From 5b81acfae24e8c113d65ce351620942c230d1cba Mon Sep 17 00:00:00 2001 From: Matthias Clasen Date: Wed, 7 Nov 2007 04:57:22 +0000 Subject: [PATCH] Update the internal copy of PCRE to 7.4 2007-11-06 Matthias Clasen * glib/pcre/*: Update the internal copy of PCRE to 7.4 svn path=/trunk/; revision=5800 --- ChangeLog | 4 + glib/pcre/Makefile.am | 5 +- glib/pcre/makefile.msc | 6 +- glib/pcre/pcre.h | 49 +++- glib/pcre/pcre_chartables.c | 9 +- glib/pcre/pcre_compile.c | 649 ++++++++++++++++++++++++++++++------------- glib/pcre/pcre_config.c | 12 + glib/pcre/pcre_dfa_exec.c | 90 ++++-- glib/pcre/pcre_exec.c | 361 ++++++++++++++++-------- glib/pcre/pcre_fullinfo.c | 18 +- glib/pcre/pcre_get.c | 6 +- glib/pcre/pcre_globals.c | 4 + glib/pcre/pcre_info.c | 8 +- glib/pcre/pcre_internal.h | 97 ++++--- glib/pcre/pcre_maketables.c | 5 +- glib/pcre/pcre_newline.c | 8 + glib/pcre/pcre_ord2utf8.c | 3 + glib/pcre/pcre_refcount.c | 5 + glib/pcre/pcre_study.c | 7 +- glib/pcre/pcre_tables.c | 254 +++++++++-------- glib/pcre/pcre_try_flipped.c | 5 + glib/pcre/pcre_version.c | 4 + glib/pcre/pcre_xclass.c | 4 + 23 files changed, 1085 insertions(+), 528 deletions(-) diff --git a/ChangeLog b/ChangeLog index eb4996f..9f11c9b 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,3 +1,7 @@ +2007-11-06 Matthias Clasen + + * glib/pcre/*: Update the internal copy of PCRE to 7.4 + 2007-10-22 Tor Lillqvist * glib/gutils.c (_glib_gettext): Plug small one-time leak on diff --git a/glib/pcre/Makefile.am b/glib/pcre/Makefile.am index 2f0b0ac..358f66e 100644 --- a/glib/pcre/Makefile.am +++ b/glib/pcre/Makefile.am @@ -9,12 +9,10 @@ INCLUDES = \ -DMAX_NAME_COUNT=10000 \ -DMAX_DUPLENGTH=30000 \ -DLINK_SIZE=2 \ - -UEBCDIC \ -DPOSIX_MALLOC_THRESHOLD=10 \ -I$(top_srcdir) \ -I$(srcdir) \ -I$(top_srcdir)/glib \ - -I$(top_builddir)/glib \ @GLIB_DEBUG_FLAGS@ \ -DG_DISABLE_DEPRECATED \ -DGLIB_COMPILATION \ @@ -28,8 +26,8 @@ noinst_LTLIBRARIES = libpcre.la libpcre_headers = libpcre_la_SOURCES = \ - pcre_chartables.c \ pcre_compile.c \ + pcre_chartables.c \ pcre_config.c \ pcre_dfa_exec.c \ pcre_exec.c \ @@ -51,7 +49,6 @@ libpcre_la_SOURCES = \ pcre.h \ pcre_internal.h \ ucp.h \ - ucpinternal.h \ $(libpcre_headers) libpcre_la_LIBADD = $(DEP_LIBS) diff --git a/glib/pcre/makefile.msc b/glib/pcre/makefile.msc index 3e4278b..a9306ad 100644 --- a/glib/pcre/makefile.msc +++ b/glib/pcre/makefile.msc @@ -22,7 +22,7 @@ DEFINES = \ -DPOSIX_MALLOC_THRESHOLD=10 OBJECTS = \ - pcre_chartables.obj \ + \.obj \ pcre_compile.obj \ pcre_config.obj \ pcre_dfa_exec.obj \ @@ -31,6 +31,7 @@ OBJECTS = \ pcre_get.obj \ pcre_globals.obj \ pcre_info.obj \ + pcre_internal.obj \ pcre_maketables.obj \ pcre_newline.obj \ pcre_ord2utf8.obj \ @@ -42,6 +43,9 @@ OBJECTS = \ pcre_valid_utf8.obj \ pcre_version.obj \ pcre_xclass.obj \ + ucp.obj \ + ucpinternal.obj \ + ucptable.h.obj \ all : pcre.lib diff --git a/glib/pcre/pcre.h b/glib/pcre/pcre.h index 7ed12be..a95f1ca 100644 --- a/glib/pcre/pcre.h +++ b/glib/pcre/pcre.h @@ -42,33 +42,45 @@ POSSIBILITY OF SUCH DAMAGE. /* The current PCRE version information. */ #define PCRE_MAJOR 7 -#define PCRE_MINOR 2 +#define PCRE_MINOR 4 #define PCRE_PRERELEASE -#define PCRE_DATE 2007-06-19 +#define PCRE_DATE 2007-09-21 /* When an application links to a PCRE DLL in Windows, the symbols that are imported have to be identified as such. When building PCRE, the appropriate export setting is defined in pcre_internal.h, which includes this file. So we -don't change an existing definition of PCRE_EXP_DECL. */ +don't change existing definitions of PCRE_EXP_DECL and PCRECPP_EXP_DECL. */ -/* But don't do that when building as part of GLib */ -#if 0 -#ifndef PCRE_EXP_DECL -# ifdef _WIN32 -# ifndef PCRE_STATIC -# define PCRE_EXP_DECL extern __declspec(dllimport) +#if defined(_WIN32) && !defined(PCRE_STATIC) +# ifndef PCRE_EXP_DECL +# define PCRE_EXP_DECL extern __declspec(dllimport) +# endif +# ifdef __cplusplus +# ifndef PCRECPP_EXP_DECL +# define PCRECPP_EXP_DECL extern __declspec(dllimport) +# endif +# ifndef PCRECPP_EXP_DEFN +# define PCRECPP_EXP_DEFN __declspec(dllimport) # endif # endif #endif -#endif /* By default, we use the standard "extern" declarations. */ #ifndef PCRE_EXP_DECL # ifdef __cplusplus -# define PCRE_EXP_DECL extern "C" +# define PCRE_EXP_DECL extern "C" # else -# define PCRE_EXP_DECL extern +# define PCRE_EXP_DECL extern +# endif +#endif + +#ifdef __cplusplus +# ifndef PCRECPP_EXP_DECL +# define PCRECPP_EXP_DECL extern +# endif +# ifndef PCRECPP_EXP_DEFN +# define PCRECPP_EXP_DEFN # endif #endif @@ -110,6 +122,8 @@ extern "C" { #define PCRE_NEWLINE_CRLF 0x00300000 #define PCRE_NEWLINE_ANY 0x00400000 #define PCRE_NEWLINE_ANYCRLF 0x00500000 +#define PCRE_BSR_ANYCRLF 0x00800000 +#define PCRE_BSR_UNICODE 0x01000000 /* Exec-time and get/set-time error codes */ @@ -135,7 +149,7 @@ extern "C" { #define PCRE_ERROR_DFA_WSSIZE (-19) #define PCRE_ERROR_DFA_RECURSE (-20) #define PCRE_ERROR_RECURSIONLIMIT (-21) -#define PCRE_ERROR_NULLWSLIMIT (-22) +#define PCRE_ERROR_NULLWSLIMIT (-22) /* No longer actually used */ #define PCRE_ERROR_BADNEWLINE (-23) /* Request types for pcre_fullinfo() */ @@ -155,6 +169,7 @@ extern "C" { #define PCRE_INFO_DEFAULT_TABLES 11 #define PCRE_INFO_OKPARTIAL 12 #define PCRE_INFO_JCHANGED 13 +#define PCRE_INFO_HASCRORLF 14 /* Request types for pcre_config(). Do not re-arrange, in order to remain compatible. */ @@ -167,6 +182,7 @@ compatible. */ #define PCRE_CONFIG_STACKRECURSE 5 #define PCRE_CONFIG_UNICODE_PROPERTIES 6 #define PCRE_CONFIG_MATCH_LIMIT_RECURSION 7 +#define PCRE_CONFIG_BSR 8 /* 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. */ @@ -226,6 +242,13 @@ typedef struct pcre_callout_block { /* ------------------------------------------------------------------ */ } 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. */ + #include "glib.h" #include "galias.h" diff --git a/glib/pcre/pcre_chartables.c b/glib/pcre/pcre_chartables.c index 6494d8e..75e46a0 100644 --- a/glib/pcre/pcre_chartables.c +++ b/glib/pcre/pcre_chartables.c @@ -1,3 +1,6 @@ +/* This file is autogenerated by ../update-pcre/update.sh during + * the update of the local copy of PCRE. + */ /************************************************* * Perl-Compatible Regular Expressions * *************************************************/ @@ -14,12 +17,16 @@ example ISO-8859-1. When dftables is run, it creates these tables in the current locale. If PCRE is configured with --enable-rebuild-chartables, this happens automatically. -The following #include is present because without it gcc 4.x may remove the +The following #includes are present because without the gcc 4.x may remove the array definition from the final binary if PCRE is built into a static library and dead code stripping is activated. This leads to link errors. Pulling in the header ensures that the array gets flagged as "someone outside this compilation unit might reference this" and so it will always be supplied to the linker. */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + #include "pcre_internal.h" const unsigned char _pcre_default_tables[] = { diff --git a/glib/pcre/pcre_compile.c b/glib/pcre/pcre_compile.c index f85eca3..26f87c3 100644 --- a/glib/pcre/pcre_compile.c +++ b/glib/pcre/pcre_compile.c @@ -42,11 +42,14 @@ POSSIBILITY OF SUCH DAMAGE. supporting internal functions that are not used by other modules. */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + #define NLBLOCK cd /* Block containing newline information */ #define PSSTART start_pattern /* Field containing processed string start */ #define PSEND end_pattern /* Field containing processed string end */ - #include "pcre_internal.h" @@ -62,6 +65,13 @@ used by pcretest. DEBUG is not defined when building a production library. */ #define SETBIT(a,b) a[b/8] |= (1 << (b%8)) +/* Maximum length value to check against when making sure that the integer that +holds the compiled pattern length does not overflow. We make it a bit less than +INT_MAX to allow for adding in group terminating bytes, so that we don't have +to check them every time. */ + +#define OFLOW_MAX (INT_MAX - 20) + /************************************************* * Code parameters and static tables * @@ -120,7 +130,7 @@ static const short int escapes[] = { /* B8 */ 0, 0, 0, 0, 0, ']', '=', '-', /* C0 */ '{',-ESC_A, -ESC_B, -ESC_C, -ESC_D,-ESC_E, 0, -ESC_G, /* C8 */-ESC_H, 0, 0, 0, 0, 0, 0, 0, -/* D0 */ '}', 0, 0, 0, 0, 0, 0, -ESC_P, +/* D0 */ '}', 0, -ESC_K, 0, 0, 0, 0, -ESC_P, /* D8 */-ESC_Q,-ESC_R, 0, 0, 0, 0, 0, 0, /* E0 */ '\\', 0, -ESC_S, 0, 0,-ESC_V, -ESC_W, -ESC_X, /* E8 */ 0,-ESC_Z, 0, 0, 0, 0, 0, 0, @@ -130,25 +140,47 @@ static const short int escapes[] = { #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. */ +/* Table of special "verbs" like (*PRUNE). This is a short table, so it is +searched linearly. Put all the names into a single string, in order to reduce +the number of relocations when a shared library is dynamically linked. */ + +typedef struct verbitem { + int len; + int op; +} verbitem; + +static const char verbnames[] = + "ACCEPT\0" + "COMMIT\0" + "F\0" + "FAIL\0" + "PRUNE\0" + "SKIP\0" + "THEN"; + +static verbitem verbs[] = { + { 6, OP_ACCEPT }, + { 6, OP_COMMIT }, + { 1, OP_FAIL }, + { 4, OP_FAIL }, + { 5, OP_PRUNE }, + { 4, OP_SKIP }, + { 4, OP_THEN } +}; + +static int verbcount = sizeof(verbs)/sizeof(verbitem); + + +/* Tables of names of POSIX character classes and their lengths. The names are +now all in a single string, to reduce the number of relocations when a shared +library is dynamically loaded. The list of lengths 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 posix_names[] = - "alpha\0" - "lower\0" - "upper\0" - "alnum\0" - "ascii\0" - "blank\0" - "cntrl\0" - "digit\0" - "graph\0" - "print\0" - "punct\0" - "space\0" - "word\0" - "xdigit"; + "alpha\0" "lower\0" "upper\0" "alnum\0" "ascii\0" "blank\0" + "cntrl\0" "digit\0" "graph\0" "print\0" "punct\0" "space\0" + "word\0" "xdigit"; static const uschar posix_name_lengths[] = { 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 }; @@ -187,11 +219,14 @@ static const int posix_class_maps[] = { /* The texts of compile-time error messages. These are "char *" because they are passed to the outside world. Do not ever re-use any error number, because they are documented. Always add a new error instead. Messages marked DEAD below -are no longer used. */ - -#define DEAD(s) "\0" - -static const char error_texts[] = +are no longer used. This used to be a table of strings, but in order to reduce +the number of relocations needed when a shared library is loaded dynamically, +it is now one long string. We cannot use a table of offsets, because the +lengths of inserts such as XSTRING(MAX_NAME_SIZE) are not known. Instead, we +simply count through to the one we want - this isn't a performance issue +because these strings are used only when there is a compilation error. */ + +static const char error_texts[] = "no error\0" "\\ at end of pattern\0" "\\c at end of pattern\0" @@ -204,7 +239,7 @@ static const char error_texts[] = "range out of order in character class\0" "nothing to repeat\0" /* 10 */ - DEAD("operand of unlimited repeat could match the empty string") + "operand of unlimited repeat could match the empty string\0" /** DEAD **/ "internal error: unexpected repeat\0" "unrecognized character after (?\0" "POSIX named classes are supported only within a class\0" @@ -214,9 +249,9 @@ static const char error_texts[] = "erroffset passed as NULL\0" "unknown option bit(s) set\0" "missing ) after comment\0" - DEAD("parentheses nested too deeply") + "parentheses nested too deeply\0" /** DEAD **/ /* 20 */ - "regular expression too large\0" + "regular expression is too large\0" "failed to get memory\0" "unmatched parentheses\0" "internal error: code overflow\0" @@ -231,7 +266,7 @@ static const char error_texts[] = "unknown POSIX class name\0" "POSIX collating elements are not supported\0" "this version of PCRE is not compiled with PCRE_UTF8 support\0" - DEAD("spare error") + "spare error\0" /** DEAD **/ "character value in \\x{...} sequence is too large\0" /* 35 */ "invalid condition (?(0)\0" @@ -252,7 +287,7 @@ static const char error_texts[] = "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)\0" "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0" /* 50 */ - "repeated subpattern is too long\0" + "repeated subpattern is too long\0" /** DEAD **/ "octal value is greater than \\377 (not in UTF-8 mode)\0" "internal error: overran compiling workspace\0" "internal error: previously-checked referenced subpattern not found\0" @@ -261,69 +296,13 @@ static const char error_texts[] = "repeating a DEFINE group is not allowed\0" "inconsistent NEWLINE options\0" "\\g is not followed by a braced name or an optionally braced non-zero number\0" - "(?+ or (?- or (?(+ or (?(- must be followed by a non-zero number"; - -static const int error_texts_offsets[] = { - 0, - 9, - 29, - 50, - 83, - 121, - 153, - 195, - 238, - 276, - 294, - 295, - 329, - 361, - 415, - 425, - 462, - 487, - 513, - 537, - 538, - 567, - 588, - 610, - 640, - 673, - 714, - 749, - 799, - 828, - 866, - 891, - 934, - 994, - 995, - 1044, - 1068, - 1107, - 1151, - 1177, - 1204, - 1243, - 1276, - 1329, - 1370, - 1391, - 1440, - 1468, - 1505, - 1557, - 1600, - 1632, - 1685, - 1729, - 1796, - 1839, - 1879, - 1908, - 1984 -}; + "(?+ or (?- or (?(+ or (?(- must be followed by a non-zero number\0" + "(*VERB) with an argument is not supported\0" + /* 60 */ + "(*VERB) not recognized\0" + "number is too big"; + + /* Definition to allow mutual recursion */ static BOOL @@ -333,6 +312,28 @@ static BOOL /************************************************* +* Find an error text * +*************************************************/ + +/* The error texts are now all in one long string, to save on relocations. As +some of the text is of unknown length, we can't use a table of offsets. +Instead, just count through the strings. This is not a performance issue +because it happens only when there has been a compilation error. + +Argument: the error number +Returns: pointer to the error string +*/ + +static const char * +find_error_text(int n) +{ +const char *s = error_texts; +for (; n > 0; n--) while (*s++ != 0); +return s; +} + + +/************************************************* * Handle escapes * *************************************************/ @@ -353,7 +354,7 @@ Arguments: Returns: zero or positive => a data character negative => a special escape sequence - on error, errorptr is set + on error, errorcodeptr is set */ static int @@ -438,10 +439,16 @@ else while (g_ascii_isdigit(ptr[1]) != 0) c = c * 10 + *(++ptr) - '0'; + if (c < 0) + { + *errorcodeptr = ERR61; + break; + } + if (c == 0 || (braced && *(++ptr) != '}')) { *errorcodeptr = ERR57; - return 0; + break; } if (negated) @@ -449,7 +456,7 @@ else if (c > bracount) { *errorcodeptr = ERR15; - return 0; + break; } c = bracount - (c - 1); } @@ -478,6 +485,11 @@ else c -= '0'; while (g_ascii_isdigit(ptr[1]) != 0) c = c * 10 + *(++ptr) - '0'; + if (c < 0) + { + *errorcodeptr = ERR61; + break; + } if (c < 10 || c <= bracount) { c = -(ESC_REF + c); @@ -573,7 +585,7 @@ else if (c == 0) { *errorcodeptr = ERR2; - return 0; + break; } #ifndef EBCDIC /* ASCII coding */ @@ -649,7 +661,7 @@ if (c == '{') *negptr = TRUE; ptr++; } - for (i = 0; i < sizeof(name) - 1; i++) + for (i = 0; i < (int)sizeof(name) - 1; i++) { c = *(++ptr); if (c == 0) goto ERROR_RETURN; @@ -678,7 +690,7 @@ top = _pcre_utt_size; while (bot < top) { i = (bot + top) >> 1; - c = strcmp(name, &_pcre_ucp_names[_pcre_utt[i].offset]); + c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset); if (c == 0) { *dptr = _pcre_utt[i].value; @@ -852,6 +864,7 @@ for (; *ptr != 0; ptr++) { while (*(++ptr) != ']') { + if (*ptr == 0) return -1; if (*ptr == '\\') { if (*(++ptr) == 0) return -1; @@ -879,7 +892,7 @@ for (; *ptr != 0; ptr++) /* An opening parens must now be a real metacharacter */ if (*ptr != '(') continue; - if (ptr[1] != '?') + if (ptr[1] != '?' && ptr[1] != '*') { count++; if (name == NULL && count == lorn) return count; @@ -1007,7 +1020,6 @@ for (;;) { int d; register int op = *cc; - switch (op) { case OP_CBRA: @@ -1096,6 +1108,7 @@ for (;;) case OP_TYPEEXACT: branchlength += GET2(cc,1); + if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2; cc += 4; break; @@ -1204,13 +1217,42 @@ for (;;) code += _pcre_OP_lengths[c]; } - /* 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 - arrange to skip the extra bytes. */ + /* Otherwise, we can get the item's length from the table, except that for + repeated character types, we have to test for \p and \P, which have an extra + two bytes of parameters. */ else { + switch(c) + { + case OP_TYPESTAR: + case OP_TYPEMINSTAR: + case OP_TYPEPLUS: + case OP_TYPEMINPLUS: + case OP_TYPEQUERY: + case OP_TYPEMINQUERY: + case OP_TYPEPOSSTAR: + case OP_TYPEPOSPLUS: + case OP_TYPEPOSQUERY: + if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2; + break; + + case OP_TYPEUPTO: + case OP_TYPEMINUPTO: + case OP_TYPEEXACT: + case OP_TYPEPOSUPTO: + if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2; + break; + } + + /* Add in the fixed length from the table */ + code += _pcre_OP_lengths[c]; + + /* 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 + arrange to skip the extra bytes. */ + #ifdef SUPPORT_UTF8 if (utf8) switch(c) { @@ -1268,14 +1310,42 @@ for (;;) if (c == OP_XCLASS) code += GET(code, 1); - /* 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 arrange to skip the extra - bytes. */ + /* Otherwise, we can get the item's length from the table, except that for + repeated character types, we have to test for \p and \P, which have an extra + two bytes of parameters. */ else { + switch(c) + { + case OP_TYPESTAR: + case OP_TYPEMINSTAR: + case OP_TYPEPLUS: + case OP_TYPEMINPLUS: + case OP_TYPEQUERY: + case OP_TYPEMINQUERY: + case OP_TYPEPOSSTAR: + case OP_TYPEPOSPLUS: + case OP_TYPEPOSQUERY: + if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2; + break; + + case OP_TYPEPOSUPTO: + case OP_TYPEUPTO: + case OP_TYPEMINUPTO: + case OP_TYPEEXACT: + if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2; + break; + } + + /* Add in the fixed length from the table */ + code += _pcre_OP_lengths[c]; + + /* 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 arrange to skip the extra bytes. */ + #ifdef SUPPORT_UTF8 if (utf8) switch(c) { @@ -1347,7 +1417,7 @@ for (code = first_significant_code(code + _pcre_OP_lengths[*code], NULL, 0, TRUE /* For other groups, scan the branches. */ - if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE) + if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND) { BOOL empty_branch; if (GET(code, 1) == 0) return TRUE; /* Hit unclosed bracket */ @@ -1371,11 +1441,15 @@ for (code = first_significant_code(code + _pcre_OP_lengths[*code], NULL, 0, TRUE switch (c) { - /* Check for quantifiers after a class */ + /* Check for quantifiers after a class. XCLASS is used for classes that + cannot be represented just by a bit map. This includes negated single + high-valued characters. The length in _pcre_OP_lengths[] is zero; the + actual length is stored in the compiled code, so we must update "code" + here. */ #ifdef SUPPORT_UTF8 case OP_XCLASS: - ccode = code + GET(code, 1); + ccode = code += GET(code, 1); goto CHECK_CLASS_REPEAT; #endif @@ -1437,6 +1511,26 @@ for (code = first_significant_code(code + _pcre_OP_lengths[*code], NULL, 0, TRUE case OP_TYPEEXACT: return FALSE; + /* These are going to continue, as they may be empty, but we have to + fudge the length for the \p and \P cases. */ + + case OP_TYPESTAR: + case OP_TYPEMINSTAR: + case OP_TYPEPOSSTAR: + case OP_TYPEQUERY: + case OP_TYPEMINQUERY: + case OP_TYPEPOSQUERY: + if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2; + break; + + /* Same for these */ + + case OP_TYPEUPTO: + case OP_TYPEMINUPTO: + case OP_TYPEPOSUPTO: + if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2; + break; + /* End of branch */ case OP_KET: @@ -1553,16 +1647,16 @@ Returns: a value representing the name, or -1 if unknown static int check_posix_name(const uschar *ptr, int len) { - int offset = 0; - int yield = 0; - while (posix_name_lengths[yield] != 0) - { - if (len == posix_name_lengths[yield] && - strcmp((const char *)ptr, posix_names + offset) == 0) return yield; - offset += posix_name_lengths[yield] + 1; - yield++; - } - return -1; +const char *pn = posix_names; +register int yield = 0; +while (posix_name_lengths[yield] != 0) + { + if (len == posix_name_lengths[yield] && + strncmp((const char *)ptr, pn, len) == 0) return yield; + pn += posix_name_lengths[yield] + 1; + yield++; + } +return -1; } @@ -1601,6 +1695,7 @@ adjust_recurse(uschar *group, int adjust, BOOL utf8, compile_data *cd, uschar *save_hwm) { uschar *ptr = group; + while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL) { int offset; @@ -2205,6 +2300,15 @@ for (;; ptr++) */ if (code < last_code) code = last_code; + + /* Paranoid check for integer overflow */ + + if (OFLOW_MAX - *lengthptr < code - last_code) + { + *errorcodeptr = ERR20; + goto FAILED; + } + *lengthptr += code - last_code; DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c)); @@ -2317,6 +2421,11 @@ for (;; ptr++) *ptrptr = ptr; if (lengthptr != NULL) { + if (OFLOW_MAX - *lengthptr < code - last_code) + { + *errorcodeptr = ERR20; + goto FAILED; + } *lengthptr += code - last_code; /* To include callout length */ DPRINTF((">> end branch\n")); } @@ -2379,16 +2488,23 @@ for (;; ptr++) goto FAILED; } - /* If the first character is '^', set the negation flag and skip it. */ + /* If the first character is '^', set the negation flag and skip it. Also, + if the first few characters (either before or after ^) are \Q\E or \E we + skip them too. This makes for compatibility with Perl. */ - if ((c = *(++ptr)) == '^') + negate_class = FALSE; + for (;;) { - negate_class = TRUE; c = *(++ptr); - } - else - { - negate_class = FALSE; + if (c == '\\') + { + if (ptr[1] == 'E') ptr++; + else if (strncmp((const char *)ptr+1, "Q\\E", 3) == 0) ptr += 3; + else break; + } + else if (!negate_class && c == '^') + negate_class = TRUE; + else break; } /* Keep a count of chars with values < 256 so that we can optimize the case @@ -2529,7 +2645,7 @@ for (;; ptr++) of the specials, which just set a flag. 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 + 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 == '\\') @@ -2549,6 +2665,7 @@ for (;; ptr++) else inescq = TRUE; continue; } + else if (-c == ESC_E) continue; /* Ignore orphan \E */ if (c < 0) { @@ -2775,6 +2892,12 @@ for (;; ptr++) oldptr = ptr; + /* Remember \r or \n */ + + if (c == '\r' || c == '\n') cd->external_flags |= PCRE_HASCRORLF; + + /* Check for range */ + if (!inescq && ptr[1] == '-') { int d; @@ -2842,6 +2965,10 @@ for (;; ptr++) if (d == c) goto LONE_SINGLE_CHARACTER; /* A few lines below */ + /* Remember \r or \n */ + + if (d == '\r' || d == '\n') cd->external_flags |= PCRE_HASCRORLF; + /* 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 @@ -2995,12 +3122,34 @@ for (;; ptr++) goto FAILED; } + +/* This code has been disabled because it would mean that \s counts as +an explicit \r or \n reference, and that's not really what is wanted. Now +we set the flag only if there is a literal "\r" or "\n" in the class. */ + +#if 0 + /* Remember whether \r or \n are in this class */ + + if (negate_class) + { + if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF; + } + else + { + if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF; + } +#endif + + /* 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. + less than 256. As long as there were no characters >= 128 and there was no + use of \p or \P, in other words, no use of any XCLASS features, we can + 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 @@ -3010,10 +3159,8 @@ for (;; ptr++) reqbyte, save the previous value for reinstating. */ #ifdef SUPPORT_UTF8 - if (class_charcount == 1 && - (!utf8 || - (!class_utf8 && (!negate_class || class_lastchar < 128)))) - + if (class_charcount == 1 && !class_utf8 && + (!utf8 || !negate_class || class_lastchar < 128)) #else if (class_charcount == 1) #endif @@ -3285,7 +3432,7 @@ for (;; ptr++) /* 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_max != 1) cd->external_flags |= PCRE_NOPARTIAL; /* Combine the op_type with the repeat_type */ @@ -3435,7 +3582,7 @@ for (;; ptr++) /* 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_max != 1) cd->external_flags |= PCRE_NOPARTIAL; if (repeat_min == 0 && repeat_max == -1) *code++ = OP_CRSTAR + repeat_type; @@ -3471,14 +3618,6 @@ for (;; ptr++) goto FAILED; } - /* This is a paranoid check to stop integer overflow later on */ - - if (len > MAX_DUPLENGTH) - { - *errorcodeptr = ERR50; - goto FAILED; - } - /* 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 @@ -3567,10 +3706,21 @@ for (;; ptr++) if (repeat_min > 1) { /* In the pre-compile phase, we don't actually do the replication. We - just adjust the length as if we had. */ + just adjust the length as if we had. Do some paranoid checks for + potential integer overflow. */ if (lengthptr != NULL) - *lengthptr += (repeat_min - 1)*length_prevgroup; + { + int delta = (repeat_min - 1)*length_prevgroup; + if ((double)(repeat_min - 1)*(double)length_prevgroup > + (double)INT_MAX || + OFLOW_MAX - *lengthptr < delta) + { + *errorcodeptr = ERR20; + goto FAILED; + } + *lengthptr += delta; + } /* This is compiling for real */ @@ -3608,11 +3758,23 @@ for (;; ptr++) /* In the pre-compile phase, we don't actually do the replication. We just adjust the length as if we had. For each repetition we must add 1 to the length for BRAZERO and for all but the last repetition we must - add 2 + 2*LINKSIZE to allow for the nesting that occurs. */ + add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some + paranoid checks to avoid integer overflow. */ if (lengthptr != NULL && repeat_max > 0) - *lengthptr += repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) - - 2 - 2*LINK_SIZE; /* Last one doesn't nest */ + { + int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) - + 2 - 2*LINK_SIZE; /* Last one doesn't nest */ + if ((double)repeat_max * + (double)(length_prevgroup + 1 + 2 + 2*LINK_SIZE) + > (double)INT_MAX || + OFLOW_MAX - *lengthptr < delta) + { + *errorcodeptr = ERR20; + goto FAILED; + } + *lengthptr += delta; + } /* This is compiling for real */ @@ -3764,9 +3926,7 @@ for (;; ptr++) /* ===================================================================*/ /* Start of nested parenthesized sub-expression, or comment or lookahead or lookbehind or option setting or condition or all the other extended - parenthesis forms. First deal with the specials; all are introduced by ?, - and the appearance of any of them means that this is not a capturing - group. */ + parenthesis forms. */ case '(': newoptions = options; @@ -3775,7 +3935,46 @@ for (;; ptr++) save_hwm = cd->hwm; reset_bracount = FALSE; - if (*(++ptr) == '?') + /* First deal with various "verbs" that can be introduced by '*'. */ + + if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0) + { + int i, namelen; + const char *vn = verbnames; + const uschar *name = ++ptr; + previous = NULL; + while ((cd->ctypes[*++ptr] & ctype_letter) != 0); + if (*ptr == ':') + { + *errorcodeptr = ERR59; /* Not supported */ + goto FAILED; + } + if (*ptr != ')') + { + *errorcodeptr = ERR60; + goto FAILED; + } + namelen = ptr - name; + for (i = 0; i < verbcount; i++) + { + if (namelen == verbs[i].len && + strncmp((char *)name, vn, namelen) == 0) + { + *code = verbs[i].op; + if (*code++ == OP_ACCEPT) cd->had_accept = TRUE; + break; + } + vn += verbs[i].len + 1; + } + if (i < verbcount) continue; + *errorcodeptr = ERR60; + goto FAILED; + } + + /* Deal with the extended parentheses; all are introduced by '?', and the + appearance of any of them means that this is not a capturing group. */ + + else if (*ptr == '?') { int i, set, unset, namelen; int *optset; @@ -4017,8 +4216,14 @@ for (;; ptr++) /* ------------------------------------------------------------ */ case '!': /* Negative lookahead */ - bravalue = OP_ASSERT_NOT; ptr++; + if (*ptr == ')') /* Optimize (?!) */ + { + *code++ = OP_FAIL; + previous = NULL; + continue; + } + bravalue = OP_ASSERT_NOT; break; @@ -4376,7 +4581,7 @@ for (;; ptr++) case 'J': /* Record that it changed in the external options */ *optset |= PCRE_DUPNAMES; - cd->external_options |= PCRE_JCHANGED; + cd->external_flags |= PCRE_JCHANGED; break; case 'i': *optset |= PCRE_CASELESS; break; @@ -4567,23 +4772,29 @@ for (;; ptr++) goto FAILED; } - /* In the pre-compile phase, update the length by the length of the nested - group, less the brackets at either end. Then reduce the compiled code to - just the brackets so that it doesn't use much memory if it is duplicated by - a quantifier. */ + /* In the pre-compile phase, update the length by the length of the group, + less the brackets at either end. Then reduce the compiled code to just a + set of non-capturing brackets so that it doesn't use much memory if it is + duplicated by a quantifier.*/ if (lengthptr != NULL) { + if (OFLOW_MAX - *lengthptr < length_prevgroup - 2 - 2*LINK_SIZE) + { + *errorcodeptr = ERR20; + goto FAILED; + } *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE; - code++; + *code++ = OP_BRA; PUTINC(code, 0, 1 + LINK_SIZE); *code++ = OP_KET; PUTINC(code, 0, 1 + LINK_SIZE); + break; /* No need to waste time with special character handling */ } /* Otherwise update the main code pointer to the end of the group. */ - else code = tempcode; + code = tempcode; /* For a DEFINE group, required and first character settings are not relevant. */ @@ -4787,6 +4998,11 @@ for (;; ptr++) *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR; for (c = 0; c < mclength; c++) *code++ = mcbuffer[c]; + /* Remember if \r or \n were seen */ + + if (mcbuffer[0] == '\r' || mcbuffer[0] == '\n') + cd->external_flags |= PCRE_HASCRORLF; + /* 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 @@ -5069,7 +5285,15 @@ for (;;) *ptrptr = ptr; *firstbyteptr = firstbyte; *reqbyteptr = reqbyte; - if (lengthptr != NULL) *lengthptr += length; + if (lengthptr != NULL) + { + if (OFLOW_MAX - *lengthptr < length) + { + *errorcodeptr = ERR20; + return FALSE; + } + *lengthptr += length; + } return TRUE; } @@ -5378,6 +5602,7 @@ real_pcre *re; int length = 1; /* For final END opcode */ int firstbyte, reqbyte, newline; int errorcode = 0; +int skipatstart = 0; #ifdef SUPPORT_UTF8 BOOL utf8; #endif @@ -5456,13 +5681,55 @@ cd->fcc = tables + fcc_offset; cd->cbits = tables + cbits_offset; cd->ctypes = tables + ctypes_offset; +/* Check for global one-time settings at the start of the pattern, and remember +the offset for later. */ + +while (ptr[skipatstart] == '(' && ptr[skipatstart+1] == '*') + { + int newnl = 0; + int newbsr = 0; + + if (strncmp((char *)(ptr+skipatstart+2), "CR)", 3) == 0) + { skipatstart += 5; newnl = PCRE_NEWLINE_CR; } + else if (strncmp((char *)(ptr+skipatstart+2), "LF)", 3) == 0) + { skipatstart += 5; newnl = PCRE_NEWLINE_LF; } + else if (strncmp((char *)(ptr+skipatstart+2), "CRLF)", 5) == 0) + { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; } + else if (strncmp((char *)(ptr+skipatstart+2), "ANY)", 4) == 0) + { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; } + else if (strncmp((char *)(ptr+skipatstart+2), "ANYCRLF)", 8) == 0) + { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; } + + else if (strncmp((char *)(ptr+skipatstart+2), "BSR_ANYCRLF)", 12) == 0) + { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; } + else if (strncmp((char *)(ptr+skipatstart+2), "BSR_UNICODE)", 12) == 0) + { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; } + + if (newnl != 0) + options = (options & ~PCRE_NEWLINE_BITS) | newnl; + else if (newbsr != 0) + options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr; + else break; + } + +/* Check validity of \R options. */ + +switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) + { + case 0: + case PCRE_BSR_ANYCRLF: + case PCRE_BSR_UNICODE: + break; + default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN; + } + /* Handle different types of newline. The three bits give seven cases. The current code allows for fixed one- or two-byte sequences, plus "any" and "anycrlf". */ -switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY)) +switch (options & PCRE_NEWLINE_BITS) { - case 0: newline = NEWLINE; break; /* Compile-time default */ + case 0: newline = NEWLINE; break; /* Build-time default */ case PCRE_NEWLINE_CR: newline = '\r'; break; case PCRE_NEWLINE_LF: newline = '\n'; break; case PCRE_NEWLINE_CR+ @@ -5525,8 +5792,8 @@ cd->hwm = cworkspace; cd->start_pattern = (const uschar *)pattern; cd->end_pattern = (const uschar *)(pattern + strlen(pattern)); cd->req_varyopt = 0; -cd->nopartial = FALSE; cd->external_options = options; +cd->external_flags = 0; /* Now do the pre-compile. On error, errorcode will be set non-zero, so we don't need to look at the result of the function here. The initial options have @@ -5534,6 +5801,7 @@ been put into the cd block so that they can be changed if an option setting is found within the regex right at the beginning. Bringing initial option settings outside can help speed up starting point checks. */ +ptr += skipatstart; code = cworkspace; *code = OP_BRA; (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS, @@ -5564,14 +5832,16 @@ if (re == NULL) goto PCRE_EARLY_ERROR_RETURN; } -/* Put in the magic number, and save the sizes, initial 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. */ +/* Put in the magic number, and save the sizes, initial options, internal +flags, 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 = cd->external_options; +re->flags = cd->external_flags; re->dummy1 = 0; re->first_byte = 0; re->req_byte = 0; @@ -5596,21 +5866,22 @@ codestart = cd->name_table + re->name_entry_size * re->name_count; cd->start_code = codestart; cd->hwm = cworkspace; cd->req_varyopt = 0; -cd->nopartial = FALSE; +cd->had_accept = 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; +ptr = (const uschar *)pattern + skipatstart; code = (uschar *)codestart; *code = OP_BRA; (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr, &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL); re->top_bracket = cd->bracount; re->top_backref = cd->top_backref; +re->flags = cd->external_flags; -if (cd->nopartial) re->options |= PCRE_NOPARTIAL; +if (cd->had_accept) reqbyte = -1; /* Must disable after (*ACCEPT) */ /* If not reached end of pattern on success, there's an excess bracket. */ @@ -5652,7 +5923,7 @@ if (errorcode != 0) PCRE_EARLY_ERROR_RETURN: *erroroffset = ptr - (const uschar *)pattern; PCRE_EARLY_ERROR_RETURN2: - *errorptr = error_texts + error_texts_offsets[errorcode]; + *errorptr = find_error_text(errorcode); if (errorcodeptr != NULL) *errorcodeptr = errorcode; return NULL; } @@ -5681,10 +5952,10 @@ if ((re->options & PCRE_ANCHORED) == 0) int ch = firstbyte & 255; re->first_byte = ((firstbyte & REQ_CASELESS) != 0 && cd->fcc[ch] == ch)? ch : firstbyte; - re->options |= PCRE_FIRSTSET; + re->flags |= PCRE_FIRSTSET; } else if (is_startline(codestart, 0, cd->backref_map)) - re->options |= PCRE_STARTLINE; + re->flags |= PCRE_STARTLINE; } } @@ -5698,7 +5969,7 @@ if (reqbyte >= 0 && int ch = reqbyte & 255; re->req_byte = ((reqbyte & REQ_CASELESS) != 0 && cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte; - re->options |= PCRE_REQCHSET; + re->flags |= PCRE_REQCHSET; } /* Print out the compiled data if debugging is enabled. This is never the @@ -5709,21 +5980,9 @@ case when building a production library. */ 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 " : ""); - } +printf("Options=%08x\n", re->options); -if ((re->options & PCRE_FIRSTSET) != 0) +if ((re->flags & PCRE_FIRSTSET) != 0) { int ch = re->first_byte & 255; const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)? @@ -5732,7 +5991,7 @@ if ((re->options & PCRE_FIRSTSET) != 0) else printf("First char = \\x%02x%s\n", ch, caseless); } -if ((re->options & PCRE_REQCHSET) != 0) +if ((re->flags & PCRE_REQCHSET) != 0) { int ch = re->req_byte & 255; const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)? @@ -5749,7 +6008,7 @@ was compiled can be seen. */ if (code - codestart > length) { (pcre_free)(re); - *errorptr = error_texts + error_texts_offsets[ERR23]; + *errorptr = find_error_text(ERR23); *erroroffset = ptr - (uschar *)pattern; if (errorcodeptr != NULL) *errorcodeptr = ERR23; return NULL; diff --git a/glib/pcre/pcre_config.c b/glib/pcre/pcre_config.c index 52d2594..220ef93 100644 --- a/glib/pcre/pcre_config.c +++ b/glib/pcre/pcre_config.c @@ -41,6 +41,10 @@ POSSIBILITY OF SUCH DAMAGE. /* This module contains the external function pcre_config(). */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + #include "pcre_internal.h" @@ -83,6 +87,14 @@ switch (what) *((int *)where) = NEWLINE; break; + case PCRE_CONFIG_BSR: +#ifdef BSR_ANYCRLF + *((int *)where) = 1; +#else + *((int *)where) = 0; +#endif + break; + case PCRE_CONFIG_LINK_SIZE: *((int *)where) = LINK_SIZE; break; diff --git a/glib/pcre/pcre_dfa_exec.c b/glib/pcre/pcre_dfa_exec.c index 87f9746..e590fbb 100644 --- a/glib/pcre/pcre_dfa_exec.c +++ b/glib/pcre/pcre_dfa_exec.c @@ -44,6 +44,10 @@ FSM). This is NOT Perl- compatible, but it has advantages in certain applications. */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + #define NLBLOCK md /* Block containing newline information */ #define PSSTART start_subject /* Field containing processed string start */ #define PSEND end_subject /* Field containing processed string end */ @@ -126,7 +130,9 @@ static uschar coptable[] = { 0, /* CREF */ 0, /* RREF */ 0, /* DEF */ - 0, 0 /* BRAZERO, BRAMINZERO */ + 0, 0, /* BRAZERO, BRAMINZERO */ + 0, 0, 0, 0, /* PRUNE, SKIP, THEN, COMMIT */ + 0, 0 /* FAIL, ACCEPT */ }; /* These 2 tables allow for compact code for testing for \D, \d, \S, \s, \W, @@ -1074,15 +1080,20 @@ for (;;) int ncount = 0; switch (c) { - case 0x000d: - if (ptr + 1 < end_subject && ptr[1] == 0x0a) ncount = 1; - /* Fall through */ - case 0x000a: case 0x000b: case 0x000c: case 0x0085: case 0x2028: case 0x2029: + if ((md->moptions & PCRE_BSR_ANYCRLF) != 0) break; + goto ANYNL01; + + case 0x000d: + if (ptr + 1 < end_subject && ptr[1] == 0x0a) ncount = 1; + /* Fall through */ + + ANYNL01: + case 0x000a: if (count > 0 && codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSPLUS) { active_count--; /* Remove non-match possibility */ @@ -1091,6 +1102,7 @@ for (;;) count++; ADD_NEW_DATA(-state_offset, count, ncount); break; + default: break; } @@ -1307,15 +1319,20 @@ for (;;) int ncount = 0; switch (c) { - case 0x000d: - if (ptr + 1 < end_subject && ptr[1] == 0x0a) ncount = 1; - /* Fall through */ - case 0x000a: case 0x000b: case 0x000c: case 0x0085: case 0x2028: case 0x2029: + if ((md->moptions & PCRE_BSR_ANYCRLF) != 0) break; + goto ANYNL02; + + case 0x000d: + if (ptr + 1 < end_subject && ptr[1] == 0x0a) ncount = 1; + /* Fall through */ + + ANYNL02: + case 0x000a: if (codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSSTAR || codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSQUERY) { @@ -1324,6 +1341,7 @@ for (;;) } ADD_NEW_DATA(-(state_offset + count), 0, ncount); break; + default: break; } @@ -1539,15 +1557,20 @@ for (;;) int ncount = 0; switch (c) { - case 0x000d: - if (ptr + 1 < end_subject && ptr[1] == 0x0a) ncount = 1; - /* Fall through */ - case 0x000a: case 0x000b: case 0x000c: case 0x0085: case 0x2028: case 0x2029: + if ((md->moptions & PCRE_BSR_ANYCRLF) != 0) break; + goto ANYNL03; + + case 0x000d: + if (ptr + 1 < end_subject && ptr[1] == 0x0a) ncount = 1; + /* Fall through */ + + ANYNL03: + case 0x000a: if (codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSUPTO) { active_count--; /* Remove non-match possibility */ @@ -1558,6 +1581,7 @@ for (;;) else { ADD_NEW_DATA(-state_offset, count, ncount); } break; + default: break; } @@ -1738,14 +1762,17 @@ for (;;) case OP_ANYNL: if (clen > 0) switch(c) { - case 0x000a: case 0x000b: case 0x000c: case 0x0085: case 0x2028: case 0x2029: + if ((md->moptions & PCRE_BSR_ANYCRLF) != 0) break; + + case 0x000a: ADD_NEW(state_offset + 1, 0); break; + case 0x000d: if (ptr + 1 < end_subject && ptr[1] == 0x0a) { @@ -2568,6 +2595,18 @@ md->end_subject = end_subject; md->moptions = options; md->poptions = re->options; +/* If the BSR option is not set at match time, copy what was set +at compile time. */ + +if ((md->moptions & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) == 0) + { + if ((re->options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) != 0) + md->moptions |= re->options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE); +#ifdef BSR_ANYCRLF + else md->moptions |= PCRE_BSR_ANYCRLF; +#endif + } + /* Handle different types of newline. The three bits give eight cases. If nothing is set at run time, whatever was used at compile time applies. */ @@ -2638,7 +2677,7 @@ if (md->tables == NULL) md->tables = _pcre_default_tables; used in a loop when finding where to start. */ lcc = md->tables + lcc_offset; -startline = (re->options & PCRE_STARTLINE) != 0; +startline = (re->flags & PCRE_STARTLINE) != 0; firstline = (re->options & PCRE_FIRSTLINE) != 0; /* Set up the first character to match, if available. The first_byte value is @@ -2649,7 +2688,7 @@ studied, there may be a bitmap of possible first characters. */ if (!anchored) { - if ((re->options & PCRE_FIRSTSET) != 0) + if ((re->flags & PCRE_FIRSTSET) != 0) { first_byte = re->first_byte & 255; if ((first_byte_caseless = ((re->first_byte & REQ_CASELESS) != 0)) == TRUE) @@ -2666,7 +2705,7 @@ if (!anchored) /* For anchored or unanchored matches, there may be a "last known required character" set. */ -if ((re->options & PCRE_REQCHSET) != 0) +if ((re->flags & PCRE_REQCHSET) != 0) { req_byte = re->req_byte & 255; req_byte_caseless = (re->req_byte & REQ_CASELESS) != 0; @@ -2836,16 +2875,17 @@ for (;;) } if (current_subject > end_subject) break; - /* If we have just passed a CR and the newline option is CRLF or ANY or - ANYCRLF, and we are now at a LF, advance the match position by one more - character. */ + /* If we have just passed a CR and we are now at a LF, and the pattern does + not contain any explicit matches for \r or \n, and the newline option is CRLF + or ANY or ANYCRLF, advance the match position by one more character. */ if (current_subject[-1] == '\r' && - (md->nltype == NLTYPE_ANY || - md->nltype == NLTYPE_ANYCRLF || - md->nllen == 2) && - current_subject < end_subject && - *current_subject == '\n') + current_subject < end_subject && + *current_subject == '\n' && + (re->flags & PCRE_HASCRORLF) == 0 && + (md->nltype == NLTYPE_ANY || + md->nltype == NLTYPE_ANYCRLF || + md->nllen == 2)) current_subject++; } /* "Bumpalong" loop */ diff --git a/glib/pcre/pcre_exec.c b/glib/pcre/pcre_exec.c index a749a21..6db7c35 100644 --- a/glib/pcre/pcre_exec.c +++ b/glib/pcre/pcre_exec.c @@ -42,6 +42,10 @@ POSSIBILITY OF SUCH DAMAGE. pattern matching using an NFA algorithm, trying to mimic Perl as closely as possible. There are also some static supporting functions. */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + #define NLBLOCK md /* Block containing newline information */ #define PSSTART start_subject /* Field containing processed string start */ #define PSEND end_subject /* Field containing processed string end */ @@ -53,16 +57,10 @@ possible. There are also some static supporting functions. */ #undef min #undef max -/* The chain of eptrblocks for tail recursions uses memory in stack workspace, -obtained at top level, the size of which is defined by EPTR_WORK_SIZE. */ - -#define EPTR_WORK_SIZE (1000) - /* Flag bits for the match() function */ #define match_condassert 0x01 /* Called to check a condition assertion */ #define match_cbegroup 0x02 /* Could-be-empty unlimited repeat group */ -#define match_tail_recursed 0x04 /* Tail recursive call */ /* Non-error returns from the match() function. Error returns are externally defined PCRE_ERROR_xxx codes, which are all negative. */ @@ -70,6 +68,14 @@ defined PCRE_ERROR_xxx codes, which are all negative. */ #define MATCH_MATCH 1 #define MATCH_NOMATCH 0 +/* Special internal returns from the match() function. Make them sufficiently +negative to avoid the external error codes. */ + +#define MATCH_COMMIT (-999) +#define MATCH_PRUNE (-998) +#define MATCH_SKIP (-997) +#define MATCH_THEN (-996) + /* 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. */ @@ -205,15 +211,15 @@ variable instead of being passed in the frame. **************************************************************************** ***************************************************************************/ - -/* Numbers for RMATCH calls */ +/* Numbers for RMATCH calls. When this list is changed, the code at HEAP_RETURN +below must be updated in sync. */ enum { RM1=1, RM2, RM3, RM4, RM5, RM6, RM7, RM8, RM9, RM10, RM11, RM12, RM13, RM14, RM15, RM16, RM17, RM18, RM19, RM20, RM21, RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30, RM31, RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40, - RM41, RM42, RM43, RM44, RM45, RM46, RM47 }; - + RM41, RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50, + RM51, RM52, RM53, RM54 }; /* These versions of the macros use the stack, as normal. There are debugging versions and production versions. Note that the "rw" argument of RMATCH isn't @@ -384,7 +390,6 @@ Arguments: match_condassert - this is an assertion condition match_cbegroup - this is the start of an unlimited repeat group that can match an empty string - match_tail_recursed - this is a tail_recursed group rdepth the recursion depth Returns: MATCH_MATCH if matched ) these values are >= 0 @@ -586,22 +591,16 @@ original_ims = ims; /* Save for resetting on ')' */ string, the match_cbegroup flag is set. When this is the case, add the current subject pointer to the chain of such remembered pointers, to be checked when we hit the closing ket, in order to break infinite loops that match no characters. -When match() is called in other circumstances, don't add to the chain. If this -is a tail recursion, use a block from the workspace, as the one on the stack is -already used. */ +When match() is called in other circumstances, don't add to the chain. The +match_cbegroup flag must NOT be used with tail recursion, because the memory +block that is used is on the stack, so a new one may be required for each +match(). */ if ((flags & match_cbegroup) != 0) { - eptrblock *p; - if ((flags & match_tail_recursed) != 0) - { - if (md->eptrn >= EPTR_WORK_SIZE) RRETURN(PCRE_ERROR_NULLWSLIMIT); - p = md->eptrchain + md->eptrn++; - } - else p = &newptrb; - p->epb_saved_eptr = eptr; - p->epb_prev = eptrb; - eptrb = p; + newptrb.epb_saved_eptr = eptr; + newptrb.epb_prev = eptrb; + eptrb = &newptrb; } /* Now start processing the opcodes. */ @@ -621,6 +620,34 @@ for (;;) switch(op) { + case OP_FAIL: + RRETURN(MATCH_NOMATCH); + + case OP_PRUNE: + RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, + ims, eptrb, flags, RM51); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + RRETURN(MATCH_PRUNE); + + case OP_COMMIT: + RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, + ims, eptrb, flags, RM52); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + RRETURN(MATCH_COMMIT); + + case OP_SKIP: + RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, + ims, eptrb, flags, RM53); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + md->start_match_ptr = eptr; /* Pass back current position */ + RRETURN(MATCH_SKIP); + + case OP_THEN: + RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, + ims, eptrb, flags, RM54); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + RRETURN(MATCH_THEN); + /* Handle a 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 @@ -662,7 +689,7 @@ for (;;) { RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, eptrb, flags, RM1); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); md->capture_last = save_capture_last; ecode += GET(ecode, 1); } @@ -677,15 +704,22 @@ for (;;) RRETURN(MATCH_NOMATCH); } - /* Insufficient room for saving captured contents. Treat as a non-capturing - bracket. */ + /* FALL THROUGH ... Insufficient room for saving captured contents. Treat + as a non-capturing bracket. */ + + /* VVVVVVVVVVVVVVVVVVVVVVVVV */ + /* VVVVVVVVVVVVVVVVVVVVVVVVV */ DPRINTF(("insufficient capture room: treat as non-capturing\n")); + /* VVVVVVVVVVVVVVVVVVVVVVVVV */ + /* VVVVVVVVVVVVVVVVVVVVVVVVV */ + /* Non-capturing bracket. Loop for all the alternatives. 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. */ + turning this into a tail recursion, except in the case when match_cbegroup + is set.*/ case OP_BRA: case OP_SBRA: @@ -693,12 +727,20 @@ for (;;) flags = (op >= OP_SBRA)? match_cbegroup : 0; for (;;) { - if (ecode[GET(ecode, 1)] != OP_ALT) + if (ecode[GET(ecode, 1)] != OP_ALT) /* Final alternative */ { - ecode += _pcre_OP_lengths[*ecode]; - flags |= match_tail_recursed; - DPRINTF(("bracket 0 tail recursion\n")); - goto TAIL_RECURSE; + if (flags == 0) /* Not a possibly empty group */ + { + ecode += _pcre_OP_lengths[*ecode]; + DPRINTF(("bracket 0 tail recursion\n")); + goto TAIL_RECURSE; + } + + /* Possibly empty group; can't use tail recursion. */ + + RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, + eptrb, flags, RM48); + RRETURN(rrc); } /* For non-final alternatives, continue the loop for a NOMATCH result; @@ -706,7 +748,7 @@ for (;;) RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, eptrb, flags, RM2); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); ecode += GET(ecode, 1); } /* Control never reaches here. */ @@ -754,7 +796,7 @@ for (;;) ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2); while (*ecode == OP_ALT) ecode += GET(ecode, 1); } - else if (rrc != MATCH_NOMATCH) + else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) { RRETURN(rrc); /* Need braces because of following else */ } @@ -766,25 +808,36 @@ for (;;) } /* 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. If the second - alternative doesn't exist, we can just plough on. */ + we can use tail recursion to avoid using another stack frame, except when + match_cbegroup is required for an unlimited repeat of a possibly empty + group. If the second alternative doesn't exist, we can just plough on. */ if (condition || *ecode == OP_ALT) { ecode += 1 + LINK_SIZE; - flags = match_tail_recursed | ((op == OP_SCOND)? match_cbegroup : 0); - goto TAIL_RECURSE; + if (op == OP_SCOND) /* Possibly empty group */ + { + RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49); + RRETURN(rrc); + } + else /* Group must match something */ + { + flags = 0; + goto TAIL_RECURSE; + } } - else + else /* Condition false & no 2nd alternative */ { ecode += 1 + LINK_SIZE; } break; - /* End of the pattern. If we are in a top-level recursion, we should - restore the offsets appropriately and continue from after the call. */ + /* End of the pattern, either real or forced. If we are in a top-level + recursion, we should restore the offsets appropriately and continue from + after the call. */ + case OP_ACCEPT: case OP_END: if (md->recursive != NULL && md->recursive->group_num == 0) { @@ -805,7 +858,7 @@ for (;;) if (md->notempty && eptr == mstart) RRETURN(MATCH_NOMATCH); md->end_match_ptr = eptr; /* Record where we ended */ md->end_offset_top = offset_top; /* and how many extracts were taken */ - md->start_match_ptr = mstart; /* and the start (\K can modify) */ + md->start_match_ptr = mstart; /* and the start (\K can modify) */ RRETURN(MATCH_MATCH); /* Change option settings */ @@ -829,7 +882,7 @@ for (;;) RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, RM4); if (rrc == MATCH_MATCH) break; - if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); ecode += GET(ecode, 1); } while (*ecode == OP_ALT); @@ -856,7 +909,7 @@ for (;;) RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, RM5); if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); ecode += GET(ecode,1); } while (*ecode == OP_ALT); @@ -880,7 +933,7 @@ for (;;) { eptr--; if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); - BACKCHAR(eptr) + BACKCHAR(eptr); } } else @@ -993,7 +1046,7 @@ for (;;) (pcre_free)(new_recursive.offset_save); RRETURN(MATCH_MATCH); } - else if (rrc != MATCH_NOMATCH) + else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) { DPRINTF(("Recursion gave error %d\n", rrc)); RRETURN(rrc); @@ -1027,10 +1080,9 @@ for (;;) do { - RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, - eptrb, 0, RM7); + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7); if (rrc == MATCH_MATCH) break; - if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); ecode += GET(ecode,1); } while (*ecode == OP_ALT); @@ -1073,11 +1125,10 @@ for (;;) if (*ecode == OP_KETRMIN) { - RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, - RM8); + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM8); if (rrc != MATCH_NOMATCH) RRETURN(rrc); ecode = prev; - flags = match_tail_recursed; + flags = 0; goto TAIL_RECURSE; } else /* OP_KETRMAX */ @@ -1085,7 +1136,7 @@ for (;;) RMATCH(eptr, prev, offset_top, md, ims, eptrb, match_cbegroup, RM9); if (rrc != MATCH_NOMATCH) RRETURN(rrc); ecode += 1 + LINK_SIZE; - flags = match_tail_recursed; + flags = 0; goto TAIL_RECURSE; } /* Control never gets here */ @@ -1216,17 +1267,21 @@ for (;;) /* 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. */ + tail recursion to avoid using another stack frame, unless we have an + unlimited repeat of a group that can match an empty string. */ flags = (*prev >= OP_SBRA)? match_cbegroup : 0; if (*ecode == OP_KETRMIN) { - RMATCH(eptr, ecode + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, - RM12); + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12); if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (flags != 0) /* Could match an empty string */ + { + RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50); + RRETURN(rrc); + } ecode = prev; - flags |= match_tail_recursed; goto TAIL_RECURSE; } else /* OP_KETRMAX */ @@ -1234,7 +1289,7 @@ for (;;) RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13); if (rrc != MATCH_NOMATCH) RRETURN(rrc); ecode += 1 + LINK_SIZE; - flags = match_tail_recursed; + flags = 0; goto TAIL_RECURSE; } /* Control never gets here */ @@ -1471,12 +1526,16 @@ for (;;) case 0x000d: if (eptr < md->end_subject && *eptr == 0x0a) eptr++; break; + case 0x000a: + break; + case 0x000b: case 0x000c: case 0x0085: case 0x2028: case 0x2029: + if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); break; } ecode++; @@ -2033,7 +2092,7 @@ for (;;) RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM21); if (rrc != MATCH_NOMATCH) RRETURN(rrc); if (eptr-- == pp) break; /* Stop if tried at original pos */ - BACKCHAR(eptr) + if (utf8) BACKCHAR(eptr); } RRETURN(MATCH_NOMATCH); } @@ -2899,12 +2958,16 @@ for (;;) case 0x000d: if (eptr < md->end_subject && *eptr == 0x0a) eptr++; break; + case 0x000a: + break; + case 0x000b: case 0x000c: case 0x0085: case 0x2028: case 0x2029: + if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); break; } } @@ -3038,9 +3101,9 @@ for (;;) for (i = 1; i <= min; i++) { if (eptr >= md->end_subject || - (*eptr < 128 && (md->ctypes[*eptr++] & ctype_space) != 0)) + (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0)) RRETURN(MATCH_NOMATCH); - while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; + while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); } break; @@ -3058,9 +3121,9 @@ for (;;) for (i = 1; i <= min; i++) { if (eptr >= md->end_subject || - (*eptr < 128 && (md->ctypes[*eptr++] & ctype_word) != 0)) + (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0)) RRETURN(MATCH_NOMATCH); - while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; + while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); } break; @@ -3117,9 +3180,12 @@ for (;;) if (eptr < md->end_subject && *eptr == 0x0a) eptr++; break; case 0x000a: + break; + case 0x000b: case 0x000c: case 0x0085: + if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); break; } } @@ -3371,11 +3437,14 @@ for (;;) if (eptr < md->end_subject && *eptr == 0x0a) eptr++; break; case 0x000a: + break; + case 0x000b: case 0x000c: case 0x0085: case 0x2028: case 0x2029: + if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); break; } break; @@ -3527,10 +3596,14 @@ for (;;) case 0x000d: if (eptr < md->end_subject && *eptr == 0x0a) eptr++; break; + case 0x000a: + break; + case 0x000b: case 0x000c: case 0x0085: + if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); break; } break; @@ -3702,7 +3775,7 @@ for (;;) RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM44); if (rrc != MATCH_NOMATCH) RRETURN(rrc); if (eptr-- == pp) break; /* Stop if tried at original pos */ - BACKCHAR(eptr); + if (utf8) BACKCHAR(eptr); } } @@ -3741,9 +3814,9 @@ for (;;) for (;;) /* Move back over one extended */ { int len = 1; - BACKCHAR(eptr); if (!utf8) c = *eptr; else { + BACKCHAR(eptr); GETCHARLEN(c, eptr, len); } prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); @@ -3764,11 +3837,6 @@ for (;;) 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) @@ -3801,15 +3869,12 @@ for (;;) { if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break; eptr++; + while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; } - break; } else { - c = max - min; - if (c > (unsigned int)(md->end_subject - eptr)) - c = md->end_subject - eptr; - eptr += c; + eptr = md->end_subject; } } break; @@ -3836,8 +3901,10 @@ for (;;) } else { - if (c != 0x000a && c != 0x000b && c != 0x000c && - c != 0x0085 && c != 0x2028 && c != 0x2029) + if (c != 0x000a && + (md->bsr_anycrlf || + (c != 0x000b && c != 0x000c && + c != 0x0085 && c != 0x2028 && c != 0x2029))) break; eptr += len; } @@ -3990,7 +4057,7 @@ for (;;) } } else -#endif +#endif /* SUPPORT_UTF8 */ /* Not UTF-8 mode */ { @@ -4027,7 +4094,9 @@ for (;;) } else { - if (c != 0x000a && c != 0x000b && c != 0x000c && c != 0x0085) + if (c != 0x000a && + (md->bsr_anycrlf || + (c != 0x000b && c != 0x000c && c != 0x0085))) break; eptr++; } @@ -4177,11 +4246,17 @@ HEAP_RETURN: switch (frame->Xwhere) { LBL( 1) LBL( 2) LBL( 3) LBL( 4) LBL( 5) LBL( 6) LBL( 7) LBL( 8) - LBL( 9) LBL(10) LBL(11) LBL(12) LBL(13) LBL(14) LBL(15) LBL(16) - LBL(17) LBL(18) LBL(19) LBL(20) LBL(21) LBL(22) LBL(23) LBL(24) - LBL(25) LBL(26) LBL(27) LBL(28) LBL(29) LBL(30) LBL(31) LBL(32) - LBL(33) LBL(34) LBL(35) LBL(36) LBL(37) LBL(38) LBL(39) LBL(40) - LBL(41) LBL(42) LBL(43) LBL(44) LBL(45) LBL(46) LBL(47) + LBL( 9) LBL(10) LBL(11) LBL(12) LBL(13) LBL(14) LBL(15) LBL(17) + LBL(19) LBL(24) LBL(25) LBL(26) LBL(27) LBL(29) LBL(31) LBL(33) + LBL(35) LBL(43) LBL(47) LBL(48) LBL(49) LBL(50) LBL(51) LBL(52) + LBL(53) LBL(54) +#ifdef SUPPORT_UTF8 + LBL(16) LBL(18) LBL(20) LBL(21) LBL(22) LBL(23) LBL(28) LBL(30) + LBL(32) LBL(34) LBL(42) LBL(46) +#ifdef SUPPORT_UCP + LBL(36) LBL(37) LBL(38) LBL(39) LBL(40) LBL(41) LBL(44) LBL(45) +#endif /* SUPPORT_UCP */ +#endif /* SUPPORT_UTF8 */ default: DPRINTF(("jump error in pcre match: label %d non-existent\n", frame->Xwhere)); return PCRE_ERROR_INTERNAL; @@ -4298,7 +4373,6 @@ const uschar *start_bits = NULL; USPTR start_match = (USPTR)subject + start_offset; USPTR end_subject; USPTR req_byte_ptr = start_match - 1; -eptrblock eptrchain[EPTR_WORK_SIZE]; pcre_study_data internal_study; const pcre_study_data *study; @@ -4361,7 +4435,7 @@ if (re->magic_number != MAGIC_NUMBER) /* Set up other data */ anchored = ((re->options | options) & PCRE_ANCHORED) != 0; -startline = (re->options & PCRE_STARTLINE) != 0; +startline = (re->flags & PCRE_STARTLINE) != 0; firstline = (re->options & PCRE_FIRSTLINE) != 0; /* The code starts after the real_pcre block and the capture name table. */ @@ -4384,16 +4458,41 @@ md->partial = (options & PCRE_PARTIAL) != 0; md->hitend = FALSE; md->recursive = NULL; /* No recursion at top level */ -md->eptrchain = eptrchain; /* Make workspace generally available */ md->lcc = tables + lcc_offset; md->ctypes = tables + ctypes_offset; +/* Handle different \R options. */ + +switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) + { + case 0: + if ((re->options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) != 0) + md->bsr_anycrlf = (re->options & PCRE_BSR_ANYCRLF) != 0; + else +#ifdef BSR_ANYCRLF + md->bsr_anycrlf = TRUE; +#else + md->bsr_anycrlf = FALSE; +#endif + break; + + case PCRE_BSR_ANYCRLF: + md->bsr_anycrlf = TRUE; + break; + + case PCRE_BSR_UNICODE: + md->bsr_anycrlf = FALSE; + break; + + default: return PCRE_ERROR_BADNEWLINE; + } + /* Handle different types of newline. The three bits give eight cases. If nothing is set at run time, whatever was used at compile time applies. */ -switch ((((options & PCRE_NEWLINE_BITS) == 0)? re->options : (pcre_uint32)options) & - PCRE_NEWLINE_BITS) +switch ((((options & PCRE_NEWLINE_BITS) == 0)? re->options : + (pcre_uint32)options) & PCRE_NEWLINE_BITS) { case 0: newline = NEWLINE; break; /* Compile-time default */ case PCRE_NEWLINE_CR: newline = '\r'; break; @@ -4432,7 +4531,7 @@ else /* Partial matching is supported only for a restricted set of regexes at the moment. */ -if (md->partial && (re->options & PCRE_NOPARTIAL) != 0) +if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0) return PCRE_ERROR_BADPARTIAL; /* Check a UTF-8 string if required. Unfortunately there's no way of passing @@ -4509,7 +4608,7 @@ studied, there may be a bitmap of possible first characters. */ if (!anchored) { - if ((re->options & PCRE_FIRSTSET) != 0) + if ((re->flags & PCRE_FIRSTSET) != 0) { first_byte = re->first_byte & 255; if ((first_byte_caseless = ((re->first_byte & REQ_CASELESS) != 0)) == TRUE) @@ -4524,7 +4623,7 @@ if (!anchored) /* For anchored or unanchored matches, there may be a "last known required character" set. */ -if ((re->options & PCRE_REQCHSET) != 0) +if ((re->flags & PCRE_REQCHSET) != 0) { req_byte = re->req_byte & 255; req_byte_caseless = (re->req_byte & REQ_CASELESS) != 0; @@ -4540,6 +4639,7 @@ the loop runs just once. */ for(;;) { USPTR save_end_subject = end_subject; + USPTR new_start_match; /* Reset the maximum number of extractions we might see. */ @@ -4680,15 +4780,48 @@ for(;;) /* OK, we can now run the match. */ - md->start_match_ptr = start_match; /* Insurance */ + md->start_match_ptr = start_match; md->match_call_count = 0; - md->eptrn = 0; /* Next free eptrchain slot */ - rc = match(start_match, md->start_code, start_match, 2, md, - ims, NULL, 0, 0); + rc = match(start_match, md->start_code, start_match, 2, md, ims, NULL, 0, 0); + + switch(rc) + { + /* NOMATCH and PRUNE advance by one character. THEN at this level acts + exactly like PRUNE. */ + + case MATCH_NOMATCH: + case MATCH_PRUNE: + case MATCH_THEN: + new_start_match = start_match + 1; +#ifdef SUPPORT_UTF8 + if (utf8) + while(new_start_match < end_subject && (*new_start_match & 0xc0) == 0x80) + new_start_match++; +#endif + break; + + /* SKIP passes back the next starting point explicitly. */ - /* Any return other than MATCH_NOMATCH breaks the loop. */ + case MATCH_SKIP: + new_start_match = md->start_match_ptr; + break; + + /* COMMIT disables the bumpalong, but otherwise behaves as NOMATCH. */ + + case MATCH_COMMIT: + rc = MATCH_NOMATCH; + goto ENDLOOP; - if (rc != MATCH_NOMATCH) break; + /* Any other return is some kind of error. */ + + default: + goto ENDLOOP; + } + + /* Control reaches here for the various types of "no match at this point" + result. Reset the code to MATCH_NOMATCH for subsequent checking. */ + + rc = MATCH_NOMATCH; /* If PCRE_FIRSTLINE is set, the match must happen before or at the first newline in the subject (though it may continue over the newline). Therefore, @@ -4696,30 +4829,26 @@ for(;;) if (firstline && IS_NEWLINE(start_match)) break; - /* Advance the match position by one character. */ + /* Advance to new matching position */ - start_match++; -#ifdef SUPPORT_UTF8 - if (utf8) - while(start_match < end_subject && (*start_match & 0xc0) == 0x80) - start_match++; -#endif + start_match = new_start_match; /* Break the loop if the pattern is anchored or if we have passed the end of the subject. */ if (anchored || start_match > end_subject) break; - /* If we have just passed a CR and the newline option is CRLF or ANY or - ANYCRLF, and we are now at a LF, advance the match position by one more - character. */ + /* If we have just passed a CR and we are now at a LF, and the pattern does + not contain any explicit matches for \r or \n, and the newline option is CRLF + or ANY or ANYCRLF, advance the match position by one more character. */ if (start_match[-1] == '\r' && - (md->nltype == NLTYPE_ANY || - md->nltype == NLTYPE_ANYCRLF || - md->nllen == 2) && - start_match < end_subject && - *start_match == '\n') + start_match < end_subject && + *start_match == '\n' && + (re->flags & PCRE_HASCRORLF) == 0 && + (md->nltype == NLTYPE_ANY || + md->nltype == NLTYPE_ANYCRLF || + md->nllen == 2)) start_match++; } /* End of for(;;) "bumpalong" loop */ @@ -4729,7 +4858,7 @@ for(;;) /* We reach here when rc is not MATCH_NOMATCH, or if one of the stopping conditions is true: -(1) The pattern is anchored; +(1) The pattern is anchored or the match was failed by (*COMMIT); (2) We are past the end of the subject; @@ -4744,6 +4873,8 @@ processing, copy those that we can. In this case there need not be overflow if certain parts of the pattern were not used, even though there are more capturing parentheses than vector slots. */ +ENDLOOP: + if (rc == MATCH_MATCH) { if (using_temporary_offsets) diff --git a/glib/pcre/pcre_fullinfo.c b/glib/pcre/pcre_fullinfo.c index 797edde..04e31f6 100644 --- a/glib/pcre/pcre_fullinfo.c +++ b/glib/pcre/pcre_fullinfo.c @@ -42,6 +42,10 @@ POSSIBILITY OF SUCH DAMAGE. information about a compiled pattern. */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + #include "pcre_internal.h" @@ -106,8 +110,8 @@ switch (what) case PCRE_INFO_FIRSTBYTE: *((int *)where) = - ((re->options & PCRE_FIRSTSET) != 0)? re->first_byte : - ((re->options & PCRE_STARTLINE) != 0)? -1 : -2; + ((re->flags & PCRE_FIRSTSET) != 0)? re->first_byte : + ((re->flags & PCRE_STARTLINE) != 0)? -1 : -2; break; /* Make sure we pass back the pointer to the bit vector in the external @@ -121,7 +125,7 @@ switch (what) case PCRE_INFO_LASTLITERAL: *((int *)where) = - ((re->options & PCRE_REQCHSET) != 0)? re->req_byte : -1; + ((re->flags & PCRE_REQCHSET) != 0)? re->req_byte : -1; break; case PCRE_INFO_NAMEENTRYSIZE: @@ -141,11 +145,15 @@ switch (what) break; case PCRE_INFO_OKPARTIAL: - *((int *)where) = (re->options & PCRE_NOPARTIAL) == 0; + *((int *)where) = (re->flags & PCRE_NOPARTIAL) == 0; break; case PCRE_INFO_JCHANGED: - *((int *)where) = (re->options & PCRE_JCHANGED) != 0; + *((int *)where) = (re->flags & PCRE_JCHANGED) != 0; + break; + + case PCRE_INFO_HASCRORLF: + *((int *)where) = (re->flags & PCRE_HASCRORLF) != 0; break; default: return PCRE_ERROR_BADOPTION; diff --git a/glib/pcre/pcre_get.c b/glib/pcre/pcre_get.c index ba0c8cb..fc283c8 100644 --- a/glib/pcre/pcre_get.c +++ b/glib/pcre/pcre_get.c @@ -43,6 +43,10 @@ from the subject string after a regex match has succeeded. The original idea for these functions came from Scott Wimer. */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + #include "pcre_internal.h" @@ -185,7 +189,7 @@ const real_pcre *re = (const real_pcre *)code; int entrysize; char *first, *last; uschar *entry; -if ((re->options & (PCRE_DUPNAMES | PCRE_JCHANGED)) == 0) +if ((re->options & PCRE_DUPNAMES) == 0 && (re->flags & PCRE_JCHANGED) == 0) return pcre_get_stringnumber(code, stringname); entrysize = pcre_get_stringtable_entries(code, stringname, &first, &last); if (entrysize <= 0) return entrysize; diff --git a/glib/pcre/pcre_globals.c b/glib/pcre/pcre_globals.c index 4e99986..56b4ce4 100644 --- a/glib/pcre/pcre_globals.c +++ b/glib/pcre/pcre_globals.c @@ -46,6 +46,10 @@ 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). */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + #include "pcre_internal.h" PCRE_EXP_DATA_DEFN int (*pcre_callout)(pcre_callout_block *) = NULL; diff --git a/glib/pcre/pcre_info.c b/glib/pcre/pcre_info.c index 52ac23f..9bcccbc 100644 --- a/glib/pcre/pcre_info.c +++ b/glib/pcre/pcre_info.c @@ -43,6 +43,10 @@ information about a compiled pattern. However, use of this function is now deprecated, as it has been superseded by pcre_fullinfo(). */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + #include "pcre_internal.h" @@ -81,8 +85,8 @@ if (re->magic_number != MAGIC_NUMBER) } 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; + *first_byte = ((re->flags & PCRE_FIRSTSET) != 0)? re->first_byte : + ((re->flags & PCRE_STARTLINE) != 0)? -1 : -2; return re->top_bracket; } diff --git a/glib/pcre/pcre_internal.h b/glib/pcre/pcre_internal.h index 30d7a52..b831366 100644 --- a/glib/pcre/pcre_internal.h +++ b/glib/pcre/pcre_internal.h @@ -67,10 +67,6 @@ be absolutely sure we get our version. */ #endif -/* Get the definitions provided by running "configure" */ - -#include "config.h" - /* Standard C headers plus the external interface definition. The only time setjmp and stdarg are used is when NO_RECURSE is set. */ @@ -112,7 +108,7 @@ PCRE_EXP_DATA_DEFN only if they are not already set. */ #ifndef PCRE_EXP_DECL # ifdef _WIN32 -# ifdef DLL_EXPORT +# ifndef PCRE_STATIC # define PCRE_EXP_DECL extern __declspec(dllexport) # define PCRE_EXP_DEFN __declspec(dllexport) # define PCRE_EXP_DATA_DEFN __declspec(dllexport) @@ -121,7 +117,6 @@ PCRE_EXP_DATA_DEFN only if they are not already set. */ # define PCRE_EXP_DEFN # define PCRE_EXP_DATA_DEFN # endif -# # else # ifdef __cplusplus # define PCRE_EXP_DECL extern "C" @@ -234,7 +229,7 @@ must begin with PCRE_. */ /* Include the public PCRE header and the definitions of UCP character property values. */ -#include +#include "pcre.h" #include "ucp.h" /* When compiling for use with the Virtual Pascal compiler, these functions @@ -363,7 +358,9 @@ capturing parenthesis numbers in back references. */ /* 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. */ +byte-mode, and more complicated ones for UTF-8 characters. BACKCHAR should +never be called in byte mode. To make sure it can never even appear when UTF-8 +support is omitted, we don't even define it. */ #ifndef SUPPORT_UTF8 #define GETCHAR(c, eptr) c = *eptr; @@ -371,7 +368,7 @@ byte-mode, and more complicated ones for UTF-8 characters. */ #define GETCHARINC(c, eptr) c = *eptr++; #define GETCHARINCTEST(c, eptr) c = *eptr++; #define GETCHARLEN(c, eptr, len) c = *eptr; -#define BACKCHAR(eptr) +/* #define BACKCHAR(eptr) */ #else /* SUPPORT_UTF8 */ @@ -464,9 +461,10 @@ if there are extra bytes. This is called when we know we are in UTF-8 mode. */ } /* If the pointer is not at the start of a character, move it back until -it is. Called only in UTF-8 mode. */ +it is. This is called only in UTF-8 mode - we don't put a test within the macro +because almost all calls are already within a block of UTF-8 only code. */ -#define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr--; +#define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr-- #endif @@ -483,17 +481,16 @@ Standard C system should have one. */ #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. */ +/* Private flags containing information about the compiled regex. They used to +live at the top end of the options word, but that got almost full, so now they +are in a 16-bit flags word. */ -#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 */ +#define PCRE_NOPARTIAL 0x0001 /* can't use partial with this regex */ +#define PCRE_FIRSTSET 0x0002 /* first_byte is set */ +#define PCRE_REQCHSET 0x0004 /* req_byte is set */ +#define PCRE_STARTLINE 0x0008 /* start after \n for multiline */ +#define PCRE_JCHANGED 0x0010 /* j option used in regex */ +#define PCRE_HASCRORLF 0x0020 /* explicit \r or \n in pattern */ /* Options for the "extra" block produced by pcre_study(). */ @@ -509,15 +506,16 @@ time, run time, or study time, respectively. */ (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_BITS) + PCRE_DUPNAMES|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE) #define PUBLIC_EXEC_OPTIONS \ (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \ - PCRE_PARTIAL|PCRE_NEWLINE_BITS) + PCRE_PARTIAL|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE) #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_BITS) + PCRE_PARTIAL|PCRE_DFA_SHORTEST|PCRE_DFA_RESTART|PCRE_NEWLINE_BITS| \ + PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE) #define PUBLIC_STUDY_OPTIONS 0 /* None defined */ @@ -607,14 +605,9 @@ enum { ESC_A = 1, ESC_G, ESC_K, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s, ESC_V, ESC_v, ESC_X, ESC_Z, ESC_z, ESC_E, ESC_Q, ESC_k, 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 +/* Opcode table: 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. -To keep stored, compiled patterns compatible, new opcodes should be added -immediately before OP_BRA, where (since release 7.0) a gap is left for this -purpose. - *** NOTE NOTE NOTE *** Whenever this list is updated, the two macro definitions that follow must also be updated to match. There is also a table called "coptable" in pcre_dfa_exec.c that must be updated. */ @@ -741,7 +734,7 @@ enum { as there's a test for >= ONCE for a subpattern that isn't an assertion. */ OP_ONCE, /* 92 Atomic group */ - OP_BRA, /* 83 Start of non-capturing bracket */ + OP_BRA, /* 93 Start of non-capturing bracket */ OP_CBRA, /* 94 Start of capturing bracket */ OP_COND, /* 95 Conditional group */ @@ -757,7 +750,19 @@ enum { OP_DEF, /* 101 The DEFINE condition */ OP_BRAZERO, /* 102 These two must remain together and in this */ - OP_BRAMINZERO /* 103 order. */ + OP_BRAMINZERO, /* 103 order. */ + + /* These are backtracking control verbs */ + + OP_PRUNE, /* 104 */ + OP_SKIP, /* 105 */ + OP_THEN, /* 106 */ + OP_COMMIT, /* 107 */ + + /* These are forced failure and success verbs */ + + OP_FAIL, /* 108 */ + OP_ACCEPT /* 109 */ }; @@ -780,8 +785,9 @@ for debugging. The macro is referenced only in pcre_printint.c. */ "class", "nclass", "xclass", "Ref", "Recurse", "Callout", \ "Alt", "Ket", "KetRmax", "KetRmin", "Assert", "Assert not", \ "AssertB", "AssertB not", "Reverse", \ - "Once", "Bra 0", "Bra", "Cond", "SBra 0", "SBra", "SCond", \ - "Cond ref", "Cond rec", "Cond def", "Brazero", "Braminzero" + "Once", "Bra", "CBra", "Cond", "SBra", "SCBra", "SCond", \ + "Cond ref", "Cond rec", "Cond def", "Brazero", "Braminzero", \ + "*PRUNE", "*SKIP", "*THEN", "*COMMIT", "*FAIL", "*ACCEPT" /* This macro defines the length of fixed length operations in the compiled @@ -845,6 +851,8 @@ in UTF-8 mode. The code that uses this table must know about such things. */ 3, /* RREF */ \ 1, /* DEF */ \ 1, 1, /* BRAZERO, BRAMINZERO */ \ + 1, 1, 1, 1, /* PRUNE, SKIP, THEN, COMMIT, */ \ + 1, 1 /* FAIL, ACCEPT */ /* A magic value for OP_RREF to indicate the "any recursion" condition. */ @@ -859,7 +867,8 @@ enum { ERR0, ERR1, ERR2, ERR3, ERR4, ERR5, ERR6, ERR7, ERR8, ERR9, 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, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58 }; + ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59, + ERR60, ERR61 }; /* 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 @@ -881,9 +890,9 @@ 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_uint32 options; /* Public options */ + pcre_uint16 flags; /* Private flags */ + pcre_uint16 dummy1; /* For future use */ pcre_uint16 top_bracket; pcre_uint16 top_backref; pcre_uint16 first_byte; @@ -926,8 +935,9 @@ typedef struct compile_data { int top_backref; /* Maximum back reference */ unsigned int backref_map; /* Bitmap of low back refs */ int external_options; /* External (initial) options */ + int external_flags; /* External flag bits to be set */ int req_varyopt; /* "After variable item" flag for reqbyte */ - BOOL nopartial; /* Set TRUE if partial won't work */ + BOOL had_accept; /* (*ACCEPT) encountered */ int nltype; /* Newline type */ int nllen; /* Newline string length */ uschar nl[4]; /* Newline string when fixed length */ @@ -986,6 +996,7 @@ typedef struct match_data { BOOL notempty; /* Empty string match not wanted */ BOOL partial; /* PARTIAL flag */ BOOL hitend; /* Hit the end of the subject at some point */ + BOOL bsr_anycrlf; /* \R is just any CRLF, not full Unicode */ const uschar *start_code; /* For use when recursing */ USPTR start_subject; /* Start of the subject string */ USPTR end_subject; /* End of the subject string */ @@ -1050,10 +1061,12 @@ total length. */ #define tables_length (ctypes_offset + 256) /* Layout of the UCP type table that translates property names into types and -codes. */ +codes. Each entry used to point directly to a name, but to reduce the number of +relocations in shared libraries, it now has an offset into a single string +instead. */ typedef struct { - pcre_uint16 offset; + pcre_uint16 name_offset; pcre_uint16 type; pcre_uint16 value; } ucp_type_table; @@ -1071,7 +1084,7 @@ extern const uschar _pcre_utf8_table4[]; extern const int _pcre_utf8_table1_size; -extern const char _pcre_ucp_names[]; +extern const char _pcre_utt_names[]; extern const ucp_type_table _pcre_utt[]; extern const int _pcre_utt_size; diff --git a/glib/pcre/pcre_maketables.c b/glib/pcre/pcre_maketables.c index 9963e13..352bea9 100644 --- a/glib/pcre/pcre_maketables.c +++ b/glib/pcre/pcre_maketables.c @@ -45,7 +45,10 @@ compilation of dftables.c, in which case the macro DFTABLES is defined. */ #ifndef DFTABLES -#include "pcre_internal.h" +# ifdef HAVE_CONFIG_H +# include "config.h" +# endif +# include "pcre_internal.h" #endif diff --git a/glib/pcre/pcre_newline.c b/glib/pcre/pcre_newline.c index ae66c73..1708d93 100644 --- a/glib/pcre/pcre_newline.c +++ b/glib/pcre/pcre_newline.c @@ -47,6 +47,10 @@ and NLTYPE_ANY. The full list of Unicode newline characters is taken from http://unicode.org/unicode/reports/tr18/. */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + #include "pcre_internal.h" @@ -124,12 +128,16 @@ _pcre_was_newline(const uschar *ptr, int type, const uschar *startptr, { int c; ptr--; +#ifdef SUPPORT_UTF8 if (utf8) { BACKCHAR(ptr); GETCHAR(c, ptr); } else c = *ptr; +#else /* no UTF-8 support */ +c = *ptr; +#endif /* SUPPORT_UTF8 */ if (type == NLTYPE_ANYCRLF) switch(c) { diff --git a/glib/pcre/pcre_ord2utf8.c b/glib/pcre/pcre_ord2utf8.c index a727612..d3904c6 100644 --- a/glib/pcre/pcre_ord2utf8.c +++ b/glib/pcre/pcre_ord2utf8.c @@ -41,6 +41,9 @@ POSSIBILITY OF SUCH DAMAGE. /* This file contains a private PCRE function that converts an ordinal character value into a UTF8 string. */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif #include "pcre_internal.h" diff --git a/glib/pcre/pcre_refcount.c b/glib/pcre/pcre_refcount.c index 8339019..b14103c 100644 --- a/glib/pcre/pcre_refcount.c +++ b/glib/pcre/pcre_refcount.c @@ -43,6 +43,11 @@ 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. */ + +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + #include "pcre_internal.h" diff --git a/glib/pcre/pcre_study.c b/glib/pcre/pcre_study.c index 3cf5b59..1c28384 100644 --- a/glib/pcre/pcre_study.c +++ b/glib/pcre/pcre_study.c @@ -42,6 +42,10 @@ POSSIBILITY OF SUCH DAMAGE. supporting functions. */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + #include "pcre_internal.h" @@ -523,7 +527,8 @@ code = (uschar *)re + re->name_table_offset + a multiline pattern that matches only at "line starts", no further processing at present. */ -if ((re->options & (PCRE_ANCHORED|PCRE_FIRSTSET|PCRE_STARTLINE)) != 0) +if ((re->options & PCRE_ANCHORED) != 0 || + (re->flags & (PCRE_FIRSTSET|PCRE_STARTLINE)) != 0) return NULL; /* Set the character tables in the block that is passed around */ diff --git a/glib/pcre/pcre_tables.c b/glib/pcre/pcre_tables.c index 9b7c300..4b14fd1 100644 --- a/glib/pcre/pcre_tables.c +++ b/glib/pcre/pcre_tables.c @@ -44,6 +44,10 @@ uses macros to change their names from _pcre_xxx to xxxx, thereby avoiding name clashes with the library. */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + #include "pcre_internal.h" @@ -83,131 +87,137 @@ const uschar _pcre_utf8_table4[] = { 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 }; -/* 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 char _pcre_ucp_names[] = - "Any\0" - "Arabic\0" - "Armenian\0" - "Balinese\0" - "Bengali\0" - "Bopomofo\0" - "Braille\0" - "Buginese\0" - "Buhid\0" - "C\0" - "Canadian_Aboriginal\0" - "Cc\0" - "Cf\0" - "Cherokee\0" - "Cn\0" - "Co\0" - "Common\0" - "Coptic\0" - "Cs\0" - "Cuneiform\0" - "Cypriot\0" - "Cyrillic\0" - "Deseret\0" - "Devanagari\0" - "Ethiopic\0" - "Georgian\0" - "Glagolitic\0" - "Gothic\0" - "Greek\0" - "Gujarati\0" - "Gurmukhi\0" - "Han\0" - "Hangul\0" - "Hanunoo\0" - "Hebrew\0" - "Hiragana\0" - "Inherited\0" - "Kannada\0" - "Katakana\0" - "Kharoshthi\0" - "Khmer\0" - "L\0" - "L&\0" - "Lao\0" - "Latin\0" - "Limbu\0" - "Linear_B\0" - "Ll\0" - "Lm\0" - "Lo\0" - "Lt\0" - "Lu\0" - "M\0" - "Malayalam\0" - "Mc\0" - "Me\0" - "Mn\0" - "Mongolian\0" - "Myanmar\0" - "N\0" - "Nd\0" - "New_Tai_Lue\0" - "Nko\0" - "Nl\0" - "No\0" - "Ogham\0" - "Old_Italic\0" - "Old_Persian\0" - "Oriya\0" - "Osmanya\0" - "P\0" - "Pc\0" - "Pd\0" - "Pe\0" - "Pf\0" - "Phags_Pa\0" - "Phoenician\0" - "Pi\0" - "Po\0" - "Ps\0" - "Runic\0" - "S\0" - "Sc\0" - "Shavian\0" - "Sinhala\0" - "Sk\0" - "Sm\0" - "So\0" - "Syloti_Nagri\0" - "Syriac\0" - "Tagalog\0" - "Tagbanwa\0" - "Tai_Le\0" - "Tamil\0" - "Telugu\0" - "Thaana\0" - "Thai\0" - "Tibetan\0" - "Tifinagh\0" - "Ugaritic\0" - "Yi\0" - "Z\0" - "Zl\0" - "Zp\0" +/* The pcre_utt[] table below translates Unicode property names into type and +code values. It is searched by binary chop, so must be in collating sequence of +name. Originally, the table contained pointers to the name strings in the first +field of each entry. However, that leads to a large number of relocations when +a shared library is dynamically loaded. A significant reduction is made by +putting all the names into a single, large string and then using offsets in the +table itself. Maintenance is more error-prone, but frequent changes to this +data is unlikely. */ + +const char _pcre_utt_names[] = + "Any\0" + "Arabic\0" + "Armenian\0" + "Balinese\0" + "Bengali\0" + "Bopomofo\0" + "Braille\0" + "Buginese\0" + "Buhid\0" + "C\0" + "Canadian_Aboriginal\0" + "Cc\0" + "Cf\0" + "Cherokee\0" + "Cn\0" + "Co\0" + "Common\0" + "Coptic\0" + "Cs\0" + "Cuneiform\0" + "Cypriot\0" + "Cyrillic\0" + "Deseret\0" + "Devanagari\0" + "Ethiopic\0" + "Georgian\0" + "Glagolitic\0" + "Gothic\0" + "Greek\0" + "Gujarati\0" + "Gurmukhi\0" + "Han\0" + "Hangul\0" + "Hanunoo\0" + "Hebrew\0" + "Hiragana\0" + "Inherited\0" + "Kannada\0" + "Katakana\0" + "Kharoshthi\0" + "Khmer\0" + "L\0" + "L&\0" + "Lao\0" + "Latin\0" + "Limbu\0" + "Linear_B\0" + "Ll\0" + "Lm\0" + "Lo\0" + "Lt\0" + "Lu\0" + "M\0" + "Malayalam\0" + "Mc\0" + "Me\0" + "Mn\0" + "Mongolian\0" + "Myanmar\0" + "N\0" + "Nd\0" + "New_Tai_Lue\0" + "Nko\0" + "Nl\0" + "No\0" + "Ogham\0" + "Old_Italic\0" + "Old_Persian\0" + "Oriya\0" + "Osmanya\0" + "P\0" + "Pc\0" + "Pd\0" + "Pe\0" + "Pf\0" + "Phags_Pa\0" + "Phoenician\0" + "Pi\0" + "Po\0" + "Ps\0" + "Runic\0" + "S\0" + "Sc\0" + "Shavian\0" + "Sinhala\0" + "Sk\0" + "Sm\0" + "So\0" + "Syloti_Nagri\0" + "Syriac\0" + "Tagalog\0" + "Tagbanwa\0" + "Tai_Le\0" + "Tamil\0" + "Telugu\0" + "Thaana\0" + "Thai\0" + "Tibetan\0" + "Tifinagh\0" + "Ugaritic\0" + "Yi\0" + "Z\0" + "Zl\0" + "Zp\0" "Zs\0"; const ucp_type_table _pcre_utt[] = { - { 0, PT_ANY, 0 }, - { 4, PT_SC, ucp_Arabic }, - { 11, PT_SC, ucp_Armenian }, - { 20, PT_SC, ucp_Balinese }, - { 29, PT_SC, ucp_Bengali }, - { 37, PT_SC, ucp_Bopomofo }, - { 46, PT_SC, ucp_Braille }, - { 54, PT_SC, ucp_Buginese }, - { 63, PT_SC, ucp_Buhid }, - { 69, PT_GC, ucp_C }, - { 71, PT_SC, ucp_Canadian_Aboriginal }, - { 91, PT_PC, ucp_Cc }, - { 94, PT_PC, ucp_Cf }, - { 97, PT_SC, ucp_Cherokee }, + { 0, PT_ANY, 0 }, + { 4, PT_SC, ucp_Arabic }, + { 11, PT_SC, ucp_Armenian }, + { 20, PT_SC, ucp_Balinese }, + { 29, PT_SC, ucp_Bengali }, + { 37, PT_SC, ucp_Bopomofo }, + { 46, PT_SC, ucp_Braille }, + { 54, PT_SC, ucp_Buginese }, + { 63, PT_SC, ucp_Buhid }, + { 69, PT_GC, ucp_C }, + { 71, PT_SC, ucp_Canadian_Aboriginal }, + { 91, PT_PC, ucp_Cc }, + { 94, PT_PC, ucp_Cf }, + { 97, PT_SC, ucp_Cherokee }, { 106, PT_PC, ucp_Cn }, { 109, PT_PC, ucp_Co }, { 112, PT_SC, ucp_Common }, diff --git a/glib/pcre/pcre_try_flipped.c b/glib/pcre/pcre_try_flipped.c index cd45968..412902b 100644 --- a/glib/pcre/pcre_try_flipped.c +++ b/glib/pcre/pcre_try_flipped.c @@ -43,6 +43,10 @@ see if it was compiled with the opposite endianness. If so, it uses an auxiliary local function to flip the appropriate bytes. */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + #include "pcre_internal.h" @@ -104,6 +108,7 @@ if (byteflip(re->magic_number, sizeof(re->magic_number)) != MAGIC_NUMBER) *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->flags = (pcre_uint16)byteflip(re->flags, sizeof(re->flags)); internal_re->top_bracket = (pcre_uint16)byteflip(re->top_bracket, sizeof(re->top_bracket)); internal_re->top_backref = diff --git a/glib/pcre/pcre_version.c b/glib/pcre/pcre_version.c index 1a9ecb2..425ab21 100644 --- a/glib/pcre/pcre_version.c +++ b/glib/pcre/pcre_version.c @@ -42,6 +42,10 @@ POSSIBILITY OF SUCH DAMAGE. string that identifies the PCRE version that is in use. */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + #include "pcre_internal.h" diff --git a/glib/pcre/pcre_xclass.c b/glib/pcre/pcre_xclass.c index 0b33351..cdf1af12 100644 --- a/glib/pcre/pcre_xclass.c +++ b/glib/pcre/pcre_xclass.c @@ -43,6 +43,10 @@ class (one that contains characters whose values are > 255). It is used by both pcre_exec() and pcre_def_exec(). */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + #include "pcre_internal.h" -- 2.7.4