--- embedaddon/pcre/pcre_compile.c 2012/02/21 23:05:51 1.1 +++ embedaddon/pcre/pcre_compile.c 2013/07/22 08:25:55 1.1.1.4 @@ -6,7 +6,7 @@ and semantics are as close as possible to those of the Perl 5 language. Written by Philip Hazel - Copyright (c) 1997-2011 University of Cambridge + Copyright (c) 1997-2013 University of Cambridge ----------------------------------------------------------------------------- Redistribution and use in source and binary forms, with or without @@ -53,18 +53,22 @@ supporting internal functions that are not used by oth #include "pcre_internal.h" -/* When PCRE_DEBUG is defined, we need the pcre_printint() function, which is -also used by pcretest. PCRE_DEBUG is not defined when building a production -library. */ +/* When PCRE_DEBUG is defined, we need the pcre(16|32)_printint() function, which +is also used by pcretest. PCRE_DEBUG is not defined when building a production +library. We do not need to select pcre16_printint.c specially, because the +COMPILE_PCREx macro will already be appropriately set. */ #ifdef PCRE_DEBUG -#include "pcre_printint.src" +/* pcre_printint.c should not include any headers */ +#define PCRE_INCLUDED +#include "pcre_printint.c" +#undef PCRE_INCLUDED #endif /* Macro for setting individual bits in class bitmaps. */ -#define SETBIT(a,b) a[b/8] |= (1 << (b%8)) +#define SETBIT(a,b) a[(b)/8] |= (1 << ((b)&7)) /* 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 @@ -73,7 +77,19 @@ to check them every time. */ #define OFLOW_MAX (INT_MAX - 20) +/* Definitions to allow mutual recursion */ +static int + add_list_to_class(pcre_uint8 *, pcre_uchar **, int, compile_data *, + const pcre_uint32 *, unsigned int); + +static BOOL + compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, int, int, + pcre_uint32 *, pcre_int32 *, pcre_uint32 *, pcre_int32 *, branch_chain *, + compile_data *, int *); + + + /************************************************* * Code parameters and static tables * *************************************************/ @@ -104,7 +120,18 @@ overrun before it actually does run off the end of the #define WORK_SIZE_SAFETY_MARGIN (100) +/* Private flags added to firstchar and reqchar. */ +#define REQ_CASELESS (1 << 0) /* Indicates caselessness */ +#define REQ_VARY (1 << 1) /* Reqchar followed non-literal item */ +/* Negative values for the firstchar and reqchar flags */ +#define REQ_UNSET (-2) +#define REQ_NONE (-1) + +/* Repeated character flags. */ + +#define UTF_LENGTH 0x10000000l /* The char contains its length. */ + /* Table for handling escaped characters in the range '0'-'z'. Positive returns are simple data values; negative values are for special things like \d and so on. Zero means further processing is needed (for things like \x), or the escape @@ -238,7 +265,7 @@ static const char posix_names[] = STRING_graph0 STRING_print0 STRING_punct0 STRING_space0 STRING_word0 STRING_xdigit; -static const uschar posix_name_lengths[] = { +static const pcre_uint8 posix_name_lengths[] = { 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 }; /* Table of class bit maps for each POSIX class. Each class is formed from a @@ -273,47 +300,101 @@ substitutes must be in the order of the names, defined both positive and negative cases. NULL means no substitute. */ #ifdef SUPPORT_UCP -static const uschar *substitutes[] = { - (uschar *)"\\P{Nd}", /* \D */ - (uschar *)"\\p{Nd}", /* \d */ - (uschar *)"\\P{Xsp}", /* \S */ /* NOTE: Xsp is Perl space */ - (uschar *)"\\p{Xsp}", /* \s */ - (uschar *)"\\P{Xwd}", /* \W */ - (uschar *)"\\p{Xwd}" /* \w */ +static const pcre_uchar string_PNd[] = { + CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, + CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_pNd[] = { + CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, + CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_PXsp[] = { + CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, + CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_pXsp[] = { + CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, + CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_PXwd[] = { + CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, + CHAR_X, CHAR_w, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_pXwd[] = { + CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, + CHAR_X, CHAR_w, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' }; + +static const pcre_uchar *substitutes[] = { + string_PNd, /* \D */ + string_pNd, /* \d */ + string_PXsp, /* \S */ /* NOTE: Xsp is Perl space */ + string_pXsp, /* \s */ + string_PXwd, /* \W */ + string_pXwd /* \w */ }; -static const uschar *posix_substitutes[] = { - (uschar *)"\\p{L}", /* alpha */ - (uschar *)"\\p{Ll}", /* lower */ - (uschar *)"\\p{Lu}", /* upper */ - (uschar *)"\\p{Xan}", /* alnum */ - NULL, /* ascii */ - (uschar *)"\\h", /* blank */ - NULL, /* cntrl */ - (uschar *)"\\p{Nd}", /* digit */ - NULL, /* graph */ - NULL, /* print */ - NULL, /* punct */ - (uschar *)"\\p{Xps}", /* space */ /* NOTE: Xps is POSIX space */ - (uschar *)"\\p{Xwd}", /* word */ - NULL, /* xdigit */ +static const pcre_uchar string_pL[] = { + CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, + CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_pLl[] = { + CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, + CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_pLu[] = { + CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, + CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_pXan[] = { + CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, + CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_h[] = { + CHAR_BACKSLASH, CHAR_h, '\0' }; +static const pcre_uchar string_pXps[] = { + CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, + CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_PL[] = { + CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, + CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_PLl[] = { + CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, + CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_PLu[] = { + CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, + CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_PXan[] = { + CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, + CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_H[] = { + CHAR_BACKSLASH, CHAR_H, '\0' }; +static const pcre_uchar string_PXps[] = { + CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, + CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' }; + +static const pcre_uchar *posix_substitutes[] = { + string_pL, /* alpha */ + string_pLl, /* lower */ + string_pLu, /* upper */ + string_pXan, /* alnum */ + NULL, /* ascii */ + string_h, /* blank */ + NULL, /* cntrl */ + string_pNd, /* digit */ + NULL, /* graph */ + NULL, /* print */ + NULL, /* punct */ + string_pXps, /* space */ /* NOTE: Xps is POSIX space */ + string_pXwd, /* word */ + NULL, /* xdigit */ /* Negated cases */ - (uschar *)"\\P{L}", /* ^alpha */ - (uschar *)"\\P{Ll}", /* ^lower */ - (uschar *)"\\P{Lu}", /* ^upper */ - (uschar *)"\\P{Xan}", /* ^alnum */ - NULL, /* ^ascii */ - (uschar *)"\\H", /* ^blank */ - NULL, /* ^cntrl */ - (uschar *)"\\P{Nd}", /* ^digit */ - NULL, /* ^graph */ - NULL, /* ^print */ - NULL, /* ^punct */ - (uschar *)"\\P{Xps}", /* ^space */ /* NOTE: Xps is POSIX space */ - (uschar *)"\\P{Xwd}", /* ^word */ - NULL /* ^xdigit */ + string_PL, /* ^alpha */ + string_PLl, /* ^lower */ + string_PLu, /* ^upper */ + string_PXan, /* ^alnum */ + NULL, /* ^ascii */ + string_H, /* ^blank */ + NULL, /* ^cntrl */ + string_PNd, /* ^digit */ + NULL, /* ^graph */ + NULL, /* ^print */ + NULL, /* ^punct */ + string_PXps, /* ^space */ /* NOTE: Xps is POSIX space */ + string_PXwd, /* ^word */ + NULL /* ^xdigit */ }; -#define POSIX_SUBSIZE (sizeof(posix_substitutes)/sizeof(uschar *)) +#define POSIX_SUBSIZE (sizeof(posix_substitutes) / sizeof(pcre_uchar *)) #endif #define STRING(a) # a @@ -372,7 +453,7 @@ static const char error_texts[] = /* 30 */ "unknown POSIX class name\0" "POSIX collating elements are not supported\0" - "this version of PCRE is not compiled with PCRE_UTF8 support\0" + "this version of PCRE is compiled without UTF support\0" "spare error\0" /** DEAD **/ "character value in \\x{...} sequence is too large\0" /* 35 */ @@ -395,7 +476,7 @@ static const char error_texts[] = "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0" /* 50 */ "repeated subpattern is too long\0" /** DEAD **/ - "octal value is greater than \\377 (not in UTF-8 mode)\0" + "octal value is greater than \\377 in 8-bit non-UTF-8 mode\0" "internal error: overran compiling workspace\0" "internal error: previously-checked referenced subpattern not found\0" "DEFINE group contains more than one branch\0" @@ -406,7 +487,7 @@ static const char error_texts[] = "a numbered reference must not be zero\0" "an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)\0" /* 60 */ - "(*VERB) not recognized\0" + "(*VERB) not recognized or malformed\0" "number is too big\0" "subpattern name expected\0" "digit expected after (?+\0" @@ -414,13 +495,20 @@ static const char error_texts[] = /* 65 */ "different names for subpatterns of the same number are not allowed\0" "(*MARK) must have an argument\0" - "this version of PCRE is not compiled with PCRE_UCP support\0" + "this version of PCRE is not compiled with Unicode property support\0" "\\c must be followed by an ASCII character\0" "\\k is not followed by a braced, angle-bracketed, or quoted name\0" /* 70 */ "internal error: unknown opcode in find_fixedlength()\0" "\\N is not supported in a class\0" "too many forward references\0" + "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0" + "invalid UTF-16 string\0" + /* 75 */ + "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0" + "character value in \\u.... sequence is too large\0" + "invalid UTF-32 string\0" + "setting UTF is disabled by the application\0" ; /* Table to identify digits and hex digits. This is used when compiling @@ -439,12 +527,18 @@ For convenience, we use the same bit definitions as in Then we can use ctype_digit and ctype_xdigit in the code. */ +/* Using a simple comparison for decimal numbers rather than a memory read +is much faster, and the resulting code is simpler (the compiler turns it +into a subtraction and unsigned comparison). */ + +#define IS_DIGIT(x) ((x) >= CHAR_0 && (x) <= CHAR_9) + #ifndef EBCDIC /* This is the "normal" case, for ASCII systems, and EBCDIC systems running in UTF-8 mode. */ -static const unsigned char digitab[] = +static const pcre_uint8 digitab[] = { 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 0- 7 */ 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 8- 15 */ @@ -483,7 +577,7 @@ static const unsigned char digitab[] = /* This is the "abnormal" case, for EBCDIC systems not running in UTF-8 mode. */ -static const unsigned char digitab[] = +static const pcre_uint8 digitab[] = { 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 0- 7 0 */ 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 8- 15 */ @@ -518,7 +612,7 @@ static const unsigned char digitab[] = 0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /* 0 - 7 F0 */ 0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/* 8 -255 */ -static const unsigned char ebcdic_chartab[] = { /* chartable partial dup */ +static const pcre_uint8 ebcdic_chartab[] = { /* chartable partial dup */ 0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /* 0- 7 */ 0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /* 8- 15 */ 0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /* 16- 23 */ @@ -554,14 +648,7 @@ static const unsigned char ebcdic_chartab[] = { /* cha #endif -/* Definition to allow mutual recursion */ -static BOOL - compile_regex(int, uschar **, const uschar **, int *, BOOL, BOOL, int, int, - int *, int *, branch_chain *, compile_data *, int *); - - - /************************************************* * Find an error text * *************************************************/ @@ -581,8 +668,8 @@ find_error_text(int n) const char *s = error_texts; for (; n > 0; n--) { - while (*s++ != 0) {}; - if (*s == 0) return "Error text not found (please report)"; + while (*s++ != CHAR_NULL) {}; + if (*s == CHAR_NULL) return "Error text not found (please report)"; } return s; } @@ -604,7 +691,7 @@ Returns: 0 if all went well, else an error number static int expand_workspace(compile_data *cd) { -uschar *newspace; +pcre_uchar *newspace; int newsize = cd->workspace_size * 2; if (newsize > COMPILE_WORK_SIZE_MAX) newsize = COMPILE_WORK_SIZE_MAX; @@ -612,13 +699,12 @@ if (cd->workspace_size >= COMPILE_WORK_SIZE_MAX || newsize - cd->workspace_size < WORK_SIZE_SAFETY_MARGIN) return ERR72; -newspace = (pcre_malloc)(newsize); +newspace = (PUBL(malloc))(IN_UCHARS(newsize)); if (newspace == NULL) return ERR21; - -memcpy(newspace, cd->start_workspace, cd->workspace_size); -cd->hwm = (uschar *)newspace + (cd->hwm - cd->start_workspace); +memcpy(newspace, cd->start_workspace, cd->workspace_size * sizeof(pcre_uchar)); +cd->hwm = (pcre_uchar *)newspace + (cd->hwm - cd->start_workspace); if (cd->workspace_size > COMPILE_WORK_SIZE) - (pcre_free)((void *)cd->start_workspace); + (PUBL(free))((void *)cd->start_workspace); cd->start_workspace = newspace; cd->workspace_size = newsize; return 0; @@ -642,17 +728,19 @@ Returns: TRUE or FALSE */ static BOOL -is_counted_repeat(const uschar *p) +is_counted_repeat(const pcre_uchar *p) { -if ((digitab[*p++] & ctype_digit) == 0) return FALSE; -while ((digitab[*p] & ctype_digit) != 0) p++; +if (!IS_DIGIT(*p)) return FALSE; +p++; +while (IS_DIGIT(*p)) p++; if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE; if (*p++ != CHAR_COMMA) return FALSE; if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE; -if ((digitab[*p++] & ctype_digit) == 0) return FALSE; -while ((digitab[*p] & ctype_digit) != 0) p++; +if (!IS_DIGIT(*p)) return FALSE; +p++; +while (IS_DIGIT(*p)) p++; return (*p == CHAR_RIGHT_CURLY_BRACKET); } @@ -664,59 +752,68 @@ return (*p == CHAR_RIGHT_CURLY_BRACKET); *************************************************/ /* This function is called when a \ has been encountered. It either returns a -positive value for a simple escape such as \n, or a negative value which -encodes one of the more complicated things such as \d. A backreference to group -n is returned as -(ESC_REF + n); ESC_REF is the highest ESC_xxx macro. When -UTF-8 is enabled, a positive value greater than 255 may be returned. On entry, -ptr is pointing at the \. On exit, it is on the final character of the escape -sequence. +positive value for a simple escape such as \n, or 0 for a data character +which will be placed in chptr. A backreference to group n is returned as +negative n. When UTF-8 is enabled, a positive value greater than 255 may +be returned in chptr. +On entry,ptr is pointing at the \. On exit, it is on the final character of the +escape sequence. Arguments: ptrptr points to the pattern position pointer + chptr points to the data character errorcodeptr points to the errorcode variable bracount number of previous extracting brackets options the options bits isclass TRUE if inside a character class -Returns: zero or positive => a data character - negative => a special escape sequence +Returns: zero => a data character + positive => a special escape sequence + negative => a back reference on error, errorcodeptr is set */ static int -check_escape(const uschar **ptrptr, int *errorcodeptr, int bracount, - int options, BOOL isclass) +check_escape(const pcre_uchar **ptrptr, pcre_uint32 *chptr, int *errorcodeptr, + int bracount, int options, BOOL isclass) { -BOOL utf8 = (options & PCRE_UTF8) != 0; -const uschar *ptr = *ptrptr + 1; -int c, i; +/* PCRE_UTF16 has the same value as PCRE_UTF8. */ +BOOL utf = (options & PCRE_UTF8) != 0; +const pcre_uchar *ptr = *ptrptr + 1; +pcre_uint32 c; +int escape = 0; +int i; GETCHARINCTEST(c, ptr); /* Get character value, increment pointer */ ptr--; /* Set pointer back to the last byte */ /* If backslash is at the end of the pattern, it's an error. */ -if (c == 0) *errorcodeptr = ERR1; +if (c == CHAR_NULL) *errorcodeptr = ERR1; /* Non-alphanumerics are literals. For digits or letters, do an initial lookup in a table. A non-zero result is something that can be returned immediately. Otherwise further processing may be required. */ #ifndef EBCDIC /* ASCII/UTF-8 coding */ -else if (c < CHAR_0 || c > CHAR_z) {} /* Not alphanumeric */ -else if ((i = escapes[c - CHAR_0]) != 0) c = i; +/* Not alphanumeric */ +else if (c < CHAR_0 || c > CHAR_z) {} +else if ((i = escapes[c - CHAR_0]) != 0) + { if (i > 0) c = (pcre_uint32)i; else escape = -i; } #else /* EBCDIC coding */ -else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {} /* Not alphanumeric */ -else if ((i = escapes[c - 0x48]) != 0) c = i; +/* Not alphanumeric */ +else if (c < CHAR_a || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {} +else if ((i = escapes[c - 0x48]) != 0) { if (i > 0) c = (pcre_uint32)i; else escape = -i; } #endif /* Escapes that need further processing, or are illegal. */ else { - const uschar *oldptr; - BOOL braced, negated; + const pcre_uchar *oldptr; + BOOL braced, negated, overflow; + int s; switch (c) { @@ -733,13 +830,15 @@ else { /* In JavaScript, \u must be followed by four hexadecimal numbers. Otherwise it is a lowercase u letter. */ - if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0 - && (digitab[ptr[3]] & ctype_xdigit) != 0 && (digitab[ptr[4]] & ctype_xdigit) != 0) + if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0 + && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0 + && MAX_255(ptr[3]) && (digitab[ptr[3]] & ctype_xdigit) != 0 + && MAX_255(ptr[4]) && (digitab[ptr[4]] & ctype_xdigit) != 0) { c = 0; for (i = 0; i < 4; ++i) { - register int cc = *(++ptr); + register pcre_uint32 cc = *(++ptr); #ifndef EBCDIC /* ASCII/UTF-8 coding */ if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */ c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10)); @@ -748,6 +847,18 @@ else c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10)); #endif } + +#if defined COMPILE_PCRE8 + if (c > (utf ? 0x10ffffU : 0xffU)) +#elif defined COMPILE_PCRE16 + if (c > (utf ? 0x10ffffU : 0xffffU)) +#elif defined COMPILE_PCRE32 + if (utf && c > 0x10ffffU) +#endif + { + *errorcodeptr = ERR76; + } + else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73; } } else @@ -774,13 +885,13 @@ else (3) For Oniguruma compatibility we also support \g followed by a name or a number either in angle brackets or in single quotes. However, these are (possibly recursive) subroutine calls, _not_ backreferences. Just return - the -ESC_g code (cf \k). */ + the ESC_g code (cf \k). */ case CHAR_g: if (isclass) break; if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE) { - c = -ESC_g; + escape = ESC_g; break; } @@ -788,12 +899,12 @@ else if (ptr[1] == CHAR_LEFT_CURLY_BRACKET) { - const uschar *p; - for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++) - if (*p != CHAR_MINUS && (digitab[*p] & ctype_digit) == 0) break; - if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET) + const pcre_uchar *p; + for (p = ptr+2; *p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET; p++) + if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break; + if (*p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET) { - c = -ESC_k; + escape = ESC_k; break; } braced = TRUE; @@ -808,12 +919,22 @@ else } else negated = FALSE; - c = 0; - while ((digitab[ptr[1]] & ctype_digit) != 0) - c = c * 10 + *(++ptr) - CHAR_0; - - if (c < 0) /* Integer overflow */ + /* The integer range is limited by the machine's int representation. */ + s = 0; + overflow = FALSE; + while (IS_DIGIT(ptr[1])) { + if (s > INT_MAX / 10 - 1) /* Integer overflow */ + { + overflow = TRUE; + break; + } + s = s * 10 + (int)(*(++ptr) - CHAR_0); + } + if (overflow) /* Integer overflow */ + { + while (IS_DIGIT(ptr[1])) + ptr++; *errorcodeptr = ERR61; break; } @@ -824,7 +945,7 @@ else break; } - if (c == 0) + if (s == 0) { *errorcodeptr = ERR58; break; @@ -832,15 +953,15 @@ else if (negated) { - if (c > bracount) + if (s > bracount) { *errorcodeptr = ERR15; break; } - c = bracount - (c - 1); + s = bracount - (s - 1); } - c = -(ESC_REF + c); + escape = -s; break; /* The handling of escape sequences consisting of a string of digits @@ -861,17 +982,28 @@ else if (!isclass) { oldptr = ptr; - c -= CHAR_0; - while ((digitab[ptr[1]] & ctype_digit) != 0) - c = c * 10 + *(++ptr) - CHAR_0; - if (c < 0) /* Integer overflow */ + /* The integer range is limited by the machine's int representation. */ + s = (int)(c -CHAR_0); + overflow = FALSE; + while (IS_DIGIT(ptr[1])) { + if (s > INT_MAX / 10 - 1) /* Integer overflow */ + { + overflow = TRUE; + break; + } + s = s * 10 + (int)(*(++ptr) - CHAR_0); + } + if (overflow) /* Integer overflow */ + { + while (IS_DIGIT(ptr[1])) + ptr++; *errorcodeptr = ERR61; break; } - if (c < 10 || c <= bracount) + if (s < 10 || s <= bracount) { - c = -(ESC_REF + c); + escape = -s; break; } ptr = oldptr; /* Put the pointer back and fall through */ @@ -891,31 +1023,34 @@ else /* \0 always starts an octal number, but we may drop through to here with a larger first octal digit. The original code used just to take the least significant 8 bits of octal numbers (I think this is what early Perls used - to do). Nowadays we allow for larger numbers in UTF-8 mode, but no more - than 3 octal digits. */ + to do). Nowadays we allow for larger numbers in UTF-8 mode and 16-bit mode, + but no more than 3 octal digits. */ case CHAR_0: c -= CHAR_0; while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7) c = c * 8 + *(++ptr) - CHAR_0; - if (!utf8 && c > 255) *errorcodeptr = ERR51; +#ifdef COMPILE_PCRE8 + if (!utf && c > 0xff) *errorcodeptr = ERR51; +#endif break; /* \x is complicated. \x{ddd} is a character number which can be greater - than 0xff in utf8 mode, but only if the ddd are hex digits. If not, { is - treated as a data character. */ + than 0xff in utf or non-8bit mode, but only if the ddd are hex digits. + If not, { is treated as a data character. */ case CHAR_x: if ((options & PCRE_JAVASCRIPT_COMPAT) != 0) { /* In JavaScript, \x must be followed by two hexadecimal numbers. Otherwise it is a lowercase x letter. */ - if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0) + if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0 + && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0) { c = 0; for (i = 0; i < 2; ++i) { - register int cc = *(++ptr); + register pcre_uint32 cc = *(++ptr); #ifndef EBCDIC /* ASCII/UTF-8 coding */ if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */ c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10)); @@ -930,16 +1065,19 @@ else if (ptr[1] == CHAR_LEFT_CURLY_BRACKET) { - const uschar *pt = ptr + 2; - int count = 0; + const pcre_uchar *pt = ptr + 2; c = 0; - while ((digitab[*pt] & ctype_xdigit) != 0) + overflow = FALSE; + while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) { - register int cc = *pt++; + register pcre_uint32 cc = *pt++; if (c == 0 && cc == CHAR_0) continue; /* Leading zeroes */ - count++; +#ifdef COMPILE_PCRE32 + if (c >= 0x10000000l) { overflow = TRUE; break; } +#endif + #ifndef EBCDIC /* ASCII/UTF-8 coding */ if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */ c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10)); @@ -947,11 +1085,25 @@ else if (cc >= CHAR_a && cc <= CHAR_z) cc += 64; /* Convert to upper case */ c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10)); #endif + +#if defined COMPILE_PCRE8 + if (c > (utf ? 0x10ffffU : 0xffU)) { overflow = TRUE; break; } +#elif defined COMPILE_PCRE16 + if (c > (utf ? 0x10ffffU : 0xffffU)) { overflow = TRUE; break; } +#elif defined COMPILE_PCRE32 + if (utf && c > 0x10ffffU) { overflow = TRUE; break; } +#endif } + if (overflow) + { + while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++; + *errorcodeptr = ERR34; + } + if (*pt == CHAR_RIGHT_CURLY_BRACKET) { - if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34; + if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73; ptr = pt; break; } @@ -963,9 +1115,9 @@ else /* Read just a single-byte hex-defined char */ c = 0; - while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0) + while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0) { - int cc; /* Some compilers don't like */ + pcre_uint32 cc; /* Some compilers don't like */ cc = *(++ptr); /* ++ in initializers */ #ifndef EBCDIC /* ASCII/UTF-8 coding */ if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */ @@ -984,7 +1136,7 @@ else case CHAR_c: c = *(++ptr); - if (c == 0) + if (c == CHAR_NULL) { *errorcodeptr = ERR2; break; @@ -1024,23 +1176,22 @@ else newline". PCRE does not support \N{name}. However, it does support quantification such as \N{2,3}. */ -if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET && +if (escape == ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET && !is_counted_repeat(ptr+2)) *errorcodeptr = ERR37; /* If PCRE_UCP is set, we change the values for \d etc. */ -if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w) - c -= (ESC_DU - ESC_D); +if ((options & PCRE_UCP) != 0 && escape >= ESC_D && escape <= ESC_w) + escape += (ESC_DU - ESC_D); /* Set the pointer to the final character before returning. */ *ptrptr = ptr; -return c; +*chptr = c; +return escape; } - - #ifdef SUPPORT_UCP /************************************************* * Handle \P and \p * @@ -1054,21 +1205,24 @@ escape sequence. Argument: ptrptr points to the pattern position pointer negptr points to a boolean that is set TRUE for negation else FALSE - dptr points to an int that is set to the detailed property value + ptypeptr points to an unsigned int that is set to the type value + pdataptr points to an unsigned int that is set to the detailed property value errorcodeptr points to the error code variable -Returns: type value from ucp_type_table, or -1 for an invalid type +Returns: TRUE if the type value was found, or FALSE for an invalid type */ -static int -get_ucp(const uschar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr) +static BOOL +get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, unsigned int *ptypeptr, + unsigned int *pdataptr, int *errorcodeptr) { -int c, i, bot, top; -const uschar *ptr = *ptrptr; -char name[32]; +pcre_uchar c; +int i, bot, top; +const pcre_uchar *ptr = *ptrptr; +pcre_uchar name[32]; c = *(++ptr); -if (c == 0) goto ERROR_RETURN; +if (c == CHAR_NULL) goto ERROR_RETURN; *negptr = FALSE; @@ -1082,10 +1236,10 @@ if (c == CHAR_LEFT_CURLY_BRACKET) *negptr = TRUE; ptr++; } - for (i = 0; i < (int)sizeof(name) - 1; i++) + for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++) { c = *(++ptr); - if (c == 0) goto ERROR_RETURN; + if (c == CHAR_NULL) goto ERROR_RETURN; if (c == CHAR_RIGHT_CURLY_BRACKET) break; name[i] = c; } @@ -1106,28 +1260,30 @@ else /* Search for a recognized property name using binary chop */ bot = 0; -top = _pcre_utt_size; +top = PRIV(utt_size); while (bot < top) { + int r; i = (bot + top) >> 1; - c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset); - if (c == 0) + r = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset); + if (r == 0) { - *dptr = _pcre_utt[i].value; - return _pcre_utt[i].type; + *ptypeptr = PRIV(utt)[i].type; + *pdataptr = PRIV(utt)[i].value; + return TRUE; } - if (c > 0) bot = i + 1; else top = i; + if (r > 0) bot = i + 1; else top = i; } *errorcodeptr = ERR47; *ptrptr = ptr; -return -1; +return FALSE; ERROR_RETURN: *errorcodeptr = ERR46; *ptrptr = ptr; -return -1; +return FALSE; } #endif @@ -1153,8 +1309,8 @@ Returns: pointer to '}' on success; current ptr on error, with errorcodeptr set non-zero */ -static const uschar * -read_repeat_counts(const uschar *p, int *minp, int *maxp, int *errorcodeptr) +static const pcre_uchar * +read_repeat_counts(const pcre_uchar *p, int *minp, int *maxp, int *errorcodeptr) { int min = 0; int max = -1; @@ -1162,7 +1318,7 @@ int max = -1; /* Read the minimum value and do a paranoid check: a negative value indicates an integer overflow. */ -while ((digitab[*p] & ctype_digit) != 0) min = min * 10 + *p++ - CHAR_0; +while (IS_DIGIT(*p)) min = min * 10 + (int)(*p++ - CHAR_0); if (min < 0 || min > 65535) { *errorcodeptr = ERR5; @@ -1177,7 +1333,7 @@ if (*p == CHAR_RIGHT_CURLY_BRACKET) max = min; else if (*(++p) != CHAR_RIGHT_CURLY_BRACKET) { max = 0; - while((digitab[*p] & ctype_digit) != 0) max = max * 10 + *p++ - CHAR_0; + while(IS_DIGIT(*p)) max = max * 10 + (int)(*p++ - CHAR_0); if (max < 0 || max > 65535) { *errorcodeptr = ERR5; @@ -1232,17 +1388,17 @@ Arguments: name name to seek, or NULL if seeking a numbered subpattern lorn name length, or subpattern number if name is NULL xmode TRUE if we are in /x mode - utf8 TRUE if we are in UTF-8 mode + utf TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode count pointer to the current capturing subpattern number (updated) Returns: the number of the named subpattern, or -1 if not found */ static int -find_parens_sub(uschar **ptrptr, compile_data *cd, const uschar *name, int lorn, - BOOL xmode, BOOL utf8, int *count) +find_parens_sub(pcre_uchar **ptrptr, compile_data *cd, const pcre_uchar *name, int lorn, + BOOL xmode, BOOL utf, int *count) { -uschar *ptr = *ptrptr; +pcre_uchar *ptr = *ptrptr; int start_count = *count; int hwm_count = start_count; BOOL dup_parens = FALSE; @@ -1254,7 +1410,11 @@ if (ptr[0] == CHAR_LEFT_PARENTHESIS) { /* Handle specials such as (*SKIP) or (*UTF8) etc. */ - if (ptr[1] == CHAR_ASTERISK) ptr += 2; + if (ptr[1] == CHAR_ASTERISK) + { + ptr += 2; + while (ptr < cd->end_pattern && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++; + } /* Handle a normal, unnamed capturing parenthesis. */ @@ -1278,7 +1438,8 @@ if (ptr[0] == CHAR_LEFT_PARENTHESIS) else if (ptr[2] == CHAR_NUMBER_SIGN) { - for (ptr += 3; *ptr != 0; ptr++) if (*ptr == CHAR_RIGHT_PARENTHESIS) break; + for (ptr += 3; *ptr != CHAR_NULL; ptr++) + if (*ptr == CHAR_RIGHT_PARENTHESIS) break; goto FAIL_EXIT; } @@ -1291,8 +1452,8 @@ if (ptr[0] == CHAR_LEFT_PARENTHESIS) ptr += 2; if (ptr[1] != CHAR_QUESTION_MARK) { - while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++; - if (*ptr != 0) ptr++; + while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++; + if (*ptr != CHAR_NULL) ptr++; } } @@ -1308,16 +1469,16 @@ if (ptr[0] == CHAR_LEFT_PARENTHESIS) if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK && ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE) { - int term; - const uschar *thisname; + pcre_uchar term; + const pcre_uchar *thisname; *count += 1; if (name == NULL && *count == lorn) return *count; term = *ptr++; if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN; thisname = ptr; while (*ptr != term) ptr++; - if (name != NULL && lorn == ptr - thisname && - strncmp((const char *)name, (const char *)thisname, lorn) == 0) + if (name != NULL && lorn == (int)(ptr - thisname) && + STRNCMP_UC_UC(name, thisname, (unsigned int)lorn) == 0) return *count; term++; } @@ -1335,11 +1496,11 @@ for (; ptr < cd->end_pattern; ptr++) if (*ptr == CHAR_BACKSLASH) { - if (*(++ptr) == 0) goto FAIL_EXIT; + if (*(++ptr) == CHAR_NULL) goto FAIL_EXIT; if (*ptr == CHAR_Q) for (;;) { - while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {}; - if (*ptr == 0) goto FAIL_EXIT; + while (*(++ptr) != CHAR_NULL && *ptr != CHAR_BACKSLASH) {}; + if (*ptr == CHAR_NULL) goto FAIL_EXIT; if (*(++ptr) == CHAR_E) break; } continue; @@ -1360,7 +1521,7 @@ for (; ptr < cd->end_pattern; ptr++) { if (ptr[2] == CHAR_E) ptr+= 2; - else if (strncmp((const char *)ptr+2, + else if (STRNCMP_UC_C8(ptr + 2, STR_Q STR_BACKSLASH STR_E, 3) == 0) ptr += 4; else @@ -1383,14 +1544,14 @@ for (; ptr < cd->end_pattern; ptr++) while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET) { - if (*ptr == 0) return -1; + if (*ptr == CHAR_NULL) return -1; if (*ptr == CHAR_BACKSLASH) { - if (*(++ptr) == 0) goto FAIL_EXIT; + if (*(++ptr) == CHAR_NULL) goto FAIL_EXIT; if (*ptr == CHAR_Q) for (;;) { - while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {}; - if (*ptr == 0) goto FAIL_EXIT; + while (*(++ptr) != CHAR_NULL && *ptr != CHAR_BACKSLASH) {}; + if (*ptr == CHAR_NULL) goto FAIL_EXIT; if (*(++ptr) == CHAR_E) break; } continue; @@ -1404,15 +1565,15 @@ for (; ptr < cd->end_pattern; ptr++) if (xmode && *ptr == CHAR_NUMBER_SIGN) { ptr++; - while (*ptr != 0) + while (*ptr != CHAR_NULL) { if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; } ptr++; -#ifdef SUPPORT_UTF8 - if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++; +#ifdef SUPPORT_UTF + if (utf) FORWARDCHAR(ptr); #endif } - if (*ptr == 0) goto FAIL_EXIT; + if (*ptr == CHAR_NULL) goto FAIL_EXIT; continue; } @@ -1420,9 +1581,9 @@ for (; ptr < cd->end_pattern; ptr++) if (*ptr == CHAR_LEFT_PARENTHESIS) { - int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, count); + int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count); if (rc > 0) return rc; - if (*ptr == 0) goto FAIL_EXIT; + if (*ptr == CHAR_NULL) goto FAIL_EXIT; } else if (*ptr == CHAR_RIGHT_PARENTHESIS) @@ -1466,16 +1627,16 @@ Arguments: name name to seek, or NULL if seeking a numbered subpattern lorn name length, or subpattern number if name is NULL xmode TRUE if we are in /x mode - utf8 TRUE if we are in UTF-8 mode + utf TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode Returns: the number of the found subpattern, or -1 if not found */ static int -find_parens(compile_data *cd, const uschar *name, int lorn, BOOL xmode, - BOOL utf8) +find_parens(compile_data *cd, const pcre_uchar *name, int lorn, BOOL xmode, + BOOL utf) { -uschar *ptr = (uschar *)cd->start_pattern; +pcre_uchar *ptr = (pcre_uchar *)cd->start_pattern; int count = 0; int rc; @@ -1486,8 +1647,8 @@ matching closing parens. That is why we have to have a for (;;) { - rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, &count); - if (rc > 0 || *ptr++ == 0) break; + rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count); + if (rc > 0 || *ptr++ == CHAR_NULL) break; } return rc; @@ -1513,8 +1674,8 @@ Arguments: Returns: pointer to the first significant opcode */ -static const uschar* -first_significant_code(const uschar *code, BOOL skipassert) +static const pcre_uchar* +first_significant_code(const pcre_uchar *code, BOOL skipassert) { for (;;) { @@ -1525,7 +1686,7 @@ for (;;) case OP_ASSERTBACK_NOT: if (!skipassert) return code; do code += GET(code, 1); while (*code == OP_ALT); - code += _pcre_OP_lengths[*code]; + code += PRIV(OP_lengths)[*code]; break; case OP_WORD_BOUNDARY: @@ -1539,7 +1700,7 @@ for (;;) case OP_RREF: case OP_NRREF: case OP_DEF: - code += _pcre_OP_lengths[*code]; + code += PRIV(OP_lengths)[*code]; break; default: @@ -1569,7 +1730,7 @@ and doing the check at the end; a flag specifies which Arguments: code points to the start of the pattern (the bracket) - utf8 TRUE in UTF-8 mode + utf TRUE in UTF-8 / UTF-16 / UTF-32 mode atend TRUE if called when the pattern is complete cd the "compile data" structure @@ -1581,12 +1742,12 @@ Returns: the fixed length, */ static int -find_fixedlength(uschar *code, BOOL utf8, BOOL atend, compile_data *cd) +find_fixedlength(pcre_uchar *code, BOOL utf, BOOL atend, compile_data *cd) { int length = -1; register int branchlength = 0; -register uschar *cc = code + 1 + LINK_SIZE; +register pcre_uchar *cc = code + 1 + LINK_SIZE; /* Scan along the opcodes for this branch. If we get to the end of the branch, check the length against that of the other branches. */ @@ -1594,8 +1755,9 @@ branch, check the length against that of the other bra for (;;) { int d; - uschar *ce, *cs; - register int op = *cc; + pcre_uchar *ce, *cs; + register pcre_uchar op = *cc; + switch (op) { /* We only need to continue for OP_CBRA (normal capturing bracket) and @@ -1608,7 +1770,7 @@ for (;;) case OP_ONCE: case OP_ONCE_NC: case OP_COND: - d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), utf8, atend, cd); + d = find_fixedlength(cc + ((op == OP_CBRA)? IMM2_SIZE : 0), utf, atend, cd); if (d < 0) return d; branchlength += d; do cc += GET(cc, 1); while (*cc == OP_ALT); @@ -1639,10 +1801,10 @@ for (;;) case OP_RECURSE: if (!atend) return -3; - cs = ce = (uschar *)cd->start_code + GET(cc, 1); /* Start subpattern */ - do ce += GET(ce, 1); while (*ce == OP_ALT); /* End subpattern */ - if (cc > cs && cc < ce) return -1; /* Recursion */ - d = find_fixedlength(cs + 2, utf8, atend, cd); + cs = ce = (pcre_uchar *)cd->start_code + GET(cc, 1); /* Start subpattern */ + do ce += GET(ce, 1); while (*ce == OP_ALT); /* End subpattern */ + if (cc > cs && cc < ce) return -1; /* Recursion */ + d = find_fixedlength(cs + IMM2_SIZE, utf, atend, cd); if (d < 0) return d; branchlength += d; cc += 1 + LINK_SIZE; @@ -1655,7 +1817,8 @@ for (;;) case OP_ASSERTBACK: case OP_ASSERTBACK_NOT: do cc += GET(cc, 1); while (*cc == OP_ALT); - /* Fall through */ + cc += PRIV(OP_lengths)[*cc]; + break; /* Skip over things that don't match chars */ @@ -1663,7 +1826,7 @@ for (;;) case OP_PRUNE_ARG: case OP_SKIP_ARG: case OP_THEN_ARG: - cc += cc[1] + _pcre_OP_lengths[*cc]; + cc += cc[1] + PRIV(OP_lengths)[*cc]; break; case OP_CALLOUT: @@ -1690,7 +1853,7 @@ for (;;) case OP_SOM: case OP_THEN: case OP_WORD_BOUNDARY: - cc += _pcre_OP_lengths[*cc]; + cc += PRIV(OP_lengths)[*cc]; break; /* Handle literal characters */ @@ -1701,8 +1864,8 @@ for (;;) case OP_NOTI: branchlength++; cc += 2; -#ifdef SUPPORT_UTF8 - if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f]; +#ifdef SUPPORT_UTF + if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); #endif break; @@ -1713,17 +1876,18 @@ for (;;) case OP_EXACTI: case OP_NOTEXACT: case OP_NOTEXACTI: - branchlength += GET2(cc,1); - cc += 4; -#ifdef SUPPORT_UTF8 - if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f]; + branchlength += (int)GET2(cc,1); + cc += 2 + IMM2_SIZE; +#ifdef SUPPORT_UTF + if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); #endif break; case OP_TYPEEXACT: branchlength += GET2(cc,1); - if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2; - cc += 4; + if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP) + cc += 2; + cc += 1 + IMM2_SIZE + 1; break; /* Handle single-char matchers */ @@ -1757,16 +1921,20 @@ for (;;) /* Check a class for variable quantification */ -#ifdef SUPPORT_UTF8 + case OP_CLASS: + case OP_NCLASS: +#if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 case OP_XCLASS: - cc += GET(cc, 1) - 33; - /* Fall through */ + /* The original code caused an unsigned overflow in 64 bit systems, + so now we use a conditional statement. */ + if (op == OP_XCLASS) + cc += GET(cc, 1); + else + cc += PRIV(OP_lengths)[OP_CLASS]; +#else + cc += PRIV(OP_lengths)[OP_CLASS]; #endif - case OP_CLASS: - case OP_NCLASS: - cc += 33; - switch (*cc) { case OP_CRPLUS: @@ -1779,9 +1947,9 @@ for (;;) case OP_CRRANGE: case OP_CRMINRANGE: - if (GET2(cc,1) != GET2(cc,3)) return -1; - branchlength += GET2(cc,1); - cc += 5; + if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1; + branchlength += (int)GET2(cc,1); + cc += 1 + 2 * IMM2_SIZE; break; default: @@ -1896,18 +2064,18 @@ length. Arguments: code points to start of expression - utf8 TRUE in UTF-8 mode + utf TRUE in UTF-8 / UTF-16 / UTF-32 mode number the required bracket number or negative to find a lookbehind Returns: pointer to the opcode for the bracket, or NULL if not found */ -const uschar * -_pcre_find_bracket(const uschar *code, BOOL utf8, int number) +const pcre_uchar * +PRIV(find_bracket)(const pcre_uchar *code, BOOL utf, int number) { for (;;) { - register int c = *code; + register pcre_uchar c = *code; if (c == OP_END) return NULL; @@ -1921,8 +2089,8 @@ for (;;) else if (c == OP_REVERSE) { - if (number < 0) return (uschar *)code; - code += _pcre_OP_lengths[c]; + if (number < 0) return (pcre_uchar *)code; + code += PRIV(OP_lengths)[c]; } /* Handle capturing bracket */ @@ -1930,9 +2098,9 @@ for (;;) else if (c == OP_CBRA || c == OP_SCBRA || c == OP_CBRAPOS || c == OP_SCBRAPOS) { - int n = GET2(code, 1+LINK_SIZE); - if (n == number) return (uschar *)code; - code += _pcre_OP_lengths[c]; + int n = (int)GET2(code, 1+LINK_SIZE); + if (n == number) return (pcre_uchar *)code; + code += PRIV(OP_lengths)[c]; } /* Otherwise, we can get the item's length from the table, except that for @@ -1960,15 +2128,13 @@ for (;;) case OP_TYPEMINUPTO: case OP_TYPEEXACT: case OP_TYPEPOSUPTO: - if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2; + if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP) + code += 2; break; case OP_MARK: case OP_PRUNE_ARG: case OP_SKIP_ARG: - code += code[1]; - break; - case OP_THEN_ARG: code += code[1]; break; @@ -1976,14 +2142,14 @@ for (;;) /* Add in the fixed length from the table */ - code += _pcre_OP_lengths[c]; + code += PRIV(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) +#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 + if (utf) switch(c) { case OP_CHAR: case OP_CHARI: @@ -2013,11 +2179,11 @@ for (;;) case OP_MINQUERYI: case OP_POSQUERY: case OP_POSQUERYI: - if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f]; + if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]); break; } #else - (void)(utf8); /* Keep compiler happy by referencing function argument */ + (void)(utf); /* Keep compiler happy by referencing function argument */ #endif } } @@ -2034,17 +2200,17 @@ instance of OP_RECURSE. Arguments: code points to start of expression - utf8 TRUE in UTF-8 mode + utf TRUE in UTF-8 / UTF-16 / UTF-32 mode Returns: pointer to the opcode for OP_RECURSE, or NULL if not found */ -static const uschar * -find_recurse(const uschar *code, BOOL utf8) +static const pcre_uchar * +find_recurse(const pcre_uchar *code, BOOL utf) { for (;;) { - register int c = *code; + register pcre_uchar c = *code; if (c == OP_END) return NULL; if (c == OP_RECURSE) return code; @@ -2079,15 +2245,13 @@ for (;;) case OP_TYPEUPTO: case OP_TYPEMINUPTO: case OP_TYPEEXACT: - if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2; + if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP) + code += 2; break; case OP_MARK: case OP_PRUNE_ARG: case OP_SKIP_ARG: - code += code[1]; - break; - case OP_THEN_ARG: code += code[1]; break; @@ -2095,48 +2259,76 @@ for (;;) /* Add in the fixed length from the table */ - code += _pcre_OP_lengths[c]; + code += PRIV(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) +#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 + if (utf) switch(c) { case OP_CHAR: case OP_CHARI: + case OP_NOT: + case OP_NOTI: case OP_EXACT: case OP_EXACTI: + case OP_NOTEXACT: + case OP_NOTEXACTI: case OP_UPTO: case OP_UPTOI: + case OP_NOTUPTO: + case OP_NOTUPTOI: case OP_MINUPTO: case OP_MINUPTOI: + case OP_NOTMINUPTO: + case OP_NOTMINUPTOI: case OP_POSUPTO: case OP_POSUPTOI: + case OP_NOTPOSUPTO: + case OP_NOTPOSUPTOI: case OP_STAR: case OP_STARI: + case OP_NOTSTAR: + case OP_NOTSTARI: case OP_MINSTAR: case OP_MINSTARI: + case OP_NOTMINSTAR: + case OP_NOTMINSTARI: case OP_POSSTAR: case OP_POSSTARI: + case OP_NOTPOSSTAR: + case OP_NOTPOSSTARI: case OP_PLUS: case OP_PLUSI: + case OP_NOTPLUS: + case OP_NOTPLUSI: case OP_MINPLUS: case OP_MINPLUSI: + case OP_NOTMINPLUS: + case OP_NOTMINPLUSI: case OP_POSPLUS: case OP_POSPLUSI: + case OP_NOTPOSPLUS: + case OP_NOTPOSPLUSI: case OP_QUERY: case OP_QUERYI: + case OP_NOTQUERY: + case OP_NOTQUERYI: case OP_MINQUERY: case OP_MINQUERYI: + case OP_NOTMINQUERY: + case OP_NOTMINQUERYI: case OP_POSQUERY: case OP_POSQUERYI: - if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f]; + case OP_NOTPOSQUERY: + case OP_NOTPOSQUERYI: + if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]); break; } #else - (void)(utf8); /* Keep compiler happy by referencing function argument */ + (void)(utf); /* Keep compiler happy by referencing function argument */ #endif } } @@ -2159,22 +2351,22 @@ bracket whose current branch will already have been sc Arguments: code points to start of search endcode points to where to stop - utf8 TRUE if in UTF8 mode + utf TRUE if in UTF-8 / UTF-16 / UTF-32 mode cd contains pointers to tables etc. Returns: TRUE if what is matched could be empty */ static BOOL -could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8, - compile_data *cd) +could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode, + BOOL utf, compile_data *cd) { -register int c; -for (code = first_significant_code(code + _pcre_OP_lengths[*code], TRUE); +register pcre_uchar c; +for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE); code < endcode; - code = first_significant_code(code + _pcre_OP_lengths[c], TRUE)) + code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE)) { - const uschar *ccode; + const pcre_uchar *ccode; c = *code; @@ -2197,13 +2389,13 @@ for (code = first_significant_code(code + _pcre_OP_len if (c == OP_RECURSE) { - const uschar *scode; + const pcre_uchar *scode; BOOL empty_branch; /* Test for forward reference */ for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE) - if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE; + if ((int)GET(scode, 0) == (int)(code + 1 - cd->start_code)) return TRUE; /* Not a forward reference, test for completed backward reference */ @@ -2215,7 +2407,7 @@ for (code = first_significant_code(code + _pcre_OP_len do { - if (could_be_empty_branch(scode, endcode, utf8, cd)) + if (could_be_empty_branch(scode, endcode, utf, cd)) { empty_branch = TRUE; break; @@ -2233,7 +2425,7 @@ for (code = first_significant_code(code + _pcre_OP_len if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO || c == OP_BRAPOSZERO) { - code += _pcre_OP_lengths[c]; + code += PRIV(OP_lengths)[c]; do code += GET(code, 1); while (*code == OP_ALT); c = *code; continue; @@ -2271,7 +2463,7 @@ for (code = first_significant_code(code + _pcre_OP_len empty_branch = FALSE; do { - if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd)) + if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd)) empty_branch = TRUE; code += GET(code, 1); } @@ -2289,11 +2481,11 @@ for (code = first_significant_code(code + _pcre_OP_len { /* 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 + high-valued characters. The length in PRIV(OP_lengths)[] is zero; the actual length is stored in the compiled code, so we must update "code" here. */ -#ifdef SUPPORT_UTF8 +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 case OP_XCLASS: ccode = code += GET(code, 1); goto CHECK_CLASS_REPEAT; @@ -2301,9 +2493,9 @@ for (code = first_significant_code(code + _pcre_OP_len case OP_CLASS: case OP_NCLASS: - ccode = code + 33; + ccode = code + PRIV(OP_lengths)[OP_CLASS]; -#ifdef SUPPORT_UTF8 +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 CHECK_CLASS_REPEAT: #endif @@ -2376,7 +2568,8 @@ for (code = first_significant_code(code + _pcre_OP_len case OP_TYPEUPTO: case OP_TYPEMINUPTO: case OP_TYPEPOSUPTO: - if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2; + if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP) + code += 2; break; /* End of branch */ @@ -2391,7 +2584,7 @@ for (code = first_significant_code(code + _pcre_OP_len /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO, MINUPTO, and POSUPTO may be followed by a multibyte character */ -#ifdef SUPPORT_UTF8 +#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 case OP_STAR: case OP_STARI: case OP_MINSTAR: @@ -2404,7 +2597,7 @@ for (code = first_significant_code(code + _pcre_OP_len case OP_MINQUERYI: case OP_POSQUERY: case OP_POSQUERYI: - if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f]; + if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]); break; case OP_UPTO: @@ -2413,7 +2606,7 @@ for (code = first_significant_code(code + _pcre_OP_len case OP_MINUPTOI: case OP_POSUPTO: case OP_POSUPTOI: - if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f]; + if (utf && HAS_EXTRALEN(code[1 + IMM2_SIZE])) code += GET_EXTRALEN(code[1 + IMM2_SIZE]); break; #endif @@ -2423,9 +2616,6 @@ for (code = first_significant_code(code + _pcre_OP_len case OP_MARK: case OP_PRUNE_ARG: case OP_SKIP_ARG: - code += code[1]; - break; - case OP_THEN_ARG: code += code[1]; break; @@ -2457,19 +2647,19 @@ Arguments: code points to start of the recursion endcode points to where to stop (current RECURSE item) bcptr points to the chain of current (unclosed) branch starts - utf8 TRUE if in UTF-8 mode + utf TRUE if in UTF-8 / UTF-16 / UTF-32 mode cd pointers to tables etc Returns: TRUE if what is matched could be empty */ static BOOL -could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr, - BOOL utf8, compile_data *cd) +could_be_empty(const pcre_uchar *code, const pcre_uchar *endcode, + branch_chain *bcptr, BOOL utf, compile_data *cd) { while (bcptr != NULL && bcptr->current_branch >= code) { - if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd)) + if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd)) return FALSE; bcptr = bcptr->outer; } @@ -2521,11 +2711,11 @@ Returns: TRUE or FALSE */ static BOOL -check_posix_syntax(const uschar *ptr, const uschar **endptr) +check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr) { -int terminator; /* Don't combine these lines; the Solaris cc */ +pcre_uchar terminator; /* Don't combine these lines; the Solaris cc */ terminator = *(++ptr); /* compiler warns about "non-constant" initializer. */ -for (++ptr; *ptr != 0; ptr++) +for (++ptr; *ptr != CHAR_NULL; ptr++) { if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) ptr++; @@ -2565,14 +2755,14 @@ Returns: a value representing the name, or -1 if u */ static int -check_posix_name(const uschar *ptr, int len) +check_posix_name(const pcre_uchar *ptr, int len) { 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; + STRNCMP_UC_C8(ptr, pn, (unsigned int)len) == 0) return yield; pn += posix_name_lengths[yield] + 1; yield++; } @@ -2604,7 +2794,7 @@ value in the reference (which is a group number). Arguments: group points to the start of the group adjust the amount by which the group is to be moved - utf8 TRUE in UTF-8 mode + utf TRUE in UTF-8 / UTF-16 / UTF-32 mode cd contains pointers to tables etc. save_hwm the hwm forward reference pointer at the start of the group @@ -2612,22 +2802,22 @@ Returns: nothing */ static void -adjust_recurse(uschar *group, int adjust, BOOL utf8, compile_data *cd, - uschar *save_hwm) +adjust_recurse(pcre_uchar *group, int adjust, BOOL utf, compile_data *cd, + pcre_uchar *save_hwm) { -uschar *ptr = group; +pcre_uchar *ptr = group; -while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL) +while ((ptr = (pcre_uchar *)find_recurse(ptr, utf)) != NULL) { int offset; - uschar *hc; + pcre_uchar *hc; /* See if this recursion is on the forward reference list. If so, adjust the reference. */ for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE) { - offset = GET(hc, 0); + offset = (int)GET(hc, 0); if (cd->start_code + offset == ptr + 1) { PUT(hc, 0, offset + adjust); @@ -2640,7 +2830,7 @@ while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NU if (hc >= cd->hwm) { - offset = GET(ptr, 1); + offset = (int)GET(ptr, 1); if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust); } @@ -2665,14 +2855,14 @@ Arguments: Returns: new code pointer */ -static uschar * -auto_callout(uschar *code, const uschar *ptr, compile_data *cd) +static pcre_uchar * +auto_callout(pcre_uchar *code, const pcre_uchar *ptr, compile_data *cd) { *code++ = OP_CALLOUT; *code++ = 255; PUT(code, 0, (int)(ptr - cd->start_pattern)); /* Pattern offset */ PUT(code, LINK_SIZE, 0); /* Default length */ -return code + 2*LINK_SIZE; +return code + 2 * LINK_SIZE; } @@ -2694,7 +2884,7 @@ Returns: nothing */ static void -complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd) +complete_callout(pcre_uchar *previous_callout, const pcre_uchar *ptr, compile_data *cd) { int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2)); PUT(previous_callout, 2 + LINK_SIZE, length); @@ -2708,9 +2898,10 @@ PUT(previous_callout, 2 + LINK_SIZE, length); *************************************************/ /* This function is passed the start and end of a class range, in UTF-8 mode -with UCP support. It searches up the characters, looking for internal ranges of +with UCP support. It searches up the characters, looking for ranges of characters in the "other" case. Each call returns the next one, updating the -start address. +start address. A character with multiple other cases is returned on its own +with a special return value. Arguments: cptr points to starting character value; updated @@ -2718,19 +2909,34 @@ Arguments: ocptr where to put start of othercase range odptr where to put end of othercase range -Yield: TRUE when range returned; FALSE when no more +Yield: -1 when no more + 0 when a range is returned + >0 the CASESET offset for char with multiple other cases + in this case, ocptr contains the original */ -static BOOL -get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr, - unsigned int *odptr) +static int +get_othercase_range(pcre_uint32 *cptr, pcre_uint32 d, pcre_uint32 *ocptr, + pcre_uint32 *odptr) { -unsigned int c, othercase, next; +pcre_uint32 c, othercase, next; +unsigned int co; +/* Find the first character that has an other case. If it has multiple other +cases, return its case offset value. */ + for (c = *cptr; c <= d; c++) - { if ((othercase = UCD_OTHERCASE(c)) != c) break; } + { + if ((co = UCD_CASESET(c)) != 0) + { + *ocptr = c++; /* Character that has the set */ + *cptr = c; /* Rest of input range */ + return (int)co; + } + if ((othercase = UCD_OTHERCASE(c)) != c) break; + } -if (c > d) return FALSE; +if (c > d) return -1; /* Reached end of range */ *ocptr = othercase; next = othercase + 1; @@ -2741,10 +2947,9 @@ for (++c; c <= d; c++) next++; } -*odptr = next - 1; -*cptr = c; - -return TRUE; +*odptr = next - 1; /* End of othercase range */ +*cptr = c; /* Rest of input range */ +return 0; } @@ -2766,9 +2971,14 @@ Returns: TRUE if auto-possessifying is OK */ static BOOL -check_char_prop(int c, int ptype, int pdata, BOOL negated) +check_char_prop(pcre_uint32 c, unsigned int ptype, unsigned int pdata, BOOL negated) { +#ifdef SUPPORT_UCP +const pcre_uint32 *p; +#endif + const ucd_record *prop = GET_UCD(c); + switch(ptype) { case PT_LAMP: @@ -2777,7 +2987,7 @@ switch(ptype) prop->chartype == ucp_Lt) == negated; case PT_GC: - return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated; + return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated; case PT_PC: return (pdata == prop->chartype) == negated; @@ -2788,25 +2998,37 @@ switch(ptype) /* These are specials */ case PT_ALNUM: - return (_pcre_ucp_gentype[prop->chartype] == ucp_L || - _pcre_ucp_gentype[prop->chartype] == ucp_N) == negated; + return (PRIV(ucp_gentype)[prop->chartype] == ucp_L || + PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated; case PT_SPACE: /* Perl space */ - return (_pcre_ucp_gentype[prop->chartype] == ucp_Z || + return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z || c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR) == negated; case PT_PXSPACE: /* POSIX space */ - return (_pcre_ucp_gentype[prop->chartype] == ucp_Z || + return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z || c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) == negated; case PT_WORD: - return (_pcre_ucp_gentype[prop->chartype] == ucp_L || - _pcre_ucp_gentype[prop->chartype] == ucp_N || + return (PRIV(ucp_gentype)[prop->chartype] == ucp_L || + PRIV(ucp_gentype)[prop->chartype] == ucp_N || c == CHAR_UNDERSCORE) == negated; + +#ifdef SUPPORT_UCP + case PT_CLIST: + p = PRIV(ucd_caseless_sets) + prop->caseset; + for (;;) + { + if (c < *p) return !negated; + if (c == *p++) return negated; + } + break; /* Control never reaches here */ +#endif } + return FALSE; } #endif /* SUPPORT_UCP */ @@ -2823,7 +3045,7 @@ sense to automatically possessify the repeated item. Arguments: previous pointer to the repeated opcode - utf8 TRUE in UTF-8 mode + utf TRUE in UTF-8 / UTF-16 / UTF-32 mode ptr next character in pattern options options bits cd contains pointers to tables etc. @@ -2832,11 +3054,13 @@ Returns: TRUE if possessifying is wanted */ static BOOL -check_auto_possessive(const uschar *previous, BOOL utf8, const uschar *ptr, - int options, compile_data *cd) +check_auto_possessive(const pcre_uchar *previous, BOOL utf, + const pcre_uchar *ptr, int options, compile_data *cd) { -int c, next; -int op_code = *previous++; +pcre_uint32 c = NOTACHAR; +pcre_uint32 next; +int escape; +pcre_uchar op_code = *previous++; /* Skip whitespace and comments in extended mode */ @@ -2844,16 +3068,16 @@ if ((options & PCRE_EXTENDED) != 0) { for (;;) { - while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++; + while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++; if (*ptr == CHAR_NUMBER_SIGN) { ptr++; - while (*ptr != 0) + while (*ptr != CHAR_NULL) { if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; } ptr++; -#ifdef SUPPORT_UTF8 - if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++; +#ifdef SUPPORT_UTF + if (utf) FORWARDCHAR(ptr); #endif } } @@ -2867,19 +3091,19 @@ value is a character, a negative value is an escape va if (*ptr == CHAR_BACKSLASH) { int temperrorcode = 0; - next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE); + escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options, + FALSE); if (temperrorcode != 0) return FALSE; ptr++; /* Point after the escape sequence */ } - -else if ((cd->ctypes[*ptr] & ctype_meta) == 0) +else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0) { -#ifdef SUPPORT_UTF8 - if (utf8) { GETCHARINC(next, ptr); } else + escape = 0; +#ifdef SUPPORT_UTF + if (utf) { GETCHARINC(next, ptr); } else #endif next = *ptr++; } - else return FALSE; /* Skip whitespace and comments in extended mode */ @@ -2888,16 +3112,16 @@ if ((options & PCRE_EXTENDED) != 0) { for (;;) { - while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++; + while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++; if (*ptr == CHAR_NUMBER_SIGN) { ptr++; - while (*ptr != 0) + while (*ptr != CHAR_NULL) { if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; } ptr++; -#ifdef SUPPORT_UTF8 - if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++; +#ifdef SUPPORT_UTF + if (utf) FORWARDCHAR(ptr); #endif } } @@ -2908,153 +3132,143 @@ if ((options & PCRE_EXTENDED) != 0) /* If the next thing is itself optional, we have to give up. */ if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK || - strncmp((char *)ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0) + STRNCMP_UC_C8(ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0) return FALSE; -/* Now compare the next item with the previous opcode. First, handle cases when -the next item is a character. */ +/* If the previous item is a character, get its value. */ -if (next >= 0) switch(op_code) +if (op_code == OP_CHAR || op_code == OP_CHARI || + op_code == OP_NOT || op_code == OP_NOTI) { - case OP_CHAR: -#ifdef SUPPORT_UTF8 +#ifdef SUPPORT_UTF GETCHARTEST(c, previous); #else c = *previous; #endif - return c != next; + } - /* For CHARI (caseless character) we must check the other case. If we have - Unicode property support, we can use it to test the other case of - high-valued characters. */ +/* Now compare the next item with the previous opcode. First, handle cases when +the next item is a character. */ - case OP_CHARI: -#ifdef SUPPORT_UTF8 - GETCHARTEST(c, previous); -#else - c = *previous; +if (escape == 0) + { + /* For a caseless UTF match, the next character may have more than one other + case, which maps to the special PT_CLIST property. Check this first. */ + +#ifdef SUPPORT_UCP + if (utf && c != NOTACHAR && (options & PCRE_CASELESS) != 0) + { + unsigned int ocs = UCD_CASESET(next); + if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT); + } #endif - if (c == next) return FALSE; -#ifdef SUPPORT_UTF8 - if (utf8) + + switch(op_code) { - unsigned int othercase; - if (next < 128) othercase = cd->fcc[next]; else + case OP_CHAR: + return c != next; + + /* For CHARI (caseless character) we must check the other case. If we have + Unicode property support, we can use it to test the other case of + high-valued characters. We know that next can have only one other case, + because multi-other-case characters are dealt with above. */ + + case OP_CHARI: + if (c == next) return FALSE; +#ifdef SUPPORT_UTF + if (utf) + { + pcre_uint32 othercase; + if (next < 128) othercase = cd->fcc[next]; else #ifdef SUPPORT_UCP - othercase = UCD_OTHERCASE((unsigned int)next); + othercase = UCD_OTHERCASE(next); #else - othercase = NOTACHAR; + othercase = NOTACHAR; #endif - return (unsigned int)c != othercase; - } - else -#endif /* SUPPORT_UTF8 */ - return (c != cd->fcc[next]); /* Non-UTF-8 mode */ + return c != othercase; + } + else +#endif /* SUPPORT_UTF */ + return (c != TABLE_GET(next, cd->fcc, next)); /* Not UTF */ - /* For OP_NOT and OP_NOTI, the data is always a single-byte character. These - opcodes are not used for multi-byte characters, because they are coded using - an XCLASS instead. */ + case OP_NOT: + return c == next; - case OP_NOT: - return (c = *previous) == next; - - case OP_NOTI: - if ((c = *previous) == next) return TRUE; -#ifdef SUPPORT_UTF8 - if (utf8) - { - unsigned int othercase; - if (next < 128) othercase = cd->fcc[next]; else + case OP_NOTI: + if (c == next) return TRUE; +#ifdef SUPPORT_UTF + if (utf) + { + pcre_uint32 othercase; + if (next < 128) othercase = cd->fcc[next]; else #ifdef SUPPORT_UCP - othercase = UCD_OTHERCASE(next); + othercase = UCD_OTHERCASE(next); #else - othercase = NOTACHAR; + othercase = NOTACHAR; #endif - return (unsigned int)c == othercase; - } - else -#endif /* SUPPORT_UTF8 */ - return (c == cd->fcc[next]); /* Non-UTF-8 mode */ + return c == othercase; + } + else +#endif /* SUPPORT_UTF */ + return (c == TABLE_GET(next, cd->fcc, next)); /* Not UTF */ - /* Note that OP_DIGIT etc. are generated only when PCRE_UCP is *not* set. - When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */ + /* Note that OP_DIGIT etc. are generated only when PCRE_UCP is *not* set. + When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */ - case OP_DIGIT: - return next > 127 || (cd->ctypes[next] & ctype_digit) == 0; + case OP_DIGIT: + return next > 255 || (cd->ctypes[next] & ctype_digit) == 0; - case OP_NOT_DIGIT: - return next <= 127 && (cd->ctypes[next] & ctype_digit) != 0; + case OP_NOT_DIGIT: + return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0; - case OP_WHITESPACE: - return next > 127 || (cd->ctypes[next] & ctype_space) == 0; + case OP_WHITESPACE: + return next > 255 || (cd->ctypes[next] & ctype_space) == 0; - case OP_NOT_WHITESPACE: - return next <= 127 && (cd->ctypes[next] & ctype_space) != 0; + case OP_NOT_WHITESPACE: + return next <= 255 && (cd->ctypes[next] & ctype_space) != 0; - case OP_WORDCHAR: - return next > 127 || (cd->ctypes[next] & ctype_word) == 0; + case OP_WORDCHAR: + return next > 255 || (cd->ctypes[next] & ctype_word) == 0; - case OP_NOT_WORDCHAR: - return next <= 127 && (cd->ctypes[next] & ctype_word) != 0; + case OP_NOT_WORDCHAR: + return next <= 255 && (cd->ctypes[next] & ctype_word) != 0; - case OP_HSPACE: - case OP_NOT_HSPACE: - switch(next) - { - case 0x09: - case 0x20: - case 0xa0: - case 0x1680: - case 0x180e: - case 0x2000: - case 0x2001: - case 0x2002: - case 0x2003: - case 0x2004: - case 0x2005: - case 0x2006: - case 0x2007: - case 0x2008: - case 0x2009: - case 0x200A: - case 0x202f: - case 0x205f: - case 0x3000: - return op_code == OP_NOT_HSPACE; - default: - return op_code != OP_NOT_HSPACE; - } + case OP_HSPACE: + case OP_NOT_HSPACE: + switch(next) + { + HSPACE_CASES: + return op_code == OP_NOT_HSPACE; - case OP_ANYNL: - case OP_VSPACE: - case OP_NOT_VSPACE: - switch(next) - { - case 0x0a: - case 0x0b: - case 0x0c: - case 0x0d: - case 0x85: - case 0x2028: - case 0x2029: - return op_code == OP_NOT_VSPACE; - default: - return op_code != OP_NOT_VSPACE; - } + default: + return op_code != OP_NOT_HSPACE; + } + case OP_ANYNL: + case OP_VSPACE: + case OP_NOT_VSPACE: + switch(next) + { + VSPACE_CASES: + return op_code == OP_NOT_VSPACE; + + default: + return op_code != OP_NOT_VSPACE; + } + #ifdef SUPPORT_UCP - case OP_PROP: - return check_char_prop(next, previous[0], previous[1], FALSE); + case OP_PROP: + return check_char_prop(next, previous[0], previous[1], FALSE); - case OP_NOTPROP: - return check_char_prop(next, previous[0], previous[1], TRUE); + case OP_NOTPROP: + return check_char_prop(next, previous[0], previous[1], TRUE); #endif - default: - return FALSE; + default: + return FALSE; + } } - /* Handle the case when the next item is \d, \s, etc. Note that when PCRE_UCP is set, \d turns into ESC_du rather than ESC_d, etc., so ESC_d etc. are generated only when PCRE_UCP is *not* set, that is, when only ASCII @@ -3065,78 +3279,51 @@ switch(op_code) { case OP_CHAR: case OP_CHARI: -#ifdef SUPPORT_UTF8 - GETCHARTEST(c, previous); -#else - c = *previous; -#endif - switch(-next) + switch(escape) { case ESC_d: - return c > 127 || (cd->ctypes[c] & ctype_digit) == 0; + return c > 255 || (cd->ctypes[c] & ctype_digit) == 0; case ESC_D: - return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0; + return c <= 255 && (cd->ctypes[c] & ctype_digit) != 0; case ESC_s: - return c > 127 || (cd->ctypes[c] & ctype_space) == 0; + return c > 255 || (cd->ctypes[c] & ctype_space) == 0; case ESC_S: - return c <= 127 && (cd->ctypes[c] & ctype_space) != 0; + return c <= 255 && (cd->ctypes[c] & ctype_space) != 0; case ESC_w: - return c > 127 || (cd->ctypes[c] & ctype_word) == 0; + return c > 255 || (cd->ctypes[c] & ctype_word) == 0; case ESC_W: - return c <= 127 && (cd->ctypes[c] & ctype_word) != 0; + return c <= 255 && (cd->ctypes[c] & ctype_word) != 0; case ESC_h: case ESC_H: switch(c) { - case 0x09: - case 0x20: - case 0xa0: - case 0x1680: - case 0x180e: - case 0x2000: - case 0x2001: - case 0x2002: - case 0x2003: - case 0x2004: - case 0x2005: - case 0x2006: - case 0x2007: - case 0x2008: - case 0x2009: - case 0x200A: - case 0x202f: - case 0x205f: - case 0x3000: - return -next != ESC_h; + HSPACE_CASES: + return escape != ESC_h; + default: - return -next == ESC_h; + return escape == ESC_h; } case ESC_v: case ESC_V: switch(c) { - case 0x0a: - case 0x0b: - case 0x0c: - case 0x0d: - case 0x85: - case 0x2028: - case 0x2029: - return -next != ESC_v; + VSPACE_CASES: + return escape != ESC_v; + default: - return -next == ESC_v; + return escape == ESC_v; } /* When PCRE_UCP is set, these values get generated for \d etc. Find their substitutions and process them. The result will always be either - -ESC_p or -ESC_P. Then fall through to process those values. */ + ESC_p or ESC_P. Then fall through to process those values. */ #ifdef SUPPORT_UCP case ESC_du: @@ -3147,8 +3334,8 @@ switch(op_code) case ESC_SU: { int temperrorcode = 0; - ptr = substitutes[-next - ESC_DU]; - next = check_escape(&ptr, &temperrorcode, 0, options, FALSE); + ptr = substitutes[escape - ESC_DU]; + escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE); if (temperrorcode != 0) return FALSE; ptr++; /* For compatibility */ } @@ -3157,12 +3344,13 @@ switch(op_code) case ESC_p: case ESC_P: { - int ptype, pdata, errorcodeptr; + unsigned int ptype = 0, pdata = 0; + int errorcodeptr; BOOL negated; ptr--; /* Make ptr point at the p or P */ - ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr); - if (ptype < 0) return FALSE; + if (!get_ucp(&ptr, &negated, &ptype, &pdata, &errorcodeptr)) + return FALSE; ptr++; /* Point past the final curly ket */ /* If the property item is optional, we have to give up. (When generated @@ -3170,12 +3358,12 @@ switch(op_code) to the original \d etc. At this point, ptr will point to a zero byte. */ if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK || - strncmp((char *)ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0) + STRNCMP_UC_C8(ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0) return FALSE; /* Do the property check. */ - return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated); + return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated); } #endif @@ -3190,39 +3378,39 @@ switch(op_code) these op-codes are never generated.) */ case OP_DIGIT: - return next == -ESC_D || next == -ESC_s || next == -ESC_W || - next == -ESC_h || next == -ESC_v || next == -ESC_R; + return escape == ESC_D || escape == ESC_s || escape == ESC_W || + escape == ESC_h || escape == ESC_v || escape == ESC_R; case OP_NOT_DIGIT: - return next == -ESC_d; + return escape == ESC_d; case OP_WHITESPACE: - return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R; + return escape == ESC_S || escape == ESC_d || escape == ESC_w; case OP_NOT_WHITESPACE: - return next == -ESC_s || next == -ESC_h || next == -ESC_v; + return escape == ESC_s || escape == ESC_h || escape == ESC_v || escape == ESC_R; case OP_HSPACE: - return next == -ESC_S || next == -ESC_H || next == -ESC_d || - next == -ESC_w || next == -ESC_v || next == -ESC_R; + return escape == ESC_S || escape == ESC_H || escape == ESC_d || + escape == ESC_w || escape == ESC_v || escape == ESC_R; case OP_NOT_HSPACE: - return next == -ESC_h; + return escape == ESC_h; /* Can't have \S in here because VT matches \S (Perl anomaly) */ case OP_ANYNL: case OP_VSPACE: - return next == -ESC_V || next == -ESC_d || next == -ESC_w; + return escape == ESC_V || escape == ESC_d || escape == ESC_w; case OP_NOT_VSPACE: - return next == -ESC_v || next == -ESC_R; + return escape == ESC_v || escape == ESC_R; case OP_WORDCHAR: - return next == -ESC_W || next == -ESC_s || next == -ESC_h || - next == -ESC_v || next == -ESC_R; + return escape == ESC_W || escape == ESC_s || escape == ESC_h || + escape == ESC_v || escape == ESC_R; case OP_NOT_WORDCHAR: - return next == -ESC_w || next == -ESC_d; + return escape == ESC_w || escape == ESC_d; default: return FALSE; @@ -3234,6 +3422,244 @@ switch(op_code) /************************************************* +* Add a character or range to a class * +*************************************************/ + +/* This function packages up the logic of adding a character or range of +characters to a class. The character values in the arguments will be within the +valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is +mutually recursive with the function immediately below. + +Arguments: + classbits the bit map for characters < 256 + uchardptr points to the pointer for extra data + options the options word + cd contains pointers to tables etc. + start start of range character + end end of range character + +Returns: the number of < 256 characters added + the pointer to extra data is updated +*/ + +static int +add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options, + compile_data *cd, pcre_uint32 start, pcre_uint32 end) +{ +pcre_uint32 c; +int n8 = 0; + +/* If caseless matching is required, scan the range and process alternate +cases. In Unicode, there are 8-bit characters that have alternate cases that +are greater than 255 and vice-versa. Sometimes we can just extend the original +range. */ + +if ((options & PCRE_CASELESS) != 0) + { +#ifdef SUPPORT_UCP + if ((options & PCRE_UTF8) != 0) + { + int rc; + pcre_uint32 oc, od; + + options &= ~PCRE_CASELESS; /* Remove for recursive calls */ + c = start; + + while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0) + { + /* Handle a single character that has more than one other case. */ + + if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd, + PRIV(ucd_caseless_sets) + rc, oc); + + /* Do nothing if the other case range is within the original range. */ + + else if (oc >= start && od <= end) continue; + + /* Extend the original range if there is overlap, noting that if oc < c, we + can't have od > end because a subrange is always shorter than the basic + range. Otherwise, use a recursive call to add the additional range. */ + + else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */ + else if (od > end && oc <= end + 1) end = od; /* Extend upwards */ + else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od); + } + } + else +#endif /* SUPPORT_UCP */ + + /* Not UTF-mode, or no UCP */ + + for (c = start; c <= end && c < 256; c++) + { + SETBIT(classbits, cd->fcc[c]); + n8++; + } + } + +/* Now handle the original range. Adjust the final value according to the bit +length - this means that the same lists of (e.g.) horizontal spaces can be used +in all cases. */ + +#if defined COMPILE_PCRE8 +#ifdef SUPPORT_UTF + if ((options & PCRE_UTF8) == 0) +#endif + if (end > 0xff) end = 0xff; + +#elif defined COMPILE_PCRE16 +#ifdef SUPPORT_UTF + if ((options & PCRE_UTF16) == 0) +#endif + if (end > 0xffff) end = 0xffff; + +#endif /* COMPILE_PCRE[8|16] */ + +/* If all characters are less than 256, use the bit map. Otherwise use extra +data. */ + +if (end < 0x100) + { + for (c = start; c <= end; c++) + { + n8++; + SETBIT(classbits, c); + } + } + +else + { + pcre_uchar *uchardata = *uchardptr; + +#ifdef SUPPORT_UTF + if ((options & PCRE_UTF8) != 0) /* All UTFs use the same flag bit */ + { + if (start < end) + { + *uchardata++ = XCL_RANGE; + uchardata += PRIV(ord2utf)(start, uchardata); + uchardata += PRIV(ord2utf)(end, uchardata); + } + else if (start == end) + { + *uchardata++ = XCL_SINGLE; + uchardata += PRIV(ord2utf)(start, uchardata); + } + } + else +#endif /* SUPPORT_UTF */ + + /* Without UTF support, character values are constrained by the bit length, + and can only be > 256 for 16-bit and 32-bit libraries. */ + +#ifdef COMPILE_PCRE8 + {} +#else + if (start < end) + { + *uchardata++ = XCL_RANGE; + *uchardata++ = start; + *uchardata++ = end; + } + else if (start == end) + { + *uchardata++ = XCL_SINGLE; + *uchardata++ = start; + } +#endif + + *uchardptr = uchardata; /* Updata extra data pointer */ + } + +return n8; /* Number of 8-bit characters */ +} + + + + +/************************************************* +* Add a list of characters to a class * +*************************************************/ + +/* This function is used for adding a list of case-equivalent characters to a +class, and also for adding a list of horizontal or vertical whitespace. If the +list is in order (which it should be), ranges of characters are detected and +handled appropriately. This function is mutually recursive with the function +above. + +Arguments: + classbits the bit map for characters < 256 + uchardptr points to the pointer for extra data + options the options word + cd contains pointers to tables etc. + p points to row of 32-bit values, terminated by NOTACHAR + except character to omit; this is used when adding lists of + case-equivalent characters to avoid including the one we + already know about + +Returns: the number of < 256 characters added + the pointer to extra data is updated +*/ + +static int +add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options, + compile_data *cd, const pcre_uint32 *p, unsigned int except) +{ +int n8 = 0; +while (p[0] < NOTACHAR) + { + int n = 0; + if (p[0] != except) + { + while(p[n+1] == p[0] + n + 1) n++; + n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]); + } + p += n + 1; + } +return n8; +} + + + +/************************************************* +* Add characters not in a list to a class * +*************************************************/ + +/* This function is used for adding the complement of a list of horizontal or +vertical whitespace to a class. The list must be in order. + +Arguments: + classbits the bit map for characters < 256 + uchardptr points to the pointer for extra data + options the options word + cd contains pointers to tables etc. + p points to row of 32-bit values, terminated by NOTACHAR + +Returns: the number of < 256 characters added + the pointer to extra data is updated +*/ + +static int +add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, + int options, compile_data *cd, const pcre_uint32 *p) +{ +BOOL utf = (options & PCRE_UTF8) != 0; +int n8 = 0; +if (p[0] > 0) + n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1); +while (p[0] < NOTACHAR) + { + while (p[1] == p[0] + 1) p++; + n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1, + (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1); + p++; + } +return n8; +} + + + +/************************************************* * Compile one branch * *************************************************/ @@ -3248,8 +3674,10 @@ Arguments: codeptr points to the pointer to the current code point ptrptr points to the current pattern pointer errorcodeptr points to error code variable - firstbyteptr set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE) - reqbyteptr set to the last literal character required, else < 0 + firstcharptr place to put the first required character + firstcharflagsptr place to put the first character flags, or a negative number + reqcharptr place to put the last required character + reqcharflagsptr place to put the last required character flags, or a negative number bcptr points to current branch chain cond_depth conditional nesting depth cd contains pointers to tables etc. @@ -3261,49 +3689,66 @@ Returns: TRUE on success */ static BOOL -compile_branch(int *optionsptr, uschar **codeptr, const uschar **ptrptr, - int *errorcodeptr, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, - int cond_depth, compile_data *cd, int *lengthptr) +compile_branch(int *optionsptr, pcre_uchar **codeptr, + const pcre_uchar **ptrptr, int *errorcodeptr, + pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr, + pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr, + branch_chain *bcptr, int cond_depth, + compile_data *cd, int *lengthptr) { int repeat_type, op_type; int repeat_min = 0, repeat_max = 0; /* To please picky compilers */ int bravalue = 0; int greedy_default, greedy_non_default; -int firstbyte, reqbyte; -int zeroreqbyte, zerofirstbyte; -int req_caseopt, reqvary, tempreqvary; +pcre_uint32 firstchar, reqchar; +pcre_int32 firstcharflags, reqcharflags; +pcre_uint32 zeroreqchar, zerofirstchar; +pcre_int32 zeroreqcharflags, zerofirstcharflags; +pcre_int32 req_caseopt, reqvary, tempreqvary; int options = *optionsptr; /* May change dynamically */ int after_manual_callout = 0; int length_prevgroup = 0; -register int c; -register uschar *code = *codeptr; -uschar *last_code = code; -uschar *orig_code = code; -uschar *tempcode; +register pcre_uint32 c; +int escape; +register pcre_uchar *code = *codeptr; +pcre_uchar *last_code = code; +pcre_uchar *orig_code = code; +pcre_uchar *tempcode; BOOL inescq = FALSE; -BOOL groupsetfirstbyte = FALSE; -const uschar *ptr = *ptrptr; -const uschar *tempptr; -const uschar *nestptr = NULL; -uschar *previous = NULL; -uschar *previous_callout = NULL; -uschar *save_hwm = NULL; -uschar classbits[32]; +BOOL groupsetfirstchar = FALSE; +const pcre_uchar *ptr = *ptrptr; +const pcre_uchar *tempptr; +const pcre_uchar *nestptr = NULL; +pcre_uchar *previous = NULL; +pcre_uchar *previous_callout = NULL; +pcre_uchar *save_hwm = NULL; +pcre_uint8 classbits[32]; /* We can fish out the UTF-8 setting once and for all into a BOOL, but we must not do this for other options (e.g. PCRE_EXTENDED) because they may change dynamically as we process the pattern. */ -#ifdef SUPPORT_UTF8 -BOOL class_utf8; -BOOL utf8 = (options & PCRE_UTF8) != 0; -uschar *class_utf8data; -uschar *class_utf8data_base; -uschar utf8_char[6]; +#ifdef SUPPORT_UTF +/* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */ +BOOL utf = (options & PCRE_UTF8) != 0; +#ifndef COMPILE_PCRE32 +pcre_uchar utf_chars[6]; +#endif #else -BOOL utf8 = FALSE; +BOOL utf = FALSE; #endif +/* Helper variables for OP_XCLASS opcode (for characters > 255). We define +class_uchardata always so that it can be passed to add_to_class() always, +though it will not be used in non-UTF 8-bit cases. This avoids having to supply +alternative calls for the different cases. */ + +pcre_uchar *class_uchardata; +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 +BOOL xclass; +pcre_uchar *class_uchardata_base; +#endif + #ifdef PCRE_DEBUG if (lengthptr != NULL) DPRINTF((">> start branch\n")); #endif @@ -3315,22 +3760,24 @@ greedy_non_default = greedy_default ^ 1; /* Initialize no first byte, no required byte. REQ_UNSET means "no char matching encountered yet". It gets changed to REQ_NONE if we hit something that -matches a non-fixed char first char; reqbyte just remains unset if we never +matches a non-fixed char first char; reqchar just remains unset if we never find one. When we hit a repeat whose minimum is zero, we may have to adjust these values to take the zero repeat into account. This is implemented by setting them to -zerofirstbyte and zeroreqbyte when such a repeat is encountered. The individual +zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual item types that can be repeated set these backoff variables appropriately. */ -firstbyte = reqbyte = zerofirstbyte = zeroreqbyte = REQ_UNSET; +firstchar = reqchar = zerofirstchar = zeroreqchar = 0; +firstcharflags = reqcharflags = zerofirstcharflags = zeroreqcharflags = REQ_UNSET; -/* The variable req_caseopt contains either the REQ_CASELESS value or zero, -according to the current setting of the caseless flag. REQ_CASELESS is a bit -value > 255. It is added into the firstbyte or reqbyte variables to record the -case status of the value. This is used only for ASCII characters. */ +/* The variable req_caseopt contains either the REQ_CASELESS value +or zero, according to the current setting of the caseless flag. The +REQ_CASELESS leaves the lower 28 bit empty. It is added into the +firstchar or reqchar variables to record the case status of the +value. This is used only for ASCII characters. */ -req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS : 0; +req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS:0; /* Switch on next character until the end of the branch */ @@ -3342,27 +3789,28 @@ for (;; ptr++) BOOL is_quantifier; BOOL is_recurse; BOOL reset_bracount; - int class_charcount; - int class_lastchar; + int class_has_8bitchar; + int class_one_char; int newoptions; int recno; int refsign; int skipbytes; - int subreqbyte; - int subfirstbyte; + pcre_uint32 subreqchar, subfirstchar; + pcre_int32 subreqcharflags, subfirstcharflags; int terminator; - int mclength; - int tempbracount; - uschar mcbuffer[8]; + unsigned int mclength; + unsigned int tempbracount; + pcre_uint32 ec; + pcre_uchar mcbuffer[8]; - /* Get next byte in the pattern */ + /* Get next character in the pattern */ c = *ptr; /* If we are at the end of a nested substitution, revert to the outer level string. Nesting only happens one level deep. */ - if (c == 0 && nestptr != NULL) + if (c == CHAR_NULL && nestptr != NULL) { ptr = nestptr; nestptr = NULL; @@ -3401,8 +3849,8 @@ for (;; ptr++) } *lengthptr += (int)(code - last_code); - DPRINTF(("length=%d added %d c=%c\n", *lengthptr, (int)(code - last_code), - c)); + DPRINTF(("length=%d added %d c=%c (0x%x)\n", *lengthptr, + (int)(code - last_code), c, c)); /* If "previous" is set and it is not at the start of the work space, move it back to there, in order to avoid filling up the work space. Otherwise, @@ -3412,7 +3860,7 @@ for (;; ptr++) { if (previous > orig_code) { - memmove(orig_code, previous, code - previous); + memmove(orig_code, previous, IN_UCHARS(code - previous)); code -= previous - orig_code; previous = orig_code; } @@ -3437,7 +3885,7 @@ for (;; ptr++) /* If in \Q...\E, check for the end; if not, we have a literal */ - if (inescq && c != 0) + if (inescq && c != CHAR_NULL) { if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E) { @@ -3481,19 +3929,19 @@ for (;; ptr++) if ((options & PCRE_EXTENDED) != 0) { - if ((cd->ctypes[c] & ctype_space) != 0) continue; + if (MAX_255(*ptr) && (cd->ctypes[c] & ctype_space) != 0) continue; if (c == CHAR_NUMBER_SIGN) { ptr++; - while (*ptr != 0) + while (*ptr != CHAR_NULL) { if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; } ptr++; -#ifdef SUPPORT_UTF8 - if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++; +#ifdef SUPPORT_UTF + if (utf) FORWARDCHAR(ptr); #endif } - if (*ptr != 0) continue; + if (*ptr != CHAR_NULL) continue; /* Else fall through to handle end of string */ c = 0; @@ -3514,8 +3962,10 @@ for (;; ptr++) case 0: /* The branch terminates at string end */ case CHAR_VERTICAL_LINE: /* or | or ) */ case CHAR_RIGHT_PARENTHESIS: - *firstbyteptr = firstbyte; - *reqbyteptr = reqbyte; + *firstcharptr = firstchar; + *firstcharflagsptr = firstcharflags; + *reqcharptr = reqchar; + *reqcharflagsptr = reqcharflags; *codeptr = code; *ptrptr = ptr; if (lengthptr != NULL) @@ -3539,7 +3989,7 @@ for (;; ptr++) previous = NULL; if ((options & PCRE_MULTILINE) != 0) { - if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE; + if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; *code++ = OP_CIRCM; } else *code++ = OP_CIRC; @@ -3551,12 +4001,14 @@ for (;; ptr++) break; /* There can never be a first char if '.' is first, whatever happens about - repeats. The value of reqbyte doesn't change either. */ + repeats. The value of reqchar doesn't change either. */ case CHAR_DOT: - if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE; - zerofirstbyte = firstbyte; - zeroreqbyte = reqbyte; + if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; + zerofirstchar = firstchar; + zerofirstcharflags = firstcharflags; + zeroreqchar = reqchar; + zeroreqcharflags = reqcharflags; previous = code; *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY; break; @@ -3611,8 +4063,7 @@ for (;; ptr++) { if (ptr[1] == CHAR_E) ptr++; - else if (strncmp((const char *)ptr+1, - STR_Q STR_BACKSLASH STR_E, 3) == 0) + else if (STRNCMP_UC_C8(ptr + 1, STR_Q STR_BACKSLASH STR_E, 3) == 0) ptr += 3; else break; @@ -3631,8 +4082,9 @@ for (;; ptr++) (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0) { *code++ = negate_class? OP_ALLANY : OP_FAIL; - if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE; - zerofirstbyte = firstbyte; + if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; + zerofirstchar = firstchar; + zerofirstcharflags = firstcharflags; break; } @@ -3642,51 +4094,55 @@ for (;; ptr++) should_flip_negation = FALSE; - /* Keep a count of chars with values < 256 so that we can optimize the case - of just a single character (as long as it's < 256). However, For higher - valued UTF-8 characters, we don't yet do any optimization. */ + /* For optimization purposes, we track some properties of the class: + class_has_8bitchar will be non-zero if the class contains at least one < + 256 character; class_one_char will be 1 if the class contains just one + character. */ - class_charcount = 0; - class_lastchar = -1; + class_has_8bitchar = 0; + class_one_char = 0; /* Initialize the 32-char bit map to all zeros. We build the map in a - temporary bit of memory, in case the class contains only 1 character (less - than 256), because in that case the compiled code doesn't use the bit map. - */ + temporary bit of memory, in case the class contains fewer than two + 8-bit characters because in that case the compiled code doesn't use the bit + map. */ - memset(classbits, 0, 32 * sizeof(uschar)); + memset(classbits, 0, 32 * sizeof(pcre_uint8)); -#ifdef SUPPORT_UTF8 - class_utf8 = FALSE; /* No chars >= 256 */ - class_utf8data = code + LINK_SIZE + 2; /* For UTF-8 items */ - class_utf8data_base = class_utf8data; /* For resetting in pass 1 */ +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + xclass = FALSE; + class_uchardata = code + LINK_SIZE + 2; /* For XCLASS items */ + class_uchardata_base = class_uchardata; /* Save the start */ #endif /* Process characters until ] is reached. By writing this as a "do" it means that an initial ] is taken as a data character. At the start of the loop, c contains the first byte of the character. */ - if (c != 0) do + if (c != CHAR_NULL) do { - const uschar *oldptr; + const pcre_uchar *oldptr; -#ifdef SUPPORT_UTF8 - if (utf8 && c > 127) +#ifdef SUPPORT_UTF + if (utf && HAS_EXTRALEN(c)) { /* Braces are required because the */ GETCHARLEN(c, ptr, ptr); /* macro generates multiple statements */ } +#endif - /* In the pre-compile phase, accumulate the length of any UTF-8 extra +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + /* In the pre-compile phase, accumulate the length of any extra data and reset the pointer. This is so that very large classes that - contain a zillion UTF-8 characters no longer overwrite the work space - (which is on the stack). */ + contain a zillion > 255 characters no longer overwrite the work space + (which is on the stack). We have to remember that there was XCLASS data, + however. */ - if (lengthptr != NULL) + if (lengthptr != NULL && class_uchardata > class_uchardata_base) { - *lengthptr += (int)(class_utf8data - class_utf8data_base); - class_utf8data = class_utf8data_base; + xclass = TRUE; + *lengthptr += class_uchardata - class_uchardata_base; + class_uchardata = class_uchardata_base; } - #endif /* Inside \Q...\E everything is literal except \E */ @@ -3714,8 +4170,8 @@ for (;; ptr++) { BOOL local_negate = FALSE; int posix_class, taboffset, tabopt; - register const uschar *cbits = cd->cbits; - uschar pbits[32]; + register const pcre_uint8 *cbits = cd->cbits; + pcre_uint8 pbits[32]; if (ptr[1] != CHAR_COLON) { @@ -3770,7 +4226,7 @@ for (;; ptr++) /* Copy in the first table (always present) */ memcpy(pbits, cbits + posix_class_maps[posix_class], - 32 * sizeof(uschar)); + 32 * sizeof(pcre_uint8)); /* If there is a second table, add or remove it as required. */ @@ -3785,7 +4241,7 @@ for (;; ptr++) for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset]; } - /* Not see if we need to remove any special characters. An option + /* Now see if we need to remove any special characters. An option value of 1 removes vertical space and 2 removes underscore. */ if (tabopt < 0) tabopt = -tabopt; @@ -3801,30 +4257,35 @@ for (;; ptr++) for (c = 0; c < 32; c++) classbits[c] |= pbits[c]; ptr = tempptr + 1; - class_charcount = 10; /* Set > 1; assumes more than 1 per class */ + /* Every class contains at least one < 256 character. */ + class_has_8bitchar = 1; + /* Every class contains at least two characters. */ + class_one_char = 2; continue; /* End of POSIX syntax handling */ } /* Backslash may introduce a single character, or it may introduce one of the specials, which just set a flag. The sequence \b is a special case. Inside a class (and only there) it is treated as backspace. We - assume that other escapes have more than one character in them, so set - class_charcount bigger than one. Unrecognized escapes fall through and - are either treated as literal characters (by default), or are faulted if + assume that other escapes have more than one character in them, so + speculatively set both class_has_8bitchar and class_one_char bigger + than one. Unrecognized escapes fall through and are either treated + as literal characters (by default), or are faulted if PCRE_EXTRA is set. */ if (c == CHAR_BACKSLASH) { - c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE); + escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, + TRUE); if (*errorcodeptr != 0) goto FAILED; - - if (-c == ESC_b) c = CHAR_BS; /* \b is backspace in a class */ - else if (-c == ESC_N) /* \N is not supported in a class */ + if (escape == 0) c = ec; + else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */ + else if (escape == ESC_N) /* \N is not supported in a class */ { *errorcodeptr = ERR71; goto FAILED; } - else if (-c == ESC_Q) /* Handle start of quoted string */ + else if (escape == ESC_Q) /* Handle start of quoted string */ { if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E) { @@ -3833,14 +4294,17 @@ for (;; ptr++) else inescq = TRUE; continue; } - else if (-c == ESC_E) continue; /* Ignore orphan \E */ + else if (escape == ESC_E) continue; /* Ignore orphan \E */ - if (c < 0) + else { - register const uschar *cbits = cd->cbits; - class_charcount += 2; /* Greater than 1 is what matters */ + register const pcre_uint8 *cbits = cd->cbits; + /* Every class contains at least two < 256 characters. */ + class_has_8bitchar++; + /* Every class contains at least two characters. */ + class_one_char += 2; - switch (-c) + switch (escape) { #ifdef SUPPORT_UCP case ESC_du: /* These are the values given for \d etc */ @@ -3850,8 +4314,8 @@ for (;; ptr++) case ESC_su: /* of the default ASCII testing. */ case ESC_SU: nestptr = ptr; - ptr = substitutes[-c - ESC_DU] - 1; /* Just before substitute */ - class_charcount -= 2; /* Undo! */ + ptr = substitutes[escape - ESC_DU] - 1; /* Just before substitute */ + class_has_8bitchar--; /* Undo! */ continue; #endif case ESC_d: @@ -3874,7 +4338,8 @@ for (;; ptr++) /* Perl 5.004 onwards omits VT from \s, but we must preserve it if it was previously set by something earlier in the character - class. */ + class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and + EBCDIC, so we lazily just adjust the appropriate bit. */ case ESC_s: classbits[0] |= cbits[cbit_space]; @@ -3888,120 +4353,26 @@ for (;; ptr++) classbits[1] |= 0x08; /* Perl 5.004 onwards omits VT from \s */ continue; + /* The rest apply in both UCP and non-UCP cases. */ + case ESC_h: - SETBIT(classbits, 0x09); /* VT */ - SETBIT(classbits, 0x20); /* SPACE */ - SETBIT(classbits, 0xa0); /* NSBP */ -#ifdef SUPPORT_UTF8 - if (utf8) - { - class_utf8 = TRUE; - *class_utf8data++ = XCL_SINGLE; - class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data); - *class_utf8data++ = XCL_SINGLE; - class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data); - *class_utf8data++ = XCL_RANGE; - class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data); - class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data); - *class_utf8data++ = XCL_SINGLE; - class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data); - *class_utf8data++ = XCL_SINGLE; - class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data); - *class_utf8data++ = XCL_SINGLE; - class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data); - } -#endif + (void)add_list_to_class(classbits, &class_uchardata, options, cd, + PRIV(hspace_list), NOTACHAR); continue; case ESC_H: - for (c = 0; c < 32; c++) - { - int x = 0xff; - switch (c) - { - case 0x09/8: x ^= 1 << (0x09%8); break; - case 0x20/8: x ^= 1 << (0x20%8); break; - case 0xa0/8: x ^= 1 << (0xa0%8); break; - default: break; - } - classbits[c] |= x; - } - -#ifdef SUPPORT_UTF8 - if (utf8) - { - class_utf8 = TRUE; - *class_utf8data++ = XCL_RANGE; - class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data); - class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data); - *class_utf8data++ = XCL_RANGE; - class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data); - class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data); - *class_utf8data++ = XCL_RANGE; - class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data); - class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data); - *class_utf8data++ = XCL_RANGE; - class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data); - class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data); - *class_utf8data++ = XCL_RANGE; - class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data); - class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data); - *class_utf8data++ = XCL_RANGE; - class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data); - class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data); - *class_utf8data++ = XCL_RANGE; - class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data); - class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data); - } -#endif + (void)add_not_list_to_class(classbits, &class_uchardata, options, + cd, PRIV(hspace_list)); continue; case ESC_v: - SETBIT(classbits, 0x0a); /* LF */ - SETBIT(classbits, 0x0b); /* VT */ - SETBIT(classbits, 0x0c); /* FF */ - SETBIT(classbits, 0x0d); /* CR */ - SETBIT(classbits, 0x85); /* NEL */ -#ifdef SUPPORT_UTF8 - if (utf8) - { - class_utf8 = TRUE; - *class_utf8data++ = XCL_RANGE; - class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data); - class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data); - } -#endif + (void)add_list_to_class(classbits, &class_uchardata, options, cd, + PRIV(vspace_list), NOTACHAR); continue; case ESC_V: - for (c = 0; c < 32; c++) - { - int x = 0xff; - switch (c) - { - case 0x0a/8: x ^= 1 << (0x0a%8); - x ^= 1 << (0x0b%8); - x ^= 1 << (0x0c%8); - x ^= 1 << (0x0d%8); - break; - case 0x85/8: x ^= 1 << (0x85%8); break; - default: break; - } - classbits[c] |= x; - } - -#ifdef SUPPORT_UTF8 - if (utf8) - { - class_utf8 = TRUE; - *class_utf8data++ = XCL_RANGE; - class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data); - class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data); - *class_utf8data++ = XCL_RANGE; - class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data); - class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data); - } -#endif + (void)add_not_list_to_class(classbits, &class_uchardata, options, + cd, PRIV(vspace_list)); continue; #ifdef SUPPORT_UCP @@ -4009,15 +4380,14 @@ for (;; ptr++) case ESC_P: { BOOL negated; - int pdata; - int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr); - if (ptype < 0) goto FAILED; - class_utf8 = TRUE; - *class_utf8data++ = ((-c == ESC_p) != negated)? + unsigned int ptype = 0, pdata = 0; + if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr)) + goto FAILED; + *class_uchardata++ = ((escape == ESC_p) != negated)? XCL_PROP : XCL_NOTPROP; - *class_utf8data++ = ptype; - *class_utf8data++ = pdata; - class_charcount -= 2; /* Not a < 256 character */ + *class_uchardata++ = ptype; + *class_uchardata++ = pdata; + class_has_8bitchar--; /* Undo! */ continue; } #endif @@ -4031,21 +4401,24 @@ for (;; ptr++) *errorcodeptr = ERR7; goto FAILED; } - class_charcount -= 2; /* Undo the default count from above */ - c = *ptr; /* Get the final character and fall through */ + class_has_8bitchar--; /* Undo the speculative increase. */ + class_one_char -= 2; /* Undo the speculative increase. */ + c = *ptr; /* Get the final character and fall through */ break; } } - /* Fall through if we have a single character (c >= 0). This may be - greater than 256 in UTF-8 mode. */ + /* Fall through if the escape just defined a single character (c >= 0). + This may be greater than 256. */ + escape = 0; + } /* End of backslash handling */ - /* A single character may be followed by '-' to form a range. However, - Perl does not permit ']' to be the end of the range. A '-' character - at the end is treated as a literal. Perl ignores orphaned \E sequences - entirely. The code for handling \Q and \E is messy. */ + /* A character may be followed by '-' to form a range. However, Perl does + not permit ']' to be the end of the range. A '-' character at the end is + treated as a literal. Perl ignores orphaned \E sequences entirely. The + code for handling \Q and \E is messy. */ CHECK_RANGE: while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E) @@ -4053,10 +4426,9 @@ for (;; ptr++) inescq = FALSE; ptr += 2; } - oldptr = ptr; - /* Remember \r or \n */ + /* Remember if \r or \n were explicitly used */ if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF; @@ -4064,7 +4436,7 @@ for (;; ptr++) if (!inescq && ptr[1] == CHAR_MINUS) { - int d; + pcre_uint32 d; ptr += 2; while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2; @@ -4080,14 +4452,19 @@ for (;; ptr++) break; } - if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET)) + /* Minus (hyphen) at the end of a class is treated as a literal, so put + back the pointer and jump to handle the character that preceded it. */ + + if (*ptr == CHAR_NULL || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET)) { ptr = oldptr; - goto LONE_SINGLE_CHARACTER; + goto CLASS_SINGLE_CHARACTER; } -#ifdef SUPPORT_UTF8 - if (utf8) + /* Otherwise, we have a potential range; pick up the next character */ + +#ifdef SUPPORT_UTF + if (utf) { /* Braces are required because the */ GETCHARLEN(d, ptr, ptr); /* macro generates multiple statements */ } @@ -4101,257 +4478,173 @@ for (;; ptr++) if (!inescq && d == CHAR_BACKSLASH) { - d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE); + int descape; + descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE); if (*errorcodeptr != 0) goto FAILED; - /* \b is backspace; any other special means the '-' was literal */ + /* \b is backspace; any other special means the '-' was literal. */ - if (d < 0) + if (descape != 0) { - if (d == -ESC_b) d = CHAR_BS; else + if (descape == ESC_b) d = CHAR_BS; else { ptr = oldptr; - goto LONE_SINGLE_CHARACTER; /* A few lines below */ + goto CLASS_SINGLE_CHARACTER; /* A few lines below */ } } } /* Check that the two values are in the correct order. Optimize - one-character ranges */ + one-character ranges. */ if (d < c) { *errorcodeptr = ERR8; goto FAILED; } + if (d == c) goto CLASS_SINGLE_CHARACTER; /* A few lines below */ - if (d == c) goto LONE_SINGLE_CHARACTER; /* A few lines below */ + /* We have found a character range, so single character optimizations + cannot be done anymore. Any value greater than 1 indicates that there + is more than one character. */ - /* Remember \r or \n */ + class_one_char = 2; + /* Remember an explicit \r or \n, and add the range to the class. */ + if (d == CHAR_CR || d == CHAR_NL) 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 - available. */ + class_has_8bitchar += + add_to_class(classbits, &class_uchardata, options, cd, c, d); -#ifdef SUPPORT_UTF8 - if (utf8 && (d > 255 || ((options & PCRE_CASELESS) != 0 && d > 127))) - { - class_utf8 = TRUE; + continue; /* Go get the next char in the class */ + } - /* With UCP support, we can find the other case equivalents of - the relevant characters. There may be several ranges. Optimize how - they fit with the basic range. */ + /* Handle a single character - we can get here for a normal non-escape + char, or after \ that introduces a single character or for an apparent + range that isn't. Only the value 1 matters for class_one_char, so don't + increase it if it is already 2 or more ... just in case there's a class + with a zillion characters in it. */ -#ifdef SUPPORT_UCP - if ((options & PCRE_CASELESS) != 0) - { - unsigned int occ, ocd; - unsigned int cc = c; - unsigned int origd = d; - while (get_othercase_range(&cc, origd, &occ, &ocd)) - { - if (occ >= (unsigned int)c && - ocd <= (unsigned int)d) - continue; /* Skip embedded ranges */ + CLASS_SINGLE_CHARACTER: + if (class_one_char < 2) class_one_char++; - if (occ < (unsigned int)c && - ocd >= (unsigned int)c - 1) /* Extend the basic range */ - { /* if there is overlap, */ - c = occ; /* noting that if occ < c */ - continue; /* we can't have ocd > d */ - } /* because a subrange is */ - if (ocd > (unsigned int)d && - occ <= (unsigned int)d + 1) /* always shorter than */ - { /* the basic range. */ - d = ocd; - continue; - } + /* If class_one_char is 1, we have the first single character in the + class, and there have been no prior ranges, or XCLASS items generated by + escapes. If this is the final character in the class, we can optimize by + turning the item into a 1-character OP_CHAR[I] if it's positive, or + OP_NOT[I] if it's negative. In the positive case, it can cause firstchar + to be set. Otherwise, there can be no first char if this item is first, + whatever repeat count may follow. In the case of reqchar, save the + previous value for reinstating. */ - if (occ == ocd) - { - *class_utf8data++ = XCL_SINGLE; - } - else - { - *class_utf8data++ = XCL_RANGE; - class_utf8data += _pcre_ord2utf8(occ, class_utf8data); - } - class_utf8data += _pcre_ord2utf8(ocd, class_utf8data); - } - } -#endif /* SUPPORT_UCP */ + if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) + { + ptr++; + zeroreqchar = reqchar; + zeroreqcharflags = reqcharflags; - /* Now record the original range, possibly modified for UCP caseless - overlapping ranges. */ + if (negate_class) + { +#ifdef SUPPORT_UCP + int d; +#endif + if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; + zerofirstchar = firstchar; + zerofirstcharflags = firstcharflags; - *class_utf8data++ = XCL_RANGE; - class_utf8data += _pcre_ord2utf8(c, class_utf8data); - class_utf8data += _pcre_ord2utf8(d, class_utf8data); + /* For caseless UTF-8 mode when UCP support is available, check + whether this character has more than one other case. If so, generate + a special OP_NOTPROP item instead of OP_NOTI. */ - /* With UCP support, we are done. Without UCP support, there is no - caseless matching for UTF-8 characters > 127; we can use the bit map - for the smaller ones. */ - #ifdef SUPPORT_UCP - continue; /* With next character in the class */ -#else - if ((options & PCRE_CASELESS) == 0 || c > 127) continue; + if (utf && (options & PCRE_CASELESS) != 0 && + (d = UCD_CASESET(c)) != 0) + { + *code++ = OP_NOTPROP; + *code++ = PT_CLIST; + *code++ = d; + } + else +#endif + /* Char has only one other case, or UCP not available */ - /* Adjust upper limit and fall through to set up the map */ - - d = 127; - -#endif /* SUPPORT_UCP */ - } -#endif /* SUPPORT_UTF8 */ - - /* We use the bit map for all cases when not in UTF-8 mode; else - ranges that lie entirely within 0-127 when there is UCP support; else - for partial ranges without UCP support. */ - - class_charcount += d - c + 1; - class_lastchar = d; - - /* We can save a bit of time by skipping this in the pre-compile. */ - - if (lengthptr == NULL) for (; c <= d; c++) - { - classbits[c/8] |= (1 << (c&7)); - if ((options & PCRE_CASELESS) != 0) { - int uc = cd->fcc[c]; /* flip case */ - classbits[uc/8] |= (1 << (uc&7)); + *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT; +#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 + if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR) + code += PRIV(ord2utf)(c, code); + else +#endif + *code++ = c; } - } - continue; /* Go get the next char in the class */ - } + /* We are finished with this character class */ - /* Handle a lone single character - we can get here for a normal - non-escape char, or after \ that introduces a single character or for an - apparent range that isn't. */ + goto END_CLASS; + } - LONE_SINGLE_CHARACTER: + /* For a single, positive character, get the value into mcbuffer, and + then we can handle this with the normal one-character code. */ - /* Handle a character that cannot go in the bit map */ - -#ifdef SUPPORT_UTF8 - if (utf8 && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127))) - { - class_utf8 = TRUE; - *class_utf8data++ = XCL_SINGLE; - class_utf8data += _pcre_ord2utf8(c, class_utf8data); - -#ifdef SUPPORT_UCP - if ((options & PCRE_CASELESS) != 0) +#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 + if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR) + mclength = PRIV(ord2utf)(c, mcbuffer); + else +#endif { - unsigned int othercase; - if ((othercase = UCD_OTHERCASE(c)) != c) - { - *class_utf8data++ = XCL_SINGLE; - class_utf8data += _pcre_ord2utf8(othercase, class_utf8data); - } + mcbuffer[0] = c; + mclength = 1; } -#endif /* SUPPORT_UCP */ + goto ONE_CHAR; + } /* End of 1-char optimization */ - } - else -#endif /* SUPPORT_UTF8 */ + /* There is more than one character in the class, or an XCLASS item + has been generated. Add this character to the class. */ - /* Handle a single-byte character */ - { - classbits[c/8] |= (1 << (c&7)); - if ((options & PCRE_CASELESS) != 0) - { - c = cd->fcc[c]; /* flip case */ - classbits[c/8] |= (1 << (c&7)); - } - class_charcount++; - class_lastchar = c; - } + class_has_8bitchar += + add_to_class(classbits, &class_uchardata, options, cd, c, c); } /* Loop until ']' reached. This "while" is the end of the "do" far above. If we are at the end of an internal nested string, revert to the outer string. */ - while (((c = *(++ptr)) != 0 || + while (((c = *(++ptr)) != CHAR_NULL || (nestptr != NULL && - (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != 0)) && + (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != CHAR_NULL)) && (c != CHAR_RIGHT_SQUARE_BRACKET || inescq)); /* Check for missing terminating ']' */ - if (c == 0) + if (c == CHAR_NULL) { *errorcodeptr = ERR6; goto FAILED; } - /* If class_charcount is 1, we saw precisely one character whose value is - 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. + /* We will need an XCLASS if data has been placed in class_uchardata. In + the second phase this is a sufficient test. However, in the pre-compile + phase, class_uchardata gets emptied to prevent workspace overflow, so it + only if the very last character in the class needs XCLASS will it contain + anything at this point. For this reason, xclass gets set TRUE above when + uchar_classdata is emptied, and that's why this code is the way it is here + instead of just doing a test on class_uchardata below. */ - 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 characters 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[I] if it's positive, or OP_NOT[I] if it's negative. - Note that OP_NOT[I] does not support multibyte characters. In the positive - case, it can cause firstbyte to be set. Otherwise, there can be no first - char if this item is first, whatever repeat count may follow. In the case - of reqbyte, save the previous value for reinstating. */ - -#ifdef SUPPORT_UTF8 - if (class_charcount == 1 && !class_utf8 && - (!utf8 || !negate_class || class_lastchar < 128)) -#else - if (class_charcount == 1) +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + if (class_uchardata > class_uchardata_base) xclass = TRUE; #endif - { - zeroreqbyte = reqbyte; - /* The OP_NOT[I] opcodes work on one-byte characters only. */ + /* If this is the first thing in the branch, there can be no first char + setting, whatever the repeat count. Any reqchar setting must remain + unchanged after any kind of repeat. */ - if (negate_class) - { - if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE; - zerofirstbyte = firstbyte; - *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT; - *code++ = class_lastchar; - break; - } + if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; + zerofirstchar = firstchar; + zerofirstcharflags = firstcharflags; + zeroreqchar = reqchar; + zeroreqcharflags = reqcharflags; - /* For a single, positive character, get the value into mcbuffer, and - then we can handle this with the normal one-character code. */ - -#ifdef SUPPORT_UTF8 - if (utf8 && class_lastchar > 127) - mclength = _pcre_ord2utf8(class_lastchar, mcbuffer); - else -#endif - { - mcbuffer[0] = class_lastchar; - mclength = 1; - } - goto ONE_CHAR; - } /* End of 1-char optimization */ - - /* The general case - not the one-char optimization. If this is the first - thing in the branch, there can be no first char setting, whatever the - repeat count. Any reqbyte setting must remain unchanged after any kind of - repeat. */ - - if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE; - zerofirstbyte = firstbyte; - zeroreqbyte = reqbyte; - /* If there are characters with values > 255, we have to compile an extended class, with its own opcode, unless there was a negated special such as \S in the class, and PCRE_UCP is not set, because in that case all @@ -4360,25 +4653,30 @@ for (;; ptr++) be listed) there are no characters < 256, we can omit the bitmap in the actual compiled code. */ -#ifdef SUPPORT_UTF8 - if (class_utf8 && (!should_flip_negation || (options & PCRE_UCP) != 0)) +#ifdef SUPPORT_UTF + if (xclass && (!should_flip_negation || (options & PCRE_UCP) != 0)) +#elif !defined COMPILE_PCRE8 + if (xclass && !should_flip_negation) +#endif +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 { - *class_utf8data++ = XCL_END; /* Marks the end of extra data */ + *class_uchardata++ = XCL_END; /* Marks the end of extra data */ *code++ = OP_XCLASS; code += LINK_SIZE; - *code = negate_class? XCL_NOT : 0; + *code = negate_class? XCL_NOT:0; /* If the map is required, move up the extra data to make room for it; otherwise just move the code pointer to the end of the extra data. */ - if (class_charcount > 0) + if (class_has_8bitchar > 0) { *code++ |= XCL_MAP; - memmove(code + 32, code, class_utf8data - code); + memmove(code + (32 / sizeof(pcre_uchar)), code, + IN_UCHARS(class_uchardata - code)); memcpy(code, classbits, 32); - code = class_utf8data + 32; + code = class_uchardata + (32 / sizeof(pcre_uchar)); } - else code = class_utf8data; + else code = class_uchardata; /* Now fill in the complete length of the item */ @@ -4394,16 +4692,15 @@ for (;; ptr++) negating it if necessary. */ *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS; - if (negate_class) + if (lengthptr == NULL) /* Save time in the pre-compile phase */ { - if (lengthptr == NULL) /* Save time in the pre-compile phase */ - for (c = 0; c < 32; c++) code[c] = ~classbits[c]; - } - else - { + if (negate_class) + for (c = 0; c < 32; c++) classbits[c] = ~classbits[c]; memcpy(code, classbits, 32); } - code += 32; + code += 32 / sizeof(pcre_uchar); + + END_CLASS: break; @@ -4440,8 +4737,10 @@ for (;; ptr++) if (repeat_min == 0) { - firstbyte = zerofirstbyte; /* Adjust for zero repeat */ - reqbyte = zeroreqbyte; /* Ditto */ + firstchar = zerofirstchar; /* Adjust for zero repeat */ + firstcharflags = zerofirstcharflags; + reqchar = zeroreqchar; /* Ditto */ + reqcharflags = zeroreqcharflags; } /* Remember whether this is a variable length repeat */ @@ -4483,7 +4782,7 @@ for (;; ptr++) if (*previous == OP_RECURSE) { - memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE); + memmove(previous + 1 + LINK_SIZE, previous, IN_UCHARS(1 + LINK_SIZE)); *previous = OP_ONCE; PUT(previous, 1, 2 + 2*LINK_SIZE); previous[2 + 2*LINK_SIZE] = OP_KET; @@ -4504,39 +4803,50 @@ for (;; ptr++) /* Now handle repetition for the different types of item. */ - /* If previous was a character match, abolish the item and generate a - repeat item instead. If a char item has a minumum of more than one, ensure - that it is set in reqbyte - it might not be if a sequence such as x{3} is - the first thing in a branch because the x will have gone into firstbyte - instead. */ + /* If previous was a character or negated character match, abolish the item + and generate a repeat item instead. If a char item has a minimum of more + than one, ensure that it is set in reqchar - it might not be if a sequence + such as x{3} is the first thing in a branch because the x will have gone + into firstchar instead. */ - if (*previous == OP_CHAR || *previous == OP_CHARI) + if (*previous == OP_CHAR || *previous == OP_CHARI + || *previous == OP_NOT || *previous == OP_NOTI) { - op_type = (*previous == OP_CHAR)? 0 : OP_STARI - OP_STAR; + switch (*previous) + { + default: /* Make compiler happy. */ + case OP_CHAR: op_type = OP_STAR - OP_STAR; break; + case OP_CHARI: op_type = OP_STARI - OP_STAR; break; + case OP_NOT: op_type = OP_NOTSTAR - OP_STAR; break; + case OP_NOTI: op_type = OP_NOTSTARI - OP_STAR; break; + } - /* Deal with UTF-8 characters that take up more than one byte. It's + /* Deal with UTF characters that take up more than one character. It's easier to write this out separately than try to macrify it. Use c to - hold the length of the character in bytes, plus 0x80 to flag that it's a - length rather than a small character. */ + hold the length of the character in bytes, plus UTF_LENGTH to flag that + it's a length rather than a small character. */ -#ifdef SUPPORT_UTF8 - if (utf8 && (code[-1] & 0x80) != 0) +#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 + if (utf && NOT_FIRSTCHAR(code[-1])) { - uschar *lastchar = code - 1; - while((*lastchar & 0xc0) == 0x80) lastchar--; + pcre_uchar *lastchar = code - 1; + BACKCHAR(lastchar); c = (int)(code - lastchar); /* Length of UTF-8 character */ - memcpy(utf8_char, lastchar, c); /* Save the char */ - c |= 0x80; /* Flag c as a length */ + memcpy(utf_chars, lastchar, IN_UCHARS(c)); /* Save the char */ + c |= UTF_LENGTH; /* Flag c as a length */ } else -#endif +#endif /* SUPPORT_UTF */ - /* Handle the case of a single byte - either with no UTF8 support, or - with UTF-8 disabled, or for a UTF-8 character < 128. */ - + /* Handle the case of a single charater - either with no UTF support, or + with UTF disabled, or for a single character UTF character. */ { c = code[-1]; - if (repeat_min > 1) reqbyte = c | req_caseopt | cd->req_varyopt; + if (*previous <= OP_CHARI && repeat_min > 1) + { + reqchar = c; + reqcharflags = req_caseopt | cd->req_varyopt; + } } /* If the repetition is unlimited, it pays to see if the next thing on @@ -4546,7 +4856,7 @@ for (;; ptr++) if (!possessive_quantifier && repeat_max < 0 && - check_auto_possessive(previous, utf8, ptr + 1, options, cd)) + check_auto_possessive(previous, utf, ptr + 1, options, cd)) { repeat_type = 0; /* Force greedy */ possessive_quantifier = TRUE; @@ -4555,26 +4865,6 @@ for (;; ptr++) goto OUTPUT_SINGLE_REPEAT; /* Code shared with single character types */ } - /* If previous was a single negated character ([^a] or similar), we use - one of the special opcodes, replacing it. The code is shared with single- - character repeats by setting opt_type to add a suitable offset into - repeat_type. We can also test for auto-possessification. OP_NOT and OP_NOTI - are currently used only for single-byte chars. */ - - else if (*previous == OP_NOT || *previous == OP_NOTI) - { - op_type = ((*previous == OP_NOT)? OP_NOTSTAR : OP_NOTSTARI) - OP_STAR; - c = previous[1]; - if (!possessive_quantifier && - repeat_max < 0 && - check_auto_possessive(previous, utf8, ptr + 1, options, cd)) - { - repeat_type = 0; /* Force greedy */ - possessive_quantifier = TRUE; - } - goto OUTPUT_SINGLE_REPEAT; - } - /* If previous was a character type match (\d or similar), abolish it and create a suitable repeat item. The code is shared with single-character repeats by setting op_type to add a suitable offset into repeat_type. Note @@ -4584,14 +4874,14 @@ for (;; ptr++) else if (*previous < OP_EODN) { - uschar *oldcode; + pcre_uchar *oldcode; int prop_type, prop_value; op_type = OP_TYPESTAR - OP_STAR; /* Use type opcodes */ c = *previous; if (!possessive_quantifier && repeat_max < 0 && - check_auto_possessive(previous, utf8, ptr + 1, options, cd)) + check_auto_possessive(previous, utf, ptr + 1, options, cd)) { repeat_type = 0; /* Force greedy */ possessive_quantifier = TRUE; @@ -4613,16 +4903,6 @@ for (;; ptr++) if (repeat_max == 0) goto END_REPEAT; - /*--------------------------------------------------------------------*/ - /* This code is obsolete from release 8.00; the restriction was finally - removed: */ - - /* 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->external_flags |= PCRE_NOPARTIAL; */ - /*--------------------------------------------------------------------*/ - /* Combine the op_type with the repeat_type */ repeat_type += op_type; @@ -4671,14 +4951,14 @@ for (;; ptr++) we have to insert the character for the previous code. For a repeated Unicode property match, there are two extra bytes that define the required property. In UTF-8 mode, long characters have their length in - c, with the 0x80 bit as a flag. */ + c, with the UTF_LENGTH bit as a flag. */ if (repeat_max < 0) { -#ifdef SUPPORT_UTF8 - if (utf8 && c >= 128) +#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 + if (utf && (c & UTF_LENGTH) != 0) { - memcpy(code, utf8_char, c & 7); + memcpy(code, utf_chars, IN_UCHARS(c & 7)); code += c & 7; } else @@ -4700,10 +4980,10 @@ for (;; ptr++) else if (repeat_max != repeat_min) { -#ifdef SUPPORT_UTF8 - if (utf8 && c >= 128) +#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 + if (utf && (c & UTF_LENGTH) != 0) { - memcpy(code, utf8_char, c & 7); + memcpy(code, utf_chars, IN_UCHARS(c & 7)); code += c & 7; } else @@ -4730,10 +5010,10 @@ for (;; ptr++) /* The character or character type itself comes last in all cases. */ -#ifdef SUPPORT_UTF8 - if (utf8 && c >= 128) +#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 + if (utf && (c & UTF_LENGTH) != 0) { - memcpy(code, utf8_char, c & 7); + memcpy(code, utf_chars, IN_UCHARS(c & 7)); code += c & 7; } else @@ -4757,7 +5037,7 @@ for (;; ptr++) else if (*previous == OP_CLASS || *previous == OP_NCLASS || -#ifdef SUPPORT_UTF8 +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 *previous == OP_XCLASS || #endif *previous == OP_REF || @@ -4769,16 +5049,6 @@ for (;; ptr++) goto END_REPEAT; } - /*--------------------------------------------------------------------*/ - /* This code is obsolete from release 8.00; the restriction was finally - removed: */ - - /* 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->external_flags |= PCRE_NOPARTIAL; */ - /*--------------------------------------------------------------------*/ - if (repeat_min == 0 && repeat_max == -1) *code++ = OP_CRSTAR + repeat_type; else if (repeat_min == 1 && repeat_max == -1) @@ -4806,8 +5076,8 @@ for (;; ptr++) { register int i; int len = (int)(code - previous); - uschar *bralink = NULL; - uschar *brazeroptr = NULL; + pcre_uchar *bralink = NULL; + pcre_uchar *brazeroptr = NULL; /* Repeating a DEFINE group is pointless, but Perl allows the syntax, so we just ignore the repeat. */ @@ -4860,8 +5130,8 @@ for (;; ptr++) if (repeat_max <= 1) /* Covers 0, 1, and unlimited */ { *code = OP_END; - adjust_recurse(previous, 1, utf8, cd, save_hwm); - memmove(previous+1, previous, len); + adjust_recurse(previous, 1, utf, cd, save_hwm); + memmove(previous + 1, previous, IN_UCHARS(len)); code++; if (repeat_max == 0) { @@ -4884,8 +5154,8 @@ for (;; ptr++) { int offset; *code = OP_END; - adjust_recurse(previous, 2 + LINK_SIZE, utf8, cd, save_hwm); - memmove(previous + 2 + LINK_SIZE, previous, len); + adjust_recurse(previous, 2 + LINK_SIZE, utf, cd, save_hwm); + memmove(previous + 2 + LINK_SIZE, previous, IN_UCHARS(len)); code += 2 + LINK_SIZE; *previous++ = OP_BRAZERO + repeat_type; *previous++ = OP_BRA; @@ -4938,13 +5208,17 @@ for (;; ptr++) else { - if (groupsetfirstbyte && reqbyte < 0) reqbyte = firstbyte; + if (groupsetfirstchar && reqcharflags < 0) + { + reqchar = firstchar; + reqcharflags = firstcharflags; + } for (i = 1; i < repeat_min; i++) { - uschar *hc; - uschar *this_hwm = cd->hwm; - memcpy(code, previous, len); + pcre_uchar *hc; + pcre_uchar *this_hwm = cd->hwm; + memcpy(code, previous, IN_UCHARS(len)); while (cd->hwm > cd->start_workspace + cd->workspace_size - WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm)) @@ -4953,8 +5227,8 @@ for (;; ptr++) int this_offset = this_hwm - cd->start_workspace; *errorcodeptr = expand_workspace(cd); if (*errorcodeptr != 0) goto FAILED; - save_hwm = (uschar *)cd->start_workspace + save_offset; - this_hwm = (uschar *)cd->start_workspace + this_offset; + save_hwm = (pcre_uchar *)cd->start_workspace + save_offset; + this_hwm = (pcre_uchar *)cd->start_workspace + this_offset; } for (hc = save_hwm; hc < this_hwm; hc += LINK_SIZE) @@ -5006,8 +5280,8 @@ for (;; ptr++) else for (i = repeat_max - 1; i >= 0; i--) { - uschar *hc; - uschar *this_hwm = cd->hwm; + pcre_uchar *hc; + pcre_uchar *this_hwm = cd->hwm; *code++ = OP_BRAZERO + repeat_type; @@ -5023,7 +5297,7 @@ for (;; ptr++) PUTINC(code, 0, offset); } - memcpy(code, previous, len); + memcpy(code, previous, IN_UCHARS(len)); /* Ensure there is enough workspace for forward references before copying them. */ @@ -5035,8 +5309,8 @@ for (;; ptr++) int this_offset = this_hwm - cd->start_workspace; *errorcodeptr = expand_workspace(cd); if (*errorcodeptr != 0) goto FAILED; - save_hwm = (uschar *)cd->start_workspace + save_offset; - this_hwm = (uschar *)cd->start_workspace + this_offset; + save_hwm = (pcre_uchar *)cd->start_workspace + save_offset; + this_hwm = (pcre_uchar *)cd->start_workspace + this_offset; } for (hc = save_hwm; hc < this_hwm; hc += LINK_SIZE) @@ -5055,7 +5329,7 @@ for (;; ptr++) { int oldlinkoffset; int offset = (int)(code - bralink + 1); - uschar *bra = code - offset; + pcre_uchar *bra = code - offset; oldlinkoffset = GET(bra, 1); bralink = (oldlinkoffset == 0)? NULL : bralink - oldlinkoffset; *code++ = OP_KET; @@ -5091,8 +5365,8 @@ for (;; ptr++) else { - uschar *ketcode = code - 1 - LINK_SIZE; - uschar *bracode = ketcode - GET(ketcode, 1); + pcre_uchar *ketcode = code - 1 - LINK_SIZE; + pcre_uchar *bracode = ketcode - GET(ketcode, 1); /* Convert possessive ONCE brackets to non-capturing */ @@ -5114,10 +5388,10 @@ for (;; ptr++) if (lengthptr == NULL) { - uschar *scode = bracode; + pcre_uchar *scode = bracode; do { - if (could_be_empty_branch(scode, ketcode, utf8, cd)) + if (could_be_empty_branch(scode, ketcode, utf, cd)) { *bracode += OP_SBRA - OP_BRA; break; @@ -5140,8 +5414,8 @@ for (;; ptr++) { int nlen = (int)(code - bracode); *code = OP_END; - adjust_recurse(bracode, 1 + LINK_SIZE, utf8, cd, save_hwm); - memmove(bracode + 1+LINK_SIZE, bracode, nlen); + adjust_recurse(bracode, 1 + LINK_SIZE, utf, cd, save_hwm); + memmove(bracode + 1 + LINK_SIZE, bracode, IN_UCHARS(nlen)); code += 1 + LINK_SIZE; nlen += 1 + LINK_SIZE; *bracode = OP_BRAPOS; @@ -5210,15 +5484,16 @@ for (;; ptr++) int len; if (*tempcode == OP_TYPEEXACT) - tempcode += _pcre_OP_lengths[*tempcode] + - ((tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP)? 2 : 0); + tempcode += PRIV(OP_lengths)[*tempcode] + + ((tempcode[1 + IMM2_SIZE] == OP_PROP + || tempcode[1 + IMM2_SIZE] == OP_NOTPROP)? 2 : 0); else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT) { - tempcode += _pcre_OP_lengths[*tempcode]; -#ifdef SUPPORT_UTF8 - if (utf8 && tempcode[-1] >= 0xc0) - tempcode += _pcre_utf8_table4[tempcode[-1] & 0x3f]; + tempcode += PRIV(OP_lengths)[*tempcode]; +#ifdef SUPPORT_UTF + if (utf && HAS_EXTRALEN(tempcode[-1])) + tempcode += GET_EXTRALEN(tempcode[-1]); #endif } @@ -5255,8 +5530,8 @@ for (;; ptr++) default: *code = OP_END; - adjust_recurse(tempcode, 1 + LINK_SIZE, utf8, cd, save_hwm); - memmove(tempcode + 1+LINK_SIZE, tempcode, len); + adjust_recurse(tempcode, 1 + LINK_SIZE, utf, cd, save_hwm); + memmove(tempcode + 1 + LINK_SIZE, tempcode, IN_UCHARS(len)); code += 1 + LINK_SIZE; len += 1 + LINK_SIZE; tempcode[0] = OP_ONCE; @@ -5268,7 +5543,7 @@ for (;; ptr++) } /* In all case we no longer have a previous item. We also set the - "follows varying string" flag for subsequently encountered reqbytes if + "follows varying string" flag for subsequently encountered reqchars if it isn't already set and we have just passed a varying length item. */ END_REPEAT: @@ -5291,16 +5566,18 @@ for (;; ptr++) /* First deal with various "verbs" that can be introduced by '*'. */ - if (*(++ptr) == CHAR_ASTERISK && - ((cd->ctypes[ptr[1]] & ctype_letter) != 0 || ptr[1] == ':')) + ptr++; + if (ptr[0] == CHAR_ASTERISK && (ptr[1] == ':' + || (MAX_255(ptr[1]) && ((cd->ctypes[ptr[1]] & ctype_letter) != 0)))) { int i, namelen; int arglen = 0; const char *vn = verbnames; - const uschar *name = ptr + 1; - const uschar *arg = NULL; + const pcre_uchar *name = ptr + 1; + const pcre_uchar *arg = NULL; previous = NULL; - while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {}; + ptr++; + while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_letter) != 0) ptr++; namelen = (int)(ptr - name); /* It appears that Perl allows any characters whatsoever, other than @@ -5310,8 +5587,13 @@ for (;; ptr++) if (*ptr == CHAR_COLON) { arg = ++ptr; - while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++; + while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++; arglen = (int)(ptr - arg); + if ((unsigned int)arglen > MAX_MARK) + { + *errorcodeptr = ERR75; + goto FAILED; + } } if (*ptr != CHAR_RIGHT_PARENTHESIS) @@ -5325,8 +5607,10 @@ for (;; ptr++) for (i = 0; i < verbcount; i++) { if (namelen == verbs[i].len && - strncmp((char *)name, vn, namelen) == 0) + STRNCMP_UC_C8(name, vn, namelen) == 0) { + int setverb; + /* Check for open captures before ACCEPT and convert it to ASSERT_ACCEPT if in an assertion. */ @@ -5344,10 +5628,11 @@ for (;; ptr++) *code++ = OP_CLOSE; PUT2INC(code, 0, oc->number); } - *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT; + setverb = *code++ = + (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT; - /* Do not set firstbyte after *ACCEPT */ - if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE; + /* Do not set firstchar after *ACCEPT */ + if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; } /* Handle other cases with/without an argument */ @@ -5359,8 +5644,7 @@ for (;; ptr++) *errorcodeptr = ERR66; goto FAILED; } - *code = verbs[i].op; - if (*code++ == OP_THEN) cd->external_flags |= PCRE_HASTHEN; + setverb = *code++ = verbs[i].op; } else @@ -5370,14 +5654,28 @@ for (;; ptr++) *errorcodeptr = ERR59; goto FAILED; } - *code = verbs[i].op_arg; - if (*code++ == OP_THEN_ARG) cd->external_flags |= PCRE_HASTHEN; + setverb = *code++ = verbs[i].op_arg; *code++ = arglen; - memcpy(code, arg, arglen); + memcpy(code, arg, IN_UCHARS(arglen)); code += arglen; *code++ = 0; } + switch (setverb) + { + case OP_THEN: + case OP_THEN_ARG: + cd->external_flags |= PCRE_HASTHEN; + break; + + case OP_PRUNE: + case OP_PRUNE_ARG: + case OP_SKIP: + case OP_SKIP_ARG: + cd->had_pruneorskip = TRUE; + break; + } + break; /* Found verb, exit loop */ } @@ -5396,15 +5694,15 @@ for (;; ptr++) { int i, set, unset, namelen; int *optset; - const uschar *name; - uschar *slot; + const pcre_uchar *name; + pcre_uchar *slot; switch (*(++ptr)) { case CHAR_NUMBER_SIGN: /* Comment; skip to ket */ ptr++; - while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++; - if (*ptr == 0) + while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++; + if (*ptr == CHAR_NULL) { *errorcodeptr = ERR18; goto FAILED; @@ -5427,6 +5725,7 @@ for (;; ptr++) /* ------------------------------------------------------------ */ case CHAR_LEFT_PARENTHESIS: bravalue = OP_COND; /* Conditional group */ + tempptr = ptr; /* A condition can be an assertion, a number (referring to a numbered group), a name (referring to a named group), or 'R', referring to @@ -5439,21 +5738,35 @@ for (;; ptr++) be the recursive thing or the name 'R' (and similarly for 'R' followed by digits), and (b) a number could be a name that consists of digits. In both cases, we look for a name first; if not found, we try the other - cases. */ + cases. + For compatibility with auto-callouts, we allow a callout to be + specified before a condition that is an assertion. First, check for the + syntax of a callout; if found, adjust the temporary pointer that is + used to check for an assertion condition. That's all that is needed! */ + + if (ptr[1] == CHAR_QUESTION_MARK && ptr[2] == CHAR_C) + { + for (i = 3;; i++) if (!IS_DIGIT(ptr[i])) break; + if (ptr[i] == CHAR_RIGHT_PARENTHESIS) + tempptr += i + 1; + } + /* For conditions that are assertions, check the syntax, and then exit the switch. This will take control down to where bracketed groups, including assertions, are processed. */ - if (ptr[1] == CHAR_QUESTION_MARK && (ptr[2] == CHAR_EQUALS_SIGN || - ptr[2] == CHAR_EXCLAMATION_MARK || ptr[2] == CHAR_LESS_THAN_SIGN)) + if (tempptr[1] == CHAR_QUESTION_MARK && + (tempptr[2] == CHAR_EQUALS_SIGN || + tempptr[2] == CHAR_EXCLAMATION_MARK || + tempptr[2] == CHAR_LESS_THAN_SIGN)) break; /* Most other conditions use OP_CREF (a couple change to OP_RREF - below), and all need to skip 3 bytes at the start of the group. */ + below), and all need to skip 1+IMM2_SIZE bytes at the start of the group. */ code[1+LINK_SIZE] = OP_CREF; - skipbytes = 3; + skipbytes = 1+IMM2_SIZE; refsign = -1; /* Check for a test for recursion in a named group. */ @@ -5480,13 +5793,13 @@ for (;; ptr++) } else { - terminator = 0; + terminator = CHAR_NULL; if (ptr[1] == CHAR_MINUS || ptr[1] == CHAR_PLUS) refsign = *(++ptr); } /* We now expect to read a name; any thing else is an error */ - if ((cd->ctypes[ptr[1]] & ctype_word) == 0) + if (!MAX_255(ptr[1]) || (cd->ctypes[ptr[1]] & ctype_word) == 0) { ptr += 1; /* To get the right offset */ *errorcodeptr = ERR28; @@ -5497,16 +5810,15 @@ for (;; ptr++) recno = 0; name = ++ptr; - while ((cd->ctypes[*ptr] & ctype_word) != 0) + while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) { if (recno >= 0) - recno = ((digitab[*ptr] & ctype_digit) != 0)? - recno * 10 + *ptr - CHAR_0 : -1; + recno = (IS_DIGIT(*ptr))? recno * 10 + (int)(*ptr - CHAR_0) : -1; ptr++; } namelen = (int)(ptr - name); - if ((terminator > 0 && *ptr++ != terminator) || + if ((terminator > 0 && *ptr++ != (pcre_uchar)terminator) || *ptr++ != CHAR_RIGHT_PARENTHESIS) { ptr--; /* Error offset */ @@ -5549,7 +5861,7 @@ for (;; ptr++) slot = cd->name_table; for (i = 0; i < cd->names_found; i++) { - if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break; + if (STRNCMP_UC_UC(name, slot+IMM2_SIZE, namelen) == 0) break; slot += cd->name_entry_size; } @@ -5565,19 +5877,19 @@ for (;; ptr++) /* Search the pattern for a forward reference */ else if ((i = find_parens(cd, name, namelen, - (options & PCRE_EXTENDED) != 0, utf8)) > 0) + (options & PCRE_EXTENDED) != 0, utf)) > 0) { PUT2(code, 2+LINK_SIZE, i); code[1+LINK_SIZE]++; } - /* If terminator == 0 it means that the name followed directly after - the opening parenthesis [e.g. (?(abc)...] and in this case there are - some further alternatives to try. For the cases where terminator != 0 - [things like (?(... or (?('name')... or (?(R&name)... ] we have + /* If terminator == CHAR_NULL it means that the name followed directly + after the opening parenthesis [e.g. (?(abc)...] and in this case there + are some further alternatives to try. For the cases where terminator != + 0 [things like (?(... or (?('name')... or (?(R&name)... ] we have now checked all the possibilities, so give an error. */ - else if (terminator != 0) + else if (terminator != CHAR_NULL) { *errorcodeptr = ERR15; goto FAILED; @@ -5591,7 +5903,7 @@ for (;; ptr++) recno = 0; for (i = 1; i < namelen; i++) { - if ((digitab[name[i]] & ctype_digit) == 0) + if (!IS_DIGIT(name[i])) { *errorcodeptr = ERR15; goto FAILED; @@ -5606,7 +5918,7 @@ for (;; ptr++) /* Similarly, check for the (?(DEFINE) "condition", which is always false. */ - else if (namelen == 6 && strncmp((char *)name, STRING_DEFINE, 6) == 0) + else if (namelen == 6 && STRNCMP_UC_C8(name, STRING_DEFINE, 6) == 0) { code[1+LINK_SIZE] = OP_DEF; skipbytes = 1; @@ -5669,7 +5981,8 @@ for (;; ptr++) break; default: /* Could be name define, else bad */ - if ((cd->ctypes[ptr[1]] & ctype_word) != 0) goto DEFINE_NAME; + if (MAX_255(ptr[1]) && (cd->ctypes[ptr[1]] & ctype_word) != 0) + goto DEFINE_NAME; ptr++; /* Correct offset for error */ *errorcodeptr = ERR24; goto FAILED; @@ -5691,8 +6004,9 @@ for (;; ptr++) *code++ = OP_CALLOUT; { int n = 0; - while ((digitab[*(++ptr)] & ctype_digit) != 0) - n = n * 10 + *ptr - CHAR_0; + ptr++; + while(IS_DIGIT(*ptr)) + n = n * 10 + *ptr++ - CHAR_0; if (*ptr != CHAR_RIGHT_PARENTHESIS) { *errorcodeptr = ERR39; @@ -5737,14 +6051,14 @@ for (;; ptr++) CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE; name = ++ptr; - while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++; + while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++; namelen = (int)(ptr - name); /* In the pre-compile phase, just do a syntax check. */ if (lengthptr != NULL) { - if (*ptr != terminator) + if (*ptr != (pcre_uchar)terminator) { *errorcodeptr = ERR42; goto FAILED; @@ -5754,9 +6068,9 @@ for (;; ptr++) *errorcodeptr = ERR49; goto FAILED; } - if (namelen + 3 > cd->name_entry_size) + if (namelen + IMM2_SIZE + 1 > cd->name_entry_size) { - cd->name_entry_size = namelen + 3; + cd->name_entry_size = namelen + IMM2_SIZE + 1; if (namelen > MAX_NAME_SIZE) { *errorcodeptr = ERR48; @@ -5785,10 +6099,10 @@ for (;; ptr++) for (i = 0; i < cd->names_found; i++) { - int crc = memcmp(name, slot+2, namelen); + int crc = memcmp(name, slot+IMM2_SIZE, IN_UCHARS(namelen)); if (crc == 0) { - if (slot[2+namelen] == 0) + if (slot[IMM2_SIZE+namelen] == 0) { if (GET2(slot, 0) != cd->bracount + 1 && (options & PCRE_DUPNAMES) == 0) @@ -5809,7 +6123,7 @@ for (;; ptr++) if (crc < 0) { memmove(slot + cd->name_entry_size, slot, - (cd->names_found - i) * cd->name_entry_size); + IN_UCHARS((cd->names_found - i) * cd->name_entry_size)); break; } @@ -5823,7 +6137,7 @@ for (;; ptr++) if (!dupname) { - uschar *cslot = cd->name_table; + pcre_uchar *cslot = cd->name_table; for (i = 0; i < cd->names_found; i++) { if (cslot != slot) @@ -5840,8 +6154,8 @@ for (;; ptr++) } PUT2(slot, 0, cd->bracount + 1); - memcpy(slot + 2, name, namelen); - slot[2+namelen] = 0; + memcpy(slot + IMM2_SIZE, name, IN_UCHARS(namelen)); + slot[IMM2_SIZE + namelen] = 0; } } @@ -5867,7 +6181,7 @@ for (;; ptr++) NAMED_REF_OR_RECURSE: name = ++ptr; - while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++; + while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++; namelen = (int)(ptr - name); /* In the pre-compile phase, do a syntax check. We used to just set @@ -5879,14 +6193,14 @@ for (;; ptr++) if (lengthptr != NULL) { - const uschar *temp; + const pcre_uchar *temp; if (namelen == 0) { *errorcodeptr = ERR62; goto FAILED; } - if (*ptr != terminator) + if (*ptr != (pcre_uchar)terminator) { *errorcodeptr = ERR42; goto FAILED; @@ -5909,7 +6223,7 @@ for (;; ptr++) temp = cd->end_pattern; cd->end_pattern = ptr; recno = find_parens(cd, name, namelen, - (options & PCRE_EXTENDED) != 0, utf8); + (options & PCRE_EXTENDED) != 0, utf); cd->end_pattern = temp; if (recno < 0) recno = 0; /* Forward ref; set dummy number */ } @@ -5924,8 +6238,8 @@ for (;; ptr++) slot = cd->name_table; for (i = 0; i < cd->names_found; i++) { - if (strncmp((char *)name, (char *)slot+2, namelen) == 0 && - slot[2+namelen] == 0) + if (STRNCMP_UC_UC(name, slot+IMM2_SIZE, namelen) == 0 && + slot[IMM2_SIZE+namelen] == 0) break; slot += cd->name_entry_size; } @@ -5936,7 +6250,7 @@ for (;; ptr++) } else if ((recno = /* Forward back reference */ find_parens(cd, name, namelen, - (options & PCRE_EXTENDED) != 0, utf8)) <= 0) + (options & PCRE_EXTENDED) != 0, utf)) <= 0) { *errorcodeptr = ERR15; goto FAILED; @@ -5961,7 +6275,7 @@ for (;; ptr++) case CHAR_0: case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4: case CHAR_5: case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9: { - const uschar *called; + const pcre_uchar *called; terminator = CHAR_RIGHT_PARENTHESIS; /* Come here from the \g<...> and \g'...' code (Oniguruma @@ -5975,7 +6289,7 @@ for (;; ptr++) if ((refsign = *ptr) == CHAR_PLUS) { ptr++; - if ((digitab[*ptr] & ctype_digit) == 0) + if (!IS_DIGIT(*ptr)) { *errorcodeptr = ERR63; goto FAILED; @@ -5983,16 +6297,16 @@ for (;; ptr++) } else if (refsign == CHAR_MINUS) { - if ((digitab[ptr[1]] & ctype_digit) == 0) + if (!IS_DIGIT(ptr[1])) goto OTHER_CHAR_AFTER_QUERY; ptr++; } recno = 0; - while((digitab[*ptr] & ctype_digit) != 0) + while(IS_DIGIT(*ptr)) recno = recno * 10 + *ptr++ - CHAR_0; - if (*ptr != terminator) + if (*ptr != (pcre_uchar)terminator) { *errorcodeptr = ERR29; goto FAILED; @@ -6040,14 +6354,14 @@ for (;; ptr++) { *code = OP_END; if (recno != 0) - called = _pcre_find_bracket(cd->start_code, utf8, recno); + called = PRIV(find_bracket)(cd->start_code, utf, recno); /* Forward reference */ if (called == NULL) { if (find_parens(cd, NULL, recno, - (options & PCRE_EXTENDED) != 0, utf8) < 0) + (options & PCRE_EXTENDED) != 0, utf) < 0) { *errorcodeptr = ERR15; goto FAILED; @@ -6077,7 +6391,7 @@ for (;; ptr++) conditional subpatterns will be picked up then. */ else if (GET(called, 1) == 0 && cond_depth <= 0 && - could_be_empty(called, code, bcptr, utf8, cd)) + could_be_empty(called, code, bcptr, utf, cd)) { *errorcodeptr = ERR40; goto FAILED; @@ -6085,18 +6399,18 @@ for (;; ptr++) } /* Insert the recursion/subroutine item. It does not have a set first - byte (relevant if it is repeated, because it will then be wrapped - with ONCE brackets). */ + character (relevant if it is repeated, because it will then be + wrapped with ONCE brackets). */ *code = OP_RECURSE; PUT(code, 1, (int)(called - cd->start_code)); code += 1 + LINK_SIZE; - groupsetfirstbyte = FALSE; + groupsetfirstchar = FALSE; } /* Can't determine a first byte now */ - if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE; + if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; continue; @@ -6153,7 +6467,7 @@ for (;; ptr++) both phases. If we are not at the pattern start, reset the greedy defaults and the - case value for firstbyte and reqbyte. */ + case value for firstchar and reqchar. */ if (*ptr == CHAR_RIGHT_PARENTHESIS) { @@ -6166,7 +6480,7 @@ for (;; ptr++) { greedy_default = ((newoptions & PCRE_UNGREEDY) != 0); greedy_non_default = greedy_default ^ 1; - req_caseopt = ((newoptions & PCRE_CASELESS) != 0)? REQ_CASELESS : 0; + req_caseopt = ((newoptions & PCRE_CASELESS) != 0)? REQ_CASELESS:0; } /* Change options at this level, and pass them back for use @@ -6203,7 +6517,7 @@ for (;; ptr++) NUMBERED_GROUP: cd->bracount += 1; PUT2(code, 1+LINK_SIZE, cd->bracount); - skipbytes = 2; + skipbytes = IMM2_SIZE; } /* Process nested bracketed regex. Assertions used not to be repeatable, @@ -6229,8 +6543,10 @@ for (;; ptr++) skipbytes, /* Skip over bracket number */ cond_depth + ((bravalue == OP_COND)?1:0), /* Depth of condition subpatterns */ - &subfirstbyte, /* For possible first char */ - &subreqbyte, /* For possible last char */ + &subfirstchar, /* For possible first char */ + &subfirstcharflags, + &subreqchar, /* For possible last char */ + &subreqcharflags, bcptr, /* Current branch chain */ cd, /* Tables block */ (lengthptr == NULL)? NULL : /* Actual compile phase */ @@ -6258,7 +6574,7 @@ for (;; ptr++) if (bravalue == OP_COND && lengthptr == NULL) { - uschar *tc = code; + pcre_uchar *tc = code; int condcount = 0; do { @@ -6281,7 +6597,7 @@ for (;; ptr++) } /* A "normal" conditional group. If there is just one branch, we must not - make use of its firstbyte or reqbyte, because this is equivalent to an + make use of its firstchar or reqchar, because this is equivalent to an empty second branch. */ else @@ -6291,7 +6607,7 @@ for (;; ptr++) *errorcodeptr = ERR27; goto FAILED; } - if (condcount == 1) subfirstbyte = subreqbyte = REQ_NONE; + if (condcount == 1) subfirstcharflags = subreqcharflags = REQ_NONE; } } @@ -6335,55 +6651,69 @@ for (;; ptr++) /* Handle updating of the required and first characters for other types of group. Update for normal brackets of all kinds, and conditions with two branches (see code above). If the bracket is followed by a quantifier with - zero repeat, we have to back off. Hence the definition of zeroreqbyte and - zerofirstbyte outside the main loop so that they can be accessed for the + zero repeat, we have to back off. Hence the definition of zeroreqchar and + zerofirstchar outside the main loop so that they can be accessed for the back off. */ - zeroreqbyte = reqbyte; - zerofirstbyte = firstbyte; - groupsetfirstbyte = FALSE; + zeroreqchar = reqchar; + zeroreqcharflags = reqcharflags; + zerofirstchar = firstchar; + zerofirstcharflags = firstcharflags; + groupsetfirstchar = FALSE; if (bravalue >= OP_ONCE) { - /* If we have not yet set a firstbyte in this branch, take it from the + /* If we have not yet set a firstchar in this branch, take it from the subpattern, remembering that it was set here so that a repeat of more - than one can replicate it as reqbyte if necessary. If the subpattern has - no firstbyte, set "none" for the whole branch. In both cases, a zero - repeat forces firstbyte to "none". */ + than one can replicate it as reqchar if necessary. If the subpattern has + no firstchar, set "none" for the whole branch. In both cases, a zero + repeat forces firstchar to "none". */ - if (firstbyte == REQ_UNSET) + if (firstcharflags == REQ_UNSET) { - if (subfirstbyte >= 0) + if (subfirstcharflags >= 0) { - firstbyte = subfirstbyte; - groupsetfirstbyte = TRUE; + firstchar = subfirstchar; + firstcharflags = subfirstcharflags; + groupsetfirstchar = TRUE; } - else firstbyte = REQ_NONE; - zerofirstbyte = REQ_NONE; + else firstcharflags = REQ_NONE; + zerofirstcharflags = REQ_NONE; } - /* If firstbyte was previously set, convert the subpattern's firstbyte - into reqbyte if there wasn't one, using the vary flag that was in + /* If firstchar was previously set, convert the subpattern's firstchar + into reqchar if there wasn't one, using the vary flag that was in existence beforehand. */ - else if (subfirstbyte >= 0 && subreqbyte < 0) - subreqbyte = subfirstbyte | tempreqvary; + else if (subfirstcharflags >= 0 && subreqcharflags < 0) + { + subreqchar = subfirstchar; + subreqcharflags = subfirstcharflags | tempreqvary; + } /* If the subpattern set a required byte (or set a first byte that isn't really the first byte - see above), set it. */ - if (subreqbyte >= 0) reqbyte = subreqbyte; + if (subreqcharflags >= 0) + { + reqchar = subreqchar; + reqcharflags = subreqcharflags; + } } - /* For a forward assertion, we take the reqbyte, if set. This can be + /* For a forward assertion, we take the reqchar, if set. This can be helpful if the pattern that follows the assertion doesn't set a different - char. For example, it's useful for /(?=abcde).+/. We can't set firstbyte + char. For example, it's useful for /(?=abcde).+/. We can't set firstchar for an assertion, however because it leads to incorrect effect for patterns - such as /(?=a)a.+/ when the "real" "a" would then become a reqbyte instead - of a firstbyte. This is overcome by a scan at the end if there's no - firstbyte, looking for an asserted first char. */ + such as /(?=a)a.+/ when the "real" "a" would then become a reqchar instead + of a firstchar. This is overcome by a scan at the end if there's no + firstchar, looking for an asserted first char. */ - else if (bravalue == OP_ASSERT && subreqbyte >= 0) reqbyte = subreqbyte; + else if (bravalue == OP_ASSERT && subreqcharflags >= 0) + { + reqchar = subreqchar; + reqcharflags = subreqcharflags; + } break; /* End of processing '(' */ @@ -6391,19 +6721,21 @@ for (;; ptr++) /* Handle metasequences introduced by \. For ones like \d, the ESC_ values are arranged to be the negation of the corresponding OP_values in the default case when PCRE_UCP is not set. For the back references, the values - are ESC_REF plus the reference number. Only back references and those types + are negative the reference number. Only back references and those types that consume a character may be repeated. We can test for values between ESC_b and ESC_Z for the latter; this may have to change if any new ones are ever created. */ case CHAR_BACKSLASH: tempptr = ptr; - c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE); + escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, FALSE); if (*errorcodeptr != 0) goto FAILED; - if (c < 0) + if (escape == 0) /* The escape coded a single character */ + c = ec; + else { - if (-c == ESC_Q) /* Handle start of quoted string */ + if (escape == ESC_Q) /* Handle start of quoted string */ { if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E) ptr += 2; /* avoid empty string */ @@ -6411,29 +6743,31 @@ for (;; ptr++) continue; } - if (-c == ESC_E) continue; /* Perl ignores an orphan \E */ + if (escape == ESC_E) continue; /* Perl ignores an orphan \E */ /* For metasequences that actually match a character, we disable the setting of a first character if it hasn't already been set. */ - if (firstbyte == REQ_UNSET && -c > ESC_b && -c < ESC_Z) - firstbyte = REQ_NONE; + if (firstcharflags == REQ_UNSET && escape > ESC_b && escape < ESC_Z) + firstcharflags = REQ_NONE; /* Set values to reset to if this is followed by a zero repeat. */ - zerofirstbyte = firstbyte; - zeroreqbyte = reqbyte; + zerofirstchar = firstchar; + zerofirstcharflags = firstcharflags; + zeroreqchar = reqchar; + zeroreqcharflags = reqcharflags; /* \g or \g'name' is a subroutine call by name and \g or \g'n' is a subroutine call by number (Oniguruma syntax). In fact, the value - -ESC_g is returned only for these cases. So we don't need to check for < - or ' if the value is -ESC_g. For the Perl syntax \g{n} the value is - -ESC_REF+n, and for the Perl syntax \g{name} the result is -ESC_k (as + ESC_g is returned only for these cases. So we don't need to check for < + or ' if the value is ESC_g. For the Perl syntax \g{n} the value is + -n, and for the Perl syntax \g{name} the result is ESC_k (as that is a synonym for a named back reference). */ - if (-c == ESC_g) + if (escape == ESC_g) { - const uschar *p; + const pcre_uchar *p; save_hwm = cd->hwm; /* Normally this is set when '(' is read */ terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)? CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE; @@ -6450,18 +6784,19 @@ for (;; ptr++) if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS) { - BOOL isnumber = TRUE; - for (p = ptr + 1; *p != 0 && *p != terminator; p++) + BOOL is_a_number = TRUE; + for (p = ptr + 1; *p != CHAR_NULL && *p != (pcre_uchar)terminator; p++) { - if ((cd->ctypes[*p] & ctype_digit) == 0) isnumber = FALSE; + if (!MAX_255(*p)) { is_a_number = FALSE; break; } + if ((cd->ctypes[*p] & ctype_digit) == 0) is_a_number = FALSE; if ((cd->ctypes[*p] & ctype_word) == 0) break; } - if (*p != terminator) + if (*p != (pcre_uchar)terminator) { *errorcodeptr = ERR57; break; } - if (isnumber) + if (is_a_number) { ptr++; goto HANDLE_NUMERICAL_RECURSION; @@ -6473,8 +6808,8 @@ for (;; ptr++) /* Test a signed number in angle brackets or quotes. */ p = ptr + 2; - while ((digitab[*p] & ctype_digit) != 0) p++; - if (*p != terminator) + while (IS_DIGIT(*p)) p++; + if (*p != (pcre_uchar)terminator) { *errorcodeptr = ERR57; break; @@ -6486,7 +6821,7 @@ for (;; ptr++) /* \k or \k'name' is a back reference by name (Perl syntax). We also support \k{name} (.NET syntax). */ - if (-c == ESC_k) + if (escape == ESC_k) { if ((ptr[1] != CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET)) @@ -6501,17 +6836,17 @@ for (;; ptr++) goto NAMED_REF_OR_RECURSE; } - /* Back references are handled specially; must disable firstbyte if + /* Back references are handled specially; must disable firstchar if not set to cope with cases like (?=(\w+))\1: which would otherwise set ':' later. */ - if (-c >= ESC_REF) + if (escape < 0) { open_capitem *oc; - recno = -c - ESC_REF; + recno = -escape; HANDLE_REFERENCE: /* Come here from named backref handling */ - if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE; + if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; previous = code; *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF; PUT2INC(code, 0, recno); @@ -6535,14 +6870,14 @@ for (;; ptr++) /* So are Unicode property matches, if supported. */ #ifdef SUPPORT_UCP - else if (-c == ESC_P || -c == ESC_p) + else if (escape == ESC_P || escape == ESC_p) { BOOL negated; - int pdata; - int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr); - if (ptype < 0) goto FAILED; + unsigned int ptype = 0, pdata = 0; + if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr)) + goto FAILED; previous = code; - *code++ = ((-c == ESC_p) != negated)? OP_PROP : OP_NOTPROP; + *code++ = ((escape == ESC_p) != negated)? OP_PROP : OP_NOTPROP; *code++ = ptype; *code++ = pdata; } @@ -6551,7 +6886,7 @@ for (;; ptr++) /* If Unicode properties are not supported, \X, \P, and \p are not allowed. */ - else if (-c == ESC_X || -c == ESC_P || -c == ESC_p) + else if (escape == ESC_X || escape == ESC_P || escape == ESC_p) { *errorcodeptr = ERR45; goto FAILED; @@ -6561,15 +6896,19 @@ for (;; ptr++) /* For the rest (including \X when Unicode properties are supported), we can obtain the OP value by negating the escape value in the default situation when PCRE_UCP is not set. When it *is* set, we substitute - Unicode property tests. */ + Unicode property tests. Note that \b and \B do a one-character + lookbehind, and \A also behaves as if it does. */ else { + if ((escape == ESC_b || escape == ESC_B || escape == ESC_A) && + cd->max_lookbehind == 0) + cd->max_lookbehind = 1; #ifdef SUPPORT_UCP - if (-c >= ESC_DU && -c <= ESC_wu) + if (escape >= ESC_DU && escape <= ESC_wu) { nestptr = ptr + 1; /* Where to resume */ - ptr = substitutes[-c - ESC_DU] - 1; /* Just before substitute */ + ptr = substitutes[escape - ESC_DU] - 1; /* Just before substitute */ } else #endif @@ -6577,8 +6916,8 @@ for (;; ptr++) so that it works in DFA mode and in lookbehinds. */ { - previous = (-c > ESC_b && -c < ESC_Z)? code : NULL; - *code++ = (!utf8 && c == -ESC_C)? OP_ALLANY : -c; + previous = (escape > ESC_b && escape < ESC_Z)? code : NULL; + *code++ = (!utf && escape == ESC_C)? OP_ALLANY : escape; } } continue; @@ -6588,9 +6927,9 @@ for (;; ptr++) a value > 127. We set its representation in the length/buffer, and then handle it as a data character. */ -#ifdef SUPPORT_UTF8 - if (utf8 && c > 127) - mclength = _pcre_ord2utf8(c, mcbuffer); +#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 + if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR) + mclength = PRIV(ord2utf)(c, mcbuffer); else #endif @@ -6611,12 +6950,9 @@ for (;; ptr++) mclength = 1; mcbuffer[0] = c; -#ifdef SUPPORT_UTF8 - if (utf8 && c >= 0xc0) - { - while ((ptr[1] & 0xc0) == 0x80) - mcbuffer[mclength++] = *(++ptr); - } +#ifdef SUPPORT_UTF + if (utf && HAS_EXTRALEN(c)) + ACROSSCHAR(TRUE, ptr[1], mcbuffer[mclength++] = *(++ptr)); #endif /* At this point we have the character's bytes in mcbuffer, and the length @@ -6624,6 +6960,28 @@ for (;; ptr++) ONE_CHAR: previous = code; + + /* For caseless UTF-8 mode when UCP support is available, check whether + this character has more than one other case. If so, generate a special + OP_PROP item instead of OP_CHARI. */ + +#ifdef SUPPORT_UCP + if (utf && (options & PCRE_CASELESS) != 0) + { + GETCHAR(c, mcbuffer); + if ((c = UCD_CASESET(c)) != 0) + { + *code++ = OP_PROP; + *code++ = PT_CLIST; + *code++ = c; + if (firstcharflags == REQ_UNSET) firstcharflags = zerofirstcharflags = REQ_NONE; + break; + } + } +#endif + + /* Caseful matches, or not one of the multicase characters. */ + *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR; for (c = 0; c < mclength; c++) *code++ = mcbuffer[c]; @@ -6634,34 +6992,47 @@ for (;; ptr++) /* 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 + Otherwise, leave the firstchar value alone, and don't change it on a zero repeat. */ - if (firstbyte == REQ_UNSET) + if (firstcharflags == REQ_UNSET) { - zerofirstbyte = REQ_NONE; - zeroreqbyte = reqbyte; + zerofirstcharflags = REQ_NONE; + zeroreqchar = reqchar; + zeroreqcharflags = reqcharflags; - /* If the character is more than one byte long, we can set firstbyte + /* If the character is more than one byte long, we can set firstchar only if it is not to be matched caselessly. */ if (mclength == 1 || req_caseopt == 0) { - firstbyte = mcbuffer[0] | req_caseopt; - if (mclength != 1) reqbyte = code[-1] | cd->req_varyopt; + firstchar = mcbuffer[0] | req_caseopt; + firstchar = mcbuffer[0]; + firstcharflags = req_caseopt; + + if (mclength != 1) + { + reqchar = code[-1]; + reqcharflags = cd->req_varyopt; + } } - else firstbyte = reqbyte = REQ_NONE; + else firstcharflags = reqcharflags = REQ_NONE; } - /* firstbyte was previously set; we can set reqbyte only if the length is + /* firstchar was previously set; we can set reqchar only if the length is 1 or the matching is caseful. */ else { - zerofirstbyte = firstbyte; - zeroreqbyte = reqbyte; + zerofirstchar = firstchar; + zerofirstcharflags = firstcharflags; + zeroreqchar = reqchar; + zeroreqcharflags = reqcharflags; if (mclength == 1 || req_caseopt == 0) - reqbyte = code[-1] | req_caseopt | cd->req_varyopt; + { + reqchar = code[-1]; + reqcharflags = req_caseopt | cd->req_varyopt; + } } break; /* End of literal character handling */ @@ -6680,7 +7051,6 @@ return FALSE; - /************************************************* * Compile sequence of alternatives * *************************************************/ @@ -6701,8 +7071,10 @@ Arguments: reset_bracount TRUE to reset the count for each branch skipbytes skip this many bytes at start (for brackets and OP_COND) cond_depth depth of nesting for conditional subpatterns - firstbyteptr place to put the first required character, or a negative number - reqbyteptr place to put the last required character, or a negative number + firstcharptr place to put the first required character + firstcharflagsptr place to put the first character flags, or a negative number + reqcharptr place to put the last required character + reqcharflagsptr place to put the last required character flags, or a negative number bcptr pointer to the chain of currently open branches cd points to the data block with tables pointers etc. lengthptr NULL during the real compile phase @@ -6712,29 +7084,34 @@ Returns: TRUE on success */ static BOOL -compile_regex(int options, uschar **codeptr, const uschar **ptrptr, +compile_regex(int options, pcre_uchar **codeptr, const pcre_uchar **ptrptr, int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes, - int cond_depth, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, - compile_data *cd, int *lengthptr) + int cond_depth, + pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr, + pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr, + branch_chain *bcptr, compile_data *cd, int *lengthptr) { -const uschar *ptr = *ptrptr; -uschar *code = *codeptr; -uschar *last_branch = code; -uschar *start_bracket = code; -uschar *reverse_count = NULL; +const pcre_uchar *ptr = *ptrptr; +pcre_uchar *code = *codeptr; +pcre_uchar *last_branch = code; +pcre_uchar *start_bracket = code; +pcre_uchar *reverse_count = NULL; open_capitem capitem; int capnumber = 0; -int firstbyte, reqbyte; -int branchfirstbyte, branchreqbyte; +pcre_uint32 firstchar, reqchar; +pcre_int32 firstcharflags, reqcharflags; +pcre_uint32 branchfirstchar, branchreqchar; +pcre_int32 branchfirstcharflags, branchreqcharflags; int length; -int orig_bracount; -int max_bracount; +unsigned int orig_bracount; +unsigned int max_bracount; branch_chain bc; bc.outer = bcptr; bc.current_branch = code; -firstbyte = reqbyte = REQ_UNSET; +firstchar = reqchar = 0; +firstcharflags = reqcharflags = REQ_UNSET; /* Accumulate the length for use in the pre-compile phase. Start with the length of the BRA and KET and any extra bytes that are required at the @@ -6793,9 +7170,9 @@ for (;;) /* Now compile the branch; in the pre-compile phase its length gets added into the length. */ - if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstbyte, - &branchreqbyte, &bc, cond_depth, cd, - (lengthptr == NULL)? NULL : &length)) + if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstchar, + &branchfirstcharflags, &branchreqchar, &branchreqcharflags, &bc, + cond_depth, cd, (lengthptr == NULL)? NULL : &length)) { *ptrptr = ptr; return FALSE; @@ -6810,43 +7187,58 @@ for (;;) if (lengthptr == NULL) { - /* If this is the first branch, the firstbyte and reqbyte values for the + /* If this is the first branch, the firstchar and reqchar values for the branch become the values for the regex. */ if (*last_branch != OP_ALT) { - firstbyte = branchfirstbyte; - reqbyte = branchreqbyte; + firstchar = branchfirstchar; + firstcharflags = branchfirstcharflags; + reqchar = branchreqchar; + reqcharflags = branchreqcharflags; } - /* If this is not the first branch, the first char and reqbyte have to + /* If this is not the first branch, the first char and reqchar have to match the values from all the previous branches, except that if the - previous value for reqbyte didn't have REQ_VARY set, it can still match, + previous value for reqchar didn't have REQ_VARY set, it can still match, and we set REQ_VARY for the regex. */ else { - /* If we previously had a firstbyte, but it doesn't match the new branch, - we have to abandon the firstbyte for the regex, but if there was - previously no reqbyte, it takes on the value of the old firstbyte. */ + /* If we previously had a firstchar, but it doesn't match the new branch, + we have to abandon the firstchar for the regex, but if there was + previously no reqchar, it takes on the value of the old firstchar. */ - if (firstbyte >= 0 && firstbyte != branchfirstbyte) + if (firstcharflags >= 0 && + (firstcharflags != branchfirstcharflags || firstchar != branchfirstchar)) { - if (reqbyte < 0) reqbyte = firstbyte; - firstbyte = REQ_NONE; + if (reqcharflags < 0) + { + reqchar = firstchar; + reqcharflags = firstcharflags; + } + firstcharflags = REQ_NONE; } - /* If we (now or from before) have no firstbyte, a firstbyte from the - branch becomes a reqbyte if there isn't a branch reqbyte. */ + /* If we (now or from before) have no firstchar, a firstchar from the + branch becomes a reqchar if there isn't a branch reqchar. */ - if (firstbyte < 0 && branchfirstbyte >= 0 && branchreqbyte < 0) - branchreqbyte = branchfirstbyte; + if (firstcharflags < 0 && branchfirstcharflags >= 0 && branchreqcharflags < 0) + { + branchreqchar = branchfirstchar; + branchreqcharflags = branchfirstcharflags; + } - /* Now ensure that the reqbytes match */ + /* Now ensure that the reqchars match */ - if ((reqbyte & ~REQ_VARY) != (branchreqbyte & ~REQ_VARY)) - reqbyte = REQ_NONE; - else reqbyte |= branchreqbyte; /* To "or" REQ_VARY */ + if (((reqcharflags & ~REQ_VARY) != (branchreqcharflags & ~REQ_VARY)) || + reqchar != branchreqchar) + reqcharflags = REQ_NONE; + else + { + reqchar = branchreqchar; + reqcharflags |= branchreqcharflags; /* To "or" REQ_VARY */ + } } /* If lookbehind, check that this branch matches a fixed-length string, and @@ -6875,7 +7267,12 @@ for (;;) *ptrptr = ptr; return FALSE; } - else { PUT(reverse_count, 0, fixed_length); } + else + { + if (fixed_length > cd->max_lookbehind) + cd->max_lookbehind = fixed_length; + PUT(reverse_count, 0, fixed_length); + } } } @@ -6916,7 +7313,7 @@ for (;;) if (cd->open_caps->flag) { memmove(start_bracket + 1 + LINK_SIZE, start_bracket, - code - start_bracket); + IN_UCHARS(code - start_bracket)); *start_bracket = OP_ONCE; code += 1 + LINK_SIZE; PUT(start_bracket, 1, (int)(code - start_bracket)); @@ -6936,8 +7333,10 @@ for (;;) *codeptr = code; *ptrptr = ptr; - *firstbyteptr = firstbyte; - *reqbyteptr = reqbyte; + *firstcharptr = firstchar; + *firstcharflagsptr = firstcharflags; + *reqcharptr = reqchar; + *reqcharflagsptr = reqcharflags; if (lengthptr != NULL) { if (OFLOW_MAX - *lengthptr < length) @@ -7007,23 +7406,27 @@ and the highest back reference was greater than or equ However, by keeping a bitmap of the first 31 back references, we can catch some of the more common cases more precisely. +... A second exception is when the .* appears inside an atomic group, because +this prevents the number of characters it matches from being adjusted. + Arguments: code points to start of expression (the bracket) bracket_map a bitmap of which brackets we are inside while testing; this handles up to substring 31; after that we just have to take the less precise approach - backref_map the back reference bitmap + cd points to the compile data block + atomcount atomic group level Returns: TRUE or FALSE */ static BOOL -is_anchored(register const uschar *code, unsigned int bracket_map, - unsigned int backref_map) +is_anchored(register const pcre_uchar *code, unsigned int bracket_map, + compile_data *cd, int atomcount) { do { - const uschar *scode = first_significant_code(code + _pcre_OP_lengths[*code], - FALSE); + const pcre_uchar *scode = first_significant_code( + code + PRIV(OP_lengths)[*code], FALSE); register int op = *scode; /* Non-capturing brackets */ @@ -7031,7 +7434,7 @@ do { if (op == OP_BRA || op == OP_BRAPOS || op == OP_SBRA || op == OP_SBRAPOS) { - if (!is_anchored(scode, bracket_map, backref_map)) return FALSE; + if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE; } /* Capturing brackets */ @@ -7041,30 +7444,40 @@ do { { int n = GET2(scode, 1+LINK_SIZE); int new_map = bracket_map | ((n < 32)? (1 << n) : 1); - if (!is_anchored(scode, new_map, backref_map)) return FALSE; + if (!is_anchored(scode, new_map, cd, atomcount)) return FALSE; } - /* Other brackets */ + /* Positive forward assertions and conditions */ - else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC || - op == OP_COND) + else if (op == OP_ASSERT || op == OP_COND) { - if (!is_anchored(scode, bracket_map, backref_map)) return FALSE; + if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE; } + /* Atomic groups */ + + else if (op == OP_ONCE || op == OP_ONCE_NC) + { + if (!is_anchored(scode, bracket_map, cd, atomcount + 1)) + return FALSE; + } + /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and - it isn't in brackets that are or may be referenced. */ + it isn't in brackets that are or may be referenced or inside an atomic + group. */ else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR)) { - if (scode[1] != OP_ALLANY || (bracket_map & backref_map) != 0) + if (scode[1] != OP_ALLANY || (bracket_map & cd->backref_map) != 0 || + atomcount > 0 || cd->had_pruneorskip) return FALSE; } /* Check for explicit anchoring */ else if (op != OP_SOD && op != OP_SOM && op != OP_CIRC) return FALSE; + code += GET(code, 1); } while (*code == OP_ALT); /* Loop for each alternative */ @@ -7082,25 +7495,28 @@ return TRUE; matching and for non-DOTALL patterns that start with .* (which must start at the beginning or after \n). As in the case of is_anchored() (see above), we have to take account of back references to capturing brackets that contain .* -because in that case we can't make the assumption. +because in that case we can't make the assumption. Also, the appearance of .* +inside atomic brackets or in a pattern that contains *PRUNE or *SKIP does not +count, because once again the assumption no longer holds. Arguments: code points to start of expression (the bracket) bracket_map a bitmap of which brackets we are inside while testing; this handles up to substring 31; after that we just have to take the less precise approach - backref_map the back reference bitmap + cd points to the compile data + atomcount atomic group level Returns: TRUE or FALSE */ static BOOL -is_startline(const uschar *code, unsigned int bracket_map, - unsigned int backref_map) +is_startline(const pcre_uchar *code, unsigned int bracket_map, + compile_data *cd, int atomcount) { do { - const uschar *scode = first_significant_code(code + _pcre_OP_lengths[*code], - FALSE); + const pcre_uchar *scode = first_significant_code( + code + PRIV(OP_lengths)[*code], FALSE); register int op = *scode; /* If we are at the start of a conditional assertion group, *both* the @@ -7111,7 +7527,7 @@ do { if (op == OP_COND) { scode += 1 + LINK_SIZE; - if (*scode == OP_CALLOUT) scode += _pcre_OP_lengths[OP_CALLOUT]; + if (*scode == OP_CALLOUT) scode += PRIV(OP_lengths)[OP_CALLOUT]; switch (*scode) { case OP_CREF: @@ -7122,7 +7538,7 @@ do { return FALSE; default: /* Assertion */ - if (!is_startline(scode, bracket_map, backref_map)) return FALSE; + if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE; do scode += GET(scode, 1); while (*scode == OP_ALT); scode += 1 + LINK_SIZE; break; @@ -7136,7 +7552,7 @@ do { if (op == OP_BRA || op == OP_BRAPOS || op == OP_SBRA || op == OP_SBRAPOS) { - if (!is_startline(scode, bracket_map, backref_map)) return FALSE; + if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE; } /* Capturing brackets */ @@ -7146,25 +7562,40 @@ do { { int n = GET2(scode, 1+LINK_SIZE); int new_map = bracket_map | ((n < 32)? (1 << n) : 1); - if (!is_startline(scode, new_map, backref_map)) return FALSE; + if (!is_startline(scode, new_map, cd, atomcount)) return FALSE; } - /* Other brackets */ + /* Positive forward assertions */ - else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC) + else if (op == OP_ASSERT) { - if (!is_startline(scode, bracket_map, backref_map)) return FALSE; + if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE; } - /* .* means "start at start or after \n" if it isn't in brackets that - may be referenced. */ + /* Atomic brackets */ + else if (op == OP_ONCE || op == OP_ONCE_NC) + { + if (!is_startline(scode, bracket_map, cd, atomcount + 1)) return FALSE; + } + + /* .* means "start at start or after \n" if it isn't in atomic brackets or + brackets that may be referenced, as long as the pattern does not contain + *PRUNE or *SKIP, because these break the feature. Consider, for example, + /.*?a(*PRUNE)b/ with the subject "aab", which matches "ab", i.e. not at the + start of a line. */ + else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR) { - if (scode[1] != OP_ANY || (bracket_map & backref_map) != 0) return FALSE; + if (scode[1] != OP_ANY || (bracket_map & cd->backref_map) != 0 || + atomcount > 0 || cd->had_pruneorskip) + return FALSE; } - /* Check for explicit circumflex */ + /* Check for explicit circumflex; anything else gives a FALSE result. Note + in particular that this includes atomic brackets OP_ONCE and OP_ONCE_NC + because the number of characters matched by .* cannot be adjusted inside + them. */ else if (op != OP_CIRC && op != OP_CIRCM) return FALSE; @@ -7192,26 +7623,33 @@ we return that char, otherwise -1. Arguments: code points to start of expression (the bracket) + flags points to the first char flags, or to REQ_NONE inassert TRUE if in an assertion -Returns: -1 or the fixed first char +Returns: the fixed first char, or 0 with REQ_NONE in flags */ -static int -find_firstassertedchar(const uschar *code, BOOL inassert) +static pcre_uint32 +find_firstassertedchar(const pcre_uchar *code, pcre_int32 *flags, + BOOL inassert) { -register int c = -1; +register pcre_uint32 c = 0; +int cflags = REQ_NONE; + +*flags = REQ_NONE; do { - int d; + pcre_uint32 d; + int dflags; int xl = (*code == OP_CBRA || *code == OP_SCBRA || - *code == OP_CBRAPOS || *code == OP_SCBRAPOS)? 2:0; - const uschar *scode = first_significant_code(code + 1+LINK_SIZE + xl, TRUE); - register int op = *scode; + *code == OP_CBRAPOS || *code == OP_SCBRAPOS)? IMM2_SIZE:0; + const pcre_uchar *scode = first_significant_code(code + 1+LINK_SIZE + xl, + TRUE); + register pcre_uchar op = *scode; switch(op) { default: - return -1; + return 0; case OP_BRA: case OP_BRAPOS: @@ -7223,41 +7661,44 @@ do { case OP_ONCE: case OP_ONCE_NC: case OP_COND: - if ((d = find_firstassertedchar(scode, op == OP_ASSERT)) < 0) - return -1; - if (c < 0) c = d; else if (c != d) return -1; + d = find_firstassertedchar(scode, &dflags, op == OP_ASSERT); + if (dflags < 0) + return 0; + if (cflags < 0) { c = d; cflags = dflags; } else if (c != d || cflags != dflags) return 0; break; case OP_EXACT: - scode += 2; + scode += IMM2_SIZE; /* Fall through */ case OP_CHAR: case OP_PLUS: case OP_MINPLUS: case OP_POSPLUS: - if (!inassert) return -1; - if (c < 0) c = scode[1]; - else if (c != scode[1]) return -1; + if (!inassert) return 0; + if (cflags < 0) { c = scode[1]; cflags = 0; } + else if (c != scode[1]) return 0; break; case OP_EXACTI: - scode += 2; + scode += IMM2_SIZE; /* Fall through */ case OP_CHARI: case OP_PLUSI: case OP_MINPLUSI: case OP_POSPLUSI: - if (!inassert) return -1; - if (c < 0) c = scode[1] | REQ_CASELESS; - else if (c != scode[1]) return -1; + if (!inassert) return 0; + if (cflags < 0) { c = scode[1]; cflags = REQ_CASELESS; } + else if (c != scode[1]) return 0; break; } code += GET(code, 1); } while (*code == OP_ALT); + +*flags = cflags; return c; } @@ -7285,28 +7726,59 @@ Returns: pointer to compiled data block, or NUL with errorptr and erroroffset set */ +#if defined COMPILE_PCRE8 PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION pcre_compile(const char *pattern, int options, const char **errorptr, int *erroroffset, const unsigned char *tables) +#elif defined COMPILE_PCRE16 +PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION +pcre16_compile(PCRE_SPTR16 pattern, int options, const char **errorptr, + int *erroroffset, const unsigned char *tables) +#elif defined COMPILE_PCRE32 +PCRE_EXP_DEFN pcre32 * PCRE_CALL_CONVENTION +pcre32_compile(PCRE_SPTR32 pattern, int options, const char **errorptr, + int *erroroffset, const unsigned char *tables) +#endif { +#if defined COMPILE_PCRE8 return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables); +#elif defined COMPILE_PCRE16 +return pcre16_compile2(pattern, options, NULL, errorptr, erroroffset, tables); +#elif defined COMPILE_PCRE32 +return pcre32_compile2(pattern, options, NULL, errorptr, erroroffset, tables); +#endif } +#if defined COMPILE_PCRE8 PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION pcre_compile2(const char *pattern, int options, int *errorcodeptr, const char **errorptr, int *erroroffset, const unsigned char *tables) +#elif defined COMPILE_PCRE16 +PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION +pcre16_compile2(PCRE_SPTR16 pattern, int options, int *errorcodeptr, + const char **errorptr, int *erroroffset, const unsigned char *tables) +#elif defined COMPILE_PCRE32 +PCRE_EXP_DEFN pcre32 * PCRE_CALL_CONVENTION +pcre32_compile2(PCRE_SPTR32 pattern, int options, int *errorcodeptr, + const char **errorptr, int *erroroffset, const unsigned char *tables) +#endif { -real_pcre *re; +REAL_PCRE *re; int length = 1; /* For final END opcode */ -int firstbyte, reqbyte, newline; +pcre_int32 firstcharflags, reqcharflags; +pcre_uint32 firstchar, reqchar; +pcre_uint32 limit_match = PCRE_UINT32_MAX; +pcre_uint32 limit_recursion = PCRE_UINT32_MAX; +int newline; int errorcode = 0; int skipatstart = 0; -BOOL utf8; +BOOL utf; +BOOL never_utf = FALSE; size_t size; -uschar *code; -const uschar *codestart; -const uschar *ptr; +pcre_uchar *code; +const pcre_uchar *codestart; +const pcre_uchar *ptr; compile_data compile_block; compile_data *cd = &compile_block; @@ -7317,11 +7789,11 @@ this purpose. The same space is used in the second pha to fill in forward references to subpatterns. That may overflow, in which case new memory is obtained from malloc(). */ -uschar cworkspace[COMPILE_WORK_SIZE]; +pcre_uchar cworkspace[COMPILE_WORK_SIZE]; /* Set this early so that early errors get offset 0. */ -ptr = (const uschar *)pattern; +ptr = (const pcre_uchar *)pattern; /* We can't pass back an error message if errorptr is NULL; I guess the best we can do is just return NULL, but we can set a code value if there is a code @@ -7348,7 +7820,7 @@ if (erroroffset == NULL) /* Set up pointers to the individual character tables */ -if (tables == NULL) tables = _pcre_default_tables; +if (tables == NULL) tables = PRIV(default_tables); cd->lcc = tables + lcc_offset; cd->fcc = tables + fcc_offset; cd->cbits = tables + cbits_offset; @@ -7362,36 +7834,97 @@ if ((options & ~PUBLIC_COMPILE_OPTIONS) != 0) goto PCRE_EARLY_ERROR_RETURN; } +/* If PCRE_NEVER_UTF is set, remember it. */ + +if ((options & PCRE_NEVER_UTF) != 0) never_utf = TRUE; + /* Check for global one-time settings at the start of the pattern, and remember the offset for later. */ +cd->external_flags = 0; /* Initialize here for LIMIT_MATCH/RECURSION */ + while (ptr[skipatstart] == CHAR_LEFT_PARENTHESIS && ptr[skipatstart+1] == CHAR_ASTERISK) { int newnl = 0; int newbsr = 0; - if (strncmp((char *)(ptr+skipatstart+2), STRING_UTF8_RIGHTPAR, 5) == 0) +/* For completeness and backward compatibility, (*UTFn) is supported in the +relevant libraries, but (*UTF) is generic and always supported. Note that +PCRE_UTF8 == PCRE_UTF16 == PCRE_UTF32. */ + +#ifdef COMPILE_PCRE8 + if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF8_RIGHTPAR, 5) == 0) { skipatstart += 7; options |= PCRE_UTF8; continue; } - else if (strncmp((char *)(ptr+skipatstart+2), STRING_UCP_RIGHTPAR, 4) == 0) +#endif +#ifdef COMPILE_PCRE16 + if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF16_RIGHTPAR, 6) == 0) + { skipatstart += 8; options |= PCRE_UTF16; continue; } +#endif +#ifdef COMPILE_PCRE32 + if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF32_RIGHTPAR, 6) == 0) + { skipatstart += 8; options |= PCRE_UTF32; continue; } +#endif + + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 4) == 0) + { skipatstart += 6; options |= PCRE_UTF8; continue; } + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UCP_RIGHTPAR, 4) == 0) { skipatstart += 6; options |= PCRE_UCP; continue; } - else if (strncmp((char *)(ptr+skipatstart+2), STRING_NO_START_OPT_RIGHTPAR, 13) == 0) + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_NO_START_OPT_RIGHTPAR, 13) == 0) { skipatstart += 15; options |= PCRE_NO_START_OPTIMIZE; continue; } - if (strncmp((char *)(ptr+skipatstart+2), STRING_CR_RIGHTPAR, 3) == 0) + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_LIMIT_MATCH_EQ, 12) == 0) + { + pcre_uint32 c = 0; + int p = skipatstart + 14; + while (isdigit(ptr[p])) + { + if (c > PCRE_UINT32_MAX / 10 - 1) break; /* Integer overflow */ + c = c*10 + ptr[p++] - CHAR_0; + } + if (ptr[p++] != CHAR_RIGHT_PARENTHESIS) break; + if (c < limit_match) + { + limit_match = c; + cd->external_flags |= PCRE_MLSET; + } + skipatstart = p; + continue; + } + + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_LIMIT_RECURSION_EQ, 16) == 0) + { + pcre_uint32 c = 0; + int p = skipatstart + 18; + while (isdigit(ptr[p])) + { + if (c > PCRE_UINT32_MAX / 10 - 1) break; /* Integer overflow check */ + c = c*10 + ptr[p++] - CHAR_0; + } + if (ptr[p++] != CHAR_RIGHT_PARENTHESIS) break; + if (c < limit_recursion) + { + limit_recursion = c; + cd->external_flags |= PCRE_RLSET; + } + skipatstart = p; + continue; + } + + if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_CR_RIGHTPAR, 3) == 0) { skipatstart += 5; newnl = PCRE_NEWLINE_CR; } - else if (strncmp((char *)(ptr+skipatstart+2), STRING_LF_RIGHTPAR, 3) == 0) + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_LF_RIGHTPAR, 3) == 0) { skipatstart += 5; newnl = PCRE_NEWLINE_LF; } - else if (strncmp((char *)(ptr+skipatstart+2), STRING_CRLF_RIGHTPAR, 5) == 0) + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_CRLF_RIGHTPAR, 5) == 0) { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; } - else if (strncmp((char *)(ptr+skipatstart+2), STRING_ANY_RIGHTPAR, 4) == 0) + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_ANY_RIGHTPAR, 4) == 0) { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; } - else if (strncmp((char *)(ptr+skipatstart+2), STRING_ANYCRLF_RIGHTPAR, 8) == 0) + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_ANYCRLF_RIGHTPAR, 8) == 0) { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; } - else if (strncmp((char *)(ptr+skipatstart+2), STRING_BSR_ANYCRLF_RIGHTPAR, 12) == 0) + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_BSR_ANYCRLF_RIGHTPAR, 12) == 0) { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; } - else if (strncmp((char *)(ptr+skipatstart+2), STRING_BSR_UNICODE_RIGHTPAR, 12) == 0) + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_BSR_UNICODE_RIGHTPAR, 12) == 0) { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; } if (newnl != 0) @@ -7401,22 +7934,34 @@ while (ptr[skipatstart] == CHAR_LEFT_PARENTHESIS && else break; } -utf8 = (options & PCRE_UTF8) != 0; +/* PCRE_UTF(16|32) have the same value as PCRE_UTF8. */ +utf = (options & PCRE_UTF8) != 0; +if (utf && never_utf) + { + errorcode = ERR78; + goto PCRE_EARLY_ERROR_RETURN2; + } -/* Can't support UTF8 unless PCRE has been compiled to include the code. The -return of an error code from _pcre_valid_utf8() is a new feature, introduced in +/* Can't support UTF unless PCRE has been compiled to include the code. The +return of an error code from PRIV(valid_utf)() is a new feature, introduced in release 8.13. It is passed back from pcre_[dfa_]exec(), but at the moment is not used here. */ -#ifdef SUPPORT_UTF8 -if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0 && - (errorcode = _pcre_valid_utf8((USPTR)pattern, -1, erroroffset)) != 0) +#ifdef SUPPORT_UTF +if (utf && (options & PCRE_NO_UTF8_CHECK) == 0 && + (errorcode = PRIV(valid_utf)((PCRE_PUCHAR)pattern, -1, erroroffset)) != 0) { +#if defined COMPILE_PCRE8 errorcode = ERR44; +#elif defined COMPILE_PCRE16 + errorcode = ERR74; +#elif defined COMPILE_PCRE32 + errorcode = ERR77; +#endif goto PCRE_EARLY_ERROR_RETURN2; } #else -if (utf8) +if (utf) { errorcode = ERR32; goto PCRE_EARLY_ERROR_RETURN; @@ -7492,7 +8037,10 @@ cd->backref_map = 0; /* Reflect pattern for debugging output */ DPRINTF(("------------------------------------------------------------------\n")); -DPRINTF(("%s\n", pattern)); +#ifdef PCRE_DEBUG +print_puchar(stdout, (PCRE_PUCHAR)pattern); +#endif +DPRINTF(("\n")); /* Pretend to compile the pattern while actually just accumulating the length of memory required. This behaviour is triggered by passing a non-NULL final @@ -7509,11 +8057,12 @@ cd->start_code = cworkspace; cd->hwm = cworkspace; cd->start_workspace = cworkspace; cd->workspace_size = COMPILE_WORK_SIZE; -cd->start_pattern = (const uschar *)pattern; -cd->end_pattern = (const uschar *)(pattern + strlen(pattern)); +cd->start_pattern = (const pcre_uchar *)pattern; +cd->end_pattern = (const pcre_uchar *)(pattern + STRLEN_UC((const pcre_uchar *)pattern)); cd->req_varyopt = 0; +cd->assert_depth = 0; +cd->max_lookbehind = 0; cd->external_options = options; -cd->external_flags = 0; cd->open_caps = NULL; /* Now do the pre-compile. On error, errorcode will be set non-zero, so we @@ -7526,11 +8075,12 @@ ptr += skipatstart; code = cworkspace; *code = OP_BRA; (void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE, - FALSE, 0, 0, &firstbyte, &reqbyte, NULL, cd, &length); + FALSE, 0, 0, &firstchar, &firstcharflags, &reqchar, &reqcharflags, NULL, + cd, &length); if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN; DPRINTF(("end pre-compile: length=%d workspace=%d\n", length, - cd->hwm - cworkspace)); + (int)(cd->hwm - cworkspace))); if (length > MAX_PATTERN_SIZE) { @@ -7543,8 +8093,8 @@ externally provided function. Integer overflow should because nowadays we limit the maximum value of cd->names_found and cd->name_entry_size. */ -size = length + sizeof(real_pcre) + cd->names_found * cd->name_entry_size; -re = (real_pcre *)(pcre_malloc)(size); +size = sizeof(REAL_PCRE) + (length + cd->names_found * cd->name_entry_size) * sizeof(pcre_uchar); +re = (REAL_PCRE *)(PUBL(malloc))(size); if (re == NULL) { @@ -7562,15 +8112,21 @@ re->magic_number = MAGIC_NUMBER; re->size = (int)size; re->options = cd->external_options; re->flags = cd->external_flags; -re->dummy1 = 0; -re->first_byte = 0; -re->req_byte = 0; -re->name_table_offset = sizeof(real_pcre); +re->limit_match = limit_match; +re->limit_recursion = limit_recursion; +re->first_char = 0; +re->req_char = 0; +re->name_table_offset = sizeof(REAL_PCRE) / sizeof(pcre_uchar); re->name_entry_size = cd->name_entry_size; re->name_count = cd->names_found; re->ref_count = 0; -re->tables = (tables == _pcre_default_tables)? NULL : tables; +re->tables = (tables == PRIV(default_tables))? NULL : tables; re->nullpad = NULL; +#ifdef COMPILE_PCRE32 +re->dummy = 0; +#else +re->dummy1 = re->dummy2 = re->dummy3 = 0; +#endif /* The starting points of the name/number translation table and of the code are passed around in the compile data block. The start/end pattern and initial @@ -7582,13 +8138,15 @@ field; this time it's used for remembering forward ref cd->final_bracount = cd->bracount; /* Save for checking forward references */ cd->assert_depth = 0; cd->bracount = 0; +cd->max_lookbehind = 0; cd->names_found = 0; -cd->name_table = (uschar *)re + re->name_table_offset; +cd->name_table = (pcre_uchar *)re + re->name_table_offset; codestart = cd->name_table + re->name_entry_size * re->name_count; cd->start_code = codestart; -cd->hwm = (uschar *)(cd->start_workspace); +cd->hwm = (pcre_uchar *)(cd->start_workspace); cd->req_varyopt = 0; cd->had_accept = FALSE; +cd->had_pruneorskip = FALSE; cd->check_lookbehind = FALSE; cd->open_caps = NULL; @@ -7596,20 +8154,25 @@ cd->open_caps = NULL; 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 + skipatstart; -code = (uschar *)codestart; +ptr = (const pcre_uchar *)pattern + skipatstart; +code = (pcre_uchar *)codestart; *code = OP_BRA; (void)compile_regex(re->options, &code, &ptr, &errorcode, FALSE, FALSE, 0, 0, - &firstbyte, &reqbyte, NULL, cd, NULL); + &firstchar, &firstcharflags, &reqchar, &reqcharflags, NULL, cd, NULL); re->top_bracket = cd->bracount; re->top_backref = cd->top_backref; -re->flags = cd->external_flags; +re->max_lookbehind = cd->max_lookbehind; +re->flags = cd->external_flags | PCRE_MODE; -if (cd->had_accept) reqbyte = REQ_NONE; /* Must disable after (*ACCEPT) */ +if (cd->had_accept) + { + reqchar = 0; /* Must disable after (*ACCEPT) */ + reqcharflags = REQ_NONE; + } /* If not reached end of pattern on success, there's an excess bracket. */ -if (errorcode == 0 && *ptr != 0) errorcode = ERR22; +if (errorcode == 0 && *ptr != CHAR_NULL) errorcode = ERR22; /* Fill in the terminating state and check for disastrous overflow, but if debugging, leave the test till after things are printed out. */ @@ -7620,13 +8183,20 @@ if debugging, leave the test till after things are pri if (code - codestart > length) errorcode = ERR23; #endif +#ifdef SUPPORT_VALGRIND +/* If the estimated length exceeds the really used length, mark the extra +allocated memory as unaddressable, so that any out-of-bound reads can be +detected. */ +VALGRIND_MAKE_MEM_NOACCESS(code, (length - (code - codestart)) * sizeof(pcre_uchar)); +#endif + /* Fill in any forward references that are required. There may be repeated references; optimize for them, as searching a large regex takes time. */ if (cd->hwm > cd->start_workspace) { int prev_recno = -1; - const uschar *groupptr = NULL; + const pcre_uchar *groupptr = NULL; while (errorcode == 0 && cd->hwm > cd->start_workspace) { int offset, recno; @@ -7635,18 +8205,18 @@ if (cd->hwm > cd->start_workspace) recno = GET(codestart, offset); if (recno != prev_recno) { - groupptr = _pcre_find_bracket(codestart, utf8, recno); + groupptr = PRIV(find_bracket)(codestart, utf, recno); prev_recno = recno; } if (groupptr == NULL) errorcode = ERR53; - else PUT(((uschar *)codestart), offset, (int)(groupptr - codestart)); + else PUT(((pcre_uchar *)codestart), offset, (int)(groupptr - codestart)); } } /* If the workspace had to be expanded, free the new memory. */ if (cd->workspace_size > COMPILE_WORK_SIZE) - (pcre_free)((void *)cd->start_workspace); + (PUBL(free))((void *)cd->start_workspace); /* Give an error if there's back reference to a non-existent capturing subpattern. */ @@ -7655,7 +8225,7 @@ if (errorcode == 0 && re->top_backref > re->top_bracke /* If there were any lookbehind assertions that contained OP_RECURSE (recursions or subroutine calls), a flag is set for them to be checked here, -because they may contain forward references. Actual recursions can't be fixed +because they may contain forward references. Actual recursions cannot be fixed length, but subroutine calls can. It is done like this so that those without OP_RECURSE that are not fixed length get a diagnosic with a useful offset. The exceptional ones forgo this. We scan the pattern to check that they are fixed @@ -7663,21 +8233,21 @@ length, and set their lengths. */ if (cd->check_lookbehind) { - uschar *cc = (uschar *)codestart; + pcre_uchar *cc = (pcre_uchar *)codestart; /* Loop, searching for OP_REVERSE items, and process those that do not have their length set. (Actually, it will also re-process any that have a length of zero, but that is a pathological case, and it does no harm.) When we find one, we temporarily terminate the branch it is in while we scan it. */ - for (cc = (uschar *)_pcre_find_bracket(codestart, utf8, -1); + for (cc = (pcre_uchar *)PRIV(find_bracket)(codestart, utf, -1); cc != NULL; - cc = (uschar *)_pcre_find_bracket(cc, utf8, -1)) + cc = (pcre_uchar *)PRIV(find_bracket)(cc, utf, -1)) { if (GET(cc, 1) == 0) { int fixed_length; - uschar *be = cc - 1 - LINK_SIZE + GET(cc, -LINK_SIZE); + pcre_uchar *be = cc - 1 - LINK_SIZE + GET(cc, -LINK_SIZE); int end_op = *be; *be = OP_END; fixed_length = find_fixedlength(cc, (re->options & PCRE_UTF8) != 0, TRUE, @@ -7690,6 +8260,7 @@ if (cd->check_lookbehind) (fixed_length == -4)? ERR70 : ERR25; break; } + if (fixed_length > cd->max_lookbehind) cd->max_lookbehind = fixed_length; PUT(cc, 1, fixed_length); } cc += 1 + LINK_SIZE; @@ -7700,9 +8271,9 @@ if (cd->check_lookbehind) if (errorcode != 0) { - (pcre_free)(re); + (PUBL(free))(re); PCRE_EARLY_ERROR_RETURN: - *erroroffset = (int)(ptr - (const uschar *)pattern); + *erroroffset = (int)(ptr - (const pcre_uchar *)pattern); PCRE_EARLY_ERROR_RETURN2: *errorptr = find_error_text(errorcode); if (errorcodeptr != NULL) *errorcodeptr = errorcode; @@ -7710,32 +8281,57 @@ if (errorcode != 0) } /* If the anchored option was not passed, set the flag if we can determine that -the pattern is anchored by virtue of ^ characters or \A or anything else (such -as starting with .* when DOTALL is set). +the pattern is anchored by virtue of ^ characters or \A or anything else, such +as starting with non-atomic .* when DOTALL is set and there are no occurrences +of *PRUNE or *SKIP. Otherwise, if we know what the first byte has to be, save it, because that speeds up unanchored matches no end. If not, see if we can set the PCRE_STARTLINE flag. This is helpful for multiline matches when all branches -start with ^. and also when all branches start with .* for non-DOTALL matches. -*/ +start with ^. and also when all branches start with non-atomic .* for +non-DOTALL matches when *PRUNE and SKIP are not present. */ if ((re->options & PCRE_ANCHORED) == 0) { - if (is_anchored(codestart, 0, cd->backref_map)) - re->options |= PCRE_ANCHORED; + if (is_anchored(codestart, 0, cd, 0)) re->options |= PCRE_ANCHORED; else { - if (firstbyte < 0) - firstbyte = find_firstassertedchar(codestart, FALSE); - if (firstbyte >= 0) /* Remove caseless flag for non-caseable chars */ + if (firstcharflags < 0) + firstchar = find_firstassertedchar(codestart, &firstcharflags, FALSE); + if (firstcharflags >= 0) /* Remove caseless flag for non-caseable chars */ { - int ch = firstbyte & 255; - re->first_byte = ((firstbyte & REQ_CASELESS) != 0 && - cd->fcc[ch] == ch)? ch : firstbyte; +#if defined COMPILE_PCRE8 + re->first_char = firstchar & 0xff; +#elif defined COMPILE_PCRE16 + re->first_char = firstchar & 0xffff; +#elif defined COMPILE_PCRE32 + re->first_char = firstchar; +#endif + if ((firstcharflags & REQ_CASELESS) != 0) + { +#if defined SUPPORT_UCP && !(defined COMPILE_PCRE8) + /* We ignore non-ASCII first chars in 8 bit mode. */ + if (utf) + { + if (re->first_char < 128) + { + if (cd->fcc[re->first_char] != re->first_char) + re->flags |= PCRE_FCH_CASELESS; + } + else if (UCD_OTHERCASE(re->first_char) != re->first_char) + re->flags |= PCRE_FCH_CASELESS; + } + else +#endif + if (MAX_255(re->first_char) + && cd->fcc[re->first_char] != re->first_char) + re->flags |= PCRE_FCH_CASELESS; + } + re->flags |= PCRE_FIRSTSET; } - else if (is_startline(codestart, 0, cd->backref_map)) - re->flags |= PCRE_STARTLINE; + + else if (is_startline(codestart, 0, cd, 0)) re->flags |= PCRE_STARTLINE; } } @@ -7743,12 +8339,36 @@ if ((re->options & PCRE_ANCHORED) == 0) variable length item in the regex. Remove the caseless flag for non-caseable bytes. */ -if (reqbyte >= 0 && - ((re->options & PCRE_ANCHORED) == 0 || (reqbyte & REQ_VARY) != 0)) +if (reqcharflags >= 0 && + ((re->options & PCRE_ANCHORED) == 0 || (reqcharflags & REQ_VARY) != 0)) { - int ch = reqbyte & 255; - re->req_byte = ((reqbyte & REQ_CASELESS) != 0 && - cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte; +#if defined COMPILE_PCRE8 + re->req_char = reqchar & 0xff; +#elif defined COMPILE_PCRE16 + re->req_char = reqchar & 0xffff; +#elif defined COMPILE_PCRE32 + re->req_char = reqchar; +#endif + if ((reqcharflags & REQ_CASELESS) != 0) + { +#if defined SUPPORT_UCP && !(defined COMPILE_PCRE8) + /* We ignore non-ASCII first chars in 8 bit mode. */ + if (utf) + { + if (re->req_char < 128) + { + if (cd->fcc[re->req_char] != re->req_char) + re->flags |= PCRE_RCH_CASELESS; + } + else if (UCD_OTHERCASE(re->req_char) != re->req_char) + re->flags |= PCRE_RCH_CASELESS; + } + else +#endif + if (MAX_255(re->req_char) && cd->fcc[re->req_char] != re->req_char) + re->flags |= PCRE_RCH_CASELESS; + } + re->flags |= PCRE_REQCHSET; } @@ -7763,38 +8383,50 @@ printf("Options=%08x\n", re->options); if ((re->flags & PCRE_FIRSTSET) != 0) { - int ch = re->first_byte & 255; - const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)? - "" : " (caseless)"; - if (isprint(ch)) printf("First char = %c%s\n", ch, caseless); + pcre_uchar ch = re->first_char; + const char *caseless = + ((re->flags & PCRE_FCH_CASELESS) == 0)? "" : " (caseless)"; + if (PRINTABLE(ch)) printf("First char = %c%s\n", ch, caseless); else printf("First char = \\x%02x%s\n", ch, caseless); } if ((re->flags & PCRE_REQCHSET) != 0) { - int ch = re->req_byte & 255; - const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)? - "" : " (caseless)"; - if (isprint(ch)) printf("Req char = %c%s\n", ch, caseless); + pcre_uchar ch = re->req_char; + const char *caseless = + ((re->flags & PCRE_RCH_CASELESS) == 0)? "" : " (caseless)"; + if (PRINTABLE(ch)) printf("Req char = %c%s\n", ch, caseless); else printf("Req char = \\x%02x%s\n", ch, caseless); } -pcre_printint(re, stdout, TRUE); +#if defined COMPILE_PCRE8 +pcre_printint((pcre *)re, stdout, TRUE); +#elif defined COMPILE_PCRE16 +pcre16_printint((pcre *)re, stdout, TRUE); +#elif defined COMPILE_PCRE32 +pcre32_printint((pcre *)re, stdout, TRUE); +#endif /* This check is done here in the debugging case so that the code that was compiled can be seen. */ if (code - codestart > length) { - (pcre_free)(re); + (PUBL(free))(re); *errorptr = find_error_text(ERR23); - *erroroffset = ptr - (uschar *)pattern; + *erroroffset = ptr - (pcre_uchar *)pattern; if (errorcodeptr != NULL) *errorcodeptr = ERR23; return NULL; } #endif /* PCRE_DEBUG */ +#if defined COMPILE_PCRE8 return (pcre *)re; +#elif defined COMPILE_PCRE16 +return (pcre16 *)re; +#elif defined COMPILE_PCRE32 +return (pcre32 *)re; +#endif } /* End of pcre_compile.c */