: TRUE on success +*/ + +static BOOL +compile_regex(int options, int oldims, int *brackets, uschar **codeptr, + const uschar **ptrptr, int *errorcodeptr, BOOL lookbehind, int skipbytes, + int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, compile_data *cd) +{ +const uschar *ptr = *ptrptr; +uschar *code = *codeptr; +uschar *last_branch = code; +uschar *start_bracket = code; +uschar *reverse_count = NULL; +int firstbyte, reqbyte; +int branchfirstbyte, branchreqbyte; +branch_chain bc; + +bc.outer = bcptr; +bc.current = code; + +firstbyte = reqbyte = REQ_UNSET; + +/* Offset is set zero to mark that this bracket is still open */ + +PUT(code, 1, 0); +code += 1 + LINK_SIZE + skipbytes; + +/* Loop for each alternative branch */ + +for (;;) + { + /* Handle a change of ims options at the start of the branch */ + + if ((options & PCRE_IMS) != oldims) + { + *code++ = OP_OPT; + *code++ = options & PCRE_IMS; + } + + /* Set up dummy OP_REVERSE if lookbehind assertion */ + + if (lookbehind) + { + *code++ = OP_REVERSE; + reverse_count = code; + PUTINC(code, 0, 0); + } + + /* Now compile the branch */ + + if (!compile_branch(&options, brackets, &code, &ptr, errorcodeptr, + &branchfirstbyte, &branchreqbyte, &bc, cd)) + { + *ptrptr = ptr; + return FALSE; + } + + /* If this is the first branch, the firstbyte and reqbyte values for the + branch become the values for the regex. */ + + if (*last_branch != OP_ALT) + { + firstbyte = branchfirstbyte; + reqbyte = branchreqbyte; + } + + /* If this is not the first branch, the first char and reqbyte have to + match the values from all the previous branches, except that if the previous + value for reqbyte didn't have REQ_VARY set, it can still match, and we set + REQ_VARY for the regex. */ + + else + { + /* If we previously had a firstbyte, but it doesn't match the new branch, + we have to abandon the firstbyte for the regex, but if there was previously + no reqbyte, it takes on the value of the old firstbyte. */ + + if (firstbyte >= 0 && firstbyte != branchfirstbyte) + { + if (reqbyte < 0) reqbyte = firstbyte; + firstbyte = REQ_NONE; + } + + /* If we (now or from before) have no firstbyte, a firstbyte from the + branch becomes a reqbyte if there isn't a branch reqbyte. */ + + if (firstbyte < 0 && branchfirstbyte >= 0 && branchreqbyte < 0) + branchreqbyte = branchfirstbyte; + + /* Now ensure that the reqbytes match */ + + if ((reqbyte & ~REQ_VARY) != (branchreqbyte & ~REQ_VARY)) + reqbyte = REQ_NONE; + else reqbyte |= branchreqbyte; /* To "or" REQ_VARY */ + } + + /* If lookbehind, check that this branch matches a fixed-length string, + and put the length into the OP_REVERSE item. Temporarily mark the end of + the branch with OP_END. */ + + if (lookbehind) + { + int length; + *code = OP_END; + length = find_fixedlength(last_branch, options); + DPRINTF(("fixed length = %d\n", length)); + if (length < 0) + { + *errorcodeptr = (length == -2)? ERR36 : ERR25; + *ptrptr = ptr; + return FALSE; + } + PUT(reverse_count, 0, length); + } + + /* Reached end of expression, either ')' or end of pattern. Go back through + the alternative branches and reverse the chain of offsets, with the field in + the BRA item now becoming an offset to the first alternative. If there are + no alternatives, it points to the end of the group. The length in the + terminating ket is always the length of the whole bracketed item. If any of + the ims options were changed inside the group, compile a resetting op-code + following, except at the very end of the pattern. Return leaving the pointer + at the terminating char. */ + + if (*ptr != '|') + { + int length = code - last_branch; + do + { + int prev_length = GET(last_branch, 1); + PUT(last_branch, 1, length); + length = prev_length; + last_branch -= length; + } + while (length > 0); + + /* Fill in the ket */ + + *code = OP_KET; + PUT(code, 1, code - start_bracket); + code += 1 + LINK_SIZE; + + /* Resetting option if needed */ + + if ((options & PCRE_IMS) != oldims && *ptr == ')') + { + *code++ = OP_OPT; + *code++ = oldims; + } + + /* Set values to pass back */ + + *codeptr = code; + *ptrptr = ptr; + *firstbyteptr = firstbyte; + *reqbyteptr = reqbyte; + return TRUE; + } + + /* Another branch follows; insert an "or" node. Its length field points back + to the previous branch while the bracket remains open. At the end the chain + is reversed. It's done like this so that the start of the bracket has a + zero offset until it is closed, making it possible to detect recursion. */ + + *code = OP_ALT; + PUT(code, 1, code - last_branch); + bc.current = last_branch = code; + code += 1 + LINK_SIZE; + ptr++; + } +/* Control never reaches here */ +} + + + + +/************************************************* +* Check for anchored expression * +*************************************************/ + +/* Try to find out if this is an anchored regular expression. Consider each +alternative branch. If they all start with OP_SOD or OP_CIRC, or with a bracket +all of whose alternatives start with OP_SOD or OP_CIRC (recurse ad lib), then +it's anchored. However, if this is a multiline pattern, then only OP_SOD +counts, since OP_CIRC can match in the middle. + +We can also consider a regex to be anchored if OP_SOM starts all its branches. +This is the code for \G, which means "match at start of match position, taking +into account the match offset". + +A branch is also implicitly anchored if it starts with .* and DOTALL is set, +because that will try the rest of the pattern at all possible matching points, +so there is no point trying again.... er .... + +.... except when the .* appears inside capturing parentheses, and there is a +subsequent back reference to those parentheses. We haven't enough information +to catch that case precisely. + +At first, the best we could do was to detect when .* was in capturing brackets +and the highest back reference was greater than or equal to that level. +However, by keeping a bitmap of the first 31 back references, we can catch some +of the more common cases more precisely. + +Arguments: + code points to start of expression (the bracket) + options points to the options setting + bracket_map a bitmap of which brackets we are inside while testing; this + handles up to substring 31; after that we just have to take + the less precise approach + backref_map the back reference bitmap + +Returns: TRUE or FALSE +*/ + +static BOOL +is_anchored(register const uschar *code, int *options, unsigned int bracket_map, + unsigned int backref_map) +{ +do { + const uschar *scode = + first_significant_code(code + 1+LINK_SIZE, options, PCRE_MULTILINE, FALSE); + register int op = *scode; + + /* Capturing brackets */ + + if (op > OP_BRA) + { + int new_map; + op -= OP_BRA; + if (op > EXTRACT_BASIC_MAX) op = GET2(scode, 2+LINK_SIZE); + new_map = bracket_map | ((op < 32)? (1 << op) : 1); + if (!is_anchored(scode, options, new_map, backref_map)) return FALSE; + } + + /* Other brackets */ + + else if (op == OP_BRA || op == OP_ASSERT || op == OP_ONCE || op == OP_COND) + { + if (!is_anchored(scode, options, bracket_map, backref_map)) return FALSE; + } + + /* .* is not anchored unless DOTALL is set and it isn't in brackets that + are or may be referenced. */ + + else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR) && + (*options & PCRE_DOTALL) != 0) + { + if (scode[1] != OP_ANY || (bracket_map & backref_map) != 0) return FALSE; + } + + /* Check for explicit anchoring */ + + else if (op != OP_SOD && op != OP_SOM && + ((*options & PCRE_MULTILINE) != 0 || op != OP_CIRC)) + return FALSE; + code += GET(code, 1); + } +while (*code == OP_ALT); /* Loop for each alternative */ +return TRUE; +} + + + +/************************************************* +* Check for starting with ^ or .* * +*************************************************/ + +/* This is called to find out if every branch starts with ^ or .* so that +"first char" processing can be done to speed things up in multiline +matching and for non-DOTALL patterns that start with .* (which must start at +the beginning or after \n). As in the case of is_anchored() (see above), we +have to take account of back references to capturing brackets that contain .* +because in that case we can't make the assumption. + +Arguments: + code points to start of expression (the bracket) + bracket_map a bitmap of which brackets we are inside while testing; this + handles up to substring 31; after that we just have to take + the less precise approach + backref_map the back reference bitmap + +Returns: TRUE or FALSE +*/ + +static BOOL +is_startline(const uschar *code, unsigned int bracket_map, + unsigned int backref_map) +{ +do { + const uschar *scode = first_significant_code(code + 1+LINK_SIZE, NULL, 0, + FALSE); + register int op = *scode; + + /* Capturing brackets */ + + if (op > OP_BRA) + { + int new_map; + op -= OP_BRA; + if (op > EXTRACT_BASIC_MAX) op = GET2(scode, 2+LINK_SIZE); + new_map = bracket_map | ((op < 32)? (1 << op) : 1); + if (!is_startline(scode, new_map, backref_map)) return FALSE; + } + + /* Other brackets */ + + else if (op == OP_BRA || op == OP_ASSERT || op == OP_ONCE || op == OP_COND) + { if (!is_startline(scode, bracket_map, backref_map)) return FALSE; } + + /* .* means "start at start or after \n" if it isn't in brackets that + may be referenced. */ + + else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR) + { + if (scode[1] != OP_ANY || (bracket_map & backref_map) != 0) return FALSE; + } + + /* Check for explicit circumflex */ + + else if (op != OP_CIRC) return FALSE; + + /* Move on to the next alternative */ + + code += GET(code, 1); + } +while (*code == OP_ALT); /* Loop for each alternative */ +return TRUE; +} + + + +/************************************************* +* Check for asserted fixed first char * +*************************************************/ + +/* During compilation, the "first char" settings from forward assertions are +discarded, because they can cause conflicts with actual literals that follow. +However, if we end up without a first char setting for an unanchored pattern, +it is worth scanning the regex to see if there is an initial asserted first +char. If all branches start with the same asserted char, or with a bracket all +of whose alternatives start with the same asserted char (recurse ad lib), then +we return that char, otherwise -1. + +Arguments: + code points to start of expression (the bracket) + options pointer to the options (used to check casing changes) + inassert TRUE if in an assertion + +Returns: -1 or the fixed first char +*/ + +static int +find_firstassertedchar(const uschar *code, int *options, BOOL inassert) +{ +register int c = -1; +do { + int d; + const uschar *scode = + first_significant_code(code + 1+LINK_SIZE, options, PCRE_CASELESS, TRUE); + register int op = *scode; + + if (op >= OP_BRA) op = OP_BRA; + + switch(op) + { + default: + return -1; + + case OP_BRA: + case OP_ASSERT: + case OP_ONCE: + case OP_COND: + if ((d = find_firstassertedchar(scode, options, op == OP_ASSERT)) < 0) + return -1; + if (c < 0) c = d; else if (c != d) return -1; + break; + + case OP_EXACT: /* Fall through */ + scode += 2; + + case OP_CHAR: + case OP_CHARNC: + case OP_PLUS: + case OP_MINPLUS: + if (!inassert) return -1; + if (c < 0) + { + c = scode[1]; + if ((*options & PCRE_CASELESS) != 0) c |= REQ_CASELESS; + } + else if (c != scode[1]) return -1; + break; + } + + code += GET(code, 1); + } +while (*code == OP_ALT); +return c; +} + + + +/************************************************* +* Compile a Regular Expression * +*************************************************/ + +/* This function takes a string and returns a pointer to a block of store +holding a compiled version of the expression. The original API for this +function had no error code return variable; it is retained for backwards +compatibility. The new function is given a new name. + +Arguments: + pattern the regular expression + options various option bits + errorcodeptr pointer to error code variable (pcre_compile2() only) + can be NULL if you don't want a code value + errorptr pointer to pointer to error text + erroroffset ptr offset in pattern where error was detected + tables pointer to character tables or NULL + +Returns: pointer to compiled data block, or NULL on error, + with errorptr and erroroffset set +*/ + +PCRE_DATA_SCOPE pcre * +pcre_compile(const char *pattern, int options, const char **errorptr, + int *erroroffset, const unsigned char *tables) +{ +return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables); +} + + + +PCRE_DATA_SCOPE pcre * +pcre_compile2(const char *pattern, int options, int *errorcodeptr, + const char **errorptr, int *erroroffset, const unsigned char *tables) +{ +real_pcre *re; +int length = 1 + LINK_SIZE; /* For initial BRA plus length */ +int c, firstbyte, reqbyte, newline; +int bracount = 0; +int branch_extra = 0; +int branch_newextra; +int item_count = -1; +int name_count = 0; +int max_name_size = 0; +int lastitemlength = 0; +int errorcode = 0; +#ifdef SUPPORT_UTF8 +BOOL utf8; +BOOL class_utf8; +#endif +BOOL inescq = FALSE; +BOOL capturing; +unsigned int brastackptr = 0; +size_t size; +uschar *code; +const uschar *codestart; +const uschar *ptr; +compile_data compile_block; +compile_data *cd = &compile_block; +int brastack[BRASTACK_SIZE]; +uschar bralenstack[BRASTACK_SIZE]; + +/* We can't pass back an error message if errorptr is NULL; I guess the best we +can do is just return NULL, but we can set a code value if there is a code +pointer. */ + +if (errorptr == NULL) + { + if (errorcodeptr != NULL) *errorcodeptr = 99; + return NULL; + } + +*errorptr = NULL; +if (errorcodeptr != NULL) *errorcodeptr = ERR0; + +/* However, we can give a message for this error */ + +if (erroroffset == NULL) + { + errorcode = ERR16; + goto PCRE_EARLY_ERROR_RETURN; + } + +*erroroffset = 0; + +/* Can't support UTF8 unless PCRE has been compiled to include the code. */ + +#ifdef SUPPORT_UTF8 +utf8 = (options & PCRE_UTF8) != 0; +if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0 && + (*erroroffset = _pcre_valid_utf8((uschar *)pattern, -1)) >= 0) + { + errorcode = ERR44; + goto PCRE_EARLY_ERROR_RETURN; + } +#else +if ((options & PCRE_UTF8) != 0) + { + errorcode = ERR32; + goto PCRE_EARLY_ERROR_RETURN; + } +#endif + +if ((options & ~PUBLIC_OPTIONS) != 0) + { + errorcode = ERR17; + goto PCRE_EARLY_ERROR_RETURN; + } + +/* Set up pointers to the individual character tables */ + +if (tables == NULL) tables = _pcre_default_tables; +cd->lcc = tables + lcc_offset; +cd->fcc = tables + fcc_offset; +cd->cbits = tables + cbits_offset; +cd->ctypes = tables + ctypes_offset; + +/* Handle different types of newline. The two bits give four cases. The current +code allows for one- or two-byte sequences. */ + +switch (options & PCRE_NEWLINE_CRLF) + { + default: newline = NEWLINE; break; /* Compile-time default */ + case PCRE_NEWLINE_CR: newline = '\r'; break; + case PCRE_NEWLINE_LF: newline = '\n'; break; + case PCRE_NEWLINE_CR+ + PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break; + } + +if (newline > 255) + { + cd->nllen = 2; + cd->nl[0] = (newline >> 8) & 255; + cd->nl[1] = newline & 255; + } +else + { + cd->nllen = 1; + cd->nl[0] = newline; + } + +/* Maximum back reference and backref bitmap. This is updated for numeric +references during the first pass, but for named references during the actual +compile pass. The bitmap records up to 31 back references to help in deciding +whether (.*) can be treated as anchored or not. */ + +cd->top_backref = 0; +cd->backref_map = 0; + +/* Reflect pattern for debugging output */ + +DPRINTF(("------------------------------------------------------------------\n")); +DPRINTF(("%s\n", pattern)); + +/* The first thing to do is to make a pass over the pattern to compute the +amount of store required to hold the compiled code. This does not have to be +perfect as long as errors are overestimates. At the same time we can detect any +flag settings right at the start, and extract them. Make an attempt to correct +for any counted white space if an "extended" flag setting appears late in the +pattern. We can't be so clever for #-comments. */ + +ptr = (const uschar *)(pattern - 1); +while ((c = *(++ptr)) != 0) + { + int min, max; + int class_optcount; + int bracket_length; + int duplength; + + /* If we are inside a \Q...\E sequence, all chars are literal */ + + if (inescq) + { + if ((options & PCRE_AUTO_CALLOUT) != 0) length += 2 + 2*LINK_SIZE; + goto NORMAL_CHAR; + } + + /* Otherwise, first check for ignored whitespace and comments */ + + if ((options & PCRE_EXTENDED) != 0) + { + if ((cd->ctypes[c] & ctype_space) != 0) continue; + if (c == '#') + { + while (*(++ptr) != 0) if (IS_NEWLINE(ptr)) break; + if (*ptr != 0) + { + ptr += cd->nllen - 1; + continue; + } + break; /* End loop at end of pattern */ + } + } + + item_count++; /* Is zero for the first non-comment item */ + + /* Allow space for auto callout before every item except quantifiers. */ + + if ((options & PCRE_AUTO_CALLOUT) != 0 && + c != '*' && c != '+' && c != '?' && + (c != '{' || !is_counted_repeat(ptr + 1))) + length += 2 + 2*LINK_SIZE; + + switch(c) + { + /* A backslashed item may be an escaped data character or it may be a + character type. */ + + case '\\': + c = check_escape(&ptr, &errorcode, bracount, options, FALSE); + if (errorcode != 0) goto PCRE_ERROR_RETURN; + + lastitemlength = 1; /* Default length of last item for repeats */ + + if (c >= 0) /* Data character */ + { + length += 2; /* For a one-byte character */ + +#ifdef SUPPORT_UTF8 + if (utf8 && c > 127) + { + int i; + for (i = 0; i < _pcre_utf8_table1_size; i++) + if (c <= _pcre_utf8_table1[i]) break; + length += i; + lastitemlength += i; + } +#endif + + continue; + } + + /* If \Q, enter "literal" mode */ + + if (-c == ESC_Q) + { + inescq = TRUE; + continue; + } + + /* \X is supported only if Unicode property support is compiled */ + +#ifndef SUPPORT_UCP + if (-c == ESC_X) + { + errorcode = ERR45; + goto PCRE_ERROR_RETURN; + } +#endif + + /* \P and \p are for Unicode properties, but only when the support has + been compiled. Each item needs 3 bytes. */ + + else if (-c == ESC_P || -c == ESC_p) + { +#ifdef SUPPORT_UCP + BOOL negated; + BOOL pdata; + length += 3; + lastitemlength = 3; + if (get_ucp(&ptr, &negated, &pdata, &errorcode) < 0) + goto PCRE_ERROR_RETURN; + continue; +#else + errorcode = ERR45; + goto PCRE_ERROR_RETURN; +#endif + } + + /* Other escapes need one byte */ + + length++; + + /* A back reference needs an additional 2 bytes, plus either one or 5 + bytes for a repeat. We also need to keep the value of the highest + back reference. */ + + if (c <= -ESC_REF) + { + int refnum = -c - ESC_REF; + cd->backref_map |= (refnum < 32)? (1 << refnum) : 1; + if (refnum > cd->top_backref) + cd->top_backref = refnum; + length += 2; /* For single back reference */ + if (ptr[1] == '{' && is_counted_repeat(ptr+2)) + { + ptr = read_repeat_counts(ptr+2, &min, &max, &errorcode); + if (errorcode != 0) goto PCRE_ERROR_RETURN; + if ((min == 0 && (max == 1 || max == -1)) || + (min == 1 && max == -1)) + length++; + else length += 5; + if (ptr[1] == '?') ptr++; + } + } + continue; + + case '^': /* Single-byte metacharacters */ + case '.': + case '$': + length++; + lastitemlength = 1; + continue; + + case '*': /* These repeats won't be after brackets; */ + case '+': /* those are handled separately */ + case '?': + length++; + goto POSESSIVE; /* A few lines below */ + + /* This covers the cases of braced repeats after a single char, metachar, + class, or back reference. */ + + case '{': + if (!is_counted_repeat(ptr+1)) goto NORMAL_CHAR; + ptr = read_repeat_counts(ptr+1, &min, &max, &errorcode); + if (errorcode != 0) goto PCRE_ERROR_RETURN; + + /* These special cases just insert one extra opcode */ + + if ((min == 0 && (max == 1 || max == -1)) || + (min == 1 && max == -1)) + length++; + + /* These cases might insert additional copies of a preceding character. */ + + else + { + if (min != 1) + { + length -= lastitemlength; /* Uncount the original char or metachar */ + if (min > 0) length += 3 + lastitemlength; + } + length += lastitemlength + ((max > 0)? 3 : 1); + } + + if (ptr[1] == '?') ptr++; /* Needs no extra length */ + + POSESSIVE: /* Test for possessive quantifier */ + if (ptr[1] == '+') + { + ptr++; + length += 2 + 2*LINK_SIZE; /* Allow for atomic brackets */ + } + continue; + + /* An alternation contains an offset to the next branch or ket. If any ims + options changed in the previous branch(es), and/or if we are in a + lookbehind assertion, extra space will be needed at the start of the + branch. This is handled by branch_extra. */ + + case '|': + length += 1 + LINK_SIZE + branch_extra; + continue; + + /* A character class uses 33 characters provided that all the character + values are less than 256. Otherwise, it uses a bit map for low valued + characters, and individual items for others. Don't worry about character + types that aren't allowed in classes - they'll get picked up during the + compile. A character class that contains only one single-byte character + uses 2 or 3 bytes, depending on whether it is negated or not. Notice this + where we can. (In UTF-8 mode we can do this only for chars < 128.) */ + + case '[': + if (*(++ptr) == '^') + { + class_optcount = 10; /* Greater than one */ + ptr++; + } + else class_optcount = 0; + +#ifdef SUPPORT_UTF8 + class_utf8 = FALSE; +#endif + + /* Written as a "do" so that an initial ']' is taken as data */ + + if (*ptr != 0) do + { + /* Inside \Q...\E everything is literal except \E */ + + if (inescq) + { + if (*ptr != '\\' || ptr[1] != 'E') goto GET_ONE_CHARACTER; + inescq = FALSE; + ptr += 1; + continue; + } + + /* Outside \Q...\E, check for escapes */ + + if (*ptr == '\\') + { + c = check_escape(&ptr, &errorcode, bracount, options, TRUE); + if (errorcode != 0) goto PCRE_ERROR_RETURN; + + /* \b is backspace inside a class; \X is literal */ + + if (-c == ESC_b) c = '\b'; + else if (-c == ESC_X) c = 'X'; + + /* \Q enters quoting mode */ + + else if (-c == ESC_Q) + { + inescq = TRUE; + continue; + } + + /* Handle escapes that turn into characters */ + + if (c >= 0) goto NON_SPECIAL_CHARACTER; + + /* Escapes that are meta-things. The normal ones just affect the + bit map, but Unicode properties require an XCLASS extended item. */ + + else + { + class_optcount = 10; /* \d, \s etc; make sure > 1 */ +#ifdef SUPPORT_UTF8 + if (-c == ESC_p || -c == ESC_P) + { + if (!class_utf8) + { + class_utf8 = TRUE; + length += LINK_SIZE + 2; + } + length += 3; + } +#endif + } + } + + /* Check the syntax for POSIX stuff. The bits we actually handle are + checked during the real compile phase. */ + + else if (*ptr == '[' && + (ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') && + check_posix_syntax(ptr, &ptr, cd)) + { + ptr++; + class_optcount = 10; /* Make sure > 1 */ + } + + /* Anything else increments the possible optimization count. We have to + detect ranges here so that we can compute the number of extra ranges for + caseless wide characters when UCP support is available. If there are wide + characters, we are going to have to use an XCLASS, even for single + characters. */ + + else + { + int d; + + GET_ONE_CHARACTER: + +#ifdef SUPPORT_UTF8 + if (utf8) + { + int extra = 0; + GETCHARLEN(c, ptr, extra); + ptr += extra; + } + else c = *ptr; +#else + c = *ptr; +#endif + + /* Come here from handling \ above when it escapes to a char value */ + + NON_SPECIAL_CHARACTER: + class_optcount++; + + d = -1; + if (ptr[1] == '-') + { + uschar const *hyptr = ptr++; + if (ptr[1] == '\\') + { + ptr++; + d = check_escape(&ptr, &errorcode, bracount, options, TRUE); + if (errorcode != 0) goto PCRE_ERROR_RETURN; + if (-d == ESC_b) d = '\b'; /* backspace */ + else if (-d == ESC_X) d = 'X'; /* literal X in a class */ + } + else if (ptr[1] != 0 && ptr[1] != ']') + { + ptr++; +#ifdef SUPPORT_UTF8 + if (utf8) + { + int extra = 0; + GETCHARLEN(d, ptr, extra); + ptr += extra; + } + else +#endif + d = *ptr; + } + if (d < 0) ptr = hyptr; /* go back to hyphen as data */ + } + + /* If d >= 0 we have a range. In UTF-8 mode, if the end is > 255, or > + 127 for caseless matching, we will need to use an XCLASS. */ + + if (d >= 0) + { + class_optcount = 10; /* Ensure > 1 */ + if (d < c) + { + errorcode = ERR8; + goto PCRE_ERROR_RETURN; + } + +#ifdef SUPPORT_UTF8 + if (utf8 && (d > 255 || ((options & PCRE_CASELESS) != 0 && d > 127))) + { + uschar buffer[6]; + if (!class_utf8) /* Allow for XCLASS overhead */ + { + class_utf8 = TRUE; + length += LINK_SIZE + 2; + } + +#ifdef SUPPORT_UCP + /* If we have UCP support, find out how many extra ranges are + needed to map the other case of characters within this range. We + have to mimic the range optimization here, because extending the + range upwards might push d over a boundary that makes is use + another byte in the UTF-8 representation. */ + + if ((options & PCRE_CASELESS) != 0) + { + int occ, ocd; + int cc = c; + int origd = d; + while (get_othercase_range(&cc, origd, &occ, &ocd)) + { + if (occ >= c && ocd <= d) continue; /* Skip embedded */ + + if (occ < c && ocd >= c - 1) /* Extend the basic range */ + { /* if there is overlap, */ + c = occ; /* noting that if occ < c */ + continue; /* we can't have ocd > d */ + } /* because a subrange is */ + if (ocd > d && occ <= d + 1) /* always shorter than */ + { /* the basic range. */ + d = ocd; + continue; + } + + /* An extra item is needed */ + + length += 1 + _pcre_ord2utf8(occ, buffer) + + ((occ == ocd)? 0 : _pcre_ord2utf8(ocd, buffer)); + } + } +#endif /* SUPPORT_UCP */ + + /* The length of the (possibly extended) range */ + + length += 1 + _pcre_ord2utf8(c, buffer) + _pcre_ord2utf8(d, buffer); + } +#endif /* SUPPORT_UTF8 */ + + } + + /* We have a single character. There is nothing to be done unless we + are in UTF-8 mode. If the char is > 255, or 127 when caseless, we must + allow for an XCL_SINGLE item, doubled for caselessness if there is UCP + support. */ + + else + { +#ifdef SUPPORT_UTF8 + if (utf8 && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127))) + { + uschar buffer[6]; + class_optcount = 10; /* Ensure > 1 */ + if (!class_utf8) /* Allow for XCLASS overhead */ + { + class_utf8 = TRUE; + length += LINK_SIZE + 2; + } +#ifdef SUPPORT_UCP + length += (((options & PCRE_CASELESS) != 0)? 2 : 1) * + (1 + _pcre_ord2utf8(c, buffer)); +#else /* SUPPORT_UCP */ + length += 1 + _pcre_ord2utf8(c, buffer); +#endif /* SUPPORT_UCP */ + } +#endif /* SUPPORT_UTF8 */ + } + } + } + while (*(++ptr) != 0 && (inescq || *ptr != ']')); /* Concludes "do" above */ + + if (*ptr == 0) /* Missing terminating ']' */ + { + errorcode = ERR6; + goto PCRE_ERROR_RETURN; + } + + /* We can optimize when there was only one optimizable character. Repeats + for positive and negated single one-byte chars are handled by the general + code. Here, we handle repeats for the class opcodes. */ + + if (class_optcount == 1) length += 3; else + { + length += 33; + + /* A repeat needs either 1 or 5 bytes. If it is a possessive quantifier, + we also need extra for wrapping the whole thing in a sub-pattern. */ + + if (*ptr != 0 && ptr[1] == '{' && is_counted_repeat(ptr+2)) + { + ptr = read_repeat_counts(ptr+2, &min, &max, &errorcode); + if (errorcode != 0) goto PCRE_ERROR_RETURN; + if ((min == 0 && (max == 1 || max == -1)) || + (min == 1 && max == -1)) + length++; + else length += 5; + if (ptr[1] == '+') + { + ptr++; + length += 2 + 2*LINK_SIZE; + } + else if (ptr[1] == '?') ptr++; + } + } + continue; + + /* Brackets may be genuine groups or special things */ + + case '(': + branch_newextra = 0; + bracket_length = 1 + LINK_SIZE; + capturing = FALSE; + + /* Handle special forms of bracket, which all start (? */ + + if (ptr[1] == '?') + { + int set, unset; + int *optset; + + switch (c = ptr[2]) + { + /* Skip over comments entirely */ + case '#': + ptr += 3; + while (*ptr != 0 && *ptr != ')') ptr++; + if (*ptr == 0) + { + errorcode = ERR18; + goto PCRE_ERROR_RETURN; + } + continue; + + /* Non-referencing groups and lookaheads just move the pointer on, and + then behave like a non-special bracket, except that they don't increment + the count of extracting brackets. Ditto for the "once only" bracket, + which is in Perl from version 5.005. */ + + case ':': + case '=': + case '!': + case '>': + ptr += 2; + break; + + /* Named subpatterns are an extension copied from Python */ + + case 'P': + ptr += 3; + + /* Handle the definition of a named subpattern */ + + if (*ptr == '<') + { + const uschar *p; /* Don't amalgamate; some compilers */ + p = ++ptr; /* grumble at autoincrement in declaration */ + while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++; + if (*ptr != '>') + { + errorcode = ERR42; + goto PCRE_ERROR_RETURN; + } + name_count++; + if (name_count > MAX_NAME_COUNT) + { + errorcode = ERR49; + goto PCRE_ERROR_RETURN; + } + if (ptr - p > max_name_size) + { + max_name_size = (ptr - p); + if (max_name_size > MAX_NAME_SIZE) + { + errorcode = ERR48; + goto PCRE_ERROR_RETURN; + } + } + capturing = TRUE; /* Named parentheses are always capturing */ + break; /* Go handle capturing parentheses */ + } + + /* Handle back references and recursive calls to named subpatterns */ + + if (*ptr == '=' || *ptr == '>') + { + length += 3 + 3*LINK_SIZE; /* Allow for the automatic "once" */ + while ((cd->ctypes[*(++ptr)] & ctype_word) != 0); + if (*ptr != ')') + { + errorcode = ERR42; + goto PCRE_ERROR_RETURN; + } + goto RECURSE_CHECK_QUANTIFIED; + } + + /* Unknown character after (?P */ + + errorcode = ERR41; + goto PCRE_ERROR_RETURN; + + /* (?R) specifies a recursive call to the regex, which is an extension + to provide the facility which can be obtained by (?p{perl-code}) in + Perl 5.6. In Perl 5.8 this has become (??{perl-code}). + + From PCRE 4.00, items such as (?3) specify subroutine-like "calls" to + the appropriate numbered brackets. This includes both recursive and + non-recursive calls. (?R) is now synonymous with (?0). */ + + case 'R': + ptr++; + + case '0': case '1': case '2': case '3': case '4': + case '5': case '6': case '7': case '8': case '9': + ptr += 2; + if (c != 'R') + while ((digitab[*(++ptr)] & ctype_digit) != 0); + if (*ptr != ')') + { + errorcode = ERR29; + goto PCRE_ERROR_RETURN; + } + length += 3 + 3*LINK_SIZE; /* Allows for the automatic "once" */ + + /* If this item is quantified, it will get wrapped inside brackets so + as to use the code for quantified brackets. We jump down and use the + code that handles this for real brackets. Come here from code for + named recursions/subroutines. */ + + RECURSE_CHECK_QUANTIFIED: + if (ptr[1] == '+' || ptr[1] == '*' || ptr[1] == '?' || ptr[1] == '{') + { + length += 2 + 2 * LINK_SIZE; /* to make bracketed */ + duplength = 5 + 3 * LINK_SIZE; + goto HANDLE_QUANTIFIED_BRACKETS; + } + continue; + + /* (?C) is an extension which provides "callout" - to provide a bit of + the functionality of the Perl (?{...}) feature. An optional number may + follow (default is zero). */ + + case 'C': + ptr += 2; + while ((digitab[*(++ptr)] & ctype_digit) != 0); + if (*ptr != ')') + { + errorcode = ERR39; + goto PCRE_ERROR_RETURN; + } + length += 2 + 2*LINK_SIZE; + continue; + + /* Lookbehinds are in Perl from version 5.005 */ + + case '<': + ptr += 3; + if (*ptr == '=' || *ptr == '!') + { + branch_newextra = 1 + LINK_SIZE; + length += 1 + LINK_SIZE; /* For the first branch */ + break; + } + errorcode = ERR24; + goto PCRE_ERROR_RETURN; + + /* Conditionals are in Perl from version 5.005. The bracket must either + be followed by a number (for bracket reference) or by an assertion + group. PCRE extends this by allowing a name to reference a named group; + unfortunately, previously 'R' was implemented for a recursion test. + When this is compiled, we look for the named group 'R' first. At this + point we just do a basic syntax check. */ + + case '(': + if ((cd->ctypes[ptr[3]] & ctype_word) != 0) + { + ptr += 4; + length += 3; + while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++; + if (*ptr != ')') + { + errorcode = ERR26; + goto PCRE_ERROR_RETURN; + } + } + else /* An assertion must follow */ + { + ptr++; /* Can treat like ':' as far as spacing is concerned */ + if (ptr[2] != '?' || + (ptr[3] != '=' && ptr[3] != '!' && ptr[3] != '<') ) + { + ptr += 2; /* To get right offset in message */ + errorcode = ERR28; + goto PCRE_ERROR_RETURN; + } + } + break; + + /* Else loop checking valid options until ) is met. Anything else is an + error. If we are without any brackets, i.e. at top level, the settings + act as if specified in the options, so massage the options immediately. + This is for backward compatibility with Perl 5.004. */ + + default: + set = unset = 0; + optset = &set; + ptr += 2; + + for (;; ptr++) + { + c = *ptr; + switch (c) + { + case 'i': + *optset |= PCRE_CASELESS; + continue; + + case 'J': + *optset |= PCRE_DUPNAMES; + options |= PCRE_JCHANGED; /* Record that it changed */ + continue; + + case 'm': + *optset |= PCRE_MULTILINE; + continue; + + case 's': + *optset |= PCRE_DOTALL; + continue; + + case 'x': + *optset |= PCRE_EXTENDED; + continue; + + case 'X': + *optset |= PCRE_EXTRA; + continue; + + case 'U': + *optset |= PCRE_UNGREEDY; + continue; + + case '-': + optset = &unset; + continue; + + /* A termination by ')' indicates an options-setting-only item; if + this is at the very start of the pattern (indicated by item_count + being zero), we use it to set the global options. This is helpful + when analyzing the pattern for first characters, etc. Otherwise + nothing is done here and it is handled during the compiling + process. + + We allow for more than one options setting at the start. If such + settings do not change the existing options, nothing is compiled. + However, we must leave space just in case something is compiled. + This can happen for pathological sequences such as (?i)(?-i) + because the global options will end up with -i set. The space is + small and not significant. (Before I did this there was a reported + bug with (?i)(?-i) in a machine-generated pattern.) + + [Historical note: Up to Perl 5.8, options settings at top level + were always global settings, wherever they appeared in the pattern. + That is, they were equivalent to an external setting. From 5.8 + onwards, they apply only to what follows (which is what you might + expect).] */ + + case ')': + if (item_count == 0) + { + options = (options | set) & (~unset); + set = unset = 0; /* To save length */ + item_count--; /* To allow for several */ + length += 2; + } + + /* Fall through */ + + /* A termination by ':' indicates the start of a nested group with + the given options set. This is again handled at compile time, but + we must allow for compiled space if any of the ims options are + set. We also have to allow for resetting space at the end of + the group, which is why 4 is added to the length and not just 2. + If there are several changes of options within the same group, this + will lead to an over-estimate on the length, but this shouldn't + matter very much. We also have to allow for resetting options at + the start of any alternations, which we do by setting + branch_newextra to 2. */ + + case ':': + if (((set|unset) & PCRE_IMS) != 0) + { + length += 4; + branch_newextra = 2; + } + goto END_OPTIONS; + + /* Unrecognized option character */ + + default: + errorcode = ERR12; + goto PCRE_ERROR_RETURN; + } + } + + /* If we hit a closing bracket, that's it - this is a freestanding + option-setting. We need to ensure that branch_extra is updated if + necessary. The only values branch_newextra can have here are 0 or 2. + If the value is 2, then branch_extra must either be 2 or 5, depending + on whether this is a lookbehind group or not. */ + + END_OPTIONS: + if (c == ')') + { + if (branch_newextra == 2 && + (branch_extra == 0 || branch_extra == 1+LINK_SIZE)) + branch_extra += branch_newextra; + continue; + } + + /* If options were terminated by ':' control comes here. This is a + non-capturing group with an options change. There is nothing more that + needs to be done because "capturing" is already set FALSE by default; + we can just fall through. */ + + } + } + + /* Ordinary parentheses, not followed by '?', are capturing unless + PCRE_NO_AUTO_CAPTURE is set. */ + + else capturing = (options & PCRE_NO_AUTO_CAPTURE) == 0; + + /* Capturing brackets must be counted so we can process escapes in a + Perlish way. If the number exceeds EXTRACT_BASIC_MAX we are going to need + an additional 3 bytes of memory per capturing bracket. */ + + if (capturing) + { + bracount++; + if (bracount > EXTRACT_BASIC_MAX) bracket_length += 3; + } + + /* Save length for computing whole length at end if there's a repeat that + requires duplication of the group. Also save the current value of + branch_extra, and start the new group with the new value. If non-zero, this + will either be 2 for a (?imsx: group, or 3 for a lookbehind assertion. */ + + if (brastackptr >= sizeof(brastack)/sizeof(int)) + { + errorcode = ERR19; + goto PCRE_ERROR_RETURN; + } + + bralenstack[brastackptr] = branch_extra; + branch_extra = branch_newextra; + + brastack[brastackptr++] = length; + length += bracket_length; + continue; + + /* Handle ket. Look for subsequent max/min; for certain sets of values we + have to replicate this bracket up to that many times. If brastackptr is + 0 this is an unmatched bracket which will generate an error, but take care + not to try to access brastack[-1] when computing the length and restoring + the branch_extra value. */ + + case ')': + length += 1 + LINK_SIZE; + if (brastackptr > 0) + { + duplength = length - brastack[--brastackptr]; + branch_extra = bralenstack[brastackptr]; + /* This is a paranoid check to stop integer overflow later on */ + if (duplength > MAX_DUPLENGTH) + { + errorcode = ERR50; + goto PCRE_ERROR_RETURN; + } + } + else duplength = 0; + + /* The following code is also used when a recursion such as (?3) is + followed by a quantifier, because in that case, it has to be wrapped inside + brackets so that the quantifier works. The value of duplength must be + set before arrival. */ + + HANDLE_QUANTIFIED_BRACKETS: + + /* Leave ptr at the final char; for read_repeat_counts this happens + automatically; for the others we need an increment. */ + + if ((c = ptr[1]) == '{' && is_counted_repeat(ptr+2)) + { + ptr = read_repeat_counts(ptr+2, &min, &max, &errorcode); + if (errorcode != 0) goto PCRE_ERROR_RETURN; + } + else if (c == '*') { min = 0; max = -1; ptr++; } + else if (c == '+') { min = 1; max = -1; ptr++; } + else if (c == '?') { min = 0; max = 1; ptr++; } + else { min = 1; max = 1; } + + /* If the minimum is zero, we have to allow for an OP_BRAZERO before the + group, and if the maximum is greater than zero, we have to replicate + maxval-1 times; each replication acquires an OP_BRAZERO plus a nesting + bracket set. */ + + if (min == 0) + { + length++; + if (max > 0) length += (max - 1) * (duplength + 3 + 2*LINK_SIZE); + } + + /* When the minimum is greater than zero, we have to replicate up to + minval-1 times, with no additions required in the copies. Then, if there + is a limited maximum we have to replicate up to maxval-1 times allowing + for a BRAZERO item before each optional copy and nesting brackets for all + but one of the optional copies. */ + + else + { + length += (min - 1) * duplength; + if (max > min) /* Need this test as max=-1 means no limit */ + length += (max - min) * (duplength + 3 + 2*LINK_SIZE) + - (2 + 2*LINK_SIZE); + } + + /* Allow space for once brackets for "possessive quantifier" */ + + if (ptr[1] == '+') + { + ptr++; + length += 2 + 2*LINK_SIZE; + } + continue; + + /* Non-special character. It won't be space or # in extended mode, so it is + always a genuine character. If we are in a \Q...\E sequence, check for the + end; if not, we have a literal. */ + + default: + NORMAL_CHAR: + + if (inescq && c == '\\' && ptr[1] == 'E') + { + inescq = FALSE; + ptr++; + continue; + } + + length += 2; /* For a one-byte character */ + lastitemlength = 1; /* Default length of last item for repeats */ + + /* In UTF-8 mode, check for additional bytes. */ + +#ifdef SUPPORT_UTF8 + if (utf8 && (c & 0xc0) == 0xc0) + { + while ((ptr[1] & 0xc0) == 0x80) /* Can't flow over the end */ + { /* because the end is marked */ + lastitemlength++; /* by a zero byte. */ + length++; + ptr++; + } + } +#endif + + continue; + } + } + +length += 2 + LINK_SIZE; /* For final KET and END */ + +if ((options & PCRE_AUTO_CALLOUT) != 0) + length += 2 + 2*LINK_SIZE; /* For final callout */ + +if (length > MAX_PATTERN_SIZE) + { + errorcode = ERR20; + goto PCRE_EARLY_ERROR_RETURN; + } + +/* Compute the size of data block needed and get it, either from malloc or +externally provided function. Integer overflow should no longer be possible +because nowadays we limit the maximum value of name_count and max_name size. */ + +size = length + sizeof(real_pcre) + name_count * (max_name_size + 3); +re = (real_pcre *)(pcre_malloc)(size); + +if (re == NULL) + { + errorcode = ERR21; + goto PCRE_EARLY_ERROR_RETURN; + } + +/* Put in the magic number, and save the sizes, options, and character table +pointer. NULL is used for the default character tables. The nullpad field is at +the end; it's there to help in the case when a regex compiled on a system with +4-byte pointers is run on another with 8-byte pointers. */ + +re->magic_number = MAGIC_NUMBER; +re->size = size; +re->options = options; +re->dummy1 = 0; +re->name_table_offset = sizeof(real_pcre); +re->name_entry_size = max_name_size + 3; +re->name_count = name_count; +re->ref_count = 0; +re->tables = (tables == _pcre_default_tables)? NULL : tables; +re->nullpad = NULL; + +/* The starting points of the name/number translation table and of the code are +passed around in the compile data block. */ + +cd->names_found = 0; +cd->name_entry_size = max_name_size + 3; +cd->name_table = (uschar *)re + re->name_table_offset; +codestart = cd->name_table + re->name_entry_size * re->name_count; +cd->start_code = codestart; +cd->start_pattern = (const uschar *)pattern; +cd->req_varyopt = 0; +cd->nopartial = FALSE; + +/* Set up a starting, non-extracting bracket, then compile the expression. On +error, errorcode will be set non-zero, so we don't need to look at the result +of the function here. */ + +ptr = (const uschar *)pattern; +code = (uschar *)codestart; +*code = OP_BRA; +bracount = 0; +(void)compile_regex(options, options & PCRE_IMS, &bracount, &code, &ptr, + &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd); +re->top_bracket = bracount; +re->top_backref = cd->top_backref; + +if (cd->nopartial) re->options |= PCRE_NOPARTIAL; + +/* If not reached end of pattern on success, there's an excess bracket. */ + +if (errorcode == 0 && *ptr != 0) errorcode = ERR22; + +/* Fill in the terminating state and check for disastrous overflow, but +if debugging, leave the test till after things are printed out. */ + +*code++ = OP_END; + +#ifndef DEBUG +if (code - codestart > length) errorcode = ERR23; +#endif + +/* Give an error if there's back reference to a non-existent capturing +subpattern. */ + +if (re->top_backref > re->top_bracket) errorcode = ERR15; + +/* Failed to compile, or error while post-processing */ + +if (errorcode != 0) + { + (pcre_free)(re); + PCRE_ERROR_RETURN: + *erroroffset = ptr - (const uschar *)pattern; + PCRE_EARLY_ERROR_RETURN: + *errorptr = error_texts[errorcode]; + if (errorcodeptr != NULL) *errorcodeptr = errorcode; + return NULL; + } + +/* If the anchored option was not passed, set the flag if we can determine that +the pattern is anchored by virtue of ^ characters or \A or anything else (such +as starting with .* when DOTALL is set). + +Otherwise, if we know what the first character has to be, save it, because that +speeds up unanchored matches no end. If not, see if we can set the +PCRE_STARTLINE flag. This is helpful for multiline matches when all branches +start with ^. and also when all branches start with .* for non-DOTALL matches. +*/ + +if ((options & PCRE_ANCHORED) == 0) + { + int temp_options = options; + if (is_anchored(codestart, &temp_options, 0, cd->backref_map)) + re->options |= PCRE_ANCHORED; + else + { + if (firstbyte < 0) + firstbyte = find_firstassertedchar(codestart, &temp_options, FALSE); + if (firstbyte >= 0) /* Remove caseless flag for non-caseable chars */ + { + int ch = firstbyte & 255; + re->first_byte = ((firstbyte & REQ_CASELESS) != 0 && + cd->fcc[ch] == ch)? ch : firstbyte; + re->options |= PCRE_FIRSTSET; + } + else if (is_startline(codestart, 0, cd->backref_map)) + re->options |= PCRE_STARTLINE; + } + } + +/* For an anchored pattern, we use the "required byte" only if it follows a +variable length item in the regex. Remove the caseless flag for non-caseable +bytes. */ + +if (reqbyte >= 0 && + ((re->options & PCRE_ANCHORED) == 0 || (reqbyte & REQ_VARY) != 0)) + { + int ch = reqbyte & 255; + re->req_byte = ((reqbyte & REQ_CASELESS) != 0 && + cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte; + re->options |= PCRE_REQCHSET; + } + +/* Print out the compiled data if debugging is enabled. This is never the +case when building a production library. */ + +#ifdef DEBUG + +printf("Length = %d top_bracket = %d top_backref = %d\n", + length, re->top_bracket, re->top_backref); + +if (re->options != 0) + { + printf("%s%s%s%s%s%s%s%s%s\n", + ((re->options & PCRE_NOPARTIAL) != 0)? "nopartial " : "", + ((re->options & PCRE_ANCHORED) != 0)? "anchored " : "", + ((re->options & PCRE_CASELESS) != 0)? "caseless " : "", + ((re->options & PCRE_EXTENDED) != 0)? "extended " : "", + ((re->options & PCRE_MULTILINE) != 0)? "multiline " : "", + ((re->options & PCRE_DOTALL) != 0)? "dotall " : "", + ((re->options & PCRE_DOLLAR_ENDONLY) != 0)? "endonly " : "", + ((re->options & PCRE_EXTRA) != 0)? "extra " : "", + ((re->options & PCRE_UNGREEDY) != 0)? "ungreedy " : ""); + } + +if ((re->options & PCRE_FIRSTSET) != 0) + { + int ch = re->first_byte & 255; + const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)? + "" : " (caseless)"; + if (isprint(ch)) printf("First char = %c%s\n", ch, caseless); + else printf("First char = \\x%02x%s\n", ch, caseless); + } + +if ((re->options & PCRE_REQCHSET) != 0) + { + int ch = re->req_byte & 255; + const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)? + "" : " (caseless)"; + if (isprint(ch)) printf("Req char = %c%s\n", ch, caseless); + else printf("Req char = \\x%02x%s\n", ch, caseless); + } + +pcre_printint(re, stdout); + +/* This check is done here in the debugging case so that the code that +was compiled can be seen. */ + +if (code - codestart > length) + { + (pcre_free)(re); + *errorptr = error_texts[ERR23]; + *erroroffset = ptr - (uschar *)pattern; + if (errorcodeptr != NULL) *errorcodeptr = ERR23; + return NULL; + } +#endif + +return (pcre *)re; +} + +/* End of pcre_compile.c */ ============================================================ --- pcre/pcre_config.c 3197ae2ed4c54cc20c770044d842038ba7e9f2b5 +++ pcre/pcre_config.c 3197ae2ed4c54cc20c770044d842038ba7e9f2b5 @@ -0,0 +1,116 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + +/* PCRE is a library of functions to support regular expressions whose syntax +and semantics are as close as possible to those of the Perl 5 language. + + Written by Philip Hazel + Copyright (c) 1997-2006 University of Cambridge + +----------------------------------------------------------------------------- +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + + * Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimer. + + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + * Neither the name of the University of Cambridge nor the names of its + contributors may be used to endorse or promote products derived from + this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE +LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR +CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +----------------------------------------------------------------------------- +*/ + + +/* This module contains the external function pcre_config(). */ + + +#include "pcre_internal.h" + + +/************************************************* +* Return info about what features are configured * +*************************************************/ + +/* This function has an extensible interface so that additional items can be +added compatibly. + +Arguments: + what what information is required + where where to put the information + +Returns: 0 if data returned, negative on error +*/ + +PCRE_DATA_SCOPE int +pcre_config(int what, void *where) +{ +switch (what) + { + case PCRE_CONFIG_UTF8: +#ifdef SUPPORT_UTF8 + *((int *)where) = 1; +#else + *((int *)where) = 0; +#endif + break; + + case PCRE_CONFIG_UNICODE_PROPERTIES: +#ifdef SUPPORT_UCP + *((int *)where) = 1; +#else + *((int *)where) = 0; +#endif + break; + + case PCRE_CONFIG_NEWLINE: + *((int *)where) = NEWLINE; + break; + + case PCRE_CONFIG_LINK_SIZE: + *((int *)where) = LINK_SIZE; + break; + + case PCRE_CONFIG_POSIX_MALLOC_THRESHOLD: + *((int *)where) = POSIX_MALLOC_THRESHOLD; + break; + + case PCRE_CONFIG_MATCH_LIMIT: + *((unsigned int *)where) = MATCH_LIMIT; + break; + + case PCRE_CONFIG_MATCH_LIMIT_RECURSION: + *((unsigned int *)where) = MATCH_LIMIT_RECURSION; + break; + + case PCRE_CONFIG_STACKRECURSE: +#ifdef NO_RECURSE + *((int *)where) = 0; +#else + *((int *)where) = 1; +#endif + break; + + default: return PCRE_ERROR_BADOPTION; + } + +return 0; +} + +/* End of pcre_config.c */ ============================================================ --- pcre/pcre_config.h cdd4d95dc55d8f805da3888d61eb630148b39eaf +++ pcre/pcre_config.h cdd4d95dc55d8f805da3888d61eb630148b39eaf @@ -0,0 +1,152 @@ +/* Formerly libpcre config.h.in. We do not need to worry about most of the + portability issues that upstream libpcre's build system concerns itself + with, so we make no attempt to wire this up to monotone's configure. */ + +/* Make sure these options are disabled, as we did not import all of their + support code. */ +#undef SUPPORT_UTF8 +#undef SUPPORT_UCP +#undef DEBUG + +/* On Unix-like systems config.in is converted by "configure" into config.h. +Some other environments also support the use of "configure". PCRE is written in +Standard C, but there are a few non-standard things it can cope with, allowing +it to run on SunOS4 and other "close to standard" systems. + +On a non-Unix-like system you should just copy this file into config.h, and set +up the macros the way you need them. You should normally change the definitions +of HAVE_STRERROR and HAVE_MEMMOVE to 1. Unfortunately, because of the way +autoconf works, these cannot be made the defaults. If your system has bcopy() +and not memmove(), change the definition of HAVE_BCOPY instead of HAVE_MEMMOVE. +If your system has neither bcopy() nor memmove(), leave them both as 0; an +emulation function will be used. */ + +/* If you are compiling for a system that uses EBCDIC instead of ASCII +character codes, define this macro as 1. On systems that can use "configure", +this can be done via --enable-ebcdic. */ + +#ifndef EBCDIC +#define EBCDIC 0 +#endif + +/* If you are compiling for a system other than a Unix-like system or Win32, +and it needs some magic to be inserted before the definition of a function that +is exported by the library, define this macro to contain the relevant magic. If +you do not define this macro, it defaults to "extern" for a C compiler and +"extern C" for a C++ compiler on non-Win32 systems. This macro apears at the +start of every exported function that is part of the external API. It does not +appear on functions that are "external" in the C sense, but which are internal +to the library. */ + +/* #define PCRE_DATA_SCOPE */ + +/* Define the following macro to empty if the "const" keyword does not work. */ + +#undef const + +/* Define the following macro to "unsigned" if does not define +size_t. */ + +#undef size_t + +/* The following two definitions are mainly for the benefit of SunOS4, which +does not have the strerror() or memmove() functions that should be present in +all Standard C libraries. The macros HAVE_STRERROR and HAVE_MEMMOVE should +normally be defined with the value 1 for other systems, but unfortunately we +cannot make this the default because "configure" files generated by autoconf +will only change 0 to 1; they won't change 1 to 0 if the functions are not +found. */ + +#define HAVE_STRERROR 1 +#define HAVE_MEMMOVE 1 + +/* There are some non-Unix-like systems that don't even have bcopy(). If this +macro is false, an emulation is used. If HAVE_MEMMOVE is set to 1, the value of +HAVE_BCOPY is not relevant. */ + +#define HAVE_BCOPY 0 + +/* The value of NEWLINE determines the newline character. The default is to +leave it up to the compiler, but some sites want to force a particular value. +On Unix-like systems, "configure" can be used to override this default. */ + +#ifndef NEWLINE +#define NEWLINE '\n' +#endif + +/* The value of LINK_SIZE determines the number of bytes used to store links as +offsets within the compiled regex. The default is 2, which allows for compiled +patterns up to 64K long. This covers the vast majority of cases. However, PCRE +can also be compiled to use 3 or 4 bytes instead. This allows for longer +patterns in extreme cases. On systems that support it, "configure" can be used +to override this default. */ + +#ifndef LINK_SIZE +#define LINK_SIZE 2 +#endif + +/* When calling PCRE via the POSIX interface, additional working storage is +required for holding the pointers to capturing substrings because PCRE requires +three integers per substring, whereas the POSIX interface provides only two. If +the number of expected substrings is small, the wrapper function uses space on +the stack, because this is faster than using malloc() for each call. The +threshold above which the stack is no longer used is defined by POSIX_MALLOC_ +THRESHOLD. On systems that support it, "configure" can be used to override this +default. */ + +#ifndef POSIX_MALLOC_THRESHOLD +#define POSIX_MALLOC_THRESHOLD 10 +#endif + +/* PCRE uses recursive function calls to handle backtracking while matching. +This can sometimes be a problem on systems that have stacks of limited size. +Define NO_RECURSE to get a version that doesn't use recursion in the match() +function; instead it creates its own stack by steam using pcre_recurse_malloc() +to obtain memory from the heap. For more detail, see the comments and other +stuff just above the match() function. On systems that support it, "configure" +can be used to set this in the Makefile (use --disable-stack-for-recursion). */ + +/* #define NO_RECURSE */ + +/* The value of MATCH_LIMIT determines the default number of times the internal +match() function can be called during a single execution of pcre_exec(). There +is a runtime interface for setting a different limit. The limit exists in order +to catch runaway regular expressions that take for ever to determine that they +do not match. The default is set very large so that it does not accidentally +catch legitimate cases. On systems that support it, "configure" can be used to +override this default default. */ + +#ifndef MATCH_LIMIT +#define MATCH_LIMIT 10000000 +#endif + +/* The above limit applies to all calls of match(), whether or not they +increase the recursion depth. In some environments it is desirable to limit the +depth of recursive calls of match() more strictly, in order to restrict the +maximum amount of stack (or heap, if NO_RECURSE is defined) that is used. The +value of MATCH_LIMIT_RECURSION applies only to recursive calls of match(). To +have any useful effect, it must be less than the value of MATCH_LIMIT. There is +a runtime method for setting a different limit. On systems that support it, +"configure" can be used to override this default default. */ + +#ifndef MATCH_LIMIT_RECURSION +#define MATCH_LIMIT_RECURSION MATCH_LIMIT +#endif + +/* These three limits are parameterized just in case anybody ever wants to +change them. Care must be taken if they are increased, because they guard +against integer overflow caused by enormously large patterns. */ + +#ifndef MAX_NAME_SIZE +#define MAX_NAME_SIZE 32 +#endif + +#ifndef MAX_NAME_COUNT +#define MAX_NAME_COUNT 10000 +#endif + +#ifndef MAX_DUPLENGTH +#define MAX_DUPLENGTH 30000 +#endif + +/* End */ ============================================================ --- pcre/pcre_dfa_exec.c e47bdb1351432a867879ff71b5439285458202b6 +++ pcre/pcre_dfa_exec.c e47bdb1351432a867879ff71b5439285458202b6 @@ -0,0 +1,2090 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + +/* PCRE is a library of functions to support regular expressions whose syntax +and semantics are as close as possible to those of the Perl 5 language. + + Written by Philip Hazel + Copyright (c) 1997-2006 University of Cambridge + +----------------------------------------------------------------------------- +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + + * Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimer. + + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + * Neither the name of the University of Cambridge nor the names of its + contributors may be used to endorse or promote products derived from + this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE +LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR +CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +----------------------------------------------------------------------------- +*/ + + +/* This module contains the external function pcre_dfa_exec(), which is an +alternative matching function that uses a DFA algorithm. This is NOT Perl- +compatible, but it has advantages in certain applications. */ + + +#define NLBLOCK md /* The block containing newline information */ +#include "pcre_internal.h" + + +/* For use to indent debugging output */ + +#define SP " " + + + +/************************************************* +* Code parameters and static tables * +*************************************************/ + +/* These are offsets that are used to turn the OP_TYPESTAR and friends opcodes +into others, under special conditions. A gap of 10 between the blocks should be +enough. */ + +#define OP_PROP_EXTRA (EXTRACT_BASIC_MAX+1) +#define OP_EXTUNI_EXTRA (EXTRACT_BASIC_MAX+11) + + +/* This table identifies those opcodes that are followed immediately by a +character that is to be tested in some way. This makes is possible to +centralize the loading of these characters. In the case of Type * etc, the +"character" is the opcode for \D, \d, \S, \s, \W, or \w, which will always be a +small value. */ + +static uschar coptable[] = { + 0, /* End */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* \A, \G, \B, \b, \D, \d, \S, \s, \W, \w */ + 0, 0, /* Any, Anybyte */ + 0, 0, 0, /* NOTPROP, PROP, EXTUNI */ + 0, 0, 0, 0, 0, /* \Z, \z, Opt, ^, $ */ + 1, /* Char */ + 1, /* Charnc */ + 1, /* not */ + /* Positive single-char repeats */ + 1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */ + 3, 3, 3, /* upto, minupto, exact */ + /* Negative single-char repeats - only for chars < 256 */ + 1, 1, 1, 1, 1, 1, /* NOT *, *?, +, +?, ?, ?? */ + 3, 3, 3, /* NOT upto, minupto, exact */ + /* Positive type repeats */ + 1, 1, 1, 1, 1, 1, /* Type *, *?, +, +?, ?, ?? */ + 3, 3, 3, /* Type upto, minupto, exact */ + /* Character class & ref repeats */ + 0, 0, 0, 0, 0, 0, /* *, *?, +, +?, ?, ?? */ + 0, 0, /* CRRANGE, CRMINRANGE */ + 0, /* CLASS */ + 0, /* NCLASS */ + 0, /* XCLASS - variable length */ + 0, /* REF */ + 0, /* RECURSE */ + 0, /* CALLOUT */ + 0, /* Alt */ + 0, /* Ket */ + 0, /* KetRmax */ + 0, /* KetRmin */ + 0, /* Assert */ + 0, /* Assert not */ + 0, /* Assert behind */ + 0, /* Assert behind not */ + 0, /* Reverse */ + 0, /* Once */ + 0, /* COND */ + 0, /* CREF */ + 0, 0, /* BRAZERO, BRAMINZERO */ + 0, /* BRANUMBER */ + 0 /* BRA */ +}; + +/* These 2 tables allow for compact code for testing for \D, \d, \S, \s, \W, +and \w */ + +static uschar toptable1[] = { + 0, 0, 0, 0, 0, + ctype_digit, ctype_digit, + ctype_space, ctype_space, + ctype_word, ctype_word, + 0 /* OP_ANY */ +}; + +static uschar toptable2[] = { + 0, 0, 0, 0, 0, + ctype_digit, 0, + ctype_space, 0, + ctype_word, 0, + 1 /* OP_ANY */ +}; + + +/* Structure for holding data about a particular state, which is in effect the +current data for an active path through the match tree. It must consist +entirely of ints because the working vector we are passed, and which we put +these structures in, is a vector of ints. */ + +typedef struct stateblock { + int offset; /* Offset to opcode */ + int count; /* Count for repeats */ + int ims; /* ims flag bits */ + int data; /* Some use extra data */ +} stateblock; + +#define INTS_PER_STATEBLOCK (sizeof(stateblock)/sizeof(int)) + + +#ifdef DEBUG +/************************************************* +* Print character string * +*************************************************/ + +/* Character string printing function for debugging. + +Arguments: + p points to string + length number of bytes + f where to print + +Returns: nothing +*/ + +static void +pchars(unsigned char *p, int length, FILE *f) +{ +int c; +while (length-- > 0) + { + if (isprint(c = *(p++))) + fprintf(f, "%c", c); + else + fprintf(f, "\\x%02x", c); + } +} +#endif + + + +/************************************************* +* Execute a Regular Expression - DFA engine * +*************************************************/ + +/* This internal function applies a compiled pattern to a subject string, +starting at a given point, using a DFA engine. This function is called from the +external one, possibly multiple times if the pattern is not anchored. The +function calls itself recursively for some kinds of subpattern. + +Arguments: + md the match_data block with fixed information + this_start_code the opening bracket of this subexpression's code + current_subject where we currently are in the subject string + start_offset start offset in the subject string + offsets vector to contain the matching string offsets + offsetcount size of same + workspace vector of workspace + wscount size of same + ims the current ims flags + rlevel function call recursion level + recursing regex recursive call level + +Returns: > 0 => + = 0 => + -1 => failed to match + < -1 => some kind of unexpected problem + +The following macros are used for adding states to the two state vectors (one +for the current character, one for the following character). */ + +#define ADD_ACTIVE(x,y) \ + if (active_count++ < wscount) \ + { \ + next_active_state->offset = (x); \ + next_active_state->count = (y); \ + next_active_state->ims = ims; \ + next_active_state++; \ + DPRINTF(("%.*sADD_ACTIVE(%d,%d)\n", rlevel*2-2, SP, (x), (y))); \ + } \ + else return PCRE_ERROR_DFA_WSSIZE + +#define ADD_ACTIVE_DATA(x,y,z) \ + if (active_count++ < wscount) \ + { \ + next_active_state->offset = (x); \ + next_active_state->count = (y); \ + next_active_state->ims = ims; \ + next_active_state->data = (z); \ + next_active_state++; \ + DPRINTF(("%.*sADD_ACTIVE_DATA(%d,%d,%d)\n", rlevel*2-2, SP, (x), (y), (z))); \ + } \ + else return PCRE_ERROR_DFA_WSSIZE + +#define ADD_NEW(x,y) \ + if (new_count++ < wscount) \ + { \ + next_new_state->offset = (x); \ + next_new_state->count = (y); \ + next_new_state->ims = ims; \ + next_new_state++; \ + DPRINTF(("%.*sADD_NEW(%d,%d)\n", rlevel*2-2, SP, (x), (y))); \ + } \ + else return PCRE_ERROR_DFA_WSSIZE + +#define ADD_NEW_DATA(x,y,z) \ + if (new_count++ < wscount) \ + { \ + next_new_state->offset = (x); \ + next_new_state->count = (y); \ + next_new_state->ims = ims; \ + next_new_state->data = (z); \ + next_new_state++; \ + DPRINTF(("%.*sADD_NEW_DATA(%d,%d,%d)\n", rlevel*2-2, SP, (x), (y), (z))); \ + } \ + else return PCRE_ERROR_DFA_WSSIZE + +/* And now, here is the code */ + +static int +internal_dfa_exec( + dfa_match_data *md, + const uschar *this_start_code, + const uschar *current_subject, + int start_offset, + int *offsets, + int offsetcount, + int *workspace, + int wscount, + int ims, + int rlevel, + int recursing) +{ +stateblock *active_states, *new_states, *temp_states; +stateblock *next_active_state, *next_new_state; + +const uschar *ctypes, *lcc, *fcc; +const uschar *ptr; +const uschar *end_code; + +int active_count, new_count, match_count; + +/* Some fields in the md block are frequently referenced, so we load them into +independent variables in the hope that this will perform better. */ + +const uschar *start_subject = md->start_subject; +const uschar *end_subject = md->end_subject; +const uschar *start_code = md->start_code; + +#ifdef SUPPORT_UTF8 +BOOL utf8 = (md->poptions & PCRE_UTF8) != 0; +#endif + +rlevel++; +offsetcount &= (-2); + +wscount -= 2; +wscount = (wscount - (wscount % (INTS_PER_STATEBLOCK * 2))) / + (2 * INTS_PER_STATEBLOCK); + +DPRINTF(("\n%.*s---------------------\n" + "%.*sCall to internal_dfa_exec f=%d r=%d\n", + rlevel*2-2, SP, rlevel*2-2, SP, rlevel, recursing)); + +ctypes = md->tables + ctypes_offset; +lcc = md->tables + lcc_offset; +fcc = md->tables + fcc_offset; + +match_count = PCRE_ERROR_NOMATCH; /* A negative number */ + +active_states = (stateblock *)(workspace + 2); +next_new_state = new_states = active_states + wscount; +new_count = 0; + +/* The first thing in any (sub) pattern is a bracket of some sort. Push all +the alternative states onto the list, and find out where the end is. This +makes is possible to use this function recursively, when we want to stop at a +matching internal ket rather than at the end. + +If the first opcode in the first alternative is OP_REVERSE, we are dealing with +a backward assertion. In that case, we have to find out the maximum amount to +move back, and set up each alternative appropriately. */ + +if (this_start_code[1+LINK_SIZE] == OP_REVERSE) + { + int max_back = 0; + int gone_back; + + end_code = this_start_code; + do + { + int back = GET(end_code, 2+LINK_SIZE); + if (back > max_back) max_back = back; + end_code += GET(end_code, 1); + } + while (*end_code == OP_ALT); + + /* If we can't go back the amount required for the longest lookbehind + pattern, go back as far as we can; some alternatives may still be viable. */ + +#ifdef SUPPORT_UTF8 + /* In character mode we have to step back character by character */ + + if (utf8) + { + for (gone_back = 0; gone_back < max_back; gone_back++) + { + if (current_subject <= start_subject) break; + current_subject--; + while (current_subject > start_subject && + (*current_subject & 0xc0) == 0x80) + current_subject--; + } + } + else +#endif + + /* In byte-mode we can do this quickly. */ + + { + gone_back = (current_subject - max_back < start_subject)? + current_subject - start_subject : max_back; + current_subject -= gone_back; + } + + /* Now we can process the individual branches. */ + + end_code = this_start_code; + do + { + int back = GET(end_code, 2+LINK_SIZE); + if (back <= gone_back) + { + int bstate = end_code - start_code + 2 + 2*LINK_SIZE; + ADD_NEW_DATA(-bstate, 0, gone_back - back); + } + end_code += GET(end_code, 1); + } + while (*end_code == OP_ALT); + } + +/* This is the code for a "normal" subpattern (not a backward assertion). The +start of a whole pattern is always one of these. If we are at the top level, +we may be asked to restart matching from the same point that we reached for a +previous partial match. We still have to scan through the top-level branches to +find the end state. */ + +else + { + end_code = this_start_code; + + /* Restarting */ + + if (rlevel == 1 && (md->moptions & PCRE_DFA_RESTART) != 0) + { + do { end_code += GET(end_code, 1); } while (*end_code == OP_ALT); + new_count = workspace[1]; + if (!workspace[0]) + memcpy(new_states, active_states, new_count * sizeof(stateblock)); + } + + /* Not restarting */ + + else + { + do + { + ADD_NEW(end_code - start_code + 1 + LINK_SIZE, 0); + end_code += GET(end_code, 1); + } + while (*end_code == OP_ALT); + } + } + +workspace[0] = 0; /* Bit indicating which vector is current */ + +DPRINTF(("%.*sEnd state = %d\n", rlevel*2-2, SP, end_code - start_code)); + +/* Loop for scanning the subject */ + +ptr = current_subject; +for (;;) + { + int i, j; + int clen, dlen; + unsigned int c, d; + + /* Make the new state list into the active state list and empty the + new state list. */ + + temp_states = active_states; + active_states = new_states; + new_states = temp_states; + active_count = new_count; + new_count = 0; + + workspace[0] ^= 1; /* Remember for the restarting feature */ + workspace[1] = active_count; + +#ifdef DEBUG + printf("%.*sNext character: rest of subject = \"", rlevel*2-2, SP); + pchars((uschar *)ptr, strlen((char *)ptr), stdout); + printf("\"\n"); + + printf("%.*sActive states: ", rlevel*2-2, SP); + for (i = 0; i < active_count; i++) + printf("%d/%d ", active_states[i].offset, active_states[i].count); + printf("\n"); +#endif + + /* Set the pointers for adding new states */ + + next_active_state = active_states + active_count; + next_new_state = new_states; + + /* Load the current character from the subject outside the loop, as many + different states may want to look at it, and we assume that at least one + will. */ + + if (ptr < end_subject) + { + clen = 1; +#ifdef SUPPORT_UTF8 + if (utf8) { GETCHARLEN(c, ptr, clen); } else +#endif /* SUPPORT_UTF8 */ + c = *ptr; + } + else + { + clen = 0; /* At end subject */ + c = -1; + } + + /* Scan up the active states and act on each one. The result of an action + may be to add more states to the currently active list (e.g. on hitting a + parenthesis) or it may be to put states on the new list, for considering + when we move the character pointer on. */ + + for (i = 0; i < active_count; i++) + { + stateblock *current_state = active_states + i; + const uschar *code; + int state_offset = current_state->offset; + int count, codevalue; + int chartype, script; + +#ifdef DEBUG + printf ("%.*sProcessing state %d c=", rlevel*2-2, SP, state_offset); + if (c < 0) printf("-1\n"); + else if (c > 32 && c < 127) printf("'%c'\n", c); + else printf("0x%02x\n", c); +#endif + + /* This variable is referred to implicity in the ADD_xxx macros. */ + + ims = current_state->ims; + + /* A negative offset is a special case meaning "hold off going to this + (negated) state until the number of characters in the data field have + been skipped". */ + + if (state_offset < 0) + { + if (current_state->data > 0) + { + DPRINTF(("%.*sSkipping this character\n", rlevel*2-2, SP)); + ADD_NEW_DATA(state_offset, current_state->count, + current_state->data - 1); + continue; + } + else + { + current_state->offset = state_offset = -state_offset; + } + } + + /* Check for a duplicate state with the same count, and skip if found. */ + + for (j = 0; j < i; j++) + { + if (active_states[j].offset == state_offset && + active_states[j].count == current_state->count) + { + DPRINTF(("%.*sDuplicate state: skipped\n", rlevel*2-2, SP)); + goto NEXT_ACTIVE_STATE; + } + } + + /* The state offset is the offset to the opcode */ + + code = start_code + state_offset; + codevalue = *code; + if (codevalue >= OP_BRA) codevalue = OP_BRA; /* All brackets are equal */ + + /* If this opcode is followed by an inline character, load it. It is + tempting to test for the presence of a subject character here, but that + is wrong, because sometimes zero repetitions of the subject are + permitted. + + We also use this mechanism for opcodes such as OP_TYPEPLUS that take an + argument that is not a data character - but is always one byte long. + Unfortunately, we have to take special action to deal with \P, \p, and + \X in this case. To keep the other cases fast, convert these ones to new + opcodes. */ + + if (coptable[codevalue] > 0) + { + dlen = 1; +#ifdef SUPPORT_UTF8 + if (utf8) { GETCHARLEN(d, (code + coptable[codevalue]), dlen); } else +#endif /* SUPPORT_UTF8 */ + d = code[coptable[codevalue]]; + if (codevalue >= OP_TYPESTAR) + { + if (d == OP_ANYBYTE) return PCRE_ERROR_DFA_UITEM; + if (d >= OP_NOTPROP) + codevalue += (d == OP_EXTUNI)? OP_EXTUNI_EXTRA : OP_PROP_EXTRA; + } + } + else + { + dlen = 0; /* Not strictly necessary, but compilers moan */ + d = -1; /* if these variables are not set. */ + } + + + /* Now process the individual opcodes */ + + switch (codevalue) + { + +/* ========================================================================== */ + /* Reached a closing bracket. If not at the end of the pattern, carry + on with the next opcode. Otherwise, unless we have an empty string and + PCRE_NOTEMPTY is set, save the match data, shifting up all previous + matches so we always have the longest first. */ + + case OP_KET: + case OP_KETRMIN: + case OP_KETRMAX: + if (code != end_code) + { + ADD_ACTIVE(state_offset + 1 + LINK_SIZE, 0); + if (codevalue != OP_KET) + { + ADD_ACTIVE(state_offset - GET(code, 1), 0); + } + } + else if (ptr > current_subject || (md->moptions & PCRE_NOTEMPTY) == 0) + { + if (match_count < 0) match_count = (offsetcount >= 2)? 1 : 0; + else if (match_count > 0 && ++match_count * 2 >= offsetcount) + match_count = 0; + count = ((match_count == 0)? offsetcount : match_count * 2) - 2; + if (count > 0) memmove(offsets + 2, offsets, count * sizeof(int)); + if (offsetcount >= 2) + { + offsets[0] = current_subject - start_subject; + offsets[1] = ptr - start_subject; + DPRINTF(("%.*sSet matched string = \"%.*s\"\n", rlevel*2-2, SP, + offsets[1] - offsets[0], current_subject)); + } + if ((md->moptions & PCRE_DFA_SHORTEST) != 0) + { + DPRINTF(("%.*sEnd of internal_dfa_exec %d: returning %d\n" + "%.*s---------------------\n\n", rlevel*2-2, SP, rlevel, + match_count, rlevel*2-2, SP)); + return match_count; + } + } + break; + +/* ========================================================================== */ + /* These opcodes add to the current list of states without looking + at the current character. */ + + /*-----------------------------------------------------------------*/ + case OP_ALT: + do { code += GET(code, 1); } while (*code == OP_ALT); + ADD_ACTIVE(code - start_code, 0); + break; + + /*-----------------------------------------------------------------*/ + case OP_BRA: + do + { + ADD_ACTIVE(code - start_code + 1 + LINK_SIZE, 0); + code += GET(code, 1); + } + while (*code == OP_ALT); + break; + + /*-----------------------------------------------------------------*/ + case OP_BRAZERO: + case OP_BRAMINZERO: + ADD_ACTIVE(state_offset + 1, 0); + code += 1 + GET(code, 2); + while (*code == OP_ALT) code += GET(code, 1); + ADD_ACTIVE(code - start_code + 1 + LINK_SIZE, 0); + break; + + /*-----------------------------------------------------------------*/ + case OP_BRANUMBER: + ADD_ACTIVE(state_offset + 1 + LINK_SIZE, 0); + break; + + /*-----------------------------------------------------------------*/ + case OP_CIRC: + if ((ptr == start_subject && (md->moptions & PCRE_NOTBOL) == 0) || + ((ims & PCRE_MULTILINE) != 0 && + ptr >= start_subject + md->nllen && + ptr != end_subject && + IS_NEWLINE(ptr - md->nllen))) + { ADD_ACTIVE(state_offset + 1, 0); } + break; + + /*-----------------------------------------------------------------*/ + case OP_EOD: + if (ptr >= end_subject) { ADD_ACTIVE(state_offset + 1, 0); } + break; + + /*-----------------------------------------------------------------*/ + case OP_OPT: + ims = code[1]; + ADD_ACTIVE(state_offset + 2, 0); + break; + + /*-----------------------------------------------------------------*/ + case OP_SOD: + if (ptr == start_subject) { ADD_ACTIVE(state_offset + 1, 0); } + break; + + /*-----------------------------------------------------------------*/ + case OP_SOM: + if (ptr == start_subject + start_offset) { ADD_ACTIVE(state_offset + 1, 0); } + break; + + +/* ========================================================================== */ + /* These opcodes inspect the next subject character, and sometimes + the previous one as well, but do not have an argument. The variable + clen contains the length of the current character and is zero if we are + at the end of the subject. */ + + /*-----------------------------------------------------------------*/ + case OP_ANY: + if (clen > 0 && ((ims & PCRE_DOTALL) != 0 || + ptr > end_subject - md->nllen || + !IS_NEWLINE(ptr))) + { ADD_NEW(state_offset + 1, 0); } + break; + + /*-----------------------------------------------------------------*/ + case OP_EODN: + if (clen == 0 || + (ptr == end_subject - md->nllen && IS_NEWLINE(ptr))) + { ADD_ACTIVE(state_offset + 1, 0); } + break; + + /*-----------------------------------------------------------------*/ + case OP_DOLL: + if ((md->moptions & PCRE_NOTEOL) == 0) + { + if (clen == 0 || + (ptr <= end_subject - md->nllen && IS_NEWLINE(ptr) && + ((ims & PCRE_MULTILINE) != 0 || ptr == end_subject - md->nllen) + )) + { ADD_ACTIVE(state_offset + 1, 0); } + } + else if ((ims & PCRE_MULTILINE) != 0 && + ptr <= end_subject - md->nllen && IS_NEWLINE(ptr)) + { ADD_ACTIVE(state_offset + 1, 0); } + break; + + /*-----------------------------------------------------------------*/ + + case OP_DIGIT: + case OP_WHITESPACE: + case OP_WORDCHAR: + if (clen > 0 && c < 256 && + ((ctypes[c] & toptable1[codevalue]) ^ toptable2[codevalue]) != 0) + { ADD_NEW(state_offset + 1, 0); } + break; + + /*-----------------------------------------------------------------*/ + case OP_NOT_DIGIT: + case OP_NOT_WHITESPACE: + case OP_NOT_WORDCHAR: + if (clen > 0 && (c >= 256 || + ((ctypes[c] & toptable1[codevalue]) ^ toptable2[codevalue]) != 0)) + { ADD_NEW(state_offset + 1, 0); } + break; + + /*-----------------------------------------------------------------*/ + case OP_WORD_BOUNDARY: + case OP_NOT_WORD_BOUNDARY: + { + int left_word, right_word; + + if (ptr > start_subject) + { + const uschar *temp = ptr - 1; +#ifdef SUPPORT_UTF8 + if (utf8) BACKCHAR(temp); +#endif + GETCHARTEST(d, temp); + left_word = d < 256 && (ctypes[d] & ctype_word) != 0; + } + else left_word = 0; + + if (clen > 0) right_word = c < 256 && (ctypes[c] & ctype_word) != 0; + else right_word = 0; + + if ((left_word == right_word) == (codevalue == OP_NOT_WORD_BOUNDARY)) + { ADD_ACTIVE(state_offset + 1, 0); } + } + break; + + +#ifdef SUPPORT_UCP + + /*-----------------------------------------------------------------*/ + /* Check the next character by Unicode property. We will get here only + if the support is in the binary; otherwise a compile-time error occurs. + */ + + case OP_PROP: + case OP_NOTPROP: + if (clen > 0) + { + BOOL OK; + int category = _pcre_ucp_findprop(c, &chartype, &script); + switch(code[1]) + { + case PT_ANY: + OK = TRUE; + break; + + case PT_LAMP: + OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt; + break; + + case PT_GC: + OK = category == code[2]; + break; + + case PT_PC: + OK = chartype == code[2]; + break; + + case PT_SC: + OK = script == code[2]; + break; + + /* Should never occur, but keep compilers from grumbling. */ + + default: + OK = codevalue != OP_PROP; + break; + } + + if (OK == (codevalue == OP_PROP)) { ADD_NEW(state_offset + 3, 0); } + } + break; +#endif + + + +/* ========================================================================== */ + /* These opcodes likewise inspect the subject character, but have an + argument that is not a data character. It is one of these opcodes: + OP_ANY, OP_DIGIT, OP_NOT_DIGIT, OP_WHITESPACE, OP_NOT_SPACE, OP_WORDCHAR, + OP_NOT_WORDCHAR. The value is loaded into d. */ + + case OP_TYPEPLUS: + case OP_TYPEMINPLUS: + count = current_state->count; /* Already matched */ + if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); } + if (clen > 0) + { + if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) || + (c < 256 && + (d != OP_ANY || + (ims & PCRE_DOTALL) != 0 || + ptr > end_subject - md->nllen || + !IS_NEWLINE(ptr) + ) && + ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0)) + { + count++; + ADD_NEW(state_offset, count); + } + } + break; + + /*-----------------------------------------------------------------*/ + case OP_TYPEQUERY: + case OP_TYPEMINQUERY: + ADD_ACTIVE(state_offset + 2, 0); + if (clen > 0) + { + if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) || + (c < 256 && + (d != OP_ANY || + (ims & PCRE_DOTALL) != 0 || + ptr > end_subject - md->nllen || + !IS_NEWLINE(ptr) + ) && + ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0)) + { + ADD_NEW(state_offset + 2, 0); + } + } + break; + + /*-----------------------------------------------------------------*/ + case OP_TYPESTAR: + case OP_TYPEMINSTAR: + ADD_ACTIVE(state_offset + 2, 0); + if (clen > 0) + { + if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) || + (c < 256 && + (d != OP_ANY || + (ims & PCRE_DOTALL) != 0 || + ptr > end_subject - md->nllen || + !IS_NEWLINE(ptr) + ) && + ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0)) + { + ADD_NEW(state_offset, 0); + } + } + break; + + /*-----------------------------------------------------------------*/ + case OP_TYPEEXACT: + case OP_TYPEUPTO: + case OP_TYPEMINUPTO: + if (codevalue != OP_TYPEEXACT) + { ADD_ACTIVE(state_offset + 4, 0); } + count = current_state->count; /* Number already matched */ + if (clen > 0) + { + if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) || + (c < 256 && + (d != OP_ANY || + (ims & PCRE_DOTALL) != 0 || + ptr > end_subject - md->nllen || + !IS_NEWLINE(ptr) + ) && + ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0)) + { + if (++count >= GET2(code, 1)) + { ADD_NEW(state_offset + 4, 0); } + else + { ADD_NEW(state_offset, count); } + } + } + break; + +/* ========================================================================== */ + /* These are virtual opcodes that are used when something like + OP_TYPEPLUS has OP_PROP, OP_NOTPROP, or OP_EXTUNI as its argument. It + keeps the code above fast for the other cases. The argument is in the + d variable. */ + +#ifdef SUPPORT_UCP + case OP_PROP_EXTRA + OP_TYPEPLUS: + case OP_PROP_EXTRA + OP_TYPEMINPLUS: + count = current_state->count; /* Already matched */ + if (count > 0) { ADD_ACTIVE(state_offset + 4, 0); } + if (clen > 0) + { + BOOL OK; + int category = _pcre_ucp_findprop(c, &chartype, &script); + switch(code[2]) + { + case PT_ANY: + OK = TRUE; + break; + + case PT_LAMP: + OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt; + break; + + case PT_GC: + OK = category == code[3]; + break; + + case PT_PC: + OK = chartype == code[3]; + break; + + case PT_SC: + OK = script == code[3]; + break; + + /* Should never occur, but keep compilers from grumbling. */ + + default: + OK = codevalue != OP_PROP; + break; + } + + if (OK == (d == OP_PROP)) { count++; ADD_NEW(state_offset, count); } + } + break; + + /*-----------------------------------------------------------------*/ + case OP_EXTUNI_EXTRA + OP_TYPEPLUS: + case OP_EXTUNI_EXTRA + OP_TYPEMINPLUS: + count = current_state->count; /* Already matched */ + if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); } + if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M) + { + const uschar *nptr = ptr + clen; + int ncount = 0; + while (nptr < end_subject) + { + int nd; + int ndlen = 1; + GETCHARLEN(nd, nptr, ndlen); + if (_pcre_ucp_findprop(nd, &chartype, &script) != ucp_M) break; + ncount++; + nptr += ndlen; + } + count++; + ADD_NEW_DATA(-state_offset, count, ncount); + } + break; + + /*-----------------------------------------------------------------*/ + case OP_PROP_EXTRA + OP_TYPEQUERY: + case OP_PROP_EXTRA + OP_TYPEMINQUERY: + count = 4; + goto QS1; + + case OP_PROP_EXTRA + OP_TYPESTAR: + case OP_PROP_EXTRA + OP_TYPEMINSTAR: + count = 0; + + QS1: + + ADD_ACTIVE(state_offset + 4, 0); + if (clen > 0) + { + BOOL OK; + int category = _pcre_ucp_findprop(c, &chartype, &script); + switch(code[2]) + { + case PT_ANY: + OK = TRUE; + break; + + case PT_LAMP: + OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt; + break; + + case PT_GC: + OK = category == code[3]; + break; + + case PT_PC: + OK = chartype == code[3]; + break; + + case PT_SC: + OK = script == code[3]; + break; + + /* Should never occur, but keep compilers from grumbling. */ + + default: + OK = codevalue != OP_PROP; + break; + } + + if (OK == (d == OP_PROP)) { ADD_NEW(state_offset + count, 0); } + } + break; + + /*-----------------------------------------------------------------*/ + case OP_EXTUNI_EXTRA + OP_TYPEQUERY: + case OP_EXTUNI_EXTRA + OP_TYPEMINQUERY: + count = 2; + goto QS2; + + case OP_EXTUNI_EXTRA + OP_TYPESTAR: + case OP_EXTUNI_EXTRA + OP_TYPEMINSTAR: + count = 0; + + QS2: + + ADD_ACTIVE(state_offset + 2, 0); + if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M) + { + const uschar *nptr = ptr + clen; + int ncount = 0; + while (nptr < end_subject) + { + int nd; + int ndlen = 1; + GETCHARLEN(nd, nptr, ndlen); + if (_pcre_ucp_findprop(nd, &chartype, &script) != ucp_M) break; + ncount++; + nptr += ndlen; + } + ADD_NEW_DATA(-(state_offset + count), 0, ncount); + } + break; + + /*-----------------------------------------------------------------*/ + case OP_PROP_EXTRA + OP_TYPEEXACT: + case OP_PROP_EXTRA + OP_TYPEUPTO: + case OP_PROP_EXTRA + OP_TYPEMINUPTO: + if (codevalue != OP_PROP_EXTRA + OP_TYPEEXACT) + { ADD_ACTIVE(state_offset + 6, 0); } + count = current_state->count; /* Number already matched */ + if (clen > 0) + { + BOOL OK; + int category = _pcre_ucp_findprop(c, &chartype, &script); + switch(code[4]) + { + case PT_ANY: + OK = TRUE; + break; + + case PT_LAMP: + OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt; + break; + + case PT_GC: + OK = category == code[5]; + break; + + case PT_PC: + OK = chartype == code[5]; + break; + + case PT_SC: + OK = script == code[5]; + break; + + /* Should never occur, but keep compilers from grumbling. */ + + default: + OK = codevalue != OP_PROP; + break; + } + + if (OK == (d == OP_PROP)) + { + if (++count >= GET2(code, 1)) + { ADD_NEW(state_offset + 6, 0); } + else + { ADD_NEW(state_offset, count); } + } + } + break; + + /*-----------------------------------------------------------------*/ + case OP_EXTUNI_EXTRA + OP_TYPEEXACT: + case OP_EXTUNI_EXTRA + OP_TYPEUPTO: + case OP_EXTUNI_EXTRA + OP_TYPEMINUPTO: + if (codevalue != OP_EXTUNI_EXTRA + OP_TYPEEXACT) + { ADD_ACTIVE(state_offset + 4, 0); } + count = current_state->count; /* Number already matched */ + if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M) + { + const uschar *nptr = ptr + clen; + int ncount = 0; + while (nptr < end_subject) + { + int nd; + int ndlen = 1; + GETCHARLEN(nd, nptr, ndlen); + if (_pcre_ucp_findprop(nd, &chartype, &script) != ucp_M) break; + ncount++; + nptr += ndlen; + } + if (++count >= GET2(code, 1)) + { ADD_NEW_DATA(-(state_offset + 4), 0, ncount); } + else + { ADD_NEW_DATA(-state_offset, count, ncount); } + } + break; +#endif + +/* ========================================================================== */ + /* These opcodes are followed by a character that is usually compared + to the current subject character; it is loaded into d. We still get + here even if there is no subject character, because in some cases zero + repetitions are permitted. */ + + /*-----------------------------------------------------------------*/ + case OP_CHAR: + if (clen > 0 && c == d) { ADD_NEW(state_offset + dlen + 1, 0); } + break; + + /*-----------------------------------------------------------------*/ + case OP_CHARNC: + if (clen == 0) break; + +#ifdef SUPPORT_UTF8 + if (utf8) + { + if (c == d) { ADD_NEW(state_offset + dlen + 1, 0); } else + { + int othercase; + if (c < 128) othercase = fcc[c]; else + + /* If we have Unicode property support, we can use it to test the + other case of the character. */ + +#ifdef SUPPORT_UCP + othercase = _pcre_ucp_othercase(c); +#else + othercase = -1; +#endif + + if (d == othercase) { ADD_NEW(state_offset + dlen + 1, 0); } + } + } + else +#endif /* SUPPORT_UTF8 */ + + /* Non-UTF-8 mode */ + { + if (lcc[c] == lcc[d]) { ADD_NEW(state_offset + 2, 0); } + } + break; + + +#ifdef SUPPORT_UCP + /*-----------------------------------------------------------------*/ + /* This is a tricky one because it can match more than one character. + Find out how many characters to skip, and then set up a negative state + to wait for them to pass before continuing. */ + + case OP_EXTUNI: + if (clen > 0 && _pcre_ucp_findprop(c, &chartype, &script) != ucp_M) + { + const uschar *nptr = ptr + clen; + int ncount = 0; + while (nptr < end_subject) + { + int nclen = 1; + GETCHARLEN(c, nptr, nclen); + if (_pcre_ucp_findprop(c, &chartype, &script) != ucp_M) break; + ncount++; + nptr += nclen; + } + ADD_NEW_DATA(-(state_offset + 1), 0, ncount); + } + break; +#endif + + /*-----------------------------------------------------------------*/ + /* Match a negated single character. This is only used for one-byte + characters, that is, we know that d < 256. The character we are + checking (c) can be multibyte. */ + + case OP_NOT: + if (clen > 0) + { + int otherd = ((ims & PCRE_CASELESS) != 0)? fcc[d] : d; + if (c != d && c != otherd) { ADD_NEW(state_offset + dlen + 1, 0); } + } + break; + + /*-----------------------------------------------------------------*/ + case OP_PLUS: + case OP_MINPLUS: + case OP_NOTPLUS: + case OP_NOTMINPLUS: + count = current_state->count; /* Already matched */ + if (count > 0) { ADD_ACTIVE(state_offset + dlen + 1, 0); } + if (clen > 0) + { + int otherd = -1; + if ((ims & PCRE_CASELESS) != 0) + { +#ifdef SUPPORT_UTF8 + if (utf8 && d >= 128) + { +#ifdef SUPPORT_UCP + otherd = _pcre_ucp_othercase(d); +#endif /* SUPPORT_UCP */ + } + else +#endif /* SUPPORT_UTF8 */ + otherd = fcc[d]; + } + if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR)) + { count++; ADD_NEW(state_offset, count); } + } + break; + + /*-----------------------------------------------------------------*/ + case OP_QUERY: + case OP_MINQUERY: + case OP_NOTQUERY: + case OP_NOTMINQUERY: + ADD_ACTIVE(state_offset + dlen + 1, 0); + if (clen > 0) + { + int otherd = -1; + if ((ims & PCRE_CASELESS) != 0) + { +#ifdef SUPPORT_UTF8 + if (utf8 && d >= 128) + { +#ifdef SUPPORT_UCP + otherd = _pcre_ucp_othercase(d); +#endif /* SUPPORT_UCP */ + } + else +#endif /* SUPPORT_UTF8 */ + otherd = fcc[d]; + } + if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR)) + { ADD_NEW(state_offset + dlen + 1, 0); } + } + break; + + /*-----------------------------------------------------------------*/ + case OP_STAR: + case OP_MINSTAR: + case OP_NOTSTAR: + case OP_NOTMINSTAR: + ADD_ACTIVE(state_offset + dlen + 1, 0); + if (clen > 0) + { + int otherd = -1; + if ((ims & PCRE_CASELESS) != 0) + { +#ifdef SUPPORT_UTF8 + if (utf8 && d >= 128) + { +#ifdef SUPPORT_UCP + otherd = _pcre_ucp_othercase(d); +#endif /* SUPPORT_UCP */ + } + else +#endif /* SUPPORT_UTF8 */ + otherd = fcc[d]; + } + if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR)) + { ADD_NEW(state_offset, 0); } + } + break; + + /*-----------------------------------------------------------------*/ + case OP_EXACT: + case OP_UPTO: + case OP_MINUPTO: + case OP_NOTEXACT: + case OP_NOTUPTO: + case OP_NOTMINUPTO: + if (codevalue != OP_EXACT && codevalue != OP_NOTEXACT) + { ADD_ACTIVE(state_offset + dlen + 3, 0); } + count = current_state->count; /* Number already matched */ + if (clen > 0) + { + int otherd = -1; + if ((ims & PCRE_CASELESS) != 0) + { +#ifdef SUPPORT_UTF8 + if (utf8 && d >= 128) + { +#ifdef SUPPORT_UCP + otherd = _pcre_ucp_othercase(d); +#endif /* SUPPORT_UCP */ + } + else +#endif /* SUPPORT_UTF8 */ + otherd = fcc[d]; + } + if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR)) + { + if (++count >= GET2(code, 1)) + { ADD_NEW(state_offset + dlen + 3, 0); } + else + { ADD_NEW(state_offset, count); } + } + } + break; + + +/* ========================================================================== */ + /* These are the class-handling opcodes */ + + case OP_CLASS: + case OP_NCLASS: + case OP_XCLASS: + { + BOOL isinclass = FALSE; + int next_state_offset; + const uschar *ecode; + + /* For a simple class, there is always just a 32-byte table, and we + can set isinclass from it. */ + + if (codevalue != OP_XCLASS) + { + ecode = code + 33; + if (clen > 0) + { + isinclass = (c > 255)? (codevalue == OP_NCLASS) : + ((code[1 + c/8] & (1 << (c&7))) != 0); + } + } + + /* An extended class may have a table or a list of single characters, + ranges, or both, and it may be positive or negative. There's a + function that sorts all this out. */ + + else + { + ecode = code + GET(code, 1); + if (clen > 0) isinclass = _pcre_xclass(c, code + 1 + LINK_SIZE); + } + + /* At this point, isinclass is set for all kinds of class, and ecode + points to the byte after the end of the class. If there is a + quantifier, this is where it will be. */ + + next_state_offset = ecode - start_code; + + switch (*ecode) + { + case OP_CRSTAR: + case OP_CRMINSTAR: + ADD_ACTIVE(next_state_offset + 1, 0); + if (isinclass) { ADD_NEW(state_offset, 0); } + break; + + case OP_CRPLUS: + case OP_CRMINPLUS: + count = current_state->count; /* Already matched */ + if (count > 0) { ADD_ACTIVE(next_state_offset + 1, 0); } + if (isinclass) { count++; ADD_NEW(state_offset, count); } + break; + + case OP_CRQUERY: + case OP_CRMINQUERY: + ADD_ACTIVE(next_state_offset + 1, 0); + if (isinclass) { ADD_NEW(next_state_offset + 1, 0); } + break; + + case OP_CRRANGE: + case OP_CRMINRANGE: + count = current_state->count; /* Already matched */ + if (count >= GET2(ecode, 1)) + { ADD_ACTIVE(next_state_offset + 5, 0); } + if (isinclass) + { + int max = GET2(ecode, 3); + if (++count >= max && max != 0) /* Max 0 => no limit */ + { ADD_NEW(next_state_offset + 5, 0); } + else + { ADD_NEW(state_offset, count); } + } + break; + + default: + if (isinclass) { ADD_NEW(next_state_offset, 0); } + break; + } + } + break; + +/* ========================================================================== */ + /* These are the opcodes for fancy brackets of various kinds. We have + to use recursion in order to handle them. */ + + case OP_ASSERT: + case OP_ASSERT_NOT: + case OP_ASSERTBACK: + case OP_ASSERTBACK_NOT: + { + int rc; + int local_offsets[2]; + int local_workspace[1000]; + const uschar *endasscode = code + GET(code, 1); + + while (*endasscode == OP_ALT) endasscode += GET(endasscode, 1); + + rc = internal_dfa_exec( + md, /* static match data */ + code, /* this subexpression's code */ + ptr, /* where we currently are */ + ptr - start_subject, /* start offset */ + local_offsets, /* offset vector */ + sizeof(local_offsets)/sizeof(int), /* size of same */ + local_workspace, /* workspace vector */ + sizeof(local_workspace)/sizeof(int), /* size of same */ + ims, /* the current ims flags */ + rlevel, /* function recursion level */ + recursing); /* pass on regex recursion */ + + if ((rc >= 0) == (codevalue == OP_ASSERT || codevalue == OP_ASSERTBACK)) + { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); } + } + break; + + /*-----------------------------------------------------------------*/ + case OP_COND: + { + int local_offsets[1000]; + int local_workspace[1000]; + int condcode = code[LINK_SIZE+1]; + + /* The only supported version of OP_CREF is for the value 0xffff, which + means "test if in a recursion". */ + + if (condcode == OP_CREF) + { + int value = GET2(code, LINK_SIZE+2); + if (value != 0xffff) return PCRE_ERROR_DFA_UCOND; + if (recursing > 0) { ADD_ACTIVE(state_offset + LINK_SIZE + 4, 0); } + else { ADD_ACTIVE(state_offset + GET(code, 1) + LINK_SIZE + 1, 0); } + } + + /* Otherwise, the condition is an assertion */ + + else + { + int rc; + const uschar *asscode = code + LINK_SIZE + 1; + const uschar *endasscode = asscode + GET(asscode, 1); + + while (*endasscode == OP_ALT) endasscode += GET(endasscode, 1); + + rc = internal_dfa_exec( + md, /* fixed match data */ + asscode, /* this subexpression's code */ + ptr, /* where we currently are */ + ptr - start_subject, /* start offset */ + local_offsets, /* offset vector */ + sizeof(local_offsets)/sizeof(int), /* size of same */ + local_workspace, /* workspace vector */ + sizeof(local_workspace)/sizeof(int), /* size of same */ + ims, /* the current ims flags */ + rlevel, /* function recursion level */ + recursing); /* pass on regex recursion */ + + if ((rc >= 0) == + (condcode == OP_ASSERT || condcode == OP_ASSERTBACK)) + { ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); } + else + { ADD_ACTIVE(state_offset + GET(code, 1) + LINK_SIZE + 1, 0); } + } + } + break; + + /*-----------------------------------------------------------------*/ + case OP_RECURSE: + { + int local_offsets[1000]; + int local_workspace[1000]; + int rc; + + DPRINTF(("%.*sStarting regex recursion %d\n", rlevel*2-2, SP, + recursing + 1)); + + rc = internal_dfa_exec( + md, /* fixed match data */ + start_code + GET(code, 1), /* this subexpression's code */ + ptr, /* where we currently are */ + ptr - start_subject, /* start offset */ + local_offsets, /* offset vector */ + sizeof(local_offsets)/sizeof(int), /* size of same */ + local_workspace, /* workspace vector */ + sizeof(local_workspace)/sizeof(int), /* size of same */ + ims, /* the current ims flags */ + rlevel, /* function recursion level */ + recursing + 1); /* regex recurse level */ + + DPRINTF(("%.*sReturn from regex recursion %d: rc=%d\n", rlevel*2-2, SP, + recursing + 1, rc)); + + /* Ran out of internal offsets */ + + if (rc == 0) return PCRE_ERROR_DFA_RECURSE; + + /* For each successful matched substring, set up the next state with a + count of characters to skip before trying it. Note that the count is in + characters, not bytes. */ + + if (rc > 0) + { + for (rc = rc*2 - 2; rc >= 0; rc -= 2) + { + const uschar *p = start_subject + local_offsets[rc]; + const uschar *pp = start_subject + local_offsets[rc+1]; + int charcount = local_offsets[rc+1] - local_offsets[rc]; + while (p < pp) if ((*p++ & 0xc0) == 0x80) charcount--; + if (charcount > 0) + { + ADD_NEW_DATA(-(state_offset + LINK_SIZE + 1), 0, (charcount - 1)); + } + else + { + ADD_ACTIVE(state_offset + LINK_SIZE + 1, 0); + } + } + } + else if (rc != PCRE_ERROR_NOMATCH) return rc; + } + break; + + /*-----------------------------------------------------------------*/ + case OP_ONCE: + { + int local_offsets[2]; + int local_workspace[1000]; + + int rc = internal_dfa_exec( + md, /* fixed match data */ + code, /* this subexpression's code */ + ptr, /* where we currently are */ + ptr - start_subject, /* start offset */ + local_offsets, /* offset vector */ + sizeof(local_offsets)/sizeof(int), /* size of same */ + local_workspace, /* workspace vector */ + sizeof(local_workspace)/sizeof(int), /* size of same */ + ims, /* the current ims flags */ + rlevel, /* function recursion level */ + recursing); /* pass on regex recursion */ + + if (rc >= 0) + { + const uschar *end_subpattern = code; + int charcount = local_offsets[1] - local_offsets[0]; + int next_state_offset, repeat_state_offset; + + do { end_subpattern += GET(end_subpattern, 1); } + while (*end_subpattern == OP_ALT); + next_state_offset = end_subpattern - start_code + LINK_SIZE + 1; + + /* If the end of this subpattern is KETRMAX or KETRMIN, we must + arrange for the repeat state also to be added to the relevant list. + Calculate the offset, or set -1 for no repeat. */ + + repeat_state_offset = (*end_subpattern == OP_KETRMAX || + *end_subpattern == OP_KETRMIN)? + end_subpattern - start_code - GET(end_subpattern, 1) : -1; + + /* If we have matched an empty string, add the next state at the + current character pointer. This is important so that the duplicate + checking kicks in, which is what breaks infinite loops that match an + empty string. */ + + if (charcount == 0) + { + ADD_ACTIVE(next_state_offset, 0); + } + + /* Optimization: if there are no more active states, and there + are no new states yet set up, then skip over the subject string + right here, to save looping. Otherwise, set up the new state to swing + into action when the end of the substring is reached. */ + + else if (i + 1 >= active_count && new_count == 0) + { + ptr += charcount; + clen = 0; + ADD_NEW(next_state_offset, 0); + + /* If we are adding a repeat state at the new character position, + we must fudge things so that it is the only current state. + Otherwise, it might be a duplicate of one we processed before, and + that would cause it to be skipped. */ + + if (repeat_state_offset >= 0) + { + next_active_state = active_states; + active_count = 0; + i = -1; + ADD_ACTIVE(repeat_state_offset, 0); + } + } + else + { + const uschar *p = start_subject + local_offsets[0]; + const uschar *pp = start_subject + local_offsets[1]; + while (p < pp) if ((*p++ & 0xc0) == 0x80) charcount--; + ADD_NEW_DATA(-next_state_offset, 0, (charcount - 1)); + if (repeat_state_offset >= 0) + { ADD_NEW_DATA(-repeat_state_offset, 0, (charcount - 1)); } + } + + } + else if (rc != PCRE_ERROR_NOMATCH) return rc; + } + break; + + +/* ========================================================================== */ + /* Handle callouts */ + + case OP_CALLOUT: + if (pcre_callout != NULL) + { + int rrc; + pcre_callout_block cb; + cb.version = 1; /* Version 1 of the callout block */ + cb.callout_number = code[1]; + cb.offset_vector = offsets; + cb.subject = (PCRE_SPTR)start_subject; + cb.subject_length = end_subject - start_subject; + cb.start_match = current_subject - start_subject; + cb.current_position = ptr - start_subject; + cb.pattern_position = GET(code, 2); + cb.next_item_length = GET(code, 2 + LINK_SIZE); + cb.capture_top = 1; + cb.capture_last = -1; + cb.callout_data = md->callout_data; + if ((rrc = (*pcre_callout)(&cb)) < 0) return rrc; /* Abandon */ + if (rrc == 0) { ADD_ACTIVE(state_offset + 2 + 2*LINK_SIZE, 0); } + } + break; + + +/* ========================================================================== */ + default: /* Unsupported opcode */ + return PCRE_ERROR_DFA_UITEM; + } + + NEXT_ACTIVE_STATE: continue; + + } /* End of loop scanning active states */ + + /* We have finished the processing at the current subject character. If no + new states have been set for the next character, we have found all the + matches that we are going to find. If we are at the top level and partial + matching has been requested, check for appropriate conditions. */ + + if (new_count <= 0) + { + if (match_count < 0 && /* No matches found */ + rlevel == 1 && /* Top level match function */ + (md->moptions & PCRE_PARTIAL) != 0 && /* Want partial matching */ + ptr >= end_subject && /* Reached end of subject */ + ptr > current_subject) /* Matched non-empty string */ + { + if (offsetcount >= 2) + { + offsets[0] = current_subject - start_subject; + offsets[1] = end_subject - start_subject; + } + match_count = PCRE_ERROR_PARTIAL; + } + + DPRINTF(("%.*sEnd of internal_dfa_exec %d: returning %d\n" + "%.*s---------------------\n\n", rlevel*2-2, SP, rlevel, match_count, + rlevel*2-2, SP)); + break; /* In effect, "return", but see the comment below */ + } + + /* One or more states are active for the next character. */ + + ptr += clen; /* Advance to next subject character */ + } /* Loop to move along the subject string */ + +/* Control gets here from "break" a few lines above. We do it this way because +if we use "return" above, we have compiler trouble. Some compilers warn if +there's nothing here because they think the function doesn't return a value. On +the other hand, if we put a dummy statement here, some more clever compilers +complain that it can't be reached. Sigh. */ + +return match_count; +} + + + + +/************************************************* +* Execute a Regular Expression - DFA engine * +*************************************************/ + +/* This external function applies a compiled re to a subject string using a DFA +engine. This function calls the internal function multiple times if the pattern +is not anchored. + +Arguments: + argument_re points to the compiled expression + extra_data points to extra data or is NULL (not currently used) + subject points to the subject string + length length of subject string (may contain binary zeros) + start_offset where to start in the subject string + options option bits + offsets vector of match offsets + offsetcount size of same + workspace workspace vector + wscount size of same + +Returns: > 0 => number of match offset pairs placed in offsets + = 0 => offsets overflowed; longest matches are present + -1 => failed to match + < -1 => some kind of unexpected problem +*/ + +PCRE_DATA_SCOPE int +pcre_dfa_exec(const pcre *argument_re, const pcre_extra *extra_data, + const char *subject, int length, int start_offset, int options, int *offsets, + int offsetcount, int *workspace, int wscount) +{ +real_pcre *re = (real_pcre *)argument_re; +dfa_match_data match_block; +dfa_match_data *md = &match_block; +BOOL utf8, anchored, startline, firstline; +const uschar *current_subject, *end_subject, *lcc; + +pcre_study_data internal_study; +const pcre_study_data *study = NULL; +real_pcre internal_re; + +const uschar *req_byte_ptr; +const uschar *start_bits = NULL; +BOOL first_byte_caseless = FALSE; +BOOL req_byte_caseless = FALSE; +int first_byte = -1; +int req_byte = -1; +int req_byte2 = -1; +int newline; + +/* Plausibility checks */ + +if ((options & ~PUBLIC_DFA_EXEC_OPTIONS) != 0) return PCRE_ERROR_BADOPTION; +if (re == NULL || subject == NULL || workspace == NULL || + (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL; +if (offsetcount < 0) return PCRE_ERROR_BADCOUNT; +if (wscount < 20) return PCRE_ERROR_DFA_WSSIZE; + +/* We need to find the pointer to any study data before we test for byte +flipping, so we scan the extra_data block first. This may set two fields in the +match block, so we must initialize them beforehand. However, the other fields +in the match block must not be set until after the byte flipping. */ + +md->tables = re->tables; +md->callout_data = NULL; + +if (extra_data != NULL) + { + unsigned int flags = extra_data->flags; + if ((flags & PCRE_EXTRA_STUDY_DATA) != 0) + study = (const pcre_study_data *)extra_data->study_data; + if ((flags & PCRE_EXTRA_MATCH_LIMIT) != 0) return PCRE_ERROR_DFA_UMLIMIT; + if ((flags & PCRE_EXTRA_MATCH_LIMIT_RECURSION) != 0) + return PCRE_ERROR_DFA_UMLIMIT; + if ((flags & PCRE_EXTRA_CALLOUT_DATA) != 0) + md->callout_data = extra_data->callout_data; + if ((flags & PCRE_EXTRA_TABLES) != 0) + md->tables = extra_data->tables; + } + +/* Check that the first field in the block is the magic number. If it is not, +test for a regex that was compiled on a host of opposite endianness. If this is +the case, flipped values are put in internal_re and internal_study if there was +study data too. */ + +if (re->magic_number != MAGIC_NUMBER) + { + re = _pcre_try_flipped(re, &internal_re, study, &internal_study); + if (re == NULL) return PCRE_ERROR_BADMAGIC; + if (study != NULL) study = &internal_study; + } + +/* Set some local values */ + +current_subject = (const unsigned char *)subject + start_offset; +end_subject = (const unsigned char *)subject + length; +req_byte_ptr = current_subject - 1; + +#ifdef SUPPORT_UTF8 +utf8 = (re->options & PCRE_UTF8) != 0; +#else +utf8 = FALSE; +#endif + +anchored = (options & (PCRE_ANCHORED|PCRE_DFA_RESTART)) != 0 || + (re->options & PCRE_ANCHORED) != 0; + +/* The remaining fixed data for passing around. */ + +md->start_code = (const uschar *)argument_re + + re->name_table_offset + re->name_count * re->name_entry_size; +md->start_subject = (const unsigned char *)subject; +md->end_subject = end_subject; +md->moptions = options; +md->poptions = re->options; + +/* Handle different types of newline. The two bits give four cases. If nothing +is set at run time, whatever was used at compile time applies. */ + +switch ((((options & PCRE_NEWLINE_CRLF) == 0)? re->options : options) & + PCRE_NEWLINE_CRLF) + { + default: newline = NEWLINE; break; /* Compile-time default */ + case PCRE_NEWLINE_CR: newline = '\r'; break; + case PCRE_NEWLINE_LF: newline = '\n'; break; + case PCRE_NEWLINE_CR+ + PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break; + } + +if (newline > 255) + { + md->nllen = 2; + md->nl[0] = (newline >> 8) & 255; + md->nl[1] = newline & 255; + } +else + { + md->nllen = 1; + md->nl[0] = newline; + } + +/* Check a UTF-8 string if required. Unfortunately there's no way of passing +back the character offset. */ + +#ifdef SUPPORT_UTF8 +if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0) + { + if (_pcre_valid_utf8((uschar *)subject, length) >= 0) + return PCRE_ERROR_BADUTF8; + if (start_offset > 0 && start_offset < length) + { + int tb = ((uschar *)subject)[start_offset]; + if (tb > 127) + { + tb &= 0xc0; + if (tb != 0 && tb != 0xc0) return PCRE_ERROR_BADUTF8_OFFSET; + } + } + } +#endif + +/* If the exec call supplied NULL for tables, use the inbuilt ones. This +is a feature that makes it possible to save compiled regex and re-use them +in other programs later. */ + +if (md->tables == NULL) md->tables = _pcre_default_tables; + +/* The lower casing table and the "must be at the start of a line" flag are +used in a loop when finding where to start. */ + +lcc = md->tables + lcc_offset; +startline = (re->options & PCRE_STARTLINE) != 0; +firstline = (re->options & PCRE_FIRSTLINE) != 0; + +/* Set up the first character to match, if available. The first_byte value is +never set for an anchored regular expression, but the anchoring may be forced +at run time, so we have to test for anchoring. The first char may be unset for +an unanchored pattern, of course. If there's no first char and the pattern was +studied, there may be a bitmap of possible first characters. */ + +if (!anchored) + { + if ((re->options & PCRE_FIRSTSET) != 0) + { + first_byte = re->first_byte & 255; + if ((first_byte_caseless = ((re->first_byte & REQ_CASELESS) != 0)) == TRUE) + first_byte = lcc[first_byte]; + } + else + { + if (startline && study != NULL && + (study->options & PCRE_STUDY_MAPPED) != 0) + start_bits = study->start_bits; + } + } + +/* For anchored or unanchored matches, there may be a "last known required +character" set. */ + +if ((re->options & PCRE_REQCHSET) != 0) + { + req_byte = re->req_byte & 255; + req_byte_caseless = (re->req_byte & REQ_CASELESS) != 0; + req_byte2 = (md->tables + fcc_offset)[req_byte]; /* case flipped */ + } + +/* Call the main matching function, looping for a non-anchored regex after a +failed match. Unless restarting, optimize by moving to the first match +character if possible, when not anchored. Then unless wanting a partial match, +check for a required later character. */ + +for (;;) + { + int rc; + + if ((options & PCRE_DFA_RESTART) == 0) + { + const uschar *save_end_subject = end_subject; + + /* Advance to a unique first char if possible. If firstline is TRUE, the + start of the match is constrained to the first line of a multiline string. + Implement this by temporarily adjusting end_subject so that we stop + scanning at a newline. If the match fails at the newline, later code breaks + this loop. */ + + if (firstline) + { + const uschar *t = current_subject; + while (t <= save_end_subject - md->nllen && !IS_NEWLINE(t)) t++; + end_subject = t; + } + + if (first_byte >= 0) + { + if (first_byte_caseless) + while (current_subject < end_subject && + lcc[*current_subject] != first_byte) + current_subject++; + else + while (current_subject < end_subject && *current_subject != first_byte) + current_subject++; + } + + /* Or to just after a linebreak for a multiline match if possible */ + + else if (startline) + { + if (current_subject > md->start_subject + md->nllen + + start_offset) + { + while (current_subject <= end_subject && + !IS_NEWLINE(current_subject - md->nllen)) + current_subject++; + } + } + + /* Or to a non-unique first char after study */ + + else if (start_bits != NULL) + { + while (current_subject < end_subject) + { + register unsigned int c = *current_subject; + if ((start_bits[c/8] & (1 << (c&7))) == 0) current_subject++; + else break; + } + } + + /* Restore fudged end_subject */ + + end_subject = save_end_subject; + } + + /* If req_byte is set, we know that that character must appear in the subject + for the match to succeed. If the first character is set, req_byte must be + later in the subject; otherwise the test starts at the match point. This + optimization can save a huge amount of work in patterns with nested unlimited + repeats that aren't going to match. Writing separate code for cased/caseless + versions makes it go faster, as does using an autoincrement and backing off + on a match. + + HOWEVER: when the subject string is very, very long, searching to its end can + take a long time, and give bad performance on quite ordinary patterns. This + showed up when somebody was matching /^C/ on a 32-megabyte string... so we + don't do this when the string is sufficiently long. + + ALSO: this processing is disabled when partial matching is requested. + */ + + if (req_byte >= 0 && + end_subject - current_subject < REQ_BYTE_MAX && + (options & PCRE_PARTIAL) == 0) + { + register const uschar *p = current_subject + ((first_byte >= 0)? 1 : 0); + + /* We don't need to repeat the search if we haven't yet reached the + place we found it at last time. */ + + if (p > req_byte_ptr) + { + if (req_byte_caseless) + { + while (p < end_subject) + { + register int pp = *p++; + if (pp == req_byte || pp == req_byte2) { p--; break; } + } + } + else + { + while (p < end_subject) + { + if (*p++ == req_byte) { p--; break; } + } + } + + /* If we can't find the required character, break the matching loop, + which will cause a return or PCRE_ERROR_NOMATCH. */ + + if (p >= end_subject) break; + + /* If we have found the required character, save the point where we + found it, so that we don't search again next time round the loop if + the start hasn't passed this character yet. */ + + req_byte_ptr = p; + } + } + + /* OK, now we can do the business */ + + rc = internal_dfa_exec( + md, /* fixed match data */ + md->start_code, /* this subexpression's code */ + current_subject, /* where we currently are */ + start_offset, /* start offset in subject */ + offsets, /* offset vector */ + offsetcount, /* size of same */ + workspace, /* workspace vector */ + wscount, /* size of same */ + re->options & (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL), /* ims flags */ + 0, /* function recurse level */ + 0); /* regex recurse level */ + + /* Anything other than "no match" means we are done, always; otherwise, carry + on only if not anchored. */ + + if (rc != PCRE_ERROR_NOMATCH || anchored) return rc; + + /* Advance to the next subject character unless we are at the end of a line + and firstline is set. */ + + if (firstline && + current_subject <= end_subject - md->nllen && + IS_NEWLINE(current_subject)) break; + current_subject++; + if (utf8) + { + while (current_subject < end_subject && (*current_subject & 0xc0) == 0x80) + current_subject++; + } + if (current_subject > end_subject) break; + } + +return PCRE_ERROR_NOMATCH; +} + +/* End of pcre_dfa_exec.c */ ============================================================ --- pcre/pcre_exec.c 67b6e933c1bf48822748163ca104e433a79208b5 +++ pcre/pcre_exec.c 67b6e933c1bf48822748163ca104e433a79208b5 @@ -0,0 +1,3933 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + +/* PCRE is a library of functions to support regular expressions whose syntax +and semantics are as close as possible to those of the Perl 5 language. + + Written by Philip Hazel + Copyright (c) 1997-2006 University of Cambridge + +----------------------------------------------------------------------------- +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + + * Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimer. + + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + * Neither the name of the University of Cambridge nor the names of its + contributors may be used to endorse or promote products derived from + this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE +LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR +CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +----------------------------------------------------------------------------- +*/ + + +/* This module contains pcre_exec(), the externally visible function that does +pattern matching using an NFA algorithm, trying to mimic Perl as closely as +possible. There are also some static supporting functions. */ + +#define NLBLOCK md /* The block containing newline information */ +#include "pcre_internal.h" + + +/* Structure for building a chain of data that actually lives on the +stack, for holding the values of the subject pointer at the start of each +subpattern, so as to detect when an empty string has been matched by a +subpattern - to break infinite loops. When NO_RECURSE is set, these blocks +are on the heap, not on the stack. */ + +typedef struct eptrblock { + struct eptrblock *epb_prev; + USPTR epb_saved_eptr; +} eptrblock; + +/* Flag bits for the match() function */ + +#define match_condassert 0x01 /* Called to check a condition assertion */ +#define match_isgroup 0x02 /* Set if start of bracketed group */ + +/* Non-error returns from the match() function. Error returns are externally +defined PCRE_ERROR_xxx codes, which are all negative. */ + +#define MATCH_MATCH 1 +#define MATCH_NOMATCH 0 + +/* Maximum number of ints of offset to save on the stack for recursive calls. +If the offset vector is bigger, malloc is used. This should be a multiple of 3, +because the offset vector is always a multiple of 3 long. */ + +#define REC_STACK_SAVE_MAX 30 + +/* Min and max values for the common repeats; for the maxima, 0 => infinity */ + +static const char rep_min[] = { 0, 0, 1, 1, 0, 0 }; +static const char rep_max[] = { 0, 0, 0, 0, 1, 1 }; + + + +#ifdef DEBUG +/************************************************* +* Debugging function to print chars * +*************************************************/ + +/* Print a sequence of chars in printable format, stopping at the end of the +subject if the requested. + +Arguments: + p points to characters + length number to print + is_subject TRUE if printing from within md->start_subject + md pointer to matching data block, if is_subject is TRUE + +Returns: nothing +*/ + +static void +pchars(const uschar *p, int length, BOOL is_subject, match_data *md) +{ +int c; +if (is_subject && length > md->end_subject - p) length = md->end_subject - p; +while (length-- > 0) + if (isprint(c = *(p++))) printf("%c", c); else printf("\\x%02x", c); +} +#endif + + + +/************************************************* +* Match a back-reference * +*************************************************/ + +/* If a back reference hasn't been set, the length that is passed is greater +than the number of characters left in the string, so the match fails. + +Arguments: + offset index into the offset vector + eptr points into the subject + length length to be matched + md points to match data block + ims the ims flags + +Returns: TRUE if matched +*/ + +static BOOL +match_ref(int offset, register USPTR eptr, int length, match_data *md, + unsigned long int ims) +{ +USPTR p = md->start_subject + md->offset_vector[offset]; + +#ifdef DEBUG +if (eptr >= md->end_subject) + printf("matching subject "); +else + { + printf("matching subject "); + pchars(eptr, length, TRUE, md); + } +printf(" against backref "); +pchars(p, length, FALSE, md); +printf("\n"); +#endif + +/* Always fail if not enough characters left */ + +if (length > md->end_subject - eptr) return FALSE; + +/* Separate the caselesss case for speed */ + +if ((ims & PCRE_CASELESS) != 0) + { + while (length-- > 0) + if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE; + } +else + { while (length-- > 0) if (*p++ != *eptr++) return FALSE; } + +return TRUE; +} + + + +/*************************************************************************** +**************************************************************************** + RECURSION IN THE match() FUNCTION + +The match() function is highly recursive, though not every recursive call +increases the recursive depth. Nevertheless, some regular expressions can cause +it to recurse to a great depth. I was writing for Unix, so I just let it call +itself recursively. This uses the stack for saving everything that has to be +saved for a recursive call. On Unix, the stack can be large, and this works +fine. + +It turns out that on some non-Unix-like systems there are problems with +programs that use a lot of stack. (This despite the fact that every last chip +has oodles of memory these days, and techniques for extending the stack have +been known for decades.) So.... + +There is a fudge, triggered by defining NO_RECURSE, which avoids recursive +calls by keeping local variables that need to be preserved in blocks of memory +obtained from malloc() instead instead of on the stack. Macros are used to +achieve this so that the actual code doesn't look very different to what it +always used to. +**************************************************************************** +***************************************************************************/ + + +/* These versions of the macros use the stack, as normal. There are debugging +versions and production versions. */ + +#ifndef NO_RECURSE +#define REGISTER register +#ifdef DEBUG +#define RMATCH(rx,ra,rb,rc,rd,re,rf,rg) \ + { \ + printf("match() called in line %d\n", __LINE__); \ + rx = match(ra,rb,rc,rd,re,rf,rg,rdepth+1); \ + printf("to line %d\n", __LINE__); \ + } +#define RRETURN(ra) \ + { \ + printf("match() returned %d from line %d ", ra, __LINE__); \ + return ra; \ + } +#else +#define RMATCH(rx,ra,rb,rc,rd,re,rf,rg) \ + rx = match(ra,rb,rc,rd,re,rf,rg,rdepth+1) +#define RRETURN(ra) return ra +#endif + +#else + + +/* These versions of the macros manage a private stack on the heap. Note +that the rd argument of RMATCH isn't actually used. It's the md argument of +match(), which never changes. */ + +#define REGISTER + +#define RMATCH(rx,ra,rb,rc,rd,re,rf,rg)\ + {\ + heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\ + if (setjmp(frame->Xwhere) == 0)\ + {\ + newframe->Xeptr = ra;\ + newframe->Xecode = rb;\ + newframe->Xoffset_top = rc;\ + newframe->Xims = re;\ + newframe->Xeptrb = rf;\ + newframe->Xflags = rg;\ + newframe->Xrdepth = frame->Xrdepth + 1;\ + newframe->Xprevframe = frame;\ + frame = newframe;\ + DPRINTF(("restarting from line %d\n", __LINE__));\ + goto HEAP_RECURSE;\ + }\ + else\ + {\ + DPRINTF(("longjumped back to line %d\n", __LINE__));\ + frame = md->thisframe;\ + rx = frame->Xresult;\ + }\ + } + +#define RRETURN(ra)\ + {\ + heapframe *newframe = frame;\ + frame = newframe->Xprevframe;\ + (pcre_stack_free)(newframe);\ + if (frame != NULL)\ + {\ + frame->Xresult = ra;\ + md->thisframe = frame;\ + longjmp(frame->Xwhere, 1);\ + }\ + return ra;\ + } + + +/* Structure for remembering the local variables in a private frame */ + +typedef struct heapframe { + struct heapframe *Xprevframe; + + /* Function arguments that may change */ + + const uschar *Xeptr; + const uschar *Xecode; + int Xoffset_top; + long int Xims; + eptrblock *Xeptrb; + int Xflags; + unsigned int Xrdepth; + + /* Function local variables */ + + const uschar *Xcallpat; + const uschar *Xcharptr; + const uschar *Xdata; + const uschar *Xnext; + const uschar *Xpp; + const uschar *Xprev; + const uschar *Xsaved_eptr; + + recursion_info Xnew_recursive; + + BOOL Xcur_is_word; + BOOL Xcondition; + BOOL Xminimize; + BOOL Xprev_is_word; + + unsigned long int Xoriginal_ims; + +#ifdef SUPPORT_UCP + int Xprop_type; + int Xprop_value; + int Xprop_fail_result; + int Xprop_category; + int Xprop_chartype; + int Xprop_script; + int *Xprop_test_variable; +#endif + + int Xctype; + int Xfc; + int Xfi; + int Xlength; + int Xmax; + int Xmin; + int Xnumber; + int Xoffset; + int Xop; + int Xsave_capture_last; + int Xsave_offset1, Xsave_offset2, Xsave_offset3; + int Xstacksave[REC_STACK_SAVE_MAX]; + + eptrblock Xnewptrb; + + /* Place to pass back result, and where to jump back to */ + + int Xresult; + jmp_buf Xwhere; + +} heapframe; + +#endif + + +/*************************************************************************** +***************************************************************************/ + + + +/************************************************* +* Match from current position * +*************************************************/ + +/* On entry ecode points to the first opcode, and eptr to the first character +in the subject string, while eptrb holds the value of eptr at the start of the +last bracketed group - used for breaking infinite loops matching zero-length +strings. This function is called recursively in many circumstances. Whenever it +returns a negative (error) response, the outer incarnation must also return the +same response. + +Performance note: It might be tempting to extract commonly used fields from the +md structure (e.g. utf8, end_subject) into individual variables to improve +performance. Tests using gcc on a SPARC disproved this; in the first case, it +made performance worse. + +Arguments: + eptr pointer in subject + ecode position in code + offset_top current top pointer + md pointer to "static" info for the match + ims current /i, /m, and /s options + eptrb pointer to chain of blocks containing eptr at start of + brackets - for testing for empty matches + flags can contain + match_condassert - this is an assertion condition + match_isgroup - this is the start of a bracketed group + rdepth the recursion depth + +Returns: MATCH_MATCH if matched ) these values are >= 0 + MATCH_NOMATCH if failed to match ) + a negative PCRE_ERROR_xxx value if aborted by an error condition + (e.g. stopped by repeated call or recursion limit) +*/ + +static int +match(REGISTER USPTR eptr, REGISTER const uschar *ecode, + int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb, + int flags, unsigned int rdepth) +{ +/* These variables do not need to be preserved over recursion in this function, +so they can be ordinary variables in all cases. Mark them with "register" +because they are used a lot in loops. */ + +register int rrc; /* Returns from recursive calls */ +register int i; /* Used for loops not involving calls to RMATCH() */ +register unsigned int c; /* Character values not kept over RMATCH() calls */ +register BOOL utf8; /* Local copy of UTF-8 flag for speed */ + +/* When recursion is not being used, all "local" variables that have to be +preserved over calls to RMATCH() are part of a "frame" which is obtained from +heap storage. Set up the top-level frame here; others are obtained from the +heap whenever RMATCH() does a "recursion". See the macro definitions above. */ + +#ifdef NO_RECURSE +heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe)); +frame->Xprevframe = NULL; /* Marks the top level */ + +/* Copy in the original argument variables */ + +frame->Xeptr = eptr; +frame->Xecode = ecode; +frame->Xoffset_top = offset_top; +frame->Xims = ims; +frame->Xeptrb = eptrb; +frame->Xflags = flags; +frame->Xrdepth = rdepth; + +/* This is where control jumps back to to effect "recursion" */ + +HEAP_RECURSE: + +/* Macros make the argument variables come from the current frame */ + +#define eptr frame->Xeptr +#define ecode frame->Xecode +#define offset_top frame->Xoffset_top +#define ims frame->Xims +#define eptrb frame->Xeptrb +#define flags frame->Xflags +#define rdepth frame->Xrdepth + +/* Ditto for the local variables */ + +#ifdef SUPPORT_UTF8 +#define charptr frame->Xcharptr +#endif +#define callpat frame->Xcallpat +#define data frame->Xdata +#define next frame->Xnext +#define pp frame->Xpp +#define prev frame->Xprev +#define saved_eptr frame->Xsaved_eptr + +#define new_recursive frame->Xnew_recursive + +#define cur_is_word frame->Xcur_is_word +#define condition frame->Xcondition +#define minimize frame->Xminimize +#define prev_is_word frame->Xprev_is_word + +#define original_ims frame->Xoriginal_ims + +#ifdef SUPPORT_UCP +#define prop_type frame->Xprop_type +#define prop_value frame->Xprop_value +#define prop_fail_result frame->Xprop_fail_result +#define prop_category frame->Xprop_category +#define prop_chartype frame->Xprop_chartype +#define prop_script frame->Xprop_script +#define prop_test_variable frame->Xprop_test_variable +#endif + +#define ctype frame->Xctype +#define fc frame->Xfc +#define fi frame->Xfi +#define length frame->Xlength +#define max frame->Xmax +#define min frame->Xmin +#define number frame->Xnumber +#define offset frame->Xoffset +#define op frame->Xop +#define save_capture_last frame->Xsave_capture_last +#define save_offset1 frame->Xsave_offset1 +#define save_offset2 frame->Xsave_offset2 +#define save_offset3 frame->Xsave_offset3 +#define stacksave frame->Xstacksave + +#define newptrb frame->Xnewptrb + +/* When recursion is being used, local variables are allocated on the stack and +get preserved during recursion in the normal way. In this environment, fi and +i, and fc and c, can be the same variables. */ + +#else +#define fi i +#define fc c + + +#ifdef SUPPORT_UTF8 /* Many of these variables are used only */ +const uschar *charptr; /* in small blocks of the code. My normal */ +#endif /* style of coding would have declared */ +const uschar *callpat; /* them within each of those blocks. */ +const uschar *data; /* However, in order to accommodate the */ +const uschar *next; /* version of this code that uses an */ +USPTR pp; /* external "stack" implemented on the */ +const uschar *prev; /* heap, it is easier to declare them all */ +USPTR saved_eptr; /* here, so the declarations can be cut */ + /* out in a block. The only declarations */ +recursion_info new_recursive; /* within blocks below are for variables */ + /* that do not have to be preserved over */ +BOOL cur_is_word; /* a recursive call to RMATCH(). */ +BOOL condition; +BOOL minimize; +BOOL prev_is_word; + +unsigned long int original_ims; + +#ifdef SUPPORT_UCP +int prop_type; +int prop_value; +int prop_fail_result; +int prop_category; +int prop_chartype; +int prop_script; +int *prop_test_variable; +#endif + +int ctype; +int length; +int max; +int min; +int number; +int offset; +int op; +int save_capture_last; +int save_offset1, save_offset2, save_offset3; +int stacksave[REC_STACK_SAVE_MAX]; + +eptrblock newptrb; +#endif + +/* These statements are here to stop the compiler complaining about unitialized +variables. */ + +#ifdef SUPPORT_UCP +prop_value = 0; +prop_fail_result = 0; +prop_test_variable = NULL; +#endif + +/* This label is used for tail recursion, which is used in a few cases even +when NO_RECURSE is not defined, in order to reduce the amount of stack that is +used. Thanks to Ian Taylor for noticing this possibility and sending the +original patch. */ + +TAIL_RECURSE: + +/* OK, now we can get on with the real code of the function. Recursive calls +are specified by the macro RMATCH and RRETURN is used to return. When +NO_RECURSE is *not* defined, these just turn into a recursive call to match() +and a "return", respectively (possibly with some debugging if DEBUG is +defined). However, RMATCH isn't like a function call because it's quite a +complicated macro. It has to be used in one particular way. This shouldn't, +however, impact performance when true recursion is being used. */ + +/* First check that we haven't called match() too many times, or that we +haven't exceeded the recursive call limit. */ + +if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT); +if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT); + +original_ims = ims; /* Save for resetting on ')' */ + +#ifdef SUPPORT_UTF8 +utf8 = md->utf8; /* Local copy of the flag */ +#else +utf8 = FALSE; +#endif + +/* At the start of a bracketed group, add the current subject pointer to the +stack of such pointers, to be re-instated at the end of the group when we hit +the closing ket. When match() is called in other circumstances, we don't add to +this stack. */ + +if ((flags & match_isgroup) != 0) + { + newptrb.epb_prev = eptrb; + newptrb.epb_saved_eptr = eptr; + eptrb = &newptrb; + } + +/* Now start processing the operations. */ + +for (;;) + { + op = *ecode; + minimize = FALSE; + + /* For partial matching, remember if we ever hit the end of the subject after + matching at least one subject character. */ + + if (md->partial && + eptr >= md->end_subject && + eptr > md->start_match) + md->hitend = TRUE; + + /* Opening capturing bracket. If there is space in the offset vector, save + the current subject position in the working slot at the top of the vector. We + mustn't change the current values of the data slot, because they may be set + from a previous iteration of this group, and be referred to by a reference + inside the group. + + If the bracket fails to match, we need to restore this value and also the + values of the final offsets, in case they were set by a previous iteration of + the same bracket. + + If there isn't enough space in the offset vector, treat this as if it were a + non-capturing bracket. Don't worry about setting the flag for the error case + here; that is handled in the code for KET. */ + + if (op > OP_BRA) + { + number = op - OP_BRA; + + /* For extended extraction brackets (large number), we have to fish out the + number from a dummy opcode at the start. */ + + if (number > EXTRACT_BASIC_MAX) + number = GET2(ecode, 2+LINK_SIZE); + offset = number << 1; + +#ifdef DEBUG + printf("start bracket %d subject=", number); + pchars(eptr, 16, TRUE, md); + printf("\n"); +#endif + + if (offset < md->offset_max) + { + save_offset1 = md->offset_vector[offset]; + save_offset2 = md->offset_vector[offset+1]; + save_offset3 = md->offset_vector[md->offset_end - number]; + save_capture_last = md->capture_last; + + DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3)); + md->offset_vector[md->offset_end - number] = eptr - md->start_subject; + + do + { + RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, + match_isgroup); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + md->capture_last = save_capture_last; + ecode += GET(ecode, 1); + } + while (*ecode == OP_ALT); + + DPRINTF(("bracket %d failed\n", number)); + + md->offset_vector[offset] = save_offset1; + md->offset_vector[offset+1] = save_offset2; + md->offset_vector[md->offset_end - number] = save_offset3; + + RRETURN(MATCH_NOMATCH); + } + + /* Insufficient room for saving captured contents */ + + else op = OP_BRA; + } + + /* Other types of node can be handled by a switch */ + + switch(op) + { + case OP_BRA: /* Non-capturing bracket: optimized */ + DPRINTF(("start bracket 0\n")); + + /* Loop for all the alternatives */ + + for (;;) + { + /* When we get to the final alternative within the brackets, we would + return the result of a recursive call to match() whatever happened. We + can reduce stack usage by turning this into a tail recursion. */ + + if (ecode[GET(ecode, 1)] != OP_ALT) + { + ecode += 1 + LINK_SIZE; + flags = match_isgroup; + DPRINTF(("bracket 0 tail recursion\n")); + goto TAIL_RECURSE; + } + + /* For non-final alternatives, continue the loop for a NOMATCH result; + otherwise return. */ + + RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, + match_isgroup); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode += GET(ecode, 1); + } + /* Control never reaches here. */ + + /* Conditional group: compilation checked that there are no more than + two branches. If the condition is false, skipping the first branch takes us + past the end if there is only one branch, but that's OK because that is + exactly what going to the ket would do. As there is only one branch to be + obeyed, we can use tail recursion to avoid using another stack frame. */ + + case OP_COND: + if (ecode[LINK_SIZE+1] == OP_CREF) /* Condition extract or recurse test */ + { + offset = GET2(ecode, LINK_SIZE+2) << 1; /* Doubled ref number */ + condition = (offset == CREF_RECURSE * 2)? + (md->recursive != NULL) : + (offset < offset_top && md->offset_vector[offset] >= 0); + ecode += condition? (LINK_SIZE + 4) : (LINK_SIZE + 1 + GET(ecode, 1)); + flags = match_isgroup; + goto TAIL_RECURSE; + } + + /* The condition is an assertion. Call match() to evaluate it - setting + the final argument TRUE causes it to stop at the end of an assertion. */ + + else + { + RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, + match_condassert | match_isgroup); + if (rrc == MATCH_MATCH) + { + ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE+2); + while (*ecode == OP_ALT) ecode += GET(ecode, 1); + } + else if (rrc != MATCH_NOMATCH) + { + RRETURN(rrc); /* Need braces because of following else */ + } + else ecode += GET(ecode, 1); + + /* We are now at the branch that is to be obeyed. As there is only one, + we can use tail recursion to avoid using another stack frame. */ + + ecode += 1 + LINK_SIZE; + flags = match_isgroup; + goto TAIL_RECURSE; + } + /* Control never reaches here */ + + /* Skip over conditional reference or large extraction number data if + encountered. */ + + case OP_CREF: + case OP_BRANUMBER: + ecode += 3; + break; + + /* End of the pattern. If we are in a recursion, we should restore the + offsets appropriately and continue from after the call. */ + + case OP_END: + if (md->recursive != NULL && md->recursive->group_num == 0) + { + recursion_info *rec = md->recursive; + DPRINTF(("End of pattern in a (?0) recursion\n")); + md->recursive = rec->prevrec; + memmove(md->offset_vector, rec->offset_save, + rec->saved_max * sizeof(int)); + md->start_match = rec->save_start; + ims = original_ims; + ecode = rec->after_call; + break; + } + + /* Otherwise, if PCRE_NOTEMPTY is set, fail if we have matched an empty + string - backtracking will then try other alternatives, if any. */ + + if (md->notempty && eptr == md->start_match) RRETURN(MATCH_NOMATCH); + md->end_match_ptr = eptr; /* Record where we ended */ + md->end_offset_top = offset_top; /* and how many extracts were taken */ + RRETURN(MATCH_MATCH); + + /* Change option settings */ + + case OP_OPT: + ims = ecode[1]; + ecode += 2; + DPRINTF(("ims set to %02lx\n", ims)); + break; + + /* Assertion brackets. Check the alternative branches in turn - the + matching won't pass the KET for an assertion. If any one branch matches, + the assertion is true. Lookbehind assertions have an OP_REVERSE item at the + start of each branch to move the current point backwards, so the code at + this level is identical to the lookahead case. */ + + case OP_ASSERT: + case OP_ASSERTBACK: + do + { + RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, + match_isgroup); + if (rrc == MATCH_MATCH) break; + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode += GET(ecode, 1); + } + while (*ecode == OP_ALT); + if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH); + + /* If checking an assertion for a condition, return MATCH_MATCH. */ + + if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH); + + /* Continue from after the assertion, updating the offsets high water + mark, since extracts may have been taken during the assertion. */ + + do ecode += GET(ecode,1); while (*ecode == OP_ALT); + ecode += 1 + LINK_SIZE; + offset_top = md->end_offset_top; + continue; + + /* Negative assertion: all branches must fail to match */ + + case OP_ASSERT_NOT: + case OP_ASSERTBACK_NOT: + do + { + RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, + match_isgroup); + if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode += GET(ecode,1); + } + while (*ecode == OP_ALT); + + if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH); + + ecode += 1 + LINK_SIZE; + continue; + + /* Move the subject pointer back. This occurs only at the start of + each branch of a lookbehind assertion. If we are too close to the start to + move back, this match function fails. When working with UTF-8 we move + back a number of characters, not bytes. */ + + case OP_REVERSE: +#ifdef SUPPORT_UTF8 + if (utf8) + { + c = GET(ecode,1); + for (i = 0; i < c; i++) + { + eptr--; + if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); + BACKCHAR(eptr) + } + } + else +#endif + + /* No UTF-8 support, or not in UTF-8 mode: count is byte count */ + + { + eptr -= GET(ecode,1); + if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); + } + + /* Skip to next op code */ + + ecode += 1 + LINK_SIZE; + break; + + /* The callout item calls an external function, if one is provided, passing + details of the match so far. This is mainly for debugging, though the + function is able to force a failure. */ + + case OP_CALLOUT: + if (pcre_callout != NULL) + { + pcre_callout_block cb; + cb.version = 1; /* Version 1 of the callout block */ + cb.callout_number = ecode[1]; + cb.offset_vector = md->offset_vector; + cb.subject = (PCRE_SPTR)md->start_subject; + cb.subject_length = md->end_subject - md->start_subject; + cb.start_match = md->start_match - md->start_subject; + cb.current_position = eptr - md->start_subject; + cb.pattern_position = GET(ecode, 2); + cb.next_item_length = GET(ecode, 2 + LINK_SIZE); + cb.capture_top = offset_top/2; + cb.capture_last = md->capture_last; + cb.callout_data = md->callout_data; + if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH); + if (rrc < 0) RRETURN(rrc); + } + ecode += 2 + 2*LINK_SIZE; + break; + + /* Recursion either matches the current regex, or some subexpression. The + offset data is the offset to the starting bracket from the start of the + whole pattern. (This is so that it works from duplicated subpatterns.) + + If there are any capturing brackets started but not finished, we have to + save their starting points and reinstate them after the recursion. However, + we don't know how many such there are (offset_top records the completed + total) so we just have to save all the potential data. There may be up to + 65535 such values, which is too large to put on the stack, but using malloc + for small numbers seems expensive. As a compromise, the stack is used when + there are no more than REC_STACK_SAVE_MAX values to store; otherwise malloc + is used. A problem is what to do if the malloc fails ... there is no way of + returning to the top level with an error. Save the top REC_STACK_SAVE_MAX + values on the stack, and accept that the rest may be wrong. + + There are also other values that have to be saved. We use a chained + sequence of blocks that actually live on the stack. Thanks to Robin Houston + for the original version of this logic. */ + + case OP_RECURSE: + { + callpat = md->start_code + GET(ecode, 1); + new_recursive.group_num = *callpat - OP_BRA; + + /* For extended extraction brackets (large number), we have to fish out + the number from a dummy opcode at the start. */ + + if (new_recursive.group_num > EXTRACT_BASIC_MAX) + new_recursive.group_num = GET2(callpat, 2+LINK_SIZE); + + /* Add to "recursing stack" */ + + new_recursive.prevrec = md->recursive; + md->recursive = &new_recursive; + + /* Find where to continue from afterwards */ + + ecode += 1 + LINK_SIZE; + new_recursive.after_call = ecode; + + /* Now save the offset data. */ + + new_recursive.saved_max = md->offset_end; + if (new_recursive.saved_max <= REC_STACK_SAVE_MAX) + new_recursive.offset_save = stacksave; + else + { + new_recursive.offset_save = + (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int)); + if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY); + } + + memcpy(new_recursive.offset_save, md->offset_vector, + new_recursive.saved_max * sizeof(int)); + new_recursive.save_start = md->start_match; + md->start_match = eptr; + + /* OK, now we can do the recursion. For each top-level alternative we + restore the offset and recursion data. */ + + DPRINTF(("Recursing into group %d\n", new_recursive.group_num)); + do + { + RMATCH(rrc, eptr, callpat + 1 + LINK_SIZE, offset_top, md, ims, + eptrb, match_isgroup); + if (rrc == MATCH_MATCH) + { + DPRINTF(("Recursion matched\n")); + md->recursive = new_recursive.prevrec; + if (new_recursive.offset_save != stacksave) + (pcre_free)(new_recursive.offset_save); + RRETURN(MATCH_MATCH); + } + else if (rrc != MATCH_NOMATCH) + { + DPRINTF(("Recursion gave error %d\n", rrc)); + RRETURN(rrc); + } + + md->recursive = &new_recursive; + memcpy(md->offset_vector, new_recursive.offset_save, + new_recursive.saved_max * sizeof(int)); + callpat += GET(callpat, 1); + } + while (*callpat == OP_ALT); + + DPRINTF(("Recursion didn't match\n")); + md->recursive = new_recursive.prevrec; + if (new_recursive.offset_save != stacksave) + (pcre_free)(new_recursive.offset_save); + RRETURN(MATCH_NOMATCH); + } + /* Control never reaches here */ + + /* "Once" brackets are like assertion brackets except that after a match, + the point in the subject string is not moved back. Thus there can never be + a move back into the brackets. Friedl calls these "atomic" subpatterns. + Check the alternative branches in turn - the matching won't pass the KET + for this kind of subpattern. If any one branch matches, we carry on as at + the end of a normal bracket, leaving the subject pointer. */ + + case OP_ONCE: + prev = ecode; + saved_eptr = eptr; + + do + { + RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, + eptrb, match_isgroup); + if (rrc == MATCH_MATCH) break; + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode += GET(ecode,1); + } + while (*ecode == OP_ALT); + + /* If hit the end of the group (which could be repeated), fail */ + + if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH); + + /* Continue as from after the assertion, updating the offsets high water + mark, since extracts may have been taken. */ + + do ecode += GET(ecode,1); while (*ecode == OP_ALT); + + offset_top = md->end_offset_top; + eptr = md->end_match_ptr; + + /* For a non-repeating ket, just continue at this level. This also + happens for a repeating ket if no characters were matched in the group. + This is the forcible breaking of infinite loops as implemented in Perl + 5.005. If there is an options reset, it will get obeyed in the normal + course of events. */ + + if (*ecode == OP_KET || eptr == saved_eptr) + { + ecode += 1+LINK_SIZE; + break; + } + + /* The repeating kets try the rest of the pattern or restart from the + preceding bracket, in the appropriate order. The second "call" of match() + uses tail recursion, to avoid using another stack frame. We need to reset + any options that changed within the bracket before re-running it, so + check the next opcode. */ + + if (ecode[1+LINK_SIZE] == OP_OPT) + { + ims = (ims & ~PCRE_IMS) | ecode[4]; + DPRINTF(("ims set to %02lx at group repeat\n", ims)); + } + + if (*ecode == OP_KETRMIN) + { + RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode = prev; + flags = match_isgroup; + goto TAIL_RECURSE; + } + else /* OP_KETRMAX */ + { + RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode += 1 + LINK_SIZE; + flags = 0; + goto TAIL_RECURSE; + } + /* Control never gets here */ + + /* An alternation is the end of a branch; scan along to find the end of the + bracketed group and go to there. */ + + case OP_ALT: + do ecode += GET(ecode,1); while (*ecode == OP_ALT); + break; + + /* BRAZERO and BRAMINZERO occur just before a bracket group, indicating + that it may occur zero times. It may repeat infinitely, or not at all - + i.e. it could be ()* or ()? in the pattern. Brackets with fixed upper + repeat limits are compiled as a number of copies, with the optional ones + preceded by BRAZERO or BRAMINZERO. */ + + case OP_BRAZERO: + { + next = ecode+1; + RMATCH(rrc, eptr, next, offset_top, md, ims, eptrb, match_isgroup); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + do next += GET(next,1); while (*next == OP_ALT); + ecode = next + 1+LINK_SIZE; + } + break; + + case OP_BRAMINZERO: + { + next = ecode+1; + do next += GET(next,1); while (*next == OP_ALT); + RMATCH(rrc, eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, + match_isgroup); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode++; + } + break; + + /* End of a group, repeated or non-repeating. If we are at the end of + an assertion "group", stop matching and return MATCH_MATCH, but record the + current high water mark for use by positive assertions. Do this also + for the "once" (not-backup up) groups. */ + + case OP_KET: + case OP_KETRMIN: + case OP_KETRMAX: + prev = ecode - GET(ecode, 1); + saved_eptr = eptrb->epb_saved_eptr; + + /* Back up the stack of bracket start pointers. */ + + eptrb = eptrb->epb_prev; + + if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT || + *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT || + *prev == OP_ONCE) + { + md->end_match_ptr = eptr; /* For ONCE */ + md->end_offset_top = offset_top; + RRETURN(MATCH_MATCH); + } + + /* In all other cases except a conditional group we have to check the + group number back at the start and if necessary complete handling an + extraction by setting the offsets and bumping the high water mark. */ + + if (*prev != OP_COND) + { + number = *prev - OP_BRA; + + /* For extended extraction brackets (large number), we have to fish out + the number from a dummy opcode at the start. */ + + if (number > EXTRACT_BASIC_MAX) number = GET2(prev, 2+LINK_SIZE); + offset = number << 1; + +#ifdef DEBUG + printf("end bracket %d", number); + printf("\n"); +#endif + + /* Test for a numbered group. This includes groups called as a result + of recursion. Note that whole-pattern recursion is coded as a recurse + into group 0, so it won't be picked up here. Instead, we catch it when + the OP_END is reached. */ + + if (number > 0) + { + md->capture_last = number; + if (offset >= md->offset_max) md->offset_overflow = TRUE; else + { + md->offset_vector[offset] = + md->offset_vector[md->offset_end - number]; + md->offset_vector[offset+1] = eptr - md->start_subject; + if (offset_top <= offset) offset_top = offset + 2; + } + + /* Handle a recursively called group. Restore the offsets + appropriately and continue from after the call. */ + + if (md->recursive != NULL && md->recursive->group_num == number) + { + recursion_info *rec = md->recursive; + DPRINTF(("Recursion (%d) succeeded - continuing\n", number)); + md->recursive = rec->prevrec; + md->start_match = rec->save_start; + memcpy(md->offset_vector, rec->offset_save, + rec->saved_max * sizeof(int)); + ecode = rec->after_call; + ims = original_ims; + break; + } + } + } + + /* Reset the value of the ims flags, in case they got changed during + the group. */ + + ims = original_ims; + DPRINTF(("ims reset to %02lx\n", ims)); + + /* For a non-repeating ket, just continue at this level. This also + happens for a repeating ket if no characters were matched in the group. + This is the forcible breaking of infinite loops as implemented in Perl + 5.005. If there is an options reset, it will get obeyed in the normal + course of events. */ + + if (*ecode == OP_KET || eptr == saved_eptr) + { + ecode += 1 + LINK_SIZE; + break; + } + + /* The repeating kets try the rest of the pattern or restart from the + preceding bracket, in the appropriate order. In the second case, we can use + tail recursion to avoid using another stack frame. */ + + if (*ecode == OP_KETRMIN) + { + RMATCH(rrc, eptr, ecode + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode = prev; + flags = match_isgroup; + goto TAIL_RECURSE; + } + else /* OP_KETRMAX */ + { + RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode += 1 + LINK_SIZE; + flags = 0; + goto TAIL_RECURSE; + } + /* Control never gets here */ + + /* Start of subject unless notbol, or after internal newline if multiline */ + + case OP_CIRC: + if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH); + if ((ims & PCRE_MULTILINE) != 0) + { + if (eptr != md->start_subject && + (eptr == md->end_subject || + eptr < md->start_subject + md->nllen || + !IS_NEWLINE(eptr - md->nllen))) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + } + /* ... else fall through */ + + /* Start of subject assertion */ + + case OP_SOD: + if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH); + ecode++; + break; + + /* Start of match assertion */ + + case OP_SOM: + if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH); + ecode++; + break; + + /* Assert before internal newline if multiline, or before a terminating + newline unless endonly is set, else end of subject unless noteol is set. */ + + case OP_DOLL: + if ((ims & PCRE_MULTILINE) != 0) + { + if (eptr < md->end_subject) + { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); } + else + { if (md->noteol) RRETURN(MATCH_NOMATCH); } + ecode++; + break; + } + else + { + if (md->noteol) RRETURN(MATCH_NOMATCH); + if (!md->endonly) + { + if (eptr != md->end_subject && + (eptr != md->end_subject - md->nllen || !IS_NEWLINE(eptr))) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + } + } + /* ... else fall through for endonly */ + + /* End of subject assertion (\z) */ + + case OP_EOD: + if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH); + ecode++; + break; + + /* End of subject or ending \n assertion (\Z) */ + + case OP_EODN: + if (eptr != md->end_subject && + (eptr != md->end_subject - md->nllen || !IS_NEWLINE(eptr))) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + /* Word boundary assertions */ + + case OP_NOT_WORD_BOUNDARY: + case OP_WORD_BOUNDARY: + { + + /* Find out if the previous and current characters are "word" characters. + It takes a bit more work in UTF-8 mode. Characters > 255 are assumed to + be "non-word" characters. */ + +#ifdef SUPPORT_UTF8 + if (utf8) + { + if (eptr == md->start_subject) prev_is_word = FALSE; else + { + const uschar *lastptr = eptr - 1; + while((*lastptr & 0xc0) == 0x80) lastptr--; + GETCHAR(c, lastptr); + prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; + } + if (eptr >= md->end_subject) cur_is_word = FALSE; else + { + GETCHAR(c, eptr); + cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; + } + } + else +#endif + + /* More streamlined when not in UTF-8 mode */ + + { + prev_is_word = (eptr != md->start_subject) && + ((md->ctypes[eptr[-1]] & ctype_word) != 0); + cur_is_word = (eptr < md->end_subject) && + ((md->ctypes[*eptr] & ctype_word) != 0); + } + + /* Now see if the situation is what we want */ + + if ((*ecode++ == OP_WORD_BOUNDARY)? + cur_is_word == prev_is_word : cur_is_word != prev_is_word) + RRETURN(MATCH_NOMATCH); + } + break; + + /* Match a single character type; inline for speed */ + + case OP_ANY: + if ((ims & PCRE_DOTALL) == 0) + { + if (eptr <= md->end_subject - md->nllen && IS_NEWLINE(eptr)) + RRETURN(MATCH_NOMATCH); + } + if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH); + if (utf8) + while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; + ecode++; + break; + + /* Match a single byte, even in UTF-8 mode. This opcode really does match + any byte, even newline, independent of the setting of PCRE_DOTALL. */ + + case OP_ANYBYTE: + if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH); + ecode++; + break; + + case OP_NOT_DIGIT: + if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINCTEST(c, eptr); + if ( +#ifdef SUPPORT_UTF8 + c < 256 && +#endif + (md->ctypes[c] & ctype_digit) != 0 + ) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + case OP_DIGIT: + if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINCTEST(c, eptr); + if ( +#ifdef SUPPORT_UTF8 + c >= 256 || +#endif + (md->ctypes[c] & ctype_digit) == 0 + ) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + case OP_NOT_WHITESPACE: + if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINCTEST(c, eptr); + if ( +#ifdef SUPPORT_UTF8 + c < 256 && +#endif + (md->ctypes[c] & ctype_space) != 0 + ) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + case OP_WHITESPACE: + if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINCTEST(c, eptr); + if ( +#ifdef SUPPORT_UTF8 + c >= 256 || +#endif + (md->ctypes[c] & ctype_space) == 0 + ) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + case OP_NOT_WORDCHAR: + if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINCTEST(c, eptr); + if ( +#ifdef SUPPORT_UTF8 + c < 256 && +#endif + (md->ctypes[c] & ctype_word) != 0 + ) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + case OP_WORDCHAR: + if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINCTEST(c, eptr); + if ( +#ifdef SUPPORT_UTF8 + c >= 256 || +#endif + (md->ctypes[c] & ctype_word) == 0 + ) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + +#ifdef SUPPORT_UCP + /* Check the next character by Unicode property. We will get here only + if the support is in the binary; otherwise a compile-time error occurs. */ + + case OP_PROP: + case OP_NOTPROP: + if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINCTEST(c, eptr); + { + int chartype, script; + int category = _pcre_ucp_findprop(c, &chartype, &script); + + switch(ecode[1]) + { + case PT_ANY: + if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH); + break; + + case PT_LAMP: + if ((chartype == ucp_Lu || + chartype == ucp_Ll || + chartype == ucp_Lt) == (op == OP_NOTPROP)) + RRETURN(MATCH_NOMATCH); + break; + + case PT_GC: + if ((ecode[2] != category) == (op == OP_PROP)) + RRETURN(MATCH_NOMATCH); + break; + + case PT_PC: + if ((ecode[2] != chartype) == (op == OP_PROP)) + RRETURN(MATCH_NOMATCH); + break; + + case PT_SC: + if ((ecode[2] != script) == (op == OP_PROP)) + RRETURN(MATCH_NOMATCH); + break; + + default: + RRETURN(PCRE_ERROR_INTERNAL); + break; + } + + ecode += 3; + } + break; + + /* Match an extended Unicode sequence. We will get here only if the support + is in the binary; otherwise a compile-time error occurs. */ + + case OP_EXTUNI: + if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINCTEST(c, eptr); + { + int chartype, script; + int category = _pcre_ucp_findprop(c, &chartype, &script); + if (category == ucp_M) RRETURN(MATCH_NOMATCH); + while (eptr < md->end_subject) + { + int len = 1; + if (!utf8) c = *eptr; else + { + GETCHARLEN(c, eptr, len); + } + category = _pcre_ucp_findprop(c, &chartype, &script); + if (category != ucp_M) break; + eptr += len; + } + } + ecode++; + break; +#endif + + + /* Match a back reference, possibly repeatedly. Look past the end of the + item to see if there is repeat information following. The code is similar + to that for character classes, but repeated for efficiency. Then obey + similar code to character type repeats - written out again for speed. + However, if the referenced string is the empty string, always treat + it as matched, any number of times (otherwise there could be infinite + loops). */ + + case OP_REF: + { + offset = GET2(ecode, 1) << 1; /* Doubled ref number */ + ecode += 3; /* Advance past item */ + + /* If the reference is unset, set the length to be longer than the amount + of subject left; this ensures that every attempt at a match fails. We + can't just fail here, because of the possibility of quantifiers with zero + minima. */ + + length = (offset >= offset_top || md->offset_vector[offset] < 0)? + md->end_subject - eptr + 1 : + md->offset_vector[offset+1] - md->offset_vector[offset]; + + /* Set up for repetition, or handle the non-repeated case */ + + switch (*ecode) + { + case OP_CRSTAR: + case OP_CRMINSTAR: + case OP_CRPLUS: + case OP_CRMINPLUS: + case OP_CRQUERY: + case OP_CRMINQUERY: + c = *ecode++ - OP_CRSTAR; + minimize = (c & 1) != 0; + min = rep_min[c]; /* Pick up values from tables; */ + max = rep_max[c]; /* zero for max => infinity */ + if (max == 0) max = INT_MAX; + break; + + case OP_CRRANGE: + case OP_CRMINRANGE: + minimize = (*ecode == OP_CRMINRANGE); + min = GET2(ecode, 1); + max = GET2(ecode, 3); + if (max == 0) max = INT_MAX; + ecode += 5; + break; + + default: /* No repeat follows */ + if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH); + eptr += length; + continue; /* With the main loop */ + } + + /* If the length of the reference is zero, just continue with the + main loop. */ + + if (length == 0) continue; + + /* First, ensure the minimum number of matches are present. We get back + the length of the reference string explicitly rather than passing the + address of eptr, so that eptr can be a register variable. */ + + for (i = 1; i <= min; i++) + { + if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH); + eptr += length; + } + + /* If min = max, continue at the same level without recursion. + They are not both allowed to be zero. */ + + if (min == max) continue; + + /* If minimizing, keep trying and advancing the pointer */ + + if (minimize) + { + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max || !match_ref(offset, eptr, length, md, ims)) + RRETURN(MATCH_NOMATCH); + eptr += length; + } + /* Control never gets here */ + } + + /* If maximizing, find the longest string and work backwards */ + + else + { + pp = eptr; + for (i = min; i < max; i++) + { + if (!match_ref(offset, eptr, length, md, ims)) break; + eptr += length; + } + while (eptr >= pp) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr -= length; + } + RRETURN(MATCH_NOMATCH); + } + } + /* Control never gets here */ + + + + /* Match a bit-mapped character class, possibly repeatedly. This op code is + used when all the characters in the class have values in the range 0-255, + and either the matching is caseful, or the characters are in the range + 0-127 when UTF-8 processing is enabled. The only difference between + OP_CLASS and OP_NCLASS occurs when a data character outside the range is + encountered. + + First, look past the end of the item to see if there is repeat information + following. Then obey similar code to character type repeats - written out + again for speed. */ + + case OP_NCLASS: + case OP_CLASS: + { + data = ecode + 1; /* Save for matching */ + ecode += 33; /* Advance past the item */ + + switch (*ecode) + { + case OP_CRSTAR: + case OP_CRMINSTAR: + case OP_CRPLUS: + case OP_CRMINPLUS: + case OP_CRQUERY: + case OP_CRMINQUERY: + c = *ecode++ - OP_CRSTAR; + minimize = (c & 1) != 0; + min = rep_min[c]; /* Pick up values from tables; */ + max = rep_max[c]; /* zero for max => infinity */ + if (max == 0) max = INT_MAX; + break; + + case OP_CRRANGE: + case OP_CRMINRANGE: + minimize = (*ecode == OP_CRMINRANGE); + min = GET2(ecode, 1); + max = GET2(ecode, 3); + if (max == 0) max = INT_MAX; + ecode += 5; + break; + + default: /* No repeat follows */ + min = max = 1; + break; + } + + /* First, ensure the minimum number of matches are present. */ + +#ifdef SUPPORT_UTF8 + /* UTF-8 mode */ + if (utf8) + { + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINC(c, eptr); + if (c > 255) + { + if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); + } + else + { + if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); + } + } + } + else +#endif + /* Not UTF-8 mode */ + { + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + c = *eptr++; + if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); + } + } + + /* If max == min we can continue with the main loop without the + need to recurse. */ + + if (min == max) continue; + + /* If minimizing, keep testing the rest of the expression and advancing + the pointer while it matches the class. */ + + if (minimize) + { +#ifdef SUPPORT_UTF8 + /* UTF-8 mode */ + if (utf8) + { + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINC(c, eptr); + if (c > 255) + { + if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); + } + else + { + if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); + } + } + } + else +#endif + /* Not UTF-8 mode */ + { + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + c = *eptr++; + if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); + } + } + /* Control never gets here */ + } + + /* If maximizing, find the longest possible run, then work backwards. */ + + else + { + pp = eptr; + +#ifdef SUPPORT_UTF8 + /* UTF-8 mode */ + if (utf8) + { + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) break; + GETCHARLEN(c, eptr, len); + if (c > 255) + { + if (op == OP_CLASS) break; + } + else + { + if ((data[c/8] & (1 << (c&7))) == 0) break; + } + eptr += len; + } + for (;;) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (eptr-- == pp) break; /* Stop if tried at original pos */ + BACKCHAR(eptr); + } + } + else +#endif + /* Not UTF-8 mode */ + { + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) break; + c = *eptr; + if ((data[c/8] & (1 << (c&7))) == 0) break; + eptr++; + } + while (eptr >= pp) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr--; + } + } + + RRETURN(MATCH_NOMATCH); + } + } + /* Control never gets here */ + + + /* Match an extended character class. This opcode is encountered only + in UTF-8 mode, because that's the only time it is compiled. */ + +#ifdef SUPPORT_UTF8 + case OP_XCLASS: + { + data = ecode + 1 + LINK_SIZE; /* Save for matching */ + ecode += GET(ecode, 1); /* Advance past the item */ + + switch (*ecode) + { + case OP_CRSTAR: + case OP_CRMINSTAR: + case OP_CRPLUS: + case OP_CRMINPLUS: + case OP_CRQUERY: + case OP_CRMINQUERY: + c = *ecode++ - OP_CRSTAR; + minimize = (c & 1) != 0; + min = rep_min[c]; /* Pick up values from tables; */ + max = rep_max[c]; /* zero for max => infinity */ + if (max == 0) max = INT_MAX; + break; + + case OP_CRRANGE: + case OP_CRMINRANGE: + minimize = (*ecode == OP_CRMINRANGE); + min = GET2(ecode, 1); + max = GET2(ecode, 3); + if (max == 0) max = INT_MAX; + ecode += 5; + break; + + default: /* No repeat follows */ + min = max = 1; + break; + } + + /* First, ensure the minimum number of matches are present. */ + + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINC(c, eptr); + if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH); + } + + /* If max == min we can continue with the main loop without the + need to recurse. */ + + if (min == max) continue; + + /* If minimizing, keep testing the rest of the expression and advancing + the pointer while it matches the class. */ + + if (minimize) + { + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINC(c, eptr); + if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + } + + /* If maximizing, find the longest possible run, then work backwards. */ + + else + { + pp = eptr; + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) break; + GETCHARLEN(c, eptr, len); + if (!_pcre_xclass(c, data)) break; + eptr += len; + } + for(;;) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (eptr-- == pp) break; /* Stop if tried at original pos */ + BACKCHAR(eptr) + } + RRETURN(MATCH_NOMATCH); + } + + /* Control never gets here */ + } +#endif /* End of XCLASS */ + + /* Match a single character, casefully */ + + case OP_CHAR: +#ifdef SUPPORT_UTF8 + if (utf8) + { + length = 1; + ecode++; + GETCHARLEN(fc, ecode, length); + if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); + while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH); + } + else +#endif + + /* Non-UTF-8 mode */ + { + if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH); + if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH); + ecode += 2; + } + break; + + /* Match a single character, caselessly */ + + case OP_CHARNC: +#ifdef SUPPORT_UTF8 + if (utf8) + { + length = 1; + ecode++; + GETCHARLEN(fc, ecode, length); + + if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); + + /* If the pattern character's value is < 128, we have only one byte, and + can use the fast lookup table. */ + + if (fc < 128) + { + if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); + } + + /* Otherwise we must pick up the subject character */ + + else + { + int dc; + GETCHARINC(dc, eptr); + ecode += length; + + /* If we have Unicode property support, we can use it to test the other + case of the character, if there is one. */ + + if (fc != dc) + { +#ifdef SUPPORT_UCP + if (dc != _pcre_ucp_othercase(fc)) +#endif + RRETURN(MATCH_NOMATCH); + } + } + } + else +#endif /* SUPPORT_UTF8 */ + + /* Non-UTF-8 mode */ + { + if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH); + if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); + ecode += 2; + } + break; + + /* Match a single character repeatedly; different opcodes share code. */ + + case OP_EXACT: + min = max = GET2(ecode, 1); + ecode += 3; + goto REPEATCHAR; + + case OP_UPTO: + case OP_MINUPTO: + min = 0; + max = GET2(ecode, 1); + minimize = *ecode == OP_MINUPTO; + ecode += 3; + goto REPEATCHAR; + + case OP_STAR: + case OP_MINSTAR: + case OP_PLUS: + case OP_MINPLUS: + case OP_QUERY: + case OP_MINQUERY: + c = *ecode++ - OP_STAR; + minimize = (c & 1) != 0; + min = rep_min[c]; /* Pick up values from tables; */ + max = rep_max[c]; /* zero for max => infinity */ + if (max == 0) max = INT_MAX; + + /* Common code for all repeated single-character matches. We can give + up quickly if there are fewer than the minimum number of characters left in + the subject. */ + + REPEATCHAR: +#ifdef SUPPORT_UTF8 + if (utf8) + { + length = 1; + charptr = ecode; + GETCHARLEN(fc, ecode, length); + if (min * length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); + ecode += length; + + /* Handle multibyte character matching specially here. There is + support for caseless matching if UCP support is present. */ + + if (length > 1) + { + int oclength = 0; + uschar occhars[8]; + +#ifdef SUPPORT_UCP + int othercase; + if ((ims & PCRE_CASELESS) != 0 && + (othercase = _pcre_ucp_othercase(fc)) >= 0 && + othercase >= 0) + oclength = _pcre_ord2utf8(othercase, occhars); +#endif /* SUPPORT_UCP */ + + for (i = 1; i <= min; i++) + { + if (memcmp(eptr, charptr, length) == 0) eptr += length; + /* Need braces because of following else */ + else if (oclength == 0) { RRETURN(MATCH_NOMATCH); } + else + { + if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH); + eptr += oclength; + } + } + + if (min == max) continue; + + if (minimize) + { + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + if (memcmp(eptr, charptr, length) == 0) eptr += length; + /* Need braces because of following else */ + else if (oclength == 0) { RRETURN(MATCH_NOMATCH); } + else + { + if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH); + eptr += oclength; + } + } + /* Control never gets here */ + } + else + { + pp = eptr; + for (i = min; i < max; i++) + { + if (eptr > md->end_subject - length) break; + if (memcmp(eptr, charptr, length) == 0) eptr += length; + else if (oclength == 0) break; + else + { + if (memcmp(eptr, occhars, oclength) != 0) break; + eptr += oclength; + } + } + while (eptr >= pp) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr -= length; + } + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + } + + /* If the length of a UTF-8 character is 1, we fall through here, and + obey the code as for non-UTF-8 characters below, though in this case the + value of fc will always be < 128. */ + } + else +#endif /* SUPPORT_UTF8 */ + + /* When not in UTF-8 mode, load a single-byte character. */ + { + if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); + fc = *ecode++; + } + + /* The value of fc at this point is always less than 256, though we may or + may not be in UTF-8 mode. The code is duplicated for the caseless and + caseful cases, for speed, since matching characters is likely to be quite + common. First, ensure the minimum number of matches are present. If min = + max, continue at the same level without recursing. Otherwise, if + minimizing, keep trying the rest of the expression and advancing one + matching character if failing, up to the maximum. Alternatively, if + maximizing, find the maximum number of characters and work backwards. */ + + DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max, + max, eptr)); + + if ((ims & PCRE_CASELESS) != 0) + { + fc = md->lcc[fc]; + for (i = 1; i <= min; i++) + if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); + if (min == max) continue; + if (minimize) + { + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max || eptr >= md->end_subject || + fc != md->lcc[*eptr++]) + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + } + else + { + pp = eptr; + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break; + eptr++; + } + while (eptr >= pp) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + eptr--; + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + } + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + } + + /* Caseful comparisons (includes all multi-byte characters) */ + + else + { + for (i = 1; i <= min; i++) if (fc != *eptr++) RRETURN(MATCH_NOMATCH); + if (min == max) continue; + if (minimize) + { + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max || eptr >= md->end_subject || fc != *eptr++) + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + } + else + { + pp = eptr; + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject || fc != *eptr) break; + eptr++; + } + while (eptr >= pp) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + eptr--; + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + } + RRETURN(MATCH_NOMATCH); + } + } + /* Control never gets here */ + + /* Match a negated single one-byte character. The character we are + checking can be multibyte. */ + + case OP_NOT: + if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + ecode++; + GETCHARINCTEST(c, eptr); + if ((ims & PCRE_CASELESS) != 0) + { +#ifdef SUPPORT_UTF8 + if (c < 256) +#endif + c = md->lcc[c]; + if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH); + } + else + { + if (*ecode++ == c) RRETURN(MATCH_NOMATCH); + } + break; + + /* Match a negated single one-byte character repeatedly. This is almost a + repeat of the code for a repeated single character, but I haven't found a + nice way of commoning these up that doesn't require a test of the + positive/negative option for each character match. Maybe that wouldn't add + very much to the time taken, but character matching *is* what this is all + about... */ + + case OP_NOTEXACT: + min = max = GET2(ecode, 1); + ecode += 3; + goto REPEATNOTCHAR; + + case OP_NOTUPTO: + case OP_NOTMINUPTO: + min = 0; + max = GET2(ecode, 1); + minimize = *ecode == OP_NOTMINUPTO; + ecode += 3; + goto REPEATNOTCHAR; + + case OP_NOTSTAR: + case OP_NOTMINSTAR: + case OP_NOTPLUS: + case OP_NOTMINPLUS: + case OP_NOTQUERY: + case OP_NOTMINQUERY: + c = *ecode++ - OP_NOTSTAR; + minimize = (c & 1) != 0; + min = rep_min[c]; /* Pick up values from tables; */ + max = rep_max[c]; /* zero for max => infinity */ + if (max == 0) max = INT_MAX; + + /* Common code for all repeated single-byte matches. We can give up quickly + if there are fewer than the minimum number of bytes left in the + subject. */ + + REPEATNOTCHAR: + if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); + fc = *ecode++; + + /* The code is duplicated for the caseless and caseful cases, for speed, + since matching characters is likely to be quite common. First, ensure the + minimum number of matches are present. If min = max, continue at the same + level without recursing. Otherwise, if minimizing, keep trying the rest of + the expression and advancing one matching character if failing, up to the + maximum. Alternatively, if maximizing, find the maximum number of + characters and work backwards. */ + + DPRINTF(("negative matching %c{%d,%d} against subject %.*s\n", fc, min, max, + max, eptr)); + + if ((ims & PCRE_CASELESS) != 0) + { + fc = md->lcc[fc]; + +#ifdef SUPPORT_UTF8 + /* UTF-8 mode */ + if (utf8) + { + register int d; + for (i = 1; i <= min; i++) + { + GETCHARINC(d, eptr); + if (d < 256) d = md->lcc[d]; + if (fc == d) RRETURN(MATCH_NOMATCH); + } + } + else +#endif + + /* Not UTF-8 mode */ + { + for (i = 1; i <= min; i++) + if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); + } + + if (min == max) continue; + + if (minimize) + { +#ifdef SUPPORT_UTF8 + /* UTF-8 mode */ + if (utf8) + { + register int d; + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + GETCHARINC(d, eptr); + if (d < 256) d = md->lcc[d]; + if (fi >= max || eptr >= md->end_subject || fc == d) + RRETURN(MATCH_NOMATCH); + } + } + else +#endif + /* Not UTF-8 mode */ + { + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++]) + RRETURN(MATCH_NOMATCH); + } + } + /* Control never gets here */ + } + + /* Maximize case */ + + else + { + pp = eptr; + +#ifdef SUPPORT_UTF8 + /* UTF-8 mode */ + if (utf8) + { + register int d; + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) break; + GETCHARLEN(d, eptr, len); + if (d < 256) d = md->lcc[d]; + if (fc == d) break; + eptr += len; + } + for(;;) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (eptr-- == pp) break; /* Stop if tried at original pos */ + BACKCHAR(eptr); + } + } + else +#endif + /* Not UTF-8 mode */ + { + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break; + eptr++; + } + while (eptr >= pp) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr--; + } + } + + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + } + + /* Caseful comparisons */ + + else + { +#ifdef SUPPORT_UTF8 + /* UTF-8 mode */ + if (utf8) + { + register int d; + for (i = 1; i <= min; i++) + { + GETCHARINC(d, eptr); + if (fc == d) RRETURN(MATCH_NOMATCH); + } + } + else +#endif + /* Not UTF-8 mode */ + { + for (i = 1; i <= min; i++) + if (fc == *eptr++) RRETURN(MATCH_NOMATCH); + } + + if (min == max) continue; + + if (minimize) + { +#ifdef SUPPORT_UTF8 + /* UTF-8 mode */ + if (utf8) + { + register int d; + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + GETCHARINC(d, eptr); + if (fi >= max || eptr >= md->end_subject || fc == d) + RRETURN(MATCH_NOMATCH); + } + } + else +#endif + /* Not UTF-8 mode */ + { + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max || eptr >= md->end_subject || fc == *eptr++) + RRETURN(MATCH_NOMATCH); + } + } + /* Control never gets here */ + } + + /* Maximize case */ + + else + { + pp = eptr; + +#ifdef SUPPORT_UTF8 + /* UTF-8 mode */ + if (utf8) + { + register int d; + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) break; + GETCHARLEN(d, eptr, len); + if (fc == d) break; + eptr += len; + } + for(;;) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (eptr-- == pp) break; /* Stop if tried at original pos */ + BACKCHAR(eptr); + } + } + else +#endif + /* Not UTF-8 mode */ + { + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject || fc == *eptr) break; + eptr++; + } + while (eptr >= pp) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr--; + } + } + + RRETURN(MATCH_NOMATCH); + } + } + /* Control never gets here */ + + /* Match a single character type repeatedly; several different opcodes + share code. This is very similar to the code for single characters, but we + repeat it in the interests of efficiency. */ + + case OP_TYPEEXACT: + min = max = GET2(ecode, 1); + minimize = TRUE; + ecode += 3; + goto REPEATTYPE; + + case OP_TYPEUPTO: + case OP_TYPEMINUPTO: + min = 0; + max = GET2(ecode, 1); + minimize = *ecode == OP_TYPEMINUPTO; + ecode += 3; + goto REPEATTYPE; + + case OP_TYPESTAR: + case OP_TYPEMINSTAR: + case OP_TYPEPLUS: + case OP_TYPEMINPLUS: + case OP_TYPEQUERY: + case OP_TYPEMINQUERY: + c = *ecode++ - OP_TYPESTAR; + minimize = (c & 1) != 0; + min = rep_min[c]; /* Pick up values from tables; */ + max = rep_max[c]; /* zero for max => infinity */ + if (max == 0) max = INT_MAX; + + /* Common code for all repeated single character type matches. Note that + in UTF-8 mode, '.' matches a character of any length, but for the other + character types, the valid characters are all one-byte long. */ + + REPEATTYPE: + ctype = *ecode++; /* Code for the character type */ + +#ifdef SUPPORT_UCP + if (ctype == OP_PROP || ctype == OP_NOTPROP) + { + prop_fail_result = ctype == OP_NOTPROP; + prop_type = *ecode++; + prop_value = *ecode++; + } + else prop_type = -1; +#endif + + /* First, ensure the minimum number of matches are present. Use inline + code for maximizing the speed, and do the type test once at the start + (i.e. keep it out of the loop). Also we can test that there are at least + the minimum number of bytes before we start. This isn't as effective in + UTF-8 mode, but it does no harm. Separate the UTF-8 code completely as that + is tidier. Also separate the UCP code, which can be the same for both UTF-8 + and single-bytes. */ + + if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); + if (min > 0) + { +#ifdef SUPPORT_UCP + if (prop_type >= 0) + { + switch(prop_type) + { + case PT_ANY: + if (prop_fail_result) RRETURN(MATCH_NOMATCH); + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINC(c, eptr); + } + break; + + case PT_LAMP: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINC(c, eptr); + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if ((prop_chartype == ucp_Lu || + prop_chartype == ucp_Ll || + prop_chartype == ucp_Lt) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + case PT_GC: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINC(c, eptr); + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if ((prop_category == prop_value) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + case PT_PC: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINC(c, eptr); + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if ((prop_chartype == prop_value) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + case PT_SC: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINC(c, eptr); + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if ((prop_script == prop_value) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + default: + RRETURN(PCRE_ERROR_INTERNAL); + break; + } + } + + /* Match extended Unicode sequences. We will get here only if the + support is in the binary; otherwise a compile-time error occurs. */ + + else if (ctype == OP_EXTUNI) + { + for (i = 1; i <= min; i++) + { + GETCHARINCTEST(c, eptr); + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH); + while (eptr < md->end_subject) + { + int len = 1; + if (!utf8) c = *eptr; else + { + GETCHARLEN(c, eptr, len); + } + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if (prop_category != ucp_M) break; + eptr += len; + } + } + } + + else +#endif /* SUPPORT_UCP */ + +/* Handle all other cases when the coding is UTF-8 */ + +#ifdef SUPPORT_UTF8 + if (utf8) switch(ctype) + { + case OP_ANY: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject || + ((ims & PCRE_DOTALL) == 0 && + eptr <= md->end_subject - md->nllen && + IS_NEWLINE(eptr))) + RRETURN(MATCH_NOMATCH); + eptr++; + while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; + } + break; + + case OP_ANYBYTE: + eptr += min; + break; + + case OP_NOT_DIGIT: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINC(c, eptr); + if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) + RRETURN(MATCH_NOMATCH); + } + break; + + case OP_DIGIT: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject || + *eptr >= 128 || (md->ctypes[*eptr++] & ctype_digit) == 0) + RRETURN(MATCH_NOMATCH); + /* No need to skip more bytes - we know it's a 1-byte character */ + } + break; + + case OP_NOT_WHITESPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject || + (*eptr < 128 && (md->ctypes[*eptr++] & ctype_space) != 0)) + RRETURN(MATCH_NOMATCH); + while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; + } + break; + + case OP_WHITESPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject || + *eptr >= 128 || (md->ctypes[*eptr++] & ctype_space) == 0) + RRETURN(MATCH_NOMATCH); + /* No need to skip more bytes - we know it's a 1-byte character */ + } + break; + + case OP_NOT_WORDCHAR: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject || + (*eptr < 128 && (md->ctypes[*eptr++] & ctype_word) != 0)) + RRETURN(MATCH_NOMATCH); + while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; + } + break; + + case OP_WORDCHAR: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject || + *eptr >= 128 || (md->ctypes[*eptr++] & ctype_word) == 0) + RRETURN(MATCH_NOMATCH); + /* No need to skip more bytes - we know it's a 1-byte character */ + } + break; + + default: + RRETURN(PCRE_ERROR_INTERNAL); + } /* End switch(ctype) */ + + else +#endif /* SUPPORT_UTF8 */ + + /* Code for the non-UTF-8 case for minimum matching of operators other + than OP_PROP and OP_NOTPROP. */ + + switch(ctype) + { + case OP_ANY: + if ((ims & PCRE_DOTALL) == 0) + { + for (i = 1; i <= min; i++) + { + if (eptr <= md->end_subject - md->nllen && IS_NEWLINE(eptr)) + RRETURN(MATCH_NOMATCH); + eptr++; + } + } + else eptr += min; + break; + + case OP_ANYBYTE: + eptr += min; + break; + + case OP_NOT_DIGIT: + for (i = 1; i <= min; i++) + if ((md->ctypes[*eptr++] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_DIGIT: + for (i = 1; i <= min; i++) + if ((md->ctypes[*eptr++] & ctype_digit) == 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_NOT_WHITESPACE: + for (i = 1; i <= min; i++) + if ((md->ctypes[*eptr++] & ctype_space) != 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_WHITESPACE: + for (i = 1; i <= min; i++) + if ((md->ctypes[*eptr++] & ctype_space) == 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_NOT_WORDCHAR: + for (i = 1; i <= min; i++) + if ((md->ctypes[*eptr++] & ctype_word) != 0) + RRETURN(MATCH_NOMATCH); + break; + + case OP_WORDCHAR: + for (i = 1; i <= min; i++) + if ((md->ctypes[*eptr++] & ctype_word) == 0) + RRETURN(MATCH_NOMATCH); + break; + + default: + RRETURN(PCRE_ERROR_INTERNAL); + } + } + + /* If min = max, continue at the same level without recursing */ + + if (min == max) continue; + + /* If minimizing, we have to test the rest of the pattern before each + subsequent match. Again, separate the UTF-8 case for speed, and also + separate the UCP cases. */ + + if (minimize) + { +#ifdef SUPPORT_UCP + if (prop_type >= 0) + { + switch(prop_type) + { + case PT_ANY: + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINC(c, eptr); + if (prop_fail_result) RRETURN(MATCH_NOMATCH); + } + break; + + case PT_LAMP: + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINC(c, eptr); + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if ((prop_chartype == ucp_Lu || + prop_chartype == ucp_Ll || + prop_chartype == ucp_Lt) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + case PT_GC: + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINC(c, eptr); + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if ((prop_category == prop_value) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + case PT_PC: + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINC(c, eptr); + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if ((prop_chartype == prop_value) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + case PT_SC: + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINC(c, eptr); + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if ((prop_script == prop_value) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + default: + RRETURN(PCRE_ERROR_INTERNAL); + break; + } + } + + /* Match extended Unicode sequences. We will get here only if the + support is in the binary; otherwise a compile-time error occurs. */ + + else if (ctype == OP_EXTUNI) + { + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); + GETCHARINCTEST(c, eptr); + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH); + while (eptr < md->end_subject) + { + int len = 1; + if (!utf8) c = *eptr; else + { + GETCHARLEN(c, eptr, len); + } + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if (prop_category != ucp_M) break; + eptr += len; + } + } + } + + else +#endif /* SUPPORT_UCP */ + +#ifdef SUPPORT_UTF8 + /* UTF-8 mode */ + if (utf8) + { + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max || eptr >= md->end_subject || + (ctype == OP_ANY && (ims & PCRE_DOTALL) == 0 && + eptr <= md->end_subject - md->nllen && IS_NEWLINE(eptr))) + RRETURN(MATCH_NOMATCH); + + GETCHARINC(c, eptr); + switch(ctype) + { + case OP_ANY: /* This is the DOTALL case */ + break; + + case OP_ANYBYTE: + break; + + case OP_NOT_DIGIT: + if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) + RRETURN(MATCH_NOMATCH); + break; + + case OP_DIGIT: + if (c >= 256 || (md->ctypes[c] & ctype_digit) == 0) + RRETURN(MATCH_NOMATCH); + break; + + case OP_NOT_WHITESPACE: + if (c < 256 && (md->ctypes[c] & ctype_space) != 0) + RRETURN(MATCH_NOMATCH); + break; + + case OP_WHITESPACE: + if (c >= 256 || (md->ctypes[c] & ctype_space) == 0) + RRETURN(MATCH_NOMATCH); + break; + + case OP_NOT_WORDCHAR: + if (c < 256 && (md->ctypes[c] & ctype_word) != 0) + RRETURN(MATCH_NOMATCH); + break; + + case OP_WORDCHAR: + if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) + RRETURN(MATCH_NOMATCH); + break; + + default: + RRETURN(PCRE_ERROR_INTERNAL); + } + } + } + else +#endif + /* Not UTF-8 mode */ + { + for (fi = min;; fi++) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max || eptr >= md->end_subject || + ((ims & PCRE_DOTALL) == 0 && + eptr <= md->end_subject - md->nllen && IS_NEWLINE(eptr))) + RRETURN(MATCH_NOMATCH); + + c = *eptr++; + switch(ctype) + { + case OP_ANY: /* This is the DOTALL case */ + break; + + case OP_ANYBYTE: + break; + + case OP_NOT_DIGIT: + if ((md->ctypes[c] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_DIGIT: + if ((md->ctypes[c] & ctype_digit) == 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_NOT_WHITESPACE: + if ((md->ctypes[c] & ctype_space) != 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_WHITESPACE: + if ((md->ctypes[c] & ctype_space) == 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_NOT_WORDCHAR: + if ((md->ctypes[c] & ctype_word) != 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_WORDCHAR: + if ((md->ctypes[c] & ctype_word) == 0) RRETURN(MATCH_NOMATCH); + break; + + default: + RRETURN(PCRE_ERROR_INTERNAL); + } + } + } + /* Control never gets here */ + } + + /* If maximizing it is worth using inline code for speed, doing the type + test once at the start (i.e. keep it out of the loop). Again, keep the + UTF-8 and UCP stuff separate. */ + + else + { + pp = eptr; /* Remember where we started */ + +#ifdef SUPPORT_UCP + if (prop_type >= 0) + { + switch(prop_type) + { + case PT_ANY: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) break; + GETCHARLEN(c, eptr, len); + if (prop_fail_result) break; + eptr+= len; + } + break; + + case PT_LAMP: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) break; + GETCHARLEN(c, eptr, len); + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if ((prop_chartype == ucp_Lu || + prop_chartype == ucp_Ll || + prop_chartype == ucp_Lt) == prop_fail_result) + break; + eptr+= len; + } + break; + + case PT_GC: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) break; + GETCHARLEN(c, eptr, len); + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if ((prop_category == prop_value) == prop_fail_result) + break; + eptr+= len; + } + break; + + case PT_PC: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) break; + GETCHARLEN(c, eptr, len); + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if ((prop_chartype == prop_value) == prop_fail_result) + break; + eptr+= len; + } + break; + + case PT_SC: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) break; + GETCHARLEN(c, eptr, len); + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if ((prop_script == prop_value) == prop_fail_result) + break; + eptr+= len; + } + break; + } + + /* eptr is now past the end of the maximum run */ + + for(;;) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (eptr-- == pp) break; /* Stop if tried at original pos */ + BACKCHAR(eptr); + } + } + + /* Match extended Unicode sequences. We will get here only if the + support is in the binary; otherwise a compile-time error occurs. */ + + else if (ctype == OP_EXTUNI) + { + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) break; + GETCHARINCTEST(c, eptr); + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if (prop_category == ucp_M) break; + while (eptr < md->end_subject) + { + int len = 1; + if (!utf8) c = *eptr; else + { + GETCHARLEN(c, eptr, len); + } + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if (prop_category != ucp_M) break; + eptr += len; + } + } + + /* eptr is now past the end of the maximum run */ + + for(;;) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (eptr-- == pp) break; /* Stop if tried at original pos */ + for (;;) /* Move back over one extended */ + { + int len = 1; + BACKCHAR(eptr); + if (!utf8) c = *eptr; else + { + GETCHARLEN(c, eptr, len); + } + prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); + if (prop_category != ucp_M) break; + eptr--; + } + } + } + + else +#endif /* SUPPORT_UCP */ + +#ifdef SUPPORT_UTF8 + /* UTF-8 mode */ + + if (utf8) + { + switch(ctype) + { + case OP_ANY: + + /* Special code is required for UTF8, but when the maximum is + unlimited we don't need it, so we repeat the non-UTF8 code. This is + probably worth it, because .* is quite a common idiom. */ + + if (max < INT_MAX) + { + if ((ims & PCRE_DOTALL) == 0) + { + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject || + (eptr <= md->end_subject - md->nllen && IS_NEWLINE(eptr))) + break; + eptr++; + while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; + } + } + else + { + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) break; + eptr++; + while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; + } + } + } + + /* Handle unlimited UTF-8 repeat */ + + else + { + if ((ims & PCRE_DOTALL) == 0) + { + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject || + (eptr <= md->end_subject - md->nllen && IS_NEWLINE(eptr))) + break; + eptr++; + } + break; + } + else + { + c = max - min; + if (c > md->end_subject - eptr) c = md->end_subject - eptr; + eptr += c; + } + } + break; + + /* The byte case is the same as non-UTF8 */ + + case OP_ANYBYTE: + c = max - min; + if (c > md->end_subject - eptr) c = md->end_subject - eptr; + eptr += c; + break; + + case OP_NOT_DIGIT: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) break; + GETCHARLEN(c, eptr, len); + if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) break; + eptr+= len; + } + break; + + case OP_DIGIT: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) break; + GETCHARLEN(c, eptr, len); + if (c >= 256 ||(md->ctypes[c] & ctype_digit) == 0) break; + eptr+= len; + } + break; + + case OP_NOT_WHITESPACE: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) break; + GETCHARLEN(c, eptr, len); + if (c < 256 && (md->ctypes[c] & ctype_space) != 0) break; + eptr+= len; + } + break; + + case OP_WHITESPACE: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) break; + GETCHARLEN(c, eptr, len); + if (c >= 256 ||(md->ctypes[c] & ctype_space) == 0) break; + eptr+= len; + } + break; + + case OP_NOT_WORDCHAR: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) break; + GETCHARLEN(c, eptr, len); + if (c < 256 && (md->ctypes[c] & ctype_word) != 0) break; + eptr+= len; + } + break; + + case OP_WORDCHAR: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) break; + GETCHARLEN(c, eptr, len); + if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) break; + eptr+= len; + } + break; + + default: + RRETURN(PCRE_ERROR_INTERNAL); + } + + /* eptr is now past the end of the maximum run */ + + for(;;) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (eptr-- == pp) break; /* Stop if tried at original pos */ + BACKCHAR(eptr); + } + } + else +#endif + + /* Not UTF-8 mode */ + { + switch(ctype) + { + case OP_ANY: + if ((ims & PCRE_DOTALL) == 0) + { + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject || + (eptr <= md->end_subject - md->nllen && IS_NEWLINE(eptr))) + break; + eptr++; + } + break; + } + /* For DOTALL case, fall through and treat as \C */ + + case OP_ANYBYTE: + c = max - min; + if (c > md->end_subject - eptr) c = md->end_subject - eptr; + eptr += c; + break; + + case OP_NOT_DIGIT: + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject || (md->ctypes[*eptr] & ctype_digit) != 0) + break; + eptr++; + } + break; + + case OP_DIGIT: + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject || (md->ctypes[*eptr] & ctype_digit) == 0) + break; + eptr++; + } + break; + + case OP_NOT_WHITESPACE: + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject || (md->ctypes[*eptr] & ctype_space) != 0) + break; + eptr++; + } + break; + + case OP_WHITESPACE: + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject || (md->ctypes[*eptr] & ctype_space) == 0) + break; + eptr++; + } + break; + + case OP_NOT_WORDCHAR: + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject || (md->ctypes[*eptr] & ctype_word) != 0) + break; + eptr++; + } + break; + + case OP_WORDCHAR: + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject || (md->ctypes[*eptr] & ctype_word) == 0) + break; + eptr++; + } + break; + + default: + RRETURN(PCRE_ERROR_INTERNAL); + } + + /* eptr is now past the end of the maximum run */ + + while (eptr >= pp) + { + RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0); + eptr--; + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + } + } + + /* Get here if we can't make it match with any permitted repetitions */ + + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + /* There's been some horrible disaster. Since all codes > OP_BRA are + for capturing brackets, and there shouldn't be any gaps between 0 and + OP_BRA, arrival here can only mean there is something seriously wrong + in the code above or the OP_xxx definitions. */ + + default: + DPRINTF(("Unknown opcode %d\n", *ecode)); + RRETURN(PCRE_ERROR_UNKNOWN_NODE); + } + + /* Do not stick any code in here without much thought; it is assumed + that "continue" in the code above comes out to here to repeat the main + loop. */ + + } /* End of main loop */ +/* Control never reaches here */ +} + + +/*************************************************************************** +**************************************************************************** + RECURSION IN THE match() FUNCTION + +Undefine all the macros that were defined above to handle this. */ + +#ifdef NO_RECURSE +#undef eptr +#undef ecode +#undef offset_top +#undef ims +#undef eptrb +#undef flags + +#undef callpat +#undef charptr +#undef data +#undef next +#undef pp +#undef prev +#undef saved_eptr + +#undef new_recursive + +#undef cur_is_word +#undef condition +#undef minimize +#undef prev_is_word + +#undef original_ims + +#undef ctype +#undef length +#undef max +#undef min +#undef number +#undef offset +#undef op +#undef save_capture_last +#undef save_offset1 +#undef save_offset2 +#undef save_offset3 +#undef stacksave + +#undef newptrb + +#endif + +/* These two are defined as macros in both cases */ + +#undef fc +#undef fi + +/*************************************************************************** +***************************************************************************/ + + + +/************************************************* +* Execute a Regular Expression * +*************************************************/ + +/* This function applies a compiled re to a subject string and picks out +portions of the string if it matches. Two elements in the vector are set for +each substring: the offsets to the start and end of the substring. + +Arguments: + argument_re points to the compiled expression + extra_data points to extra data or is NULL + subject points to the subject string + length length of subject string (may contain binary zeros) + start_offset where to start in the subject string + options option bits + offsets points to a vector of ints to be filled in with offsets + offsetcount the number of elements in the vector + +Returns: > 0 => success; value is the number of elements filled in + = 0 => success, but offsets is not big enough + -1 => failed to match + < -1 => some kind of unexpected problem +*/ + +PCRE_DATA_SCOPE int +pcre_exec(const pcre *argument_re, const pcre_extra *extra_data, + PCRE_SPTR subject, int length, int start_offset, int options, int *offsets, + int offsetcount) +{ +int rc, resetcount, ocount; +int first_byte = -1; +int req_byte = -1; +int req_byte2 = -1; +int newline; +unsigned long int ims; +BOOL using_temporary_offsets = FALSE; +BOOL anchored; +BOOL startline; +BOOL firstline; +BOOL first_byte_caseless = FALSE; +BOOL req_byte_caseless = FALSE; +match_data match_block; +match_data *md = &match_block; +const uschar *tables; +const uschar *start_bits = NULL; +USPTR start_match = (USPTR)subject + start_offset; +USPTR end_subject; +USPTR req_byte_ptr = start_match - 1; + +pcre_study_data internal_study; +const pcre_study_data *study; + +real_pcre internal_re; +const real_pcre *external_re = (const real_pcre *)argument_re; +const real_pcre *re = external_re; + +/* Plausibility checks */ + +if ((options & ~PUBLIC_EXEC_OPTIONS) != 0) return PCRE_ERROR_BADOPTION; +if (re == NULL || subject == NULL || + (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL; +if (offsetcount < 0) return PCRE_ERROR_BADCOUNT; + +/* Fish out the optional data from the extra_data structure, first setting +the default values. */ + +study = NULL; +md->match_limit = MATCH_LIMIT; +md->match_limit_recursion = MATCH_LIMIT_RECURSION; +md->callout_data = NULL; + +/* The table pointer is always in native byte order. */ + +tables = external_re->tables; + +if (extra_data != NULL) + { + register unsigned int flags = extra_data->flags; + if ((flags & PCRE_EXTRA_STUDY_DATA) != 0) + study = (const pcre_study_data *)extra_data->study_data; + if ((flags & PCRE_EXTRA_MATCH_LIMIT) != 0) + md->match_limit = extra_data->match_limit; + if ((flags & PCRE_EXTRA_MATCH_LIMIT_RECURSION) != 0) + md->match_limit_recursion = extra_data->match_limit_recursion; + if ((flags & PCRE_EXTRA_CALLOUT_DATA) != 0) + md->callout_data = extra_data->callout_data; + if ((flags & PCRE_EXTRA_TABLES) != 0) tables = extra_data->tables; + } + +/* If the exec call supplied NULL for tables, use the inbuilt ones. This +is a feature that makes it possible to save compiled regex and re-use them +in other programs later. */ + +if (tables == NULL) tables = _pcre_default_tables; + +/* Check that the first field in the block is the magic number. If it is not, +test for a regex that was compiled on a host of opposite endianness. If this is +the case, flipped values are put in internal_re and internal_study if there was +study data too. */ + +if (re->magic_number != MAGIC_NUMBER) + { + re = _pcre_try_flipped(re, &internal_re, study, &internal_study); + if (re == NULL) return PCRE_ERROR_BADMAGIC; + if (study != NULL) study = &internal_study; + } + +/* Set up other data */ + +anchored = ((re->options | options) & PCRE_ANCHORED) != 0; +startline = (re->options & PCRE_STARTLINE) != 0; +firstline = (re->options & PCRE_FIRSTLINE) != 0; + +/* The code starts after the real_pcre block and the capture name table. */ + +md->start_code = (const uschar *)external_re + re->name_table_offset + + re->name_count * re->name_entry_size; + +md->start_subject = (USPTR)subject; +md->start_offset = start_offset; +md->end_subject = md->start_subject + length; +end_subject = md->end_subject; + +md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0; +md->utf8 = (re->options & PCRE_UTF8) != 0; + +md->notbol = (options & PCRE_NOTBOL) != 0; +md->noteol = (options & PCRE_NOTEOL) != 0; +md->notempty = (options & PCRE_NOTEMPTY) != 0; +md->partial = (options & PCRE_PARTIAL) != 0; +md->hitend = FALSE; + +md->recursive = NULL; /* No recursion at top level */ + +md->lcc = tables + lcc_offset; +md->ctypes = tables + ctypes_offset; + +/* Handle different types of newline. The two bits give four cases. If nothing +is set at run time, whatever was used at compile time applies. */ + +switch ((((options & PCRE_NEWLINE_CRLF) == 0)? re->options : options) & + PCRE_NEWLINE_CRLF) + { + default: newline = NEWLINE; break; /* Compile-time default */ + case PCRE_NEWLINE_CR: newline = '\r'; break; + case PCRE_NEWLINE_LF: newline = '\n'; break; + case PCRE_NEWLINE_CR+ + PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break; + } + +if (newline > 255) + { + md->nllen = 2; + md->nl[0] = (newline >> 8) & 255; + md->nl[1] = newline & 255; + } +else + { + md->nllen = 1; + md->nl[0] = newline; + } + +/* Partial matching is supported only for a restricted set of regexes at the +moment. */ + +if (md->partial && (re->options & PCRE_NOPARTIAL) != 0) + return PCRE_ERROR_BADPARTIAL; + +/* Check a UTF-8 string if required. Unfortunately there's no way of passing +back the character offset. */ + +#ifdef SUPPORT_UTF8 +if (md->utf8 && (options & PCRE_NO_UTF8_CHECK) == 0) + { + if (_pcre_valid_utf8((uschar *)subject, length) >= 0) + return PCRE_ERROR_BADUTF8; + if (start_offset > 0 && start_offset < length) + { + int tb = ((uschar *)subject)[start_offset]; + if (tb > 127) + { + tb &= 0xc0; + if (tb != 0 && tb != 0xc0) return PCRE_ERROR_BADUTF8_OFFSET; + } + } + } +#endif + +/* The ims options can vary during the matching as a result of the presence +of (?ims) items in the pattern. They are kept in a local variable so that +restoring at the exit of a group is easy. */ + +ims = re->options & (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL); + +/* If the expression has got more back references than the offsets supplied can +hold, we get a temporary chunk of working store to use during the matching. +Otherwise, we can use the vector supplied, rounding down its size to a multiple +of 3. */ + +ocount = offsetcount - (offsetcount % 3); + +if (re->top_backref > 0 && re->top_backref >= ocount/3) + { + ocount = re->top_backref * 3 + 3; + md->offset_vector = (int *)(pcre_malloc)(ocount * sizeof(int)); + if (md->offset_vector == NULL) return PCRE_ERROR_NOMEMORY; + using_temporary_offsets = TRUE; + DPRINTF(("Got memory to hold back references\n")); + } +else md->offset_vector = offsets; + +md->offset_end = ocount; +md->offset_max = (2*ocount)/3; +md->offset_overflow = FALSE; +md->capture_last = -1; + +/* Compute the minimum number of offsets that we need to reset each time. Doing +this makes a huge difference to execution time when there aren't many brackets +in the pattern. */ + +resetcount = 2 + re->top_bracket * 2; +if (resetcount > offsetcount) resetcount = ocount; + +/* Reset the working variable associated with each extraction. These should +never be used unless previously set, but they get saved and restored, and so we +initialize them to avoid reading uninitialized locations. */ + +if (md->offset_vector != NULL) + { + register int *iptr = md->offset_vector + ocount; + register int *iend = iptr - resetcount/2 + 1; + while (--iptr >= iend) *iptr = -1; + } + +/* Set up the first character to match, if available. The first_byte value is +never set for an anchored regular expression, but the anchoring may be forced +at run time, so we have to test for anchoring. The first char may be unset for +an unanchored pattern, of course. If there's no first char and the pattern was +studied, there may be a bitmap of possible first characters. */ + +if (!anchored) + { + if ((re->options & PCRE_FIRSTSET) != 0) + { + first_byte = re->first_byte & 255; + if ((first_byte_caseless = ((re->first_byte & REQ_CASELESS) != 0)) == TRUE) + first_byte = md->lcc[first_byte]; + } + else + if (!startline && study != NULL && + (study->options & PCRE_STUDY_MAPPED) != 0) + start_bits = study->start_bits; + } + +/* For anchored or unanchored matches, there may be a "last known required +character" set. */ + +if ((re->options & PCRE_REQCHSET) != 0) + { + req_byte = re->req_byte & 255; + req_byte_caseless = (re->req_byte & REQ_CASELESS) != 0; + req_byte2 = (tables + fcc_offset)[req_byte]; /* case flipped */ + } + +/* Loop for handling unanchored repeated matching attempts; for anchored regexs +the loop runs just once. */ + +do + { + USPTR save_end_subject = end_subject; + + /* Reset the maximum number of extractions we might see. */ + + if (md->offset_vector != NULL) + { + register int *iptr = md->offset_vector; + register int *iend = iptr + resetcount; + while (iptr < iend) *iptr++ = -1; + } + + /* Advance to a unique first char if possible. If firstline is TRUE, the + start of the match is constrained to the first line of a multiline string. + Implement this by temporarily adjusting end_subject so that we stop scanning + at a newline. If the match fails at the newline, later code breaks this loop. + */ + + if (firstline) + { + USPTR t = start_match; + while (t <= save_end_subject - md->nllen && !IS_NEWLINE(t)) t++; + end_subject = t; + } + + /* Now test for a unique first byte */ + + if (first_byte >= 0) + { + if (first_byte_caseless) + while (start_match < end_subject && + md->lcc[*start_match] != first_byte) + start_match++; + else + while (start_match < end_subject && *start_match != first_byte) + start_match++; + } + + /* Or to just after a linebreak for a multiline match if possible */ + + else if (startline) + { + if (start_match >= md->start_subject + md->nllen + + start_offset) + { + while (start_match <= end_subject && + !IS_NEWLINE(start_match - md->nllen)) + start_match++; + } + } + + /* Or to a non-unique first char after study */ + + else if (start_bits != NULL) + { + while (start_match < end_subject) + { + register unsigned int c = *start_match; + if ((start_bits[c/8] & (1 << (c&7))) == 0) start_match++; else break; + } + } + + /* Restore fudged end_subject */ + + end_subject = save_end_subject; + +#ifdef DEBUG /* Sigh. Some compilers never learn. */ + printf(">>>> Match against: "); + pchars(start_match, end_subject - start_match, TRUE, md); + printf("\n"); +#endif + + /* If req_byte is set, we know that that character must appear in the subject + for the match to succeed. If the first character is set, req_byte must be + later in the subject; otherwise the test starts at the match point. This + optimization can save a huge amount of backtracking in patterns with nested + unlimited repeats that aren't going to match. Writing separate code for + cased/caseless versions makes it go faster, as does using an autoincrement + and backing off on a match. + + HOWEVER: when the subject string is very, very long, searching to its end can + take a long time, and give bad performance on quite ordinary patterns. This + showed up when somebody was matching /^C/ on a 32-megabyte string... so we + don't do this when the string is sufficiently long. + + ALSO: this processing is disabled when partial matching is requested. + */ + + if (req_byte >= 0 && + end_subject - start_match < REQ_BYTE_MAX && + !md->partial) + { + register USPTR p = start_match + ((first_byte >= 0)? 1 : 0); + + /* We don't need to repeat the search if we haven't yet reached the + place we found it at last time. */ + + if (p > req_byte_ptr) + { + if (req_byte_caseless) + { + while (p < end_subject) + { + register int pp = *p++; + if (pp == req_byte || pp == req_byte2) { p--; break; } + } + } + else + { + while (p < end_subject) + { + if (*p++ == req_byte) { p--; break; } + } + } + + /* If we can't find the required character, break the matching loop */ + + if (p >= end_subject) break; + + /* If we have found the required character, save the point where we + found it, so that we don't search again next time round the loop if + the start hasn't passed this character yet. */ + + req_byte_ptr = p; + } + } + + /* When a match occurs, substrings will be set for all internal extractions; + we just need to set up the whole thing as substring 0 before returning. If + there were too many extractions, set the return code to zero. In the case + where we had to get some local store to hold offsets for backreferences, copy + those back references that we can. In this case there need not be overflow + if certain parts of the pattern were not used. */ + + md->start_match = start_match; + md->match_call_count = 0; + + rc = match(start_match, md->start_code, 2, md, ims, NULL, match_isgroup, 0); + + /* When the result is no match, if the subject's first character was a + newline and the PCRE_FIRSTLINE option is set, break (which will return + PCRE_ERROR_NOMATCH). The option requests that a match occur before the first + newline in the subject. Otherwise, advance the pointer to the next character + and continue - but the continuation will actually happen only when the + pattern is not anchored. */ + + if (rc == MATCH_NOMATCH) + { + if (firstline && + start_match <= md->end_subject - md->nllen && + IS_NEWLINE(start_match)) + break; + start_match++; +#ifdef SUPPORT_UTF8 + if (md->utf8) + while(start_match < end_subject && (*start_match & 0xc0) == 0x80) + start_match++; +#endif + continue; + } + + if (rc != MATCH_MATCH) + { + DPRINTF((">>>> error: returning %d\n", rc)); + return rc; + } + + /* We have a match! Copy the offset information from temporary store if + necessary */ + + if (using_temporary_offsets) + { + if (offsetcount >= 4) + { + memcpy(offsets + 2, md->offset_vector + 2, + (offsetcount - 2) * sizeof(int)); + DPRINTF(("Copied offsets from temporary memory\n")); + } + if (md->end_offset_top > offsetcount) + md->offset_overflow = TRUE; + + DPRINTF(("Freeing temporary memory\n")); + (pcre_free)(md->offset_vector); + } + + rc = md->offset_overflow? 0 : md->end_offset_top/2; + + if (offsetcount < 2) rc = 0; else + { + offsets[0] = start_match - md->start_subject; + offsets[1] = md->end_match_ptr - md->start_subject; + } + + DPRINTF((">>>> returning %d\n", rc)); + return rc; + } + +/* This "while" is the end of the "do" above */ + +while (!anchored && start_match <= end_subject); + +if (using_temporary_offsets) + { + DPRINTF(("Freeing temporary memory\n")); + (pcre_free)(md->offset_vector); + } + +if (md->partial && md->hitend) + { + DPRINTF((">>>> returning PCRE_ERROR_PARTIAL\n")); + return PCRE_ERROR_PARTIAL; + } +else + { + DPRINTF((">>>> returning PCRE_ERROR_NOMATCH\n")); + return PCRE_ERROR_NOMATCH; + } +} + +/* End of pcre_exec.c */ ============================================================ --- pcre/pcre_fullinfo.c 65c02930182c0ca0a12b7a2cd46cce76a0f89457 +++ pcre/pcre_fullinfo.c 65c02930182c0ca0a12b7a2cd46cce76a0f89457 @@ -0,0 +1,149 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + +/*PCRE is a library of functions to support regular expressions whose syntax +and semantics are as close as possible to those of the Perl 5 language. + + Written by Philip Hazel + Copyright (c) 1997-2006 University of Cambridge + +----------------------------------------------------------------------------- +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + + * Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimer. + + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + * Neither the name of the University of Cambridge nor the names of its + contributors may be used to endorse or promote products derived from + this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE +LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR +CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +----------------------------------------------------------------------------- +*/ + + +/* This module contains the external function pcre_fullinfo(), which returns +information about a compiled pattern. */ + + +#include "pcre_internal.h" + + +/************************************************* +* Return info about compiled pattern * +*************************************************/ + +/* This is a newer "info" function which has an extensible interface so +that additional items can be added compatibly. + +Arguments: + argument_re points to compiled code + extra_data points extra data, or NULL + what what information is required + where where to put the information + +Returns: 0 if data returned, negative on error +*/ + +PCRE_DATA_SCOPE int +pcre_fullinfo(const pcre *argument_re, const pcre_extra *extra_data, int what, + void *where) +{ +real_pcre internal_re; +pcre_study_data internal_study; +const real_pcre *re = (const real_pcre *)argument_re; +const pcre_study_data *study = NULL; + +if (re == NULL || where == NULL) return PCRE_ERROR_NULL; + +if (extra_data != NULL && (extra_data->flags & PCRE_EXTRA_STUDY_DATA) != 0) + study = (const pcre_study_data *)extra_data->study_data; + +if (re->magic_number != MAGIC_NUMBER) + { + re = _pcre_try_flipped(re, &internal_re, study, &internal_study); + if (re == NULL) return PCRE_ERROR_BADMAGIC; + if (study != NULL) study = &internal_study; + } + +switch (what) + { + case PCRE_INFO_OPTIONS: + *((unsigned long int *)where) = re->options & PUBLIC_OPTIONS; + break; + + case PCRE_INFO_SIZE: + *((size_t *)where) = re->size; + break; + + case PCRE_INFO_STUDYSIZE: + *((size_t *)where) = (study == NULL)? 0 : study->size; + break; + + case PCRE_INFO_CAPTURECOUNT: + *((int *)where) = re->top_bracket; + break; + + case PCRE_INFO_BACKREFMAX: + *((int *)where) = re->top_backref; + break; + + case PCRE_INFO_FIRSTBYTE: + *((int *)where) = + ((re->options & PCRE_FIRSTSET) != 0)? re->first_byte : + ((re->options & PCRE_STARTLINE) != 0)? -1 : -2; + break; + + /* Make sure we pass back the pointer to the bit vector in the external + block, not the internal copy (with flipped integer fields). */ + + case PCRE_INFO_FIRSTTABLE: + *((const uschar **)where) = + (study != NULL && (study->options & PCRE_STUDY_MAPPED) != 0)? + ((const pcre_study_data *)extra_data->study_data)->start_bits : NULL; + break; + + case PCRE_INFO_LASTLITERAL: + *((int *)where) = + ((re->options & PCRE_REQCHSET) != 0)? re->req_byte : -1; + break; + + case PCRE_INFO_NAMEENTRYSIZE: + *((int *)where) = re->name_entry_size; + break; + + case PCRE_INFO_NAMECOUNT: + *((int *)where) = re->name_count; + break; + + case PCRE_INFO_NAMETABLE: + *((const uschar **)where) = (const uschar *)re + re->name_table_offset; + break; + + case PCRE_INFO_DEFAULT_TABLES: + *((const uschar **)where) = (const uschar *)(_pcre_default_tables); + break; + + default: return PCRE_ERROR_BADOPTION; + } + +return 0; +} + +/* End of pcre_fullinfo.c */ ============================================================ --- pcre/pcre_get.c 4dd0a6df9fa0c3a4fb9a3356ad2b9623d9febf57 +++ pcre/pcre_get.c 4dd0a6df9fa0c3a4fb9a3356ad2b9623d9febf57 @@ -0,0 +1,462 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + +/* PCRE is a library of functions to support regular expressions whose syntax +and semantics are as close as possible to those of the Perl 5 language. + + Written by Philip Hazel + Copyright (c) 1997-2006 University of Cambridge + +----------------------------------------------------------------------------- +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + + * Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimer. + + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + * Neither the name of the University of Cambridge nor the names of its + contributors may be used to endorse or promote products derived from + this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE +LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR +CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +----------------------------------------------------------------------------- +*/ + + +/* This module contains some convenience functions for extracting substrings +from the subject string after a regex match has succeeded. The original idea +for these functions came from Scott Wimer. */ + + +#include "pcre_internal.h" + + +/************************************************* +* Find number for named string * +*************************************************/ + +/* This function is used by the get_first_set() function below, as well +as being generally available. It assumes that names are unique. + +Arguments: + code the compiled regex + stringname the name whose number is required + +Returns: the number of the named parentheses, or a negative number + (PCRE_ERROR_NOSUBSTRING) if not found +*/ + +int +pcre_get_stringnumber(const pcre *code, const char *stringname) +{ +int rc; +int entrysize; +int top, bot; +uschar *nametable; + +if ((rc = pcre_fullinfo(code, NULL, PCRE_INFO_NAMECOUNT, &top)) != 0) + return rc; +if (top <= 0) return PCRE_ERROR_NOSUBSTRING; + +if ((rc = pcre_fullinfo(code, NULL, PCRE_INFO_NAMEENTRYSIZE, &entrysize)) != 0) + return rc; +if ((rc = pcre_fullinfo(code, NULL, PCRE_INFO_NAMETABLE, &nametable)) != 0) + return rc; + +bot = 0; +while (top > bot) + { + int mid = (top + bot) / 2; + uschar *entry = nametable + entrysize*mid; + int c = strcmp(stringname, (char *)(entry + 2)); + if (c == 0) return (entry[0] << 8) + entry[1]; + if (c > 0) bot = mid + 1; else top = mid; + } + +return PCRE_ERROR_NOSUBSTRING; +} + + + +/************************************************* +* Find (multiple) entries for named string * +*************************************************/ + +/* This is used by the get_first_set() function below, as well as being +generally available. It is used when duplicated names are permitted. + +Arguments: + code the compiled regex + stringname the name whose entries required + firstptr where to put the pointer to the first entry + lastptr where to put the pointer to the last entry + +Returns: the length of each entry, or a negative number + (PCRE_ERROR_NOSUBSTRING) if not found +*/ + +int +pcre_get_stringtable_entries(const pcre *code, const char *stringname, + char **firstptr, char **lastptr) +{ +int rc; +int entrysize; +int top, bot; +uschar *nametable, *lastentry; + +if ((rc = pcre_fullinfo(code, NULL, PCRE_INFO_NAMECOUNT, &top)) != 0) + return rc; +if (top <= 0) return PCRE_ERROR_NOSUBSTRING; + +if ((rc = pcre_fullinfo(code, NULL, PCRE_INFO_NAMEENTRYSIZE, &entrysize)) != 0) + return rc; +if ((rc = pcre_fullinfo(code, NULL, PCRE_INFO_NAMETABLE, &nametable)) != 0) + return rc; + +lastentry = nametable + entrysize * (top - 1); +bot = 0; +while (top > bot) + { + int mid = (top + bot) / 2; + uschar *entry = nametable + entrysize*mid; + int c = strcmp(stringname, (char *)(entry + 2)); + if (c == 0) + { + uschar *first = entry; + uschar *last = entry; + while (first > nametable) + { + if (strcmp(stringname, (char *)(first - entrysize + 2)) != 0) break; + first -= entrysize; + } + while (last < lastentry) + { + if (strcmp(stringname, (char *)(last + entrysize + 2)) != 0) break; + last += entrysize; + } + *firstptr = (char *)first; + *lastptr = (char *)last; + return entrysize; + } + if (c > 0) bot = mid + 1; else top = mid; + } + +return PCRE_ERROR_NOSUBSTRING; +} + + + +/************************************************* +* Find first set of multiple named strings * +*************************************************/ + +/* This function allows for duplicate names in the table of named substrings. +It returns the number of the first one that was set in a pattern match. + +Arguments: + code the compiled regex + stringname the name of the capturing substring + ovector the vector of matched substrings + +Returns: the number of the first that is set, + or the number of the last one if none are set, + or a negative number on error +*/ + +static int +get_first_set(const pcre *code, const char *stringname, int *ovector) +{ +const real_pcre *re = (const real_pcre *)code; +int entrysize; +char *first, *last; +uschar *entry; +if ((re->options & (PCRE_DUPNAMES | PCRE_JCHANGED)) == 0) + return pcre_get_stringnumber(code, stringname); +entrysize = pcre_get_stringtable_entries(code, stringname, &first, &last); +if (entrysize <= 0) return entrysize; +for (entry = (uschar *)first; entry <= (uschar *)last; entry += entrysize) + { + int n = (entry[0] << 8) + entry[1]; + if (ovector[n*2] >= 0) return n; + } +return (first[0] << 8) + first[1]; +} + + + + +/************************************************* +* Copy captured string to given buffer * +*************************************************/ + +/* This function copies a single captured substring into a given buffer. +Note that we use memcpy() rather than strncpy() in case there are binary zeros +in the string. + +Arguments: + subject the subject string that was matched + ovector pointer to the offsets table + stringcount the number of substrings that were captured + (i.e. the yield of the pcre_exec call, unless + that was zero, in which case it should be 1/3 + of the offset table size) + stringnumber the number of the required substring + buffer where to put the substring + size the size of the buffer + +Returns: if successful: + the length of the copied string, not including the zero + that is put on the end; can be zero + if not successful: + PCRE_ERROR_NOMEMORY (-6) buffer too small + PCRE_ERROR_NOSUBSTRING (-7) no such captured substring +*/ + +int +pcre_copy_substring(const char *subject, int *ovector, int stringcount, + int stringnumber, char *buffer, int size) +{ +int yield; +if (stringnumber < 0 || stringnumber >= stringcount) + return PCRE_ERROR_NOSUBSTRING; +stringnumber *= 2; +yield = ovector[stringnumber+1] - ovector[stringnumber]; +if (size < yield + 1) return PCRE_ERROR_NOMEMORY; +memcpy(buffer, subject + ovector[stringnumber], yield); +buffer[yield] = 0; +return yield; +} + + + +/************************************************* +* Copy named captured string to given buffer * +*************************************************/ + +/* This function copies a single captured substring into a given buffer, +identifying it by name. If the regex permits duplicate names, the first +substring that is set is chosen. + +Arguments: + code the compiled regex + subject the subject string that was matched + ovector pointer to the offsets table + stringcount the number of substrings that were captured + (i.e. the yield of the pcre_exec call, unless + that was zero, in which case it should be 1/3 + of the offset table size) + stringname the name of the required substring + buffer where to put the substring + size the size of the buffer + +Returns: if successful: + the length of the copied string, not including the zero + that is put on the end; can be zero + if not successful: + PCRE_ERROR_NOMEMORY (-6) buffer too small + PCRE_ERROR_NOSUBSTRING (-7) no such captured substring +*/ + +int +pcre_copy_named_substring(const pcre *code, const char *subject, int *ovector, + int stringcount, const char *stringname, char *buffer, int size) +{ +int n = get_first_set(code, stringname, ovector); +if (n <= 0) return n; +return pcre_copy_substring(subject, ovector, stringcount, n, buffer, size); +} + + + +/************************************************* +* Copy all captured strings to new store * +*************************************************/ + +/* This function gets one chunk of store and builds a list of pointers and all +of the captured substrings in it. A NULL pointer is put on the end of the list. + +Arguments: + subject the subject string that was matched + ovector pointer to the offsets table + stringcount the number of substrings that were captured + (i.e. the yield of the pcre_exec call, unless + that was zero, in which case it should be 1/3 + of the offset table size) + listptr set to point to the list of pointers + +Returns: if successful: 0 + if not successful: + PCRE_ERROR_NOMEMORY (-6) failed to get store +*/ + +int +pcre_get_substring_list(const char *subject, int *ovector, int stringcount, + const char ***listptr) +{ +int i; +int size = sizeof(char *); +int double_count = stringcount * 2; +char **stringlist; +char *p; + +for (i = 0; i < double_count; i += 2) + size += sizeof(char *) + ovector[i+1] - ovector[i] + 1; + +stringlist = (char **)(pcre_malloc)(size); +if (stringlist == NULL) return PCRE_ERROR_NOMEMORY; + +*listptr = (const char **)stringlist; +p = (char *)(stringlist + stringcount + 1); + +for (i = 0; i < double_count; i += 2) + { + int len = ovector[i+1] - ovector[i]; + memcpy(p, subject + ovector[i], len); + *stringlist++ = p; + p += len; + *p++ = 0; + } + +*stringlist = NULL; +return 0; +} + + + +/************************************************* +* Free store obtained by get_substring_list * +*************************************************/ + +/* This function exists for the benefit of people calling PCRE from non-C +programs that can call its functions, but not free() or (pcre_free)() directly. + +Argument: the result of a previous pcre_get_substring_list() +Returns: nothing +*/ + +void +pcre_free_substring_list(const char **pointer) +{ +(pcre_free)((void *)pointer); +} + + + +/************************************************* +* Copy captured string to new store * +*************************************************/ + +/* This function copies a single captured substring into a piece of new +store + +Arguments: + subject the subject string that was matched + ovector pointer to the offsets table + stringcount the number of substrings that were captured + (i.e. the yield of the pcre_exec call, unless + that was zero, in which case it should be 1/3 + of the offset table size) + stringnumber the number of the required substring + stringptr where to put a pointer to the substring + +Returns: if successful: + the length of the string, not including the zero that + is put on the end; can be zero + if not successful: + PCRE_ERROR_NOMEMORY (-6) failed to get store + PCRE_ERROR_NOSUBSTRING (-7) substring not present +*/ + +int +pcre_get_substring(const char *subject, int *ovector, int stringcount, + int stringnumber, const char **stringptr) +{ +int yield; +char *substring; +if (stringnumber < 0 || stringnumber >= stringcount) + return PCRE_ERROR_NOSUBSTRING; +stringnumber *= 2; +yield = ovector[stringnumber+1] - ovector[stringnumber]; +substring = (char *)(pcre_malloc)(yield + 1); +if (substring == NULL) return PCRE_ERROR_NOMEMORY; +memcpy(substring, subject + ovector[stringnumber], yield); +substring[yield] = 0; +*stringptr = substring; +return yield; +} + + + +/************************************************* +* Copy named captured string to new store * +*************************************************/ + +/* This function copies a single captured substring, identified by name, into +new store. If the regex permits duplicate names, the first substring that is +set is chosen. + +Arguments: + code the compiled regex + subject the subject string that was matched + ovector pointer to the offsets table + stringcount the number of substrings that were captured + (i.e. the yield of the pcre_exec call, unless + that was zero, in which case it should be 1/3 + of the offset table size) + stringname the name of the required substring + stringptr where to put the pointer + +Returns: if successful: + the length of the copied string, not including the zero + that is put on the end; can be zero + if not successful: + PCRE_ERROR_NOMEMORY (-6) couldn't get memory + PCRE_ERROR_NOSUBSTRING (-7) no such captured substring +*/ + +int +pcre_get_named_substring(const pcre *code, const char *subject, int *ovector, + int stringcount, const char *stringname, const char **stringptr) +{ +int n = get_first_set(code, stringname, ovector); +if (n <= 0) return n; +return pcre_get_substring(subject, ovector, stringcount, n, stringptr); + +} + + + + +/************************************************* +* Free store obtained by get_substring * +*************************************************/ + +/* This function exists for the benefit of people calling PCRE from non-C +programs that can call its functions, but not free() or (pcre_free)() directly. + +Argument: the result of a previous pcre_get_substring() +Returns: nothing +*/ + +void +pcre_free_substring(const char *pointer) +{ +(pcre_free)((void *)pointer); +} + +/* End of pcre_get.c */ ============================================================ --- pcre/pcre_globals.c 4a27067a587c5cd92ad9671107a3541f6429082c +++ pcre/pcre_globals.c 4a27067a587c5cd92ad9671107a3541f6429082c @@ -0,0 +1,69 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + +/* PCRE is a library of functions to support regular expressions whose syntax +and semantics are as close as possible to those of the Perl 5 language. + + Written by Philip Hazel + Copyright (c) 1997-2006 University of Cambridge + +----------------------------------------------------------------------------- +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + + * Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimer. + + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + * Neither the name of the University of Cambridge nor the names of its + contributors may be used to endorse or promote products derived from + this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE +LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR +CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +----------------------------------------------------------------------------- +*/ + + +/* This module contains global variables that are exported by the PCRE library. +PCRE is thread-clean and doesn't use any global variables in the normal sense. +However, it calls memory allocation and freeing functions via the four +indirections below, and it can optionally do callouts, using the fifth +indirection. These values can be changed by the caller, but are shared between +all threads. However, when compiling for Virtual Pascal, things are done +differently, and global variables are not used (see pcre.in). */ + + +#include "pcre_internal.h" + + +#ifndef VPCOMPAT +#ifdef __cplusplus +extern "C" void *(*pcre_malloc)(size_t) = malloc; +extern "C" void (*pcre_free)(void *) = free; +extern "C" void *(*pcre_stack_malloc)(size_t) = malloc; +extern "C" void (*pcre_stack_free)(void *) = free; +extern "C" int (*pcre_callout)(pcre_callout_block *) = NULL; +#else +void *(*pcre_malloc)(size_t) = malloc; +void (*pcre_free)(void *) = free; +void *(*pcre_stack_malloc)(size_t) = malloc; +void (*pcre_stack_free)(void *) = free; +int (*pcre_callout)(pcre_callout_block *) = NULL; +#endif +#endif + +/* End of pcre_globals.c */ ============================================================ --- pcre/pcre_info.c 14399144255ceaa0d09a0d16ef030e3dc91cff5b +++ pcre/pcre_info.c 14399144255ceaa0d09a0d16ef030e3dc91cff5b @@ -0,0 +1,89 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + +/* PCRE is a library of functions to support regular expressions whose syntax +and semantics are as close as possible to those of the Perl 5 language. + + Written by Philip Hazel + Copyright (c) 1997-2006 University of Cambridge + +----------------------------------------------------------------------------- +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + + * Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimer. + + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + * Neither the name of the University of Cambridge nor the names of its + contributors may be used to endorse or promote products derived from + this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE +LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR +CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +----------------------------------------------------------------------------- +*/ + + +/* This module contains the external function pcre_info(), which gives some +information about a compiled pattern. However, use of this function is now +deprecated, as it has been superseded by pcre_fullinfo(). */ + + +#include "pcre_internal.h" + + +/************************************************* +* (Obsolete) Return info about compiled pattern * +*************************************************/ + +/* This is the original "info" function. It picks potentially useful data out +of the private structure, but its interface was too rigid. It remains for +backwards compatibility. The public options are passed back in an int - though +the re->options field has been expanded to a long int, all the public options +at the low end of it, and so even on 16-bit systems this will still be OK. +Therefore, I haven't changed the API for pcre_info(). + +Arguments: + argument_re points to compiled code + optptr where to pass back the options + first_byte where to pass back the first character, + or -1 if multiline and all branches start ^, + or -2 otherwise + +Returns: number of capturing subpatterns + or negative values on error +*/ + +PCRE_DATA_SCOPE int +pcre_info(const pcre *argument_re, int *optptr, int *first_byte) +{ +real_pcre internal_re; +const real_pcre *re = (const real_pcre *)argument_re; +if (re == NULL) return PCRE_ERROR_NULL; +if (re->magic_number != MAGIC_NUMBER) + { + re = _pcre_try_flipped(re, &internal_re, NULL, NULL); + if (re == NULL) return PCRE_ERROR_BADMAGIC; + } +if (optptr != NULL) *optptr = (int)(re->options & PUBLIC_OPTIONS); +if (first_byte != NULL) + *first_byte = ((re->options & PCRE_FIRSTSET) != 0)? re->first_byte : + ((re->options & PCRE_STARTLINE) != 0)? -1 : -2; +return re->top_bracket; +} + +/* End of pcre_info.c */ ============================================================ --- pcre/pcre_internal.h d6b78d25230e9f7d3cdf183f2fb45b6de45dc062 +++ pcre/pcre_internal.h d6b78d25230e9f7d3cdf183f2fb45b6de45dc062 @@ -0,0 +1,952 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + + +/* PCRE is a library of functions to support regular expressions whose syntax +and semantics are as close as possible to those of the Perl 5 language. + + Written by Philip Hazel + Copyright (c) 1997-2006 University of Cambridge + +----------------------------------------------------------------------------- +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + + * Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimer. + + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + * Neither the name of the University of Cambridge nor the names of its + contributors may be used to endorse or promote products derived from + this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE +LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR +CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +----------------------------------------------------------------------------- +*/ + +/* This header contains definitions that are shared between the different +modules, but which are not relevant to the exported API. This includes some +functions whose names all begin with "_pcre_". */ + +#ifndef PCRE_INTERNAL_H +#define PCRE_INTERNAL_H + +/* Define DEBUG to get debugging output on stdout. */ + +#if 0 +#define DEBUG +#endif + +/* Use a macro for debugging printing, 'cause that eliminates the use of #ifdef +inline, and there are *still* stupid compilers about that don't like indented +pre-processor statements, or at least there were when I first wrote this. After +all, it had only been about 10 years then... */ + +#ifdef DEBUG +#define DPRINTF(p) printf p +#else +#define DPRINTF(p) /*nothing*/ +#endif + + +/* Get the definitions provided by running "configure" */ + +#include "pcre_config.h" + +/* Standard C headers plus the external interface definition. The only time +setjmp and stdarg are used is when NO_RECURSE is set. */ + +#include +#include +#include +#include +#include +#include +#include +#include + +#ifndef PCRE_SPY +#define PCRE_DEFINITION /* Win32 __declspec(export) trigger for .dll */ +#endif + +/* We need to have types that specify unsigned 16-bit and 32-bit integers. We +cannot determine these outside the compilation (e.g. by running a program as +part of "configure") because PCRE is often cross-compiled for use on other +systems. Instead we make use of the maximum sizes that are available at +preprocessor time in standard C environments. */ + +#if USHRT_MAX == 65535 + typedef unsigned short pcre_uint16; +#elif UINT_MAX == 65535 + typedef unsigned int pcre_uint16; +#else + #error Cannot determine a type for 16-bit unsigned integers +#endif + +#if UINT_MAX == 4294967295 + typedef unsigned int pcre_uint32; +#elif ULONG_MAX == 4294967295 + typedef unsigned long int pcre_uint32; +#else + #error Cannot determine a type for 32-bit unsigned integers +#endif + +/* All character handling must be done as unsigned characters. Otherwise there +are problems with top-bit-set characters and functions such as isspace(). +However, we leave the interface to the outside world as char *, because that +should make things easier for callers. We define a short type for unsigned char +to save lots of typing. I tried "uchar", but it causes problems on Digital +Unix, where it is defined in sys/types, so use "uschar" instead. */ + +typedef unsigned char uschar; + +/* PCRE is able to support 3 different kinds of newline (CR, LF, CRLF). The +following macro is used to package up testing for newlines. NLBLOCK is defined +in the various modules to indicate in which datablock the parameters exist. */ + +#define IS_NEWLINE(p) \ + ((p)[0] == NLBLOCK->nl[0] && \ + (NLBLOCK->nllen == 1 || (p)[1] == NLBLOCK->nl[1])) + +/* When PCRE is compiled as a C++ library, the subject pointer can be replaced +with a custom type. This makes it possible, for example, to allow pcre_exec() +to process subject strings that are discontinuous by using a smart pointer +class. It must always be possible to inspect all of the subject string in +pcre_exec() because of the way it backtracks. Two macros are required in the +normal case, for sign-unspecified and unsigned char pointers. The former is +used for the external interface and appears in pcre.h, which is why its name +must begin with PCRE_. */ + +#ifdef CUSTOM_SUBJECT_PTR +#define PCRE_SPTR CUSTOM_SUBJECT_PTR +#define USPTR CUSTOM_SUBJECT_PTR +#else +#define PCRE_SPTR const char * +#define USPTR const unsigned char * +#endif + +/* Include the public PCRE header and the definitions of UCP character property +values. */ + +#include "pcre.h" +#ifdef SUPPORT_UCP +#include "ucp.h" +#endif + +/* When compiling for use with the Virtual Pascal compiler, these functions +need to have their names changed. PCRE must be compiled with the -DVPCOMPAT +option on the command line. */ + +#ifdef VPCOMPAT +#define strncmp(s1,s2,m) _strncmp(s1,s2,m) +#define memcpy(d,s,n) _memcpy(d,s,n) +#define memmove(d,s,n) _memmove(d,s,n) +#define memset(s,c,n) _memset(s,c,n) +#else /* VPCOMPAT */ + +/* To cope with SunOS4 and other systems that lack memmove() but have bcopy(), +define a macro for memmove() if HAVE_MEMMOVE is false, provided that HAVE_BCOPY +is set. Otherwise, include an emulating function for those systems that have +neither (there some non-Unix environments where this is the case). This assumes +that all calls to memmove are moving strings upwards in store, which is the +case in PCRE. */ + +#if ! HAVE_MEMMOVE +#undef memmove /* some systems may have a macro */ +#if HAVE_BCOPY +#define memmove(a, b, c) bcopy(b, a, c) +#else /* HAVE_BCOPY */ +static void * +pcre_memmove(unsigned char *dest, const unsigned char *src, size_t n) +{ +size_t i; +dest += n; +src += n; +for (i = 0; i < n; ++i) *(--dest) = *(--src); +return dest; +} +#define memmove(a, b, c) pcre_memmove(a, b, c) +#endif /* not HAVE_BCOPY */ +#endif /* not HAVE_MEMMOVE */ +#endif /* not VPCOMPAT */ + + +/* PCRE keeps offsets in its compiled code as 2-byte quantities (always stored +in big-endian order) by default. These are used, for example, to link from the +start of a subpattern to its alternatives and its end. The use of 2 bytes per +offset limits the size of the compiled regex to around 64K, which is big enough +for almost everybody. However, I received a request for an even bigger limit. +For this reason, and also to make the code easier to maintain, the storing and +loading of offsets from the byte string is now handled by the macros that are +defined here. + +The macros are controlled by the value of LINK_SIZE. This defaults to 2 in +the config.h file, but can be overridden by using -D on the command line. This +is automated on Unix systems via the "configure" command. */ + +#if LINK_SIZE == 2 + +#define PUT(a,n,d) \ + (a[n] = (d) >> 8), \ + (a[(n)+1] = (d) & 255) + +#define GET(a,n) \ + (((a)[n] << 8) | (a)[(n)+1]) + +#define MAX_PATTERN_SIZE (1 << 16) + + +#elif LINK_SIZE == 3 + +#define PUT(a,n,d) \ + (a[n] = (d) >> 16), \ + (a[(n)+1] = (d) >> 8), \ + (a[(n)+2] = (d) & 255) + +#define GET(a,n) \ + (((a)[n] << 16) | ((a)[(n)+1] << 8) | (a)[(n)+2]) + +#define MAX_PATTERN_SIZE (1 << 24) + + +#elif LINK_SIZE == 4 + +#define PUT(a,n,d) \ + (a[n] = (d) >> 24), \ + (a[(n)+1] = (d) >> 16), \ + (a[(n)+2] = (d) >> 8), \ + (a[(n)+3] = (d) & 255) + +#define GET(a,n) \ + (((a)[n] << 24) | ((a)[(n)+1] << 16) | ((a)[(n)+2] << 8) | (a)[(n)+3]) + +#define MAX_PATTERN_SIZE (1 << 30) /* Keep it positive */ + + +#else +#error LINK_SIZE must be either 2, 3, or 4 +#endif + + +/* Convenience macro defined in terms of the others */ + +#define PUTINC(a,n,d) PUT(a,n,d), a += LINK_SIZE + + +/* PCRE uses some other 2-byte quantities that do not change when the size of +offsets changes. There are used for repeat counts and for other things such as +capturing parenthesis numbers in back references. */ + +#define PUT2(a,n,d) \ + a[n] = (d) >> 8; \ + a[(n)+1] = (d) & 255 + +#define GET2(a,n) \ + (((a)[n] << 8) | (a)[(n)+1]) + +#define PUT2INC(a,n,d) PUT2(a,n,d), a += 2 + + +/* When UTF-8 encoding is being used, a character is no longer just a single +byte. The macros for character handling generate simple sequences when used in +byte-mode, and more complicated ones for UTF-8 characters. */ + +#ifndef SUPPORT_UTF8 +#define GETCHAR(c, eptr) c = *eptr; +#define GETCHARTEST(c, eptr) c = *eptr; +#define GETCHARINC(c, eptr) c = *eptr++; +#define GETCHARINCTEST(c, eptr) c = *eptr++; +#define GETCHARLEN(c, eptr, len) c = *eptr; +#define BACKCHAR(eptr) + +#else /* SUPPORT_UTF8 */ + +/* Get the next UTF-8 character, not advancing the pointer. This is called when +we know we are in UTF-8 mode. */ + +#define GETCHAR(c, eptr) \ + c = *eptr; \ + if ((c & 0xc0) == 0xc0) \ + { \ + int gcii; \ + int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \ + int gcss = 6*gcaa; \ + c = (c & _pcre_utf8_table3[gcaa]) << gcss; \ + for (gcii = 1; gcii <= gcaa; gcii++) \ + { \ + gcss -= 6; \ + c |= (eptr[gcii] & 0x3f) << gcss; \ + } \ + } + +/* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the +pointer. */ + +#define GETCHARTEST(c, eptr) \ + c = *eptr; \ + if (utf8 && (c & 0xc0) == 0xc0) \ + { \ + int gcii; \ + int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \ + int gcss = 6*gcaa; \ + c = (c & _pcre_utf8_table3[gcaa]) << gcss; \ + for (gcii = 1; gcii <= gcaa; gcii++) \ + { \ + gcss -= 6; \ + c |= (eptr[gcii] & 0x3f) << gcss; \ + } \ + } + +/* Get the next UTF-8 character, advancing the pointer. This is called when we +know we are in UTF-8 mode. */ + +#define GETCHARINC(c, eptr) \ + c = *eptr++; \ + if ((c & 0xc0) == 0xc0) \ + { \ + int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \ + int gcss = 6*gcaa; \ + c = (c & _pcre_utf8_table3[gcaa]) << gcss; \ + while (gcaa-- > 0) \ + { \ + gcss -= 6; \ + c |= (*eptr++ & 0x3f) << gcss; \ + } \ + } + +/* Get the next character, testing for UTF-8 mode, and advancing the pointer */ + +#define GETCHARINCTEST(c, eptr) \ + c = *eptr++; \ + if (utf8 && (c & 0xc0) == 0xc0) \ + { \ + int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \ + int gcss = 6*gcaa; \ + c = (c & _pcre_utf8_table3[gcaa]) << gcss; \ + while (gcaa-- > 0) \ + { \ + gcss -= 6; \ + c |= (*eptr++ & 0x3f) << gcss; \ + } \ + } + +/* Get the next UTF-8 character, not advancing the pointer, incrementing length +if there are extra bytes. This is called when we know we are in UTF-8 mode. */ + +#define GETCHARLEN(c, eptr, len) \ + c = *eptr; \ + if ((c & 0xc0) == 0xc0) \ + { \ + int gcii; \ + int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \ + int gcss = 6*gcaa; \ + c = (c & _pcre_utf8_table3[gcaa]) << gcss; \ + for (gcii = 1; gcii <= gcaa; gcii++) \ + { \ + gcss -= 6; \ + c |= (eptr[gcii] & 0x3f) << gcss; \ + } \ + len += gcaa; \ + } + +/* If the pointer is not at the start of a character, move it back until +it is. Called only in UTF-8 mode. */ + +#define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr--; + +#endif + + +/* In case there is no definition of offsetof() provided - though any proper +Standard C system should have one. */ + +#ifndef offsetof +#define offsetof(p_type,field) ((size_t)&(((p_type *)0)->field)) +#endif + + +/* These are the public options that can change during matching. */ + +#define PCRE_IMS (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL) + +/* Private options flags start at the most significant end of the four bytes. +The public options defined in pcre.h start at the least significant end. Make +sure they don't overlap! The bits are getting a bit scarce now -- when we run +out, there is a dummy word in the structure that could be used for the private +bits. */ + +#define PCRE_NOPARTIAL 0x80000000 /* can't use partial with this regex */ +#define PCRE_FIRSTSET 0x40000000 /* first_byte is set */ +#define PCRE_REQCHSET 0x20000000 /* req_byte is set */ +#define PCRE_STARTLINE 0x10000000 /* start after \n for multiline */ +#define PCRE_JCHANGED 0x08000000 /* j option changes within regex */ + +/* Options for the "extra" block produced by pcre_study(). */ + +#define PCRE_STUDY_MAPPED 0x01 /* a map of starting chars exists */ + +/* Masks for identifying the public options that are permitted at compile +time, run time, or study time, respectively. */ + +#define PUBLIC_OPTIONS \ + (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \ + PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \ + PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \ + PCRE_DUPNAMES|PCRE_NEWLINE_CR|PCRE_NEWLINE_LF) + +#define PUBLIC_EXEC_OPTIONS \ + (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \ + PCRE_PARTIAL|PCRE_NEWLINE_CR|PCRE_NEWLINE_LF) + +#define PUBLIC_DFA_EXEC_OPTIONS \ + (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \ + PCRE_PARTIAL|PCRE_DFA_SHORTEST|PCRE_DFA_RESTART|PCRE_NEWLINE_CR| \ + PCRE_NEWLINE_LF) + +#define PUBLIC_STUDY_OPTIONS 0 /* None defined */ + +/* Magic number to provide a small check against being handed junk. Also used +to detect whether a pattern was compiled on a host of different endianness. */ + +#define MAGIC_NUMBER 0x50435245UL /* 'PCRE' */ + +/* Negative values for the firstchar and reqchar variables */ + +#define REQ_UNSET (-2) +#define REQ_NONE (-1) + +/* The maximum remaining length of subject we are prepared to search for a +req_byte match. */ + +#define REQ_BYTE_MAX 1000 + +/* Flags added to firstbyte or reqbyte; a "non-literal" item is either a +variable-length repeat, or a anything other than literal characters. */ + +#define REQ_CASELESS 0x0100 /* indicates caselessness */ +#define REQ_VARY 0x0200 /* reqbyte followed non-literal item */ + +/* Miscellaneous definitions */ + +typedef int BOOL; + +#define FALSE 0 +#define TRUE 1 + +/* Escape items that are just an encoding of a particular data value. Note that +ESC_n is defined as yet another macro, which is set in config.h to either \n +(the default) or \r (which some people want). */ + +#ifndef ESC_e +#define ESC_e 27 +#endif + +#ifndef ESC_f +#define ESC_f '\f' +#endif + +#ifndef ESC_n +#define ESC_n NEWLINE +#endif + +#ifndef ESC_r +#define ESC_r '\r' +#endif + +/* We can't officially use ESC_t because it is a POSIX reserved identifier +(presumably because of all the others like size_t). */ + +#ifndef ESC_tee +#define ESC_tee '\t' +#endif + +/* Codes for different types of Unicode property */ + +#define PT_ANY 0 /* Any property - matches all chars */ +#define PT_LAMP 1 /* L& - the union of Lu, Ll, Lt */ +#define PT_GC 2 /* General characteristic (e.g. L) */ +#define PT_PC 3 /* Particular characteristic (e.g. Lu) */ +#define PT_SC 4 /* Script (e.g. Han) */ + +/* Flag bits and data types for the extended class (OP_XCLASS) for classes that +contain UTF-8 characters with values greater than 255. */ + +#define XCL_NOT 0x01 /* Flag: this is a negative class */ +#define XCL_MAP 0x02 /* Flag: a 32-byte map is present */ + +#define XCL_END 0 /* Marks end of individual items */ +#define XCL_SINGLE 1 /* Single item (one multibyte char) follows */ +#define XCL_RANGE 2 /* A range (two multibyte chars) follows */ +#define XCL_PROP 3 /* Unicode property (2-byte property code follows) */ +#define XCL_NOTPROP 4 /* Unicode inverted property (ditto) */ + +/* These are escaped items that aren't just an encoding of a particular data +value such as \n. They must have non-zero values, as check_escape() returns +their negation. Also, they must appear in the same order as in the opcode +definitions below, up to ESC_z. There's a dummy for OP_ANY because it +corresponds to "." rather than an escape sequence. The final one must be +ESC_REF as subsequent values are used for \1, \2, \3, etc. There is are two +tests in the code for an escape greater than ESC_b and less than ESC_Z to +detect the types that may be repeated. These are the types that consume +characters. If any new escapes are put in between that don't consume a +character, that code will have to change. */ + +enum { ESC_A = 1, ESC_G, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s, ESC_W, + ESC_w, ESC_dum1, ESC_C, ESC_P, ESC_p, ESC_X, ESC_Z, ESC_z, ESC_E, + ESC_Q, ESC_REF }; + +/* Opcode table: OP_BRA must be last, as all values >= it are used for brackets +that extract substrings. Starting from 1 (i.e. after OP_END), the values up to +OP_EOD must correspond in order to the list of escapes immediately above. +Note that whenever this list is updated, the two macro definitions that follow +must also be updated to match. */ + +enum { + OP_END, /* 0 End of pattern */ + + /* Values corresponding to backslashed metacharacters */ + + OP_SOD, /* 1 Start of data: \A */ + OP_SOM, /* 2 Start of match (subject + offset): \G */ + OP_NOT_WORD_BOUNDARY, /* 3 \B */ + OP_WORD_BOUNDARY, /* 4 \b */ + OP_NOT_DIGIT, /* 5 \D */ + OP_DIGIT, /* 6 \d */ + OP_NOT_WHITESPACE, /* 7 \S */ + OP_WHITESPACE, /* 8 \s */ + OP_NOT_WORDCHAR, /* 9 \W */ + OP_WORDCHAR, /* 10 \w */ + OP_ANY, /* 11 Match any character */ + OP_ANYBYTE, /* 12 Match any byte (\C); different to OP_ANY for UTF-8 */ + OP_NOTPROP, /* 13 \P (not Unicode property) */ + OP_PROP, /* 14 \p (Unicode property) */ + OP_EXTUNI, /* 15 \X (extended Unicode sequence */ + OP_EODN, /* 16 End of data or \n at end of data: \Z. */ + OP_EOD, /* 17 End of data: \z */ + + OP_OPT, /* 18 Set runtime options */ + OP_CIRC, /* 19 Start of line - varies with multiline switch */ + OP_DOLL, /* 20 End of line - varies with multiline switch */ + OP_CHAR, /* 21 Match one character, casefully */ + OP_CHARNC, /* 22 Match one character, caselessly */ + OP_NOT, /* 23 Match one character, not the following one */ + + OP_STAR, /* 24 The maximizing and minimizing versions of */ + OP_MINSTAR, /* 25 all these opcodes must come in pairs, with */ + OP_PLUS, /* 26 the minimizing one second. */ + OP_MINPLUS, /* 27 This first set applies to single characters */ + OP_QUERY, /* 28 */ + OP_MINQUERY, /* 29 */ + OP_UPTO, /* 30 From 0 to n matches */ + OP_MINUPTO, /* 31 */ + OP_EXACT, /* 32 Exactly n matches */ + + OP_NOTSTAR, /* 33 The maximizing and minimizing versions of */ + OP_NOTMINSTAR, /* 34 all these opcodes must come in pairs, with */ + OP_NOTPLUS, /* 35 the minimizing one second. */ + OP_NOTMINPLUS, /* 36 This set applies to "not" single characters */ + OP_NOTQUERY, /* 37 */ + OP_NOTMINQUERY, /* 38 */ + OP_NOTUPTO, /* 39 From 0 to n matches */ + OP_NOTMINUPTO, /* 40 */ + OP_NOTEXACT, /* 41 Exactly n matches */ + + OP_TYPESTAR, /* 42 The maximizing and minimizing versions of */ + OP_TYPEMINSTAR, /* 43 all these opcodes must come in pairs, with */ + OP_TYPEPLUS, /* 44 the minimizing one second. These codes must */ + OP_TYPEMINPLUS, /* 45 be in exactly the same order as those above. */ + OP_TYPEQUERY, /* 46 This set applies to character types such as \d */ + OP_TYPEMINQUERY, /* 47 */ + OP_TYPEUPTO, /* 48 From 0 to n matches */ + OP_TYPEMINUPTO, /* 49 */ + OP_TYPEEXACT, /* 50 Exactly n matches */ + + OP_CRSTAR, /* 51 The maximizing and minimizing versions of */ + OP_CRMINSTAR, /* 52 all these opcodes must come in pairs, with */ + OP_CRPLUS, /* 53 the minimizing one second. These codes must */ + OP_CRMINPLUS, /* 54 be in exactly the same order as those above. */ + OP_CRQUERY, /* 55 These are for character classes and back refs */ + OP_CRMINQUERY, /* 56 */ + OP_CRRANGE, /* 57 These are different to the three sets above. */ + OP_CRMINRANGE, /* 58 */ + + OP_CLASS, /* 59 Match a character class, chars < 256 only */ + OP_NCLASS, /* 60 Same, but the bitmap was created from a negative + class - the difference is relevant only when a UTF-8 + character > 255 is encountered. */ + + OP_XCLASS, /* 61 Extended class for handling UTF-8 chars within the + class. This does both positive and negative. */ + + OP_REF, /* 62 Match a back reference */ + OP_RECURSE, /* 63 Match a numbered subpattern (possibly recursive) */ + OP_CALLOUT, /* 64 Call out to external function if provided */ + + OP_ALT, /* 65 Start of alternation */ + OP_KET, /* 66 End of group that doesn't have an unbounded repeat */ + OP_KETRMAX, /* 67 These two must remain together and in this */ + OP_KETRMIN, /* 68 order. They are for groups the repeat for ever. */ + + /* The assertions must come before ONCE and COND */ + + OP_ASSERT, /* 69 Positive lookahead */ + OP_ASSERT_NOT, /* 70 Negative lookahead */ + OP_ASSERTBACK, /* 71 Positive lookbehind */ + OP_ASSERTBACK_NOT, /* 72 Negative lookbehind */ + OP_REVERSE, /* 73 Move pointer back - used in lookbehind assertions */ + + /* ONCE and COND must come after the assertions, with ONCE first, as there's + a test for >= ONCE for a subpattern that isn't an assertion. */ + + OP_ONCE, /* 74 Once matched, don't back up into the subpattern */ + OP_COND, /* 75 Conditional group */ + OP_CREF, /* 76 Used to hold an extraction string number (cond ref) */ + + OP_BRAZERO, /* 77 These two must remain together and in this */ + OP_BRAMINZERO, /* 78 order. */ + + OP_BRANUMBER, /* 79 Used for extracting brackets whose number is greater + than can fit into an opcode. */ + + OP_BRA /* 80 This and greater values are used for brackets that + extract substrings up to EXTRACT_BASIC_MAX. After + that, use is made of OP_BRANUMBER. */ +}; + +/* WARNING WARNING WARNING: There is an implicit assumption in pcre.c and +study.c that all opcodes are less than 128 in value. This makes handling UTF-8 +character sequences easier. */ + +/* The highest extraction number before we have to start using additional +bytes. (Originally PCRE didn't have support for extraction counts highter than +this number.) The value is limited by the number of opcodes left after OP_BRA, +i.e. 255 - OP_BRA. We actually set it a bit lower to leave room for additional +opcodes. */ + +#define EXTRACT_BASIC_MAX 100 + + +/* This macro defines textual names for all the opcodes. These are used only +for debugging. The macro is referenced only in pcre_printint.c. */ + +#define OP_NAME_LIST \ + "End", "\\A", "\\G", "\\B", "\\b", "\\D", "\\d", \ + "\\S", "\\s", "\\W", "\\w", "Any", "Anybyte", \ + "notprop", "prop", "extuni", \ + "\\Z", "\\z", \ + "Opt", "^", "$", "char", "charnc", "not", \ + "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \ + "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \ + "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \ + "*", "*?", "+", "+?", "?", "??", "{", "{", \ + "class", "nclass", "xclass", "Ref", "Recurse", "Callout", \ + "Alt", "Ket", "KetRmax", "KetRmin", "Assert", "Assert not", \ + "AssertB", "AssertB not", "Reverse", "Once", "Cond", "Cond ref",\ + "Brazero", "Braminzero", "Branumber", "Bra" + + +/* This macro defines the length of fixed length operations in the compiled +regex. The lengths are used when searching for specific things, and also in the +debugging printing of a compiled regex. We use a macro so that it can be +defined close to the definitions of the opcodes themselves. + +As things have been extended, some of these are no longer fixed lenths, but are +minima instead. For example, the length of a single-character repeat may vary +in UTF-8 mode. The code that uses this table must know about such things. */ + +#define OP_LENGTHS \ + 1, /* End */ \ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* \A, \G, \B, \B, \D, \d, \S, \s, \W, \w */ \ + 1, 1, /* Any, Anybyte */ \ + 3, 3, 1, /* NOTPROP, PROP, EXTUNI */ \ + 1, 1, 2, 1, 1, /* \Z, \z, Opt, ^, $ */ \ + 2, /* Char - the minimum length */ \ + 2, /* Charnc - the minimum length */ \ + 2, /* not */ \ + /* Positive single-char repeats ** These are */ \ + 2, 2, 2, 2, 2, 2, /* *, *?, +, +?, ?, ?? ** minima in */ \ + 4, 4, 4, /* upto, minupto, exact ** UTF-8 mode */ \ + /* Negative single-char repeats - only for chars < 256 */ \ + 2, 2, 2, 2, 2, 2, /* NOT *, *?, +, +?, ?, ?? */ \ + 4, 4, 4, /* NOT upto, minupto, exact */ \ + /* Positive type repeats */ \ + 2, 2, 2, 2, 2, 2, /* Type *, *?, +, +?, ?, ?? */ \ + 4, 4, 4, /* Type upto, minupto, exact */ \ + /* Character class & ref repeats */ \ + 1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */ \ + 5, 5, /* CRRANGE, CRMINRANGE */ \ + 33, /* CLASS */ \ + 33, /* NCLASS */ \ + 0, /* XCLASS - variable length */ \ + 3, /* REF */ \ + 1+LINK_SIZE, /* RECURSE */ \ + 2+2*LINK_SIZE, /* CALLOUT */ \ + 1+LINK_SIZE, /* Alt */ \ + 1+LINK_SIZE, /* Ket */ \ + 1+LINK_SIZE, /* KetRmax */ \ + 1+LINK_SIZE, /* KetRmin */ \ + 1+LINK_SIZE, /* Assert */ \ + 1+LINK_SIZE, /* Assert not */ \ + 1+LINK_SIZE, /* Assert behind */ \ + 1+LINK_SIZE, /* Assert behind not */ \ + 1+LINK_SIZE, /* Reverse */ \ + 1+LINK_SIZE, /* Once */ \ + 1+LINK_SIZE, /* COND */ \ + 3, /* CREF */ \ + 1, 1, /* BRAZERO, BRAMINZERO */ \ + 3, /* BRANUMBER */ \ + 1+LINK_SIZE /* BRA */ \ + + +/* A magic value for OP_CREF to indicate the "in recursion" condition. */ + +#define CREF_RECURSE 0xffff + +/* Error code numbers. They are given names so that they can more easily be +tracked. */ + +enum { ERR0, ERR1, ERR2, ERR3, ERR4, ERR5, ERR6, ERR7, ERR8, ERR9, + ERR10, ERR11, ERR12, ERR13, ERR14, ERR15, ERR16, ERR17, ERR18, ERR19, + ERR20, ERR21, ERR22, ERR23, ERR24, ERR25, ERR26, ERR27, ERR28, ERR29, + ERR30, ERR31, ERR32, ERR33, ERR34, ERR35, ERR36, ERR37, ERR38, ERR39, + ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49, + ERR50, ERR51 }; + +/* The real format of the start of the pcre block; the index of names and the +code vector run on as long as necessary after the end. We store an explicit +offset to the name table so that if a regex is compiled on one host, saved, and +then run on another where the size of pointers is different, all might still +be well. For the case of compiled-on-4 and run-on-8, we include an extra +pointer that is always NULL. For future-proofing, a few dummy fields were +originally included - even though you can never get this planning right - but +there is only one left now. + +NOTE NOTE NOTE: +Because people can now save and re-use compiled patterns, any additions to this +structure should be made at the end, and something earlier (e.g. a new +flag in the options or one of the dummy fields) should indicate that the new +fields are present. Currently PCRE always sets the dummy fields to zero. +NOTE NOTE NOTE: +*/ + +typedef struct real_pcre { + pcre_uint32 magic_number; + pcre_uint32 size; /* Total that was malloced */ + pcre_uint32 options; + pcre_uint32 dummy1; /* For future use, maybe */ + + pcre_uint16 top_bracket; + pcre_uint16 top_backref; + pcre_uint16 first_byte; + pcre_uint16 req_byte; + pcre_uint16 name_table_offset; /* Offset to name table that follows */ + pcre_uint16 name_entry_size; /* Size of any name items */ + pcre_uint16 name_count; /* Number of name items */ + pcre_uint16 ref_count; /* Reference count */ + + const unsigned char *tables; /* Pointer to tables or NULL for std */ + const unsigned char *nullpad; /* NULL padding */ +} real_pcre; + +/* The format of the block used to store data from pcre_study(). The same +remark (see NOTE above) about extending this structure applies. */ + +typedef struct pcre_study_data { + pcre_uint32 size; /* Total that was malloced */ + pcre_uint32 options; + uschar start_bits[32]; +} pcre_study_data; + +/* Structure for passing "static" information around between the functions +doing the compiling, so that they are thread-safe. */ + +typedef struct compile_data { + const uschar *lcc; /* Points to lower casing table */ + const uschar *fcc; /* Points to case-flipping table */ + const uschar *cbits; /* Points to character type table */ + const uschar *ctypes; /* Points to table of type maps */ + const uschar *start_code; /* The start of the compiled code */ + const uschar *start_pattern; /* The start of the pattern */ + uschar *name_table; /* The name/number table */ + int names_found; /* Number of entries so far */ + int name_entry_size; /* Size of each entry */ + int top_backref; /* Maximum back reference */ + unsigned int backref_map; /* Bitmap of low back refs */ + int req_varyopt; /* "After variable item" flag for reqbyte */ + BOOL nopartial; /* Set TRUE if partial won't work */ + int nllen; /* 1 or 2 for newline string length */ + uschar nl[4]; /* Newline string */ +} compile_data; + +/* Structure for maintaining a chain of pointers to the currently incomplete +branches, for testing for left recursion. */ + +typedef struct branch_chain { + struct branch_chain *outer; + uschar *current; +} branch_chain; + +/* Structure for items in a linked list that represents an explicit recursive +call within the pattern. */ + +typedef struct recursion_info { + struct recursion_info *prevrec; /* Previous recursion record (or NULL) */ + int group_num; /* Number of group that was called */ + const uschar *after_call; /* "Return value": points after the call in the expr */ + USPTR save_start; /* Old value of md->start_match */ + int *offset_save; /* Pointer to start of saved offsets */ + int saved_max; /* Number of saved offsets */ +} recursion_info; + +/* When compiling in a mode that doesn't use recursive calls to match(), +a structure is used to remember local variables on the heap. It is defined in +pcre_exec.c, close to the match() function, so that it is easy to keep it in +step with any changes of local variable. However, the pointer to the current +frame must be saved in some "static" place over a longjmp(). We declare the +structure here so that we can put a pointer in the match_data structure. NOTE: +This isn't used for a "normal" compilation of pcre. */ + +struct heapframe; + +/* Structure for passing "static" information around between the functions +doing traditional NFA matching, so that they are thread-safe. */ + +typedef struct match_data { + unsigned long int match_call_count; /* As it says */ + unsigned long int match_limit; /* As it says */ + unsigned long int match_limit_recursion; /* As it says */ + int *offset_vector; /* Offset vector */ + int offset_end; /* One past the end */ + int offset_max; /* The maximum usable for return data */ + int nllen; /* 1 or 2 for newline string length */ + uschar nl[4]; /* Newline string */ + const uschar *lcc; /* Points to lower casing table */ + const uschar *ctypes; /* Points to table of type maps */ + BOOL offset_overflow; /* Set if too many extractions */ + BOOL notbol; /* NOTBOL flag */ + BOOL noteol; /* NOTEOL flag */ + BOOL utf8; /* UTF8 flag */ + BOOL endonly; /* Dollar not before final \n */ + BOOL notempty; /* Empty string match not wanted */ + BOOL partial; /* PARTIAL flag */ + BOOL hitend; /* Hit the end of the subject at some point */ + const uschar *start_code; /* For use when recursing */ + USPTR start_subject; /* Start of the subject string */ + USPTR end_subject; /* End of the subject string */ + USPTR start_match; /* Start of this match attempt */ + USPTR end_match_ptr; /* Subject position at end match */ + int end_offset_top; /* Highwater mark at end of match */ + int capture_last; /* Most recent capture number */ + int start_offset; /* The start offset value */ + recursion_info *recursive; /* Linked list of recursion data */ + void *callout_data; /* To pass back to callouts */ + struct heapframe *thisframe; /* Used only when compiling for no recursion */ +} match_data; + +/* A similar structure is used for the same purpose by the DFA matching +functions. */ + +typedef struct dfa_match_data { + const uschar *start_code; /* Start of the compiled pattern */ + const uschar *start_subject; /* Start of the subject string */ + const uschar *end_subject; /* End of subject string */ + const uschar *tables; /* Character tables */ + int moptions; /* Match options */ + int poptions; /* Pattern options */ + int nllen; /* 1 or 2 for newline string length */ + uschar nl[4]; /* Newline string */ + void *callout_data; /* To pass back to callouts */ +} dfa_match_data; + +/* Bit definitions for entries in the pcre_ctypes table. */ + +#define ctype_space 0x01 +#define ctype_letter 0x02 +#define ctype_digit 0x04 +#define ctype_xdigit 0x08 +#define ctype_word 0x10 /* alphameric or '_' */ +#define ctype_meta 0x80 /* regexp meta char or zero (end pattern) */ + +/* Offsets for the bitmap tables in pcre_cbits. Each table contains a set +of bits for a class map. Some classes are built by combining these tables. */ + +#define cbit_space 0 /* [:space:] or \s */ +#define cbit_xdigit 32 /* [:xdigit:] */ +#define cbit_digit 64 /* [:digit:] or \d */ +#define cbit_upper 96 /* [:upper:] */ +#define cbit_lower 128 /* [:lower:] */ +#define cbit_word 160 /* [:word:] or \w */ +#define cbit_graph 192 /* [:graph:] */ +#define cbit_print 224 /* [:print:] */ +#define cbit_punct 256 /* [:punct:] */ +#define cbit_cntrl 288 /* [:cntrl:] */ +#define cbit_length 320 /* Length of the cbits table */ + +/* Offsets of the various tables from the base tables pointer, and +total length. */ + +#define lcc_offset 0 +#define fcc_offset 256 +#define cbits_offset 512 +#define ctypes_offset (cbits_offset + cbit_length) +#define tables_length (ctypes_offset + 256) + +/* Layout of the UCP type table that translates property names into types and +codes. */ + +typedef struct { + const char *name; + pcre_uint16 type; + pcre_uint16 value; +} ucp_type_table; + + +/* Internal shared data tables. These are tables that are used by more than one +of the exported public functions. They have to be "external" in the C sense, +but are not part of the PCRE public API. The data for these tables is in the +pcre_tables.c module. */ + +extern const int _pcre_utf8_table1[]; +extern const int _pcre_utf8_table2[]; +extern const int _pcre_utf8_table3[]; +extern const uschar _pcre_utf8_table4[]; + +extern const int _pcre_utf8_table1_size; + +extern const ucp_type_table _pcre_utt[]; +extern const int _pcre_utt_size; + +extern const uschar _pcre_default_tables[]; + +extern const uschar _pcre_OP_lengths[]; + + +/* Internal shared functions. These are functions that are used by more than +one of the exported public functions. They have to be "external" in the C +sense, but are not part of the PCRE public API. */ + +extern int _pcre_ord2utf8(int, uschar *); +extern real_pcre * _pcre_try_flipped(const real_pcre *, real_pcre *, + const pcre_study_data *, pcre_study_data *); +extern int _pcre_ucp_findprop(const unsigned int, int *, int *); +extern int _pcre_ucp_othercase(const int); +extern int _pcre_valid_utf8(const uschar *, int); +extern BOOL _pcre_xclass(int, const uschar *); + +#endif + +/* End of pcre_internal.h */ ============================================================ --- pcre/pcre_maketables.c 03f5e59dd47ef2874398251712ff6ee454d76d13 +++ pcre/pcre_maketables.c 03f5e59dd47ef2874398251712ff6ee454d76d13 @@ -0,0 +1,140 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + +/* PCRE is a library of functions to support regular expressions whose syntax +and semantics are as close as possible to those of the Perl 5 language. + + Written by Philip Hazel + Copyright (c) 1997-2006 University of Cambridge + +----------------------------------------------------------------------------- +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + + * Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimer. + + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + * Neither the name of the University of Cambridge nor the names of its + contributors may be used to endorse or promote products derived from + this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE +LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR +CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +----------------------------------------------------------------------------- +*/ + + +/* This module contains the external function pcre_maketables(), which builds +character tables for PCRE in the current locale. The file is compiled on its +own as part of the PCRE library. However, it is also included in the +compilation of dftables.c, in which case the macro DFTABLES is defined. */ + + +#ifndef DFTABLES +#include "pcre_internal.h" +#endif + + +/************************************************* +* Create PCRE character tables * +*************************************************/ + +/* This function builds a set of character tables for use by PCRE and returns +a pointer to them. They are build using the ctype functions, and consequently +their contents will depend upon the current locale setting. When compiled as +part of the library, the store is obtained via pcre_malloc(), but when compiled +inside dftables, use malloc(). + +Arguments: none +Returns: pointer to the contiguous block of data +*/ + +const unsigned char * +pcre_maketables(void) +{ +unsigned char *yield, *p; +int i; + +#ifndef DFTABLES +yield = (unsigned char*)(pcre_malloc)(tables_length); +#else +yield = (unsigned char*)malloc(tables_length); +#endif + +if (yield == NULL) return NULL; +p = yield; + +/* First comes the lower casing table */ + +for (i = 0; i < 256; i++) *p++ = tolower(i); + +/* Next the case-flipping table */ + +for (i = 0; i < 256; i++) *p++ = islower(i)? toupper(i) : tolower(i); + +/* Then the character class tables. Don't try to be clever and save effort on +exclusive ones - in some locales things may be different. Note that the table +for "space" includes everything "isspace" gives, including VT in the default +locale. This makes it work for the POSIX class [:space:]. Note also that it is +possible for a character to be alnum or alpha without being lower or upper, +such as "male and female ordinals" (\xAA and \xBA) in the fr_FR locale (at +least under Debian Linux's locales as of 12/2005). So we must test for alnum +specially. */ + +memset(p, 0, cbit_length); +for (i = 0; i < 256; i++) + { + if (isdigit(i)) p[cbit_digit + i/8] |= 1 << (i&7); + if (isupper(i)) p[cbit_upper + i/8] |= 1 << (i&7); + if (islower(i)) p[cbit_lower + i/8] |= 1 << (i&7); + if (isalnum(i)) p[cbit_word + i/8] |= 1 << (i&7); + if (i == '_') p[cbit_word + i/8] |= 1 << (i&7); + if (isspace(i)) p[cbit_space + i/8] |= 1 << (i&7); + if (isxdigit(i))p[cbit_xdigit + i/8] |= 1 << (i&7); + if (isgraph(i)) p[cbit_graph + i/8] |= 1 << (i&7); + if (isprint(i)) p[cbit_print + i/8] |= 1 << (i&7); + if (ispunct(i)) p[cbit_punct + i/8] |= 1 << (i&7); + if (iscntrl(i)) p[cbit_cntrl + i/8] |= 1 << (i&7); + } +p += cbit_length; + +/* Finally, the character type table. In this, we exclude VT from the white +space chars, because Perl doesn't recognize it as such for \s and for comments +within regexes. */ + +for (i = 0; i < 256; i++) + { + int x = 0; + if (i != 0x0b && isspace(i)) x += ctype_space; + if (isalpha(i)) x += ctype_letter; + if (isdigit(i)) x += ctype_digit; + if (isxdigit(i)) x += ctype_xdigit; + if (isalnum(i) || i == '_') x += ctype_word; + + /* Note: strchr includes the terminating zero in the characters it considers. + In this instance, that is ok because we want binary zero to be flagged as a + meta-character, which in this sense is any character that terminates a run + of data characters. */ + + if (strchr("*+?{^.$|()[", i) != 0) x += ctype_meta; + *p++ = x; + } + +return yield; +} + +/* End of pcre_maketables.c */ ============================================================ --- pcre/pcre_refcount.c ebd0c3dfce2e8dc81a179e2b033364fcec82f55b +++ pcre/pcre_refcount.c ebd0c3dfce2e8dc81a179e2b033364fcec82f55b @@ -0,0 +1,77 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + +/* PCRE is a library of functions to support regular expressions whose syntax +and semantics are as close as possible to those of the Perl 5 language. + + Written by Philip Hazel + Copyright (c) 1997-2006 University of Cambridge + +----------------------------------------------------------------------------- +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + + * Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimer. + + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + * Neither the name of the University of Cambridge nor the names of its + contributors may be used to endorse or promote products derived from + this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE +LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR +CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +----------------------------------------------------------------------------- +*/ + + +/* This module contains the external function pcre_refcount(), which is an +auxiliary function that can be used to maintain a reference count in a compiled +pattern data block. This might be helpful in applications where the block is +shared by different users. */ + +#include "pcre_internal.h" + + +/************************************************* +* Maintain reference count * +*************************************************/ + +/* The reference count is a 16-bit field, initialized to zero. It is not +possible to transfer a non-zero count from one host to a different host that +has a different byte order - though I can't see why anyone in their right mind +would ever want to do that! + +Arguments: + argument_re points to compiled code + adjust value to add to the count + +Returns: the (possibly updated) count value (a non-negative number), or + a negative error number +*/ + +PCRE_DATA_SCOPE int +pcre_refcount(pcre *argument_re, int adjust) +{ +real_pcre *re = (real_pcre *)argument_re; +if (re == NULL) return PCRE_ERROR_NULL; +re->ref_count = (-adjust > re->ref_count)? 0 : + (adjust + re->ref_count > 65535)? 65535 : + re->ref_count + adjust; +return re->ref_count; +} + +/* End of pcre_refcount.c */ ============================================================ --- pcre/pcre_study.c 9a88f30ab85c1a4df719efb21352da8ec0ff0a60 +++ pcre/pcre_study.c 9a88f30ab85c1a4df719efb21352da8ec0ff0a60 @@ -0,0 +1,525 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + +/* PCRE is a library of functions to support regular expressions whose syntax +and semantics are as close as possible to those of the Perl 5 language. + + Written by Philip Hazel + Copyright (c) 1997-2006 University of Cambridge + +----------------------------------------------------------------------------- +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + + * Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimer. + + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + * Neither the name of the University of Cambridge nor the names of its + contributors may be used to endorse or promote products derived from + this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE +LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR +CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +----------------------------------------------------------------------------- +*/ + + +/* This module contains the external function pcre_study(), along with local +supporting functions. */ + + +#include "pcre_internal.h" + + +/************************************************* +* Set a bit and maybe its alternate case * +*************************************************/ + +/* Given a character, set its bit in the table, and also the bit for the other +version of a letter if we are caseless. + +Arguments: + start_bits points to the bit map + c is the character + caseless the caseless flag + cd the block with char table pointers + +Returns: nothing +*/ + +static void +set_bit(uschar *start_bits, unsigned int c, BOOL caseless, compile_data *cd) +{ +start_bits[c/8] |= (1 << (c&7)); +if (caseless && (cd->ctypes[c] & ctype_letter) != 0) + start_bits[cd->fcc[c]/8] |= (1 << (cd->fcc[c]&7)); +} + + + +/************************************************* +* Create bitmap of starting chars * +*************************************************/ + +/* This function scans a compiled unanchored expression and attempts to build a +bitmap of the set of initial characters. If it can't, it returns FALSE. As time +goes by, we may be able to get more clever at doing this. + +Arguments: + code points to an expression + start_bits points to a 32-byte table, initialized to 0 + caseless the current state of the caseless flag + utf8 TRUE if in UTF-8 mode + cd the block with char table pointers + +Returns: TRUE if table built, FALSE otherwise +*/ + +static BOOL +set_start_bits(const uschar *code, uschar *start_bits, BOOL caseless, + BOOL utf8, compile_data *cd) +{ +register int c; + +#if 0 +/* ========================================================================= */ +/* The following comment and code was inserted in January 1999. In May 2006, +when it was observed to cause compiler warnings about unused values, I took it +out again. If anybody is still using OS/2, they will have to put it back +manually. */ + +/* This next statement and the later reference to dummy are here in order to +trick the optimizer of the IBM C compiler for OS/2 into generating correct +code. Apparently IBM isn't going to fix the problem, and we would rather not +disable optimization (in this module it actually makes a big difference, and +the pcre module can use all the optimization it can get). */ + +volatile int dummy; +/* ========================================================================= */ +#endif + +do + { + const uschar *tcode = code + 1 + LINK_SIZE; + BOOL try_next = TRUE; + + while (try_next) + { + /* If a branch starts with a bracket or a positive lookahead assertion, + recurse to set bits from within them. That's all for this branch. */ + + if ((int)*tcode >= OP_BRA || *tcode == OP_ASSERT) + { + if (!set_start_bits(tcode, start_bits, caseless, utf8, cd)) + return FALSE; + try_next = FALSE; + } + + else switch(*tcode) + { + default: + return FALSE; + + /* Skip over callout */ + + case OP_CALLOUT: + tcode += 2 + 2*LINK_SIZE; + break; + + /* Skip over extended extraction bracket number */ + + case OP_BRANUMBER: + tcode += 3; + break; + + /* Skip over lookbehind and negative lookahead assertions */ + + case OP_ASSERT_NOT: + case OP_ASSERTBACK: + case OP_ASSERTBACK_NOT: + do tcode += GET(tcode, 1); while (*tcode == OP_ALT); + tcode += 1+LINK_SIZE; + break; + + /* Skip over an option setting, changing the caseless flag */ + + case OP_OPT: + caseless = (tcode[1] & PCRE_CASELESS) != 0; + tcode += 2; + break; + + /* BRAZERO does the bracket, but carries on. */ + + case OP_BRAZERO: + case OP_BRAMINZERO: + if (!set_start_bits(++tcode, start_bits, caseless, utf8, cd)) + return FALSE; +/* ========================================================================= + See the comment at the head of this function concerning the next line, + which was an old fudge for the benefit of OS/2. + dummy = 1; + ========================================================================= */ + do tcode += GET(tcode,1); while (*tcode == OP_ALT); + tcode += 1+LINK_SIZE; + break; + + /* Single-char * or ? sets the bit and tries the next item */ + + case OP_STAR: + case OP_MINSTAR: + case OP_QUERY: + case OP_MINQUERY: + set_bit(start_bits, tcode[1], caseless, cd); + tcode += 2; +#ifdef SUPPORT_UTF8 + if (utf8) while ((*tcode & 0xc0) == 0x80) tcode++; +#endif + break; + + /* Single-char upto sets the bit and tries the next */ + + case OP_UPTO: + case OP_MINUPTO: + set_bit(start_bits, tcode[3], caseless, cd); + tcode += 4; +#ifdef SUPPORT_UTF8 + if (utf8) while ((*tcode & 0xc0) == 0x80) tcode++; +#endif + break; + + /* At least one single char sets the bit and stops */ + + case OP_EXACT: /* Fall through */ + tcode += 2; + + case OP_CHAR: + case OP_CHARNC: + case OP_PLUS: + case OP_MINPLUS: + set_bit(start_bits, tcode[1], caseless, cd); + try_next = FALSE; + break; + + /* Single character type sets the bits and stops */ + + case OP_NOT_DIGIT: + for (c = 0; c < 32; c++) + start_bits[c] |= ~cd->cbits[c+cbit_digit]; + try_next = FALSE; + break; + + case OP_DIGIT: + for (c = 0; c < 32; c++) + start_bits[c] |= cd->cbits[c+cbit_digit]; + try_next = FALSE; + break; + + /* The cbit_space table has vertical tab as whitespace; we have to + discard it. */ + + case OP_NOT_WHITESPACE: + for (c = 0; c < 32; c++) + { + int d = cd->cbits[c+cbit_space]; + if (c == 1) d &= ~0x08; + start_bits[c] |= ~d; + } + try_next = FALSE; + break; + + /* The cbit_space table has vertical tab as whitespace; we have to + discard it. */ + + case OP_WHITESPACE: + for (c = 0; c < 32; c++) + { + int d = cd->cbits[c+cbit_space]; + if (c == 1) d &= ~0x08; + start_bits[c] |= d; + } + try_next = FALSE; + break; + + case OP_NOT_WORDCHAR: + for (c = 0; c < 32; c++) + start_bits[c] |= ~cd->cbits[c+cbit_word]; + try_next = FALSE; + break; + + case OP_WORDCHAR: + for (c = 0; c < 32; c++) + start_bits[c] |= cd->cbits[c+cbit_word]; + try_next = FALSE; + break; + + /* One or more character type fudges the pointer and restarts, knowing + it will hit a single character type and stop there. */ + + case OP_TYPEPLUS: + case OP_TYPEMINPLUS: + tcode++; + break; + + case OP_TYPEEXACT: + tcode += 3; + break; + + /* Zero or more repeats of character types set the bits and then + try again. */ + + case OP_TYPEUPTO: + case OP_TYPEMINUPTO: + tcode += 2; /* Fall through */ + + case OP_TYPESTAR: + case OP_TYPEMINSTAR: + case OP_TYPEQUERY: + case OP_TYPEMINQUERY: + switch(tcode[1]) + { + case OP_ANY: + return FALSE; + + case OP_NOT_DIGIT: + for (c = 0; c < 32; c++) + start_bits[c] |= ~cd->cbits[c+cbit_digit]; + break; + + case OP_DIGIT: + for (c = 0; c < 32; c++) + start_bits[c] |= cd->cbits[c+cbit_digit]; + break; + + /* The cbit_space table has vertical tab as whitespace; we have to + discard it. */ + + case OP_NOT_WHITESPACE: + for (c = 0; c < 32; c++) + { + int d = cd->cbits[c+cbit_space]; + if (c == 1) d &= ~0x08; + start_bits[c] |= ~d; + } + break; + + /* The cbit_space table has vertical tab as whitespace; we have to + discard it. */ + + case OP_WHITESPACE: + for (c = 0; c < 32; c++) + { + int d = cd->cbits[c+cbit_space]; + if (c == 1) d &= ~0x08; + start_bits[c] |= d; + } + break; + + case OP_NOT_WORDCHAR: + for (c = 0; c < 32; c++) + start_bits[c] |= ~cd->cbits[c+cbit_word]; + break; + + case OP_WORDCHAR: + for (c = 0; c < 32; c++) + start_bits[c] |= cd->cbits[c+cbit_word]; + break; + } + + tcode += 2; + break; + + /* Character class where all the information is in a bit map: set the + bits and either carry on or not, according to the repeat count. If it was + a negative class, and we are operating with UTF-8 characters, any byte + with a value >= 0xc4 is a potentially valid starter because it starts a + character with a value > 255. */ + + case OP_NCLASS: + if (utf8) + { + start_bits[24] |= 0xf0; /* Bits for 0xc4 - 0xc8 */ + memset(start_bits+25, 0xff, 7); /* Bits for 0xc9 - 0xff */ + } + /* Fall through */ + + case OP_CLASS: + { + tcode++; + + /* In UTF-8 mode, the bits in a bit map correspond to character + values, not to byte values. However, the bit map we are constructing is + for byte values. So we have to do a conversion for characters whose + value is > 127. In fact, there are only two possible starting bytes for + characters in the range 128 - 255. */ + + if (utf8) + { + for (c = 0; c < 16; c++) start_bits[c] |= tcode[c]; + for (c = 128; c < 256; c++) + { + if ((tcode[c/8] && (1 << (c&7))) != 0) + { + int d = (c >> 6) | 0xc0; /* Set bit for this starter */ + start_bits[d/8] |= (1 << (d&7)); /* and then skip on to the */ + c = (c & 0xc0) + 0x40 - 1; /* next relevant character. */ + } + } + } + + /* In non-UTF-8 mode, the two bit maps are completely compatible. */ + + else + { + for (c = 0; c < 32; c++) start_bits[c] |= tcode[c]; + } + + /* Advance past the bit map, and act on what follows */ + + tcode += 32; + switch (*tcode) + { + case OP_CRSTAR: + case OP_CRMINSTAR: + case OP_CRQUERY: + case OP_CRMINQUERY: + tcode++; + break; + + case OP_CRRANGE: + case OP_CRMINRANGE: + if (((tcode[1] << 8) + tcode[2]) == 0) tcode += 5; + else try_next = FALSE; + break; + + default: + try_next = FALSE; + break; + } + } + break; /* End of bitmap class handling */ + + } /* End of switch */ + } /* End of try_next loop */ + + code += GET(code, 1); /* Advance to next branch */ + } +while (*code == OP_ALT); +return TRUE; +} + + + +/************************************************* +* Study a compiled expression * +*************************************************/ + +/* This function is handed a compiled expression that it must study to produce +information that will speed up the matching. It returns a pcre_extra block +which then gets handed back to pcre_exec(). + +Arguments: + re points to the compiled expression + options contains option bits + errorptr points to where to place error messages; + set NULL unless error + +Returns: pointer to a pcre_extra block, with study_data filled in and the + appropriate flag set; + NULL on error or if no optimization possible +*/ + +PCRE_DATA_SCOPE pcre_extra * +pcre_study(const pcre *external_re, int options, const char **errorptr) +{ +uschar start_bits[32]; +pcre_extra *extra; +pcre_study_data *study; +const uschar *tables; +uschar *code; +compile_data compile_block; +const real_pcre *re = (const real_pcre *)external_re; + +*errorptr = NULL; + +if (re == NULL || re->magic_number != MAGIC_NUMBER) + { + *errorptr = "argument is not a compiled regular expression"; + return NULL; + } + +if ((options & ~PUBLIC_STUDY_OPTIONS) != 0) + { + *errorptr = "unknown or incorrect option bit(s) set"; + return NULL; + } + +code = (uschar *)re + re->name_table_offset + + (re->name_count * re->name_entry_size); + +/* For an anchored pattern, or an unanchored pattern that has a first char, or +a multiline pattern that matches only at "line starts", no further processing +at present. */ + +if ((re->options & (PCRE_ANCHORED|PCRE_FIRSTSET|PCRE_STARTLINE)) != 0) + return NULL; + +/* Set the character tables in the block that is passed around */ + +tables = re->tables; +if (tables == NULL) + (void)pcre_fullinfo(external_re, NULL, PCRE_INFO_DEFAULT_TABLES, + (void *)(&tables)); + +compile_block.lcc = tables + lcc_offset; +compile_block.fcc = tables + fcc_offset; +compile_block.cbits = tables + cbits_offset; +compile_block.ctypes = tables + ctypes_offset; + +/* See if we can find a fixed set of initial characters for the pattern. */ + +memset(start_bits, 0, 32 * sizeof(uschar)); +if (!set_start_bits(code, start_bits, (re->options & PCRE_CASELESS) != 0, + (re->options & PCRE_UTF8) != 0, &compile_block)) return NULL; + +/* Get a pcre_extra block and a pcre_study_data block. The study data is put in +the latter, which is pointed to by the former, which may also get additional +data set later by the calling program. At the moment, the size of +pcre_study_data is fixed. We nevertheless save it in a field for returning via +the pcre_fullinfo() function so that if it becomes variable in the future, we +don't have to change that code. */ + +extra = (pcre_extra *)(pcre_malloc) + (sizeof(pcre_extra) + sizeof(pcre_study_data)); + +if (extra == NULL) + { + *errorptr = "failed to get memory"; + return NULL; + } + +study = (pcre_study_data *)((char *)extra + sizeof(pcre_extra)); +extra->flags = PCRE_EXTRA_STUDY_DATA; +extra->study_data = study; + +study->size = sizeof(pcre_study_data); +study->options = PCRE_STUDY_MAPPED; +memcpy(study->start_bits, start_bits, sizeof(start_bits)); + +return extra; +} + +/* End of pcre_study.c */ ============================================================ --- pcre/pcre_tables.c f504519f4ad83001af74dca4ada66d1fe68f64aa +++ pcre/pcre_tables.c f504519f4ad83001af74dca4ada66d1fe68f64aa @@ -0,0 +1,198 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + +/* PCRE is a library of functions to support regular expressions whose syntax +and semantics are as close as possible to those of the Perl 5 language. + + Written by Philip Hazel + Copyright (c) 1997-2006 University of Cambridge + +----------------------------------------------------------------------------- +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + + * Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimer. + + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + * Neither the name of the University of Cambridge nor the names of its + contributors may be used to endorse or promote products derived from + this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE +LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR +CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +----------------------------------------------------------------------------- +*/ + + +/* This module contains some fixed tables that are used by more than one of the +PCRE code modules. The tables are also #included by the pcretest program, which +uses macros to change their names from _pcre_xxx to xxxx, thereby avoiding name +clashes with the library. */ + + +#include "pcre_internal.h" + + +/* Table of sizes for the fixed-length opcodes. It's defined in a macro so that +the definition is next to the definition of the opcodes in pcre_internal.h. */ + +const uschar _pcre_OP_lengths[] = { OP_LENGTHS }; + + + +/************************************************* +* Tables for UTF-8 support * +*************************************************/ + +#ifdef SUPPORT_UTF8 +/* These are the breakpoints for different numbers of bytes in a UTF-8 +character. */ + +const int _pcre_utf8_table1[] = + { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff}; + +const int _pcre_utf8_table1_size = sizeof(_pcre_utf8_table1)/sizeof(int); + +/* These are the indicator bits and the mask for the data bits to set in the +first byte of a character, indexed by the number of additional bytes. */ + +const int _pcre_utf8_table2[] = { 0, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc}; +const int _pcre_utf8_table3[] = { 0xff, 0x1f, 0x0f, 0x07, 0x03, 0x01}; + +/* Table of the number of extra characters, indexed by the first character +masked with 0x3f. The highest number for a valid UTF-8 character is in fact +0x3d. */ + +const uschar _pcre_utf8_table4[] = { + 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, + 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, + 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, + 3,3,3,3,3,3,3,3,4,4,4,4,5,5,5,5 }; +#endif + +#ifdef SUPPORT_UCP + +/* This table translates Unicode property names into type and code values. It +is searched by binary chop, so must be in collating sequence of name. */ + +const ucp_type_table _pcre_utt[] = { + { "Any", PT_ANY, 0 }, + { "Arabic", PT_SC, ucp_Arabic }, + { "Armenian", PT_SC, ucp_Armenian }, + { "Bengali", PT_SC, ucp_Bengali }, + { "Bopomofo", PT_SC, ucp_Bopomofo }, + { "Braille", PT_SC, ucp_Braille }, + { "Buginese", PT_SC, ucp_Buginese }, + { "Buhid", PT_SC, ucp_Buhid }, + { "C", PT_GC, ucp_C }, + { "Canadian_Aboriginal", PT_SC, ucp_Canadian_Aboriginal }, + { "Cc", PT_PC, ucp_Cc }, + { "Cf", PT_PC, ucp_Cf }, + { "Cherokee", PT_SC, ucp_Cherokee }, + { "Cn", PT_PC, ucp_Cn }, + { "Co", PT_PC, ucp_Co }, + { "Common", PT_SC, ucp_Common }, + { "Coptic", PT_SC, ucp_Coptic }, + { "Cs", PT_PC, ucp_Cs }, + { "Cypriot", PT_SC, ucp_Cypriot }, + { "Cyrillic", PT_SC, ucp_Cyrillic }, + { "Deseret", PT_SC, ucp_Deseret }, + { "Devanagari", PT_SC, ucp_Devanagari }, + { "Ethiopic", PT_SC, ucp_Ethiopic }, + { "Georgian", PT_SC, ucp_Georgian }, + { "Glagolitic", PT_SC, ucp_Glagolitic }, + { "Gothic", PT_SC, ucp_Gothic }, + { "Greek", PT_SC, ucp_Greek }, + { "Gujarati", PT_SC, ucp_Gujarati }, + { "Gurmukhi", PT_SC, ucp_Gurmukhi }, + { "Han", PT_SC, ucp_Han }, + { "Hangul", PT_SC, ucp_Hangul }, + { "Hanunoo", PT_SC, ucp_Hanunoo }, + { "Hebrew", PT_SC, ucp_Hebrew }, + { "Hiragana", PT_SC, ucp_Hiragana }, + { "Inherited", PT_SC, ucp_Inherited }, + { "Kannada", PT_SC, ucp_Kannada }, + { "Katakana", PT_SC, ucp_Katakana }, + { "Kharoshthi", PT_SC, ucp_Kharoshthi }, + { "Khmer", PT_SC, ucp_Khmer }, + { "L", PT_GC, ucp_L }, + { "L&", PT_LAMP, 0 }, + { "Lao", PT_SC, ucp_Lao }, + { "Latin", PT_SC, ucp_Latin }, + { "Limbu", PT_SC, ucp_Limbu }, + { "Linear_B", PT_SC, ucp_Linear_B }, + { "Ll", PT_PC, ucp_Ll }, + { "Lm", PT_PC, ucp_Lm }, + { "Lo", PT_PC, ucp_Lo }, + { "Lt", PT_PC, ucp_Lt }, + { "Lu", PT_PC, ucp_Lu }, + { "M", PT_GC, ucp_M }, + { "Malayalam", PT_SC, ucp_Malayalam }, + { "Mc", PT_PC, ucp_Mc }, + { "Me", PT_PC, ucp_Me }, + { "Mn", PT_PC, ucp_Mn }, + { "Mongolian", PT_SC, ucp_Mongolian }, + { "Myanmar", PT_SC, ucp_Myanmar }, + { "N", PT_GC, ucp_N }, + { "Nd", PT_PC, ucp_Nd }, + { "New_Tai_Lue", PT_SC, ucp_New_Tai_Lue }, + { "Nl", PT_PC, ucp_Nl }, + { "No", PT_PC, ucp_No }, + { "Ogham", PT_SC, ucp_Ogham }, + { "Old_Italic", PT_SC, ucp_Old_Italic }, + { "Old_Persian", PT_SC, ucp_Old_Persian }, + { "Oriya", PT_SC, ucp_Oriya }, + { "Osmanya", PT_SC, ucp_Osmanya }, + { "P", PT_GC, ucp_P }, + { "Pc", PT_PC, ucp_Pc }, + { "Pd", PT_PC, ucp_Pd }, + { "Pe", PT_PC, ucp_Pe }, + { "Pf", PT_PC, ucp_Pf }, + { "Pi", PT_PC, ucp_Pi }, + { "Po", PT_PC, ucp_Po }, + { "Ps", PT_PC, ucp_Ps }, + { "Runic", PT_SC, ucp_Runic }, + { "S", PT_GC, ucp_S }, + { "Sc", PT_PC, ucp_Sc }, + { "Shavian", PT_SC, ucp_Shavian }, + { "Sinhala", PT_SC, ucp_Sinhala }, + { "Sk", PT_PC, ucp_Sk }, + { "Sm", PT_PC, ucp_Sm }, + { "So", PT_PC, ucp_So }, + { "Syloti_Nagri", PT_SC, ucp_Syloti_Nagri }, + { "Syriac", PT_SC, ucp_Syriac }, + { "Tagalog", PT_SC, ucp_Tagalog }, + { "Tagbanwa", PT_SC, ucp_Tagbanwa }, + { "Tai_Le", PT_SC, ucp_Tai_Le }, + { "Tamil", PT_SC, ucp_Tamil }, + { "Telugu", PT_SC, ucp_Telugu }, + { "Thaana", PT_SC, ucp_Thaana }, + { "Thai", PT_SC, ucp_Thai }, + { "Tibetan", PT_SC, ucp_Tibetan }, + { "Tifinagh", PT_SC, ucp_Tifinagh }, + { "Ugaritic", PT_SC, ucp_Ugaritic }, + { "Yi", PT_SC, ucp_Yi }, + { "Z", PT_GC, ucp_Z }, + { "Zl", PT_PC, ucp_Zl }, + { "Zp", PT_PC, ucp_Zp }, + { "Zs", PT_PC, ucp_Zs } +}; + +const int _pcre_utt_size = sizeof(_pcre_utt)/sizeof(ucp_type_table); +#endif + +/* End of pcre_tables.c */ ============================================================ --- pcre/pcre_try_flipped.c 4d22ad726b0e1decb6c0ebaecd607b7def17d642 +++ pcre/pcre_try_flipped.c 4d22ad726b0e1decb6c0ebaecd607b7def17d642 @@ -0,0 +1,132 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + +/* PCRE is a library of functions to support regular expressions whose syntax +and semantics are as close as possible to those of the Perl 5 language. + + Written by Philip Hazel + Copyright (c) 1997-2006 University of Cambridge + +----------------------------------------------------------------------------- +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + + * Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimer. + + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + * Neither the name of the University of Cambridge nor the names of its + contributors may be used to endorse or promote products derived from + this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE +LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR +CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +----------------------------------------------------------------------------- +*/ + + +/* This module contains an internal function that tests a compiled pattern to +see if it was compiled with the opposite endianness. If so, it uses an +auxiliary local function to flip the appropriate bytes. */ + + +#include "pcre_internal.h" + + +/************************************************* +* Flip bytes in an integer * +*************************************************/ + +/* This function is called when the magic number in a regex doesn't match, in +order to flip its bytes to see if we are dealing with a pattern that was +compiled on a host of different endianness. If so, this function is used to +flip other byte values. + +Arguments: + value the number to flip + n the number of bytes to flip (assumed to be 2 or 4) + +Returns: the flipped value +*/ + +static unsigned long int +byteflip(unsigned long int value, int n) +{ +if (n == 2) return ((value & 0x00ff) << 8) | ((value & 0xff00) >> 8); +return ((value & 0x000000ff) << 24) | + ((value & 0x0000ff00) << 8) | + ((value & 0x00ff0000) >> 8) | + ((value & 0xff000000) >> 24); +} + + + +/************************************************* +* Test for a byte-flipped compiled regex * +*************************************************/ + +/* This function is called from pcre_exec(), pcre_dfa_exec(), and also from +pcre_fullinfo(). Its job is to test whether the regex is byte-flipped - that +is, it was compiled on a system of opposite endianness. The function is called +only when the native MAGIC_NUMBER test fails. If the regex is indeed flipped, +we flip all the relevant values into a different data block, and return it. + +Arguments: + re points to the regex + study points to study data, or NULL + internal_re points to a new regex block + internal_study points to a new study block + +Returns: the new block if is is indeed a byte-flipped regex + NULL if it is not +*/ + +real_pcre * +_pcre_try_flipped(const real_pcre *re, real_pcre *internal_re, + const pcre_study_data *study, pcre_study_data *internal_study) +{ +if (byteflip(re->magic_number, sizeof(re->magic_number)) != MAGIC_NUMBER) + return NULL; + +*internal_re = *re; /* To copy other fields */ +internal_re->size = byteflip(re->size, sizeof(re->size)); +internal_re->options = byteflip(re->options, sizeof(re->options)); +internal_re->top_bracket = + (pcre_uint16)byteflip(re->top_bracket, sizeof(re->top_bracket)); +internal_re->top_backref = + (pcre_uint16)byteflip(re->top_backref, sizeof(re->top_backref)); +internal_re->first_byte = + (pcre_uint16)byteflip(re->first_byte, sizeof(re->first_byte)); +internal_re->req_byte = + (pcre_uint16)byteflip(re->req_byte, sizeof(re->req_byte)); +internal_re->name_table_offset = + (pcre_uint16)byteflip(re->name_table_offset, sizeof(re->name_table_offset)); +internal_re->name_entry_size = + (pcre_uint16)byteflip(re->name_entry_size, sizeof(re->name_entry_size)); +internal_re->name_count = + (pcre_uint16)byteflip(re->name_count, sizeof(re->name_count)); + +if (study != NULL) + { + *internal_study = *study; /* To copy other fields */ + internal_study->size = byteflip(study->size, sizeof(study->size)); + internal_study->options = byteflip(study->options, sizeof(study->options)); + } + +return internal_re; +} + +/* End of pcre_tryflipped.c */ ============================================================ --- pcre/pcre_version.c 0e4d6ac360d5acfba16e4d071b69d954488e7a5f +++ pcre/pcre_version.c 0e4d6ac360d5acfba16e4d071b69d954488e7a5f @@ -0,0 +1,64 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + +/* PCRE is a library of functions to support regular expressions whose syntax +and semantics are as close as possible to those of the Perl 5 language. + + Written by Philip Hazel + Copyright (c) 1997-2006 University of Cambridge + +----------------------------------------------------------------------------- +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + + * Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimer. + + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + * Neither the name of the University of Cambridge nor the names of its + contributors may be used to endorse or promote products derived from + this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE +LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR +CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +----------------------------------------------------------------------------- +*/ + + +/* This module contains the external function pcre_version(), which returns a +string that identifies the PCRE version that is in use. */ + + +#include "pcre_internal.h" + + +/************************************************* +* Return version string * +*************************************************/ + +#define STRING(a) # a +#define XSTRING(s) STRING(s) + +PCRE_DATA_SCOPE const char * +pcre_version(void) +{ +return XSTRING(PCRE_MAJOR) + "." XSTRING(PCRE_MINOR) + XSTRING(PCRE_PRERELEASE) + " " XSTRING(PCRE_DATE); +} + +/* End of pcre_version.c */ ============================================================ --- pcre/pcre_xclass.c d672ac5b6a31841662bca6917df7d31537ae7f22 +++ pcre/pcre_xclass.c d672ac5b6a31841662bca6917df7d31537ae7f22 @@ -0,0 +1,144 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + +/* PCRE is a library of functions to support regular expressions whose syntax +and semantics are as close as possible to those of the Perl 5 language. + + Written by Philip Hazel + Copyright (c) 1997-2006 University of Cambridge + +----------------------------------------------------------------------------- +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + + * Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimer. + + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + + * Neither the name of the University of Cambridge nor the names of its + contributors may be used to endorse or promote products derived from + this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE +LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR +CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +----------------------------------------------------------------------------- +*/ + + +/* This module contains an internal function that is used to match an extended +class (one that contains characters whose values are > 255). It is used by both +pcre_exec() and pcre_def_exec(). */ + + +#include "pcre_internal.h" + + +/************************************************* +* Match character against an XCLASS * +*************************************************/ + +/* This function is called to match a character against an extended class that +might contain values > 255. + +Arguments: + c the character + data points to the flag byte of the XCLASS data + +Returns: TRUE if character matches, else FALSE +*/ + +BOOL +_pcre_xclass(int c, const uschar *data) +{ +int t; +BOOL negated = (*data & XCL_NOT) != 0; + +/* Character values < 256 are matched against a bitmap, if one is present. If +not, we still carry on, because there may be ranges that start below 256 in the +additional data. */ + +if (c < 256) + { + if ((*data & XCL_MAP) != 0 && (data[1 + c/8] & (1 << (c&7))) != 0) + return !negated; /* char found */ + } + +/* First skip the bit map if present. Then match against the list of Unicode +properties or large chars or ranges that end with a large char. We won't ever +encounter XCL_PROP or XCL_NOTPROP when UCP support is not compiled. */ + +if ((*data++ & XCL_MAP) != 0) data += 32; + +while ((t = *data++) != XCL_END) + { + int x, y; + if (t == XCL_SINGLE) + { + GETCHARINC(x, data); + if (c == x) return !negated; + } + else if (t == XCL_RANGE) + { + GETCHARINC(x, data); + GETCHARINC(y, data); + if (c >= x && c <= y) return !negated; + } + +#ifdef SUPPORT_UCP + else /* XCL_PROP & XCL_NOTPROP */ + { + int chartype, script; + int category = _pcre_ucp_findprop(c, &chartype, &script); + + switch(*data) + { + case PT_ANY: + if (t == XCL_PROP) return !negated; + break; + + case PT_LAMP: + if ((chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt) == + (t == XCL_PROP)) return !negated; + break; + + case PT_GC: + if ((data[1] == category) == (t == XCL_PROP)) return !negated; + break; + + case PT_PC: + if ((data[1] == chartype) == (t == XCL_PROP)) return !negated; + break; + + case PT_SC: + if ((data[1] == script) == (t == XCL_PROP)) return !negated; + break; + + /* This should never occur, but compilers may mutter if there is no + default. */ + + default: + return FALSE; + } + + data += 2; + } +#endif /* SUPPORT_UCP */ + } + +return negated; /* char did not match */ +} + +/* End of pcre_xclass.c */ ============================================================ --- .mtn-ignore 53ba0a32a8d3d257787d12f2e37b8c1329c664b4 +++ .mtn-ignore b587a7b3db89e9f0549ddcae609f5ccf3853f1eb @@ -5,11 +5,14 @@ tester_dir \.dirstamp ^txt2c$ ^unit_tests$ +^unit_tests\.log$ ^mtn$ ^testsuite$ ^testsuite\.log$ ^tester$ ^tester\.h$ +^run_lua_tests$ +^run_tester_tests$ ^package_full_revision(\.c|\.txt|_dist\.txt|_raw\.txt)$ ^package_revision_raw.txt$ ^config\. ============================================================ --- ChangeLog a7d6d3a553ba406d13b20f6e11d251ca4158a339 +++ ChangeLog 49c9aa7b2d19c2e0d8a55800e9f9082df054ec9d @@ -1,3 +1,16 @@ +2006-12-18 Zack Weinberg + + * pcre: New directory, contains trimmed-down version of PCRE library. + * Makefile.am: Add support for compiling either with our bundled + PCRE or with a system-provided library. + * m4/pcre.m4: New file, provides MTN_LIB_PCRE. + * configure.ac: Use MTN_LIB_PCRE. + * debian/control: Adjust Build-Depends. + + * m4/boost.m4: Use libboost_filesystem, not libboost_regex, in + BOOST_STATIC_LINK_OPTION. Kill MTN_BOOST_LIB_REGEX. + * .mtn-ignore: Add a few missing files. + 2006-11-27 Zack Weinberg * packet.cc (feed_packet_consumer): Convert from functor object ============================================================ --- Makefile.am cdd7815c1e675ef9cb034b2a0fd9d2bdb36a1f55 +++ Makefile.am 2cdac9ee2ebeba3ebcf3ff0855c993416e227207 @@ -257,6 +257,15 @@ IDNA_SOURCES = \ idna/gunicomp.h idna/idna.c idna/idn-int.h idna/profiles.c \ idna/punycode.h idna/stringprep.c idna/toutf8.c +PCRE_SOURCES = \ + pcre/pcre_chartables.c pcre/pcre_compile.c pcre/pcre_config.c \ + pcre/pcre_config.h pcre/pcre_dfa_exec.c pcre/pcre_exec.c \ + pcre/pcre_fullinfo.c pcre/pcre_get.c pcre/pcre_globals.c \ + pcre/pcre.h pcre/pcre_info.c pcre/pcre_internal.h \ + pcre/pcre_maketables.c pcre/pcre_refcount.c pcre/pcre_study.c \ + pcre/pcre_tables.c pcre/pcre_try_flipped.c pcre/pcre_version.c \ + pcre/pcre_xclass.c + UNIX_PLATFORM_SOURCES = \ unix/read_password.cc unix/get_system_flavour.cc \ unix/process.cc unix/terminal.cc unix/inodeprint.cc \ @@ -292,6 +301,10 @@ lib3rdparty_a_SOURCES = $(BOOST_SANDBOX_ $(LUA_SOURCES) \ $(SQLITE_SOURCES) +if INCLUDED_PCRE +lib3rdparty_a_SOURCES += $(PCRE_SOURCES) +endif + htmldir = $(datadir)/doc/monotone html_DATA = monotone.html @@ -351,6 +364,21 @@ tester_LDADD = lib3rdparty.a $(BOOSTLIBS tester_CXXFLAGS = $(AM_CXXFLAGS) tester_LDADD = lib3rdparty.a $(BOOSTLIBS) libplatform.a $(LIBICONV) $(LIBINTL) +if INCLUDED_PCRE + lib3rdparty_a_CPPFLAGS += -I$(top_srcdir)/pcre + mtn_CPPFLAGS += -I$(top_srcdir)/pcre + unit_tests_CPPFLAGS += -I$(top_srcdir)/pcre + tester_CPPFLAGS += -I$(top_srcdir)/pcre +else + mtn_CPPFLAGS += $(PCRE_CFLAGS) + unit_tests_CPPFLAGS += $(PCRE_CFLAGS) + tester_CPPFLAGS += $(PCRE_CFLAGS) + + mtn_LDADD += $(PCRE_LIBS) + unit_tests_LDADD += $(PCRE_LIBS) + tester_LDADD += $(PCRE_LIBS) +endif + if WIN32_PLATFORM libplatform_a_SOURCES += $(WIN32_PLATFORM_SOURCES) mtn_SOURCES += win32/main.cc ============================================================ --- configure.ac 661fc6adca568307e46e8d53d8dbd51c6ead71f0 +++ configure.ac 741803702d24a7cc1ebb7889cd061bfa35b0aaed @@ -97,8 +97,7 @@ AC_SEARCH_LIBS([deflate], [z], , AC_MSG_ # simple library checks AC_SEARCH_LIBS([deflate], [z], , AC_MSG_FAILURE([zlib is required])) -# FIXME Use pcre-config; bundled pcre support. -AC_SEARCH_LIBS([pcre_exec], [pcre], , AC_MSG_FAILURE([libpcre is required])) +MTN_LIB_PCRE # check for all things boost-related BOOST_VERSION_CHECK ============================================================ --- debian/control 72d4ce11e07e8653805b1872da91c6b52cc4acb3 +++ debian/control a04bbd93b10b7bc711c39b63bad977f1a20cb081 @@ -1,10 +1,10 @@ Build-Depends: cdbs (>= 0.4.28), debhelp Source: monotone Section: devel Priority: optional Maintainer: Shaun Jackman Build-Depends: cdbs (>= 0.4.28), debhelper (>= 4.0.0), autotools-dev, - libboost-date-time-dev, libboost-filesystem-dev, libboost-regex-dev, - libboost-test-dev, libboost-dev, libz-dev, + libboost-date-time-dev, libboost-filesystem-dev, + libboost-test-dev, libboost-dev, libpcre3-dev, libz-dev, Build-Depends-Indep: ps2eps, tetex-bin, texinfo, xpdf-utils Standards-Version: 3.7.2.2 ============================================================ --- m4/boost.m4 005410a3f9afe57b52fd5bfc18330155d400ac6f +++ m4/boost.m4 d98c8d3dc1f0acf7d6ca96acf4418eaa259d04cf @@ -92,11 +92,11 @@ AC_DEFUN([BOOST_STATIC_LINK_OPTION], # pick the one that matches the code the compiler generates). # Fortunately, this is C++, so we'll get link errors if we get # it wrong. - if test -f $i/lib64/libboost_regex${s}.a + if test -f $i/lib64/libboost_filesystem${s}.a then BOOST_LIBDIR=$i/lib64 fi - if test -f $i/lib/libboost_regex${s}.a + if test -f $i/lib/libboost_filesystem${s}.a then BOOST_LIBDIR=$i/lib fi @@ -183,15 +183,6 @@ AC_DEFUN([MTN_BOOST_LIB_DATE_TIME], std::cout << to_iso_extended_string(second_clock::universal_time()); ]])])]) -AC_DEFUN([MTN_BOOST_LIB_REGEX], -[MTN_NEED_BOOST_LIB([regex], - [AC_LANG_PROGRAM([[ - #include - using namespace boost; - ]],[[ - regex expr("foo"); - ]])])]) - dnl Unlike all the others, if we don't have this library we can still dnl build monotone; we just can't do all the tests. AC_DEFUN([MTN_BOOST_LIB_UNIT_TEST_FRAMEWORK],