--- embedaddon/pcre/pcre_compile.c 2012/02/21 23:05:51 1.1.1.1 +++ embedaddon/pcre/pcre_compile.c 2014/06/15 19:46:03 1.1.1.5 @@ -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 * *************************************************/ @@ -99,12 +115,30 @@ kicks in at the same number of forward references in a #define COMPILE_WORK_SIZE (2048*LINK_SIZE) #define COMPILE_WORK_SIZE_MAX (100*COMPILE_WORK_SIZE) +/* This value determines the size of the initial vector that is used for +remembering named groups during the pre-compile. It is allocated on the stack, +but if it is too small, it is expanded using malloc(), in a similar way to the +workspace. The value is the number of slots in the list. */ + +#define NAMED_GROUP_LIST_SIZE 20 + /* The overrun tests check for a slightly smaller size so that they detect the overrun before it actually does run off the end of the data block. */ #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 @@ -226,11 +260,25 @@ static const verbitem verbs[] = { static const int verbcount = sizeof(verbs)/sizeof(verbitem); +/* Substitutes for [[:<:]] and [[:>:]], which mean start and end of word in +another regex library. */ + +static const pcre_uchar sub_start_of_word[] = { + CHAR_BACKSLASH, CHAR_b, CHAR_LEFT_PARENTHESIS, CHAR_QUESTION_MARK, + CHAR_EQUALS_SIGN, CHAR_BACKSLASH, CHAR_w, CHAR_RIGHT_PARENTHESIS, '\0' }; + +static const pcre_uchar sub_end_of_word[] = { + CHAR_BACKSLASH, CHAR_b, CHAR_LEFT_PARENTHESIS, CHAR_QUESTION_MARK, + CHAR_LESS_THAN_SIGN, CHAR_EQUALS_SIGN, CHAR_BACKSLASH, CHAR_w, + CHAR_RIGHT_PARENTHESIS, '\0' }; + + /* Tables of names of POSIX character classes and their lengths. The names are now all in a single string, to reduce the number of relocations when a shared library is dynamically loaded. The list of lengths is terminated by a zero length entry. The first three must be alpha, lower, upper, as this is assumed -for handling case independence. */ +for handling case independence. The indices for graph, print, and punct are +needed, so identify them. */ static const char posix_names[] = STRING_alpha0 STRING_lower0 STRING_upper0 STRING_alnum0 @@ -238,9 +286,14 @@ 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 }; +#define PC_GRAPH 8 +#define PC_PRINT 9 +#define PC_PUNCT 10 + + /* Table of class bit maps for each POSIX class. Each class is formed from a base map, with an optional addition or removal of another map. Then, for some classes, there is some additional tweaking: for [:blank:] the vertical space @@ -268,52 +321,111 @@ static const int posix_class_maps[] = { cbit_xdigit,-1, 0 /* xdigit */ }; -/* Table of substitutes for \d etc when PCRE_UCP is set. The POSIX class -substitutes must be in the order of the names, defined above, and there are -both positive and negative cases. NULL means no substitute. */ +/* Table of substitutes for \d etc when PCRE_UCP is set. They are replaced by +Unicode property escapes. */ #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 */ /* Xsp is Perl space, but from 8.34, Perl */ + string_pXsp, /* \s */ /* space and POSIX space are the same. */ + 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 */ +/* The POSIX class substitutes must be in the order of the POSIX class names, +defined above, and there are both positive and negative cases. NULL means no +general substitute of a Unicode property escape (\p or \P). However, for some +POSIX classes (e.g. graph, print, punct) a special property code is compiled +directly. */ + +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 */ /* Xps is POSIX space, but from 8.34 */ + string_pXwd, /* word */ /* Perl and POSIX space are the same */ + 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 */ /* Xps is POSIX space, but from 8.34 */ + string_PXwd, /* ^word */ /* Perl and POSIX space are the same */ + NULL /* ^xdigit */ }; -#define POSIX_SUBSIZE (sizeof(posix_substitutes)/sizeof(uschar *)) +#define POSIX_SUBSIZE (sizeof(posix_substitutes) / sizeof(pcre_uchar *)) #endif #define STRING(a) # a @@ -372,9 +484,9 @@ 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" + "character value in \\x{} or \\o{} is too large\0" /* 35 */ "invalid condition (?(0)\0" "\\C not allowed in lookbehind assertion\0" @@ -395,7 +507,7 @@ 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 +518,7 @@ 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 +526,27 @@ 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" + "non-hex character in \\x{} (closing brace missing?)\0" + /* 80 */ + "non-octal character in \\o{} (closing brace missing?)\0" + "missing opening brace after \\o\0" + "parentheses are too deeply nested\0" + "invalid range in character class\0" + "group name must start with a non-digit\0" ; /* Table to identify digits and hex digits. This is used when compiling @@ -439,12 +565,18 @@ 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 +615,7 @@ 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 +650,7 @@ 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 +686,184 @@ static const unsigned char ebcdic_chartab[] = { /* cha #endif -/* Definition to allow mutual recursion */ +/* This table is used to check whether auto-possessification is possible +between adjacent character-type opcodes. The left-hand (repeated) opcode is +used to select the row, and the right-hand opcode is use to select the column. +A value of 1 means that auto-possessification is OK. For example, the second +value in the first row means that \D+\d can be turned into \D++\d. -static BOOL - compile_regex(int, uschar **, const uschar **, int *, BOOL, BOOL, int, int, - int *, int *, branch_chain *, compile_data *, int *); +The Unicode property types (\P and \p) have to be present to fill out the table +because of what their opcode values are, but the table values should always be +zero because property types are handled separately in the code. The last four +columns apply to items that cannot be repeated, so there is no need to have +rows for them. 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. */ +#define APTROWS (LAST_AUTOTAB_LEFT_OP - FIRST_AUTOTAB_OP + 1) +#define APTCOLS (LAST_AUTOTAB_RIGHT_OP - FIRST_AUTOTAB_OP + 1) +static const pcre_uint8 autoposstab[APTROWS][APTCOLS] = { +/* \D \d \S \s \W \w . .+ \C \P \p \R \H \h \V \v \X \Z \z $ $M */ + { 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 }, /* \D */ + { 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 1 }, /* \d */ + { 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 1 }, /* \S */ + { 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 }, /* \s */ + { 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 }, /* \W */ + { 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 1 }, /* \w */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0 }, /* . */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 }, /* .+ */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 }, /* \C */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* \P */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* \p */ + { 0, 1, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0 }, /* \R */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0 }, /* \H */ + { 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 0, 1, 0, 0 }, /* \h */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 1, 0, 0 }, /* \V */ + { 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 0 }, /* \v */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 } /* \X */ +}; + +/* This table is used to check whether auto-possessification is possible +between adjacent Unicode property opcodes (OP_PROP and OP_NOTPROP). The +left-hand (repeated) opcode is used to select the row, and the right-hand +opcode is used to select the column. The values are as follows: + + 0 Always return FALSE (never auto-possessify) + 1 Character groups are distinct (possessify if both are OP_PROP) + 2 Check character categories in the same group (general or particular) + 3 TRUE if the two opcodes are not the same (PROP vs NOTPROP) + + 4 Check left general category vs right particular category + 5 Check right general category vs left particular category + + 6 Left alphanum vs right general category + 7 Left space vs right general category + 8 Left word vs right general category + + 9 Right alphanum vs left general category + 10 Right space vs left general category + 11 Right word vs left general category + + 12 Left alphanum vs right particular category + 13 Left space vs right particular category + 14 Left word vs right particular category + + 15 Right alphanum vs left particular category + 16 Right space vs left particular category + 17 Right word vs left particular category +*/ + +static const pcre_uint8 propposstab[PT_TABSIZE][PT_TABSIZE] = { +/* ANY LAMP GC PC SC ALNUM SPACE PXSPACE WORD CLIST UCNC */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* PT_ANY */ + { 0, 3, 0, 0, 0, 3, 1, 1, 0, 0, 0 }, /* PT_LAMP */ + { 0, 0, 2, 4, 0, 9, 10, 10, 11, 0, 0 }, /* PT_GC */ + { 0, 0, 5, 2, 0, 15, 16, 16, 17, 0, 0 }, /* PT_PC */ + { 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0 }, /* PT_SC */ + { 0, 3, 6, 12, 0, 3, 1, 1, 0, 0, 0 }, /* PT_ALNUM */ + { 0, 1, 7, 13, 0, 1, 3, 3, 1, 0, 0 }, /* PT_SPACE */ + { 0, 1, 7, 13, 0, 1, 3, 3, 1, 0, 0 }, /* PT_PXSPACE */ + { 0, 0, 8, 14, 0, 0, 1, 1, 3, 0, 0 }, /* PT_WORD */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* PT_CLIST */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3 } /* PT_UCNC */ +}; + +/* This table is used to check whether auto-possessification is possible +between adjacent Unicode property opcodes (OP_PROP and OP_NOTPROP) when one +specifies a general category and the other specifies a particular category. The +row is selected by the general category and the column by the particular +category. The value is 1 if the particular category is not part of the general +category. */ + +static const pcre_uint8 catposstab[7][30] = { +/* Cc Cf Cn Co Cs Ll Lm Lo Lt Lu Mc Me Mn Nd Nl No Pc Pd Pe Pf Pi Po Ps Sc Sk Sm So Zl Zp Zs */ + { 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 }, /* C */ + { 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 }, /* L */ + { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 }, /* M */ + { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 }, /* N */ + { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1 }, /* P */ + { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1 }, /* S */ + { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0 } /* Z */ +}; + +/* This table is used when checking ALNUM, (PX)SPACE, SPACE, and WORD against +a general or particular category. The properties in each row are those +that apply to the character set in question. Duplication means that a little +unnecessary work is done when checking, but this keeps things much simpler +because they can all use the same code. For more details see the comment where +this table is used. + +Note: SPACE and PXSPACE used to be different because Perl excluded VT from +"space", but from Perl 5.18 it's included, so both categories are treated the +same here. */ + +static const pcre_uint8 posspropstab[3][4] = { + { ucp_L, ucp_N, ucp_N, ucp_Nl }, /* ALNUM, 3rd and 4th values redundant */ + { ucp_Z, ucp_Z, ucp_C, ucp_Cc }, /* SPACE and PXSPACE, 2nd value redundant */ + { ucp_L, ucp_N, ucp_P, ucp_Po } /* WORD */ +}; + +/* This table is used when converting repeating opcodes into possessified +versions as a result of an explicit possessive quantifier such as ++. A zero +value means there is no possessified version - in those cases the item in +question must be wrapped in ONCE brackets. The table is truncated at OP_CALLOUT +because all relevant opcodes are less than that. */ + +static const pcre_uint8 opcode_possessify[] = { + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 0 - 15 */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 16 - 31 */ + + 0, /* NOTI */ + OP_POSSTAR, 0, /* STAR, MINSTAR */ + OP_POSPLUS, 0, /* PLUS, MINPLUS */ + OP_POSQUERY, 0, /* QUERY, MINQUERY */ + OP_POSUPTO, 0, /* UPTO, MINUPTO */ + 0, /* EXACT */ + 0, 0, 0, 0, /* POS{STAR,PLUS,QUERY,UPTO} */ + + OP_POSSTARI, 0, /* STARI, MINSTARI */ + OP_POSPLUSI, 0, /* PLUSI, MINPLUSI */ + OP_POSQUERYI, 0, /* QUERYI, MINQUERYI */ + OP_POSUPTOI, 0, /* UPTOI, MINUPTOI */ + 0, /* EXACTI */ + 0, 0, 0, 0, /* POS{STARI,PLUSI,QUERYI,UPTOI} */ + + OP_NOTPOSSTAR, 0, /* NOTSTAR, NOTMINSTAR */ + OP_NOTPOSPLUS, 0, /* NOTPLUS, NOTMINPLUS */ + OP_NOTPOSQUERY, 0, /* NOTQUERY, NOTMINQUERY */ + OP_NOTPOSUPTO, 0, /* NOTUPTO, NOTMINUPTO */ + 0, /* NOTEXACT */ + 0, 0, 0, 0, /* NOTPOS{STAR,PLUS,QUERY,UPTO} */ + + OP_NOTPOSSTARI, 0, /* NOTSTARI, NOTMINSTARI */ + OP_NOTPOSPLUSI, 0, /* NOTPLUSI, NOTMINPLUSI */ + OP_NOTPOSQUERYI, 0, /* NOTQUERYI, NOTMINQUERYI */ + OP_NOTPOSUPTOI, 0, /* NOTUPTOI, NOTMINUPTOI */ + 0, /* NOTEXACTI */ + 0, 0, 0, 0, /* NOTPOS{STARI,PLUSI,QUERYI,UPTOI} */ + + OP_TYPEPOSSTAR, 0, /* TYPESTAR, TYPEMINSTAR */ + OP_TYPEPOSPLUS, 0, /* TYPEPLUS, TYPEMINPLUS */ + OP_TYPEPOSQUERY, 0, /* TYPEQUERY, TYPEMINQUERY */ + OP_TYPEPOSUPTO, 0, /* TYPEUPTO, TYPEMINUPTO */ + 0, /* TYPEEXACT */ + 0, 0, 0, 0, /* TYPEPOS{STAR,PLUS,QUERY,UPTO} */ + + OP_CRPOSSTAR, 0, /* CRSTAR, CRMINSTAR */ + OP_CRPOSPLUS, 0, /* CRPLUS, CRMINPLUS */ + OP_CRPOSQUERY, 0, /* CRQUERY, CRMINQUERY */ + OP_CRPOSRANGE, 0, /* CRRANGE, CRMINRANGE */ + 0, 0, 0, 0, /* CRPOS{STAR,PLUS,QUERY,RANGE} */ + + 0, 0, 0, /* CLASS, NCLASS, XCLASS */ + 0, 0, /* REF, REFI */ + 0, 0, /* DNREF, DNREFI */ + 0, 0 /* RECURSE, CALLOUT */ +}; + + + /************************************************* * Find an error text * *************************************************/ @@ -581,13 +883,14 @@ 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; } + /************************************************* * Expand the workspace * *************************************************/ @@ -604,7 +907,7 @@ 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 +915,12 @@ 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 +944,19 @@ 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 +968,67 @@ 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 a returned 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 +1045,15 @@ 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 +1062,18 @@ 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 +1100,13 @@ 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 +1114,12 @@ 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 +1134,22 @@ 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 +1160,7 @@ else break; } - if (c == 0) + if (s == 0) { *errorcodeptr = ERR58; break; @@ -832,90 +1168,141 @@ 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 - starting with one that is not zero is not straightforward. By experiment, - the way Perl works seems to be as follows: + starting with one that is not zero is not straightforward. Perl has changed + over the years. Nowadays \g{} for backreferences and \o{} for octal are + recommended to avoid the ambiguities in the old syntax. Outside a character class, the digits are read as a decimal number. If the - number is less than 10, or if there are that many previous extracting - left brackets, then it is a back reference. Otherwise, up to three octal - digits are read to form an escaped byte. Thus \123 is likely to be octal - 123 (cf \0123, which is octal 012 followed by the literal 3). If the octal - value is greater than 377, the least significant 8 bits are taken. Inside a - character class, \ followed by a digit is always an octal number. */ + number is less than 8 (used to be 10), or if there are that many previous + extracting left brackets, then it is a back reference. Otherwise, up to + three octal digits are read to form an escaped byte. Thus \123 is likely to + be octal 123 (cf \0123, which is octal 012 followed by the literal 3). If + the octal value is greater than 377, the least significant 8 bits are + taken. \8 and \9 are treated as the literal characters 8 and 9. + Inside a character class, \ followed by a digit is always either a literal + 8 or 9 or an octal number. */ + 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: 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 < 8 || s <= bracount) /* Check for back reference */ { - c = -(ESC_REF + c); + escape = -s; break; } ptr = oldptr; /* Put the pointer back and fall through */ } - /* Handle an octal number following \. If the first digit is 8 or 9, Perl - generates a binary zero byte and treats the digit as a following literal. - Thus we have to pull back the pointer by one. */ + /* Handle a digit following \ when the number is not a back reference. If + the first digit is 8 or 9, Perl used to generate a binary zero byte and + then treat the digit as a following literal. At least by Perl 5.18 this + changed so as not to insert the binary zero. */ - if ((c = *ptr) >= CHAR_8) - { - ptr--; - c = 0; - break; - } + if ((c = *ptr) >= CHAR_8) break; + /* Fall through with a digit less than 8 */ + /* \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. */ + /* \o is a relatively new Perl feature, supporting a more general way of + specifying character codes in octal. The only supported form is \o{ddd}. */ + case CHAR_o: + if (ptr[1] != CHAR_LEFT_CURLY_BRACKET) *errorcodeptr = ERR81; else + { + ptr += 2; + c = 0; + overflow = FALSE; + while (*ptr >= CHAR_0 && *ptr <= CHAR_7) + { + register pcre_uint32 cc = *ptr++; + if (c == 0 && cc == CHAR_0) continue; /* Leading zeroes */ +#ifdef COMPILE_PCRE32 + if (c >= 0x20000000l) { overflow = TRUE; break; } +#endif + c = (c << 3) + cc - CHAR_0 ; +#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 (*ptr >= CHAR_0 && *ptr <= CHAR_7) ptr++; + *errorcodeptr = ERR34; + } + else if (*ptr == CHAR_RIGHT_CURLY_BRACKET) + { + if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73; + } + else *errorcodeptr = ERR80; + } + break; + + /* \x is complicated. In JavaScript, \x must be followed by two hexadecimal + numbers. Otherwise it is a lowercase x letter. */ + 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)); @@ -925,56 +1312,86 @@ else #endif } } - break; - } + } /* End JavaScript handling */ - if (ptr[1] == CHAR_LEFT_CURLY_BRACKET) - { - const uschar *pt = ptr + 2; - int count = 0; + /* Handle \x in Perl's style. \x{ddd} is a character number which can be + greater than 0xff in utf or non-8bit mode, but only if the ddd are hex + digits. If not, { used to be treated as a data character. However, Perl + seems to read hex digits up to the first non-such, and ignore the rest, so + that, for example \x{zz} matches a binary zero. This seems crazy, so PCRE + now gives an error. */ - c = 0; - while ((digitab[*pt] & ctype_xdigit) != 0) + else + { + if (ptr[1] == CHAR_LEFT_CURLY_BRACKET) { - register int cc = *pt++; - if (c == 0 && cc == CHAR_0) continue; /* Leading zeroes */ - count++; + ptr += 2; + c = 0; + overflow = FALSE; + while (MAX_255(*ptr) && (digitab[*ptr] & ctype_xdigit) != 0) + { + register pcre_uint32 cc = *ptr++; + if (c == 0 && cc == CHAR_0) continue; /* Leading zeroes */ +#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)); + if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */ + c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10)); #else /* EBCDIC coding */ - 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)); + 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 (*pt == CHAR_RIGHT_CURLY_BRACKET) - { - if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34; - ptr = pt; - break; - } +#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 the sequence of hex digits does not end with '}', then we don't - recognize this construct; fall through to the normal \x handling. */ - } + if (overflow) + { + while (MAX_255(*ptr) && (digitab[*ptr] & ctype_xdigit) != 0) ptr++; + *errorcodeptr = ERR34; + } - /* Read just a single-byte hex-defined char */ + else if (*ptr == CHAR_RIGHT_CURLY_BRACKET) + { + if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73; + } - c = 0; - while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0) - { - int cc; /* Some compilers don't like */ - cc = *(++ptr); /* ++ in initializers */ + /* If the sequence of hex digits does not end with '}', give an error. + We used just to recognize this construct and fall through to the normal + \x handling, but nowadays Perl gives an error, which seems much more + sensible, so we do too. */ + + else *errorcodeptr = ERR79; + } /* End of \x{} processing */ + + /* Read a single-byte hex-defined char (up to two hex digits after \x) */ + + else + { + c = 0; + while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0) + { + 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 */ - c = c * 16 + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10)); + if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */ + c = c * 16 + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10)); #else /* EBCDIC coding */ - if (cc <= CHAR_z) cc += 64; /* Convert to upper case */ - c = c * 16 + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10)); + if (cc <= CHAR_z) cc += 64; /* Convert to upper case */ + c = c * 16 + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10)); #endif - } + } + } /* End of \xdd handling */ + } /* End of Perl-style \x handling */ break; /* For \c, a following letter is upper-cased; then the 0x40 bit is flipped. @@ -984,7 +1401,7 @@ else case CHAR_c: c = *(++ptr); - if (c == 0) + if (c == CHAR_NULL) { *errorcodeptr = ERR2; break; @@ -1024,19 +1441,20 @@ 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; } @@ -1054,21 +1472,24 @@ 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 +1503,10 @@ 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; } @@ -1106,34 +1527,35 @@ 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 - /************************************************* * Read repeat counts * *************************************************/ @@ -1153,8 +1575,8 @@ 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 +1584,7 @@ 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 +1599,7 @@ 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; @@ -1202,301 +1624,6 @@ return p; /************************************************* -* Subroutine for finding forward reference * -*************************************************/ - -/* This recursive function is called only from find_parens() below. The -top-level call starts at the beginning of the pattern. All other calls must -start at a parenthesis. It scans along a pattern's text looking for capturing -subpatterns, and counting them. If it finds a named pattern that matches the -name it is given, it returns its number. Alternatively, if the name is NULL, it -returns when it reaches a given numbered subpattern. Recursion is used to keep -track of subpatterns that reset the capturing group numbers - the (?| feature. - -This function was originally called only from the second pass, in which we know -that if (?< or (?' or (?P< is encountered, the name will be correctly -terminated because that is checked in the first pass. There is now one call to -this function in the first pass, to check for a recursive back reference by -name (so that we can make the whole group atomic). In this case, we need check -only up to the current position in the pattern, and that is still OK because -and previous occurrences will have been checked. To make this work, the test -for "end of pattern" is a check against cd->end_pattern in the main loop, -instead of looking for a binary zero. This means that the special first-pass -call can adjust cd->end_pattern temporarily. (Checks for binary zero while -processing items within the loop are OK, because afterwards the main loop will -terminate.) - -Arguments: - ptrptr address of the current character pointer (updated) - cd compile background data - 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 - 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) -{ -uschar *ptr = *ptrptr; -int start_count = *count; -int hwm_count = start_count; -BOOL dup_parens = FALSE; - -/* If the first character is a parenthesis, check on the type of group we are -dealing with. The very first call may not start with a parenthesis. */ - -if (ptr[0] == CHAR_LEFT_PARENTHESIS) - { - /* Handle specials such as (*SKIP) or (*UTF8) etc. */ - - if (ptr[1] == CHAR_ASTERISK) ptr += 2; - - /* Handle a normal, unnamed capturing parenthesis. */ - - else if (ptr[1] != CHAR_QUESTION_MARK) - { - *count += 1; - if (name == NULL && *count == lorn) return *count; - ptr++; - } - - /* All cases now have (? at the start. Remember when we are in a group - where the parenthesis numbers are duplicated. */ - - else if (ptr[2] == CHAR_VERTICAL_LINE) - { - ptr += 3; - dup_parens = TRUE; - } - - /* Handle comments; all characters are allowed until a ket is reached. */ - - else if (ptr[2] == CHAR_NUMBER_SIGN) - { - for (ptr += 3; *ptr != 0; ptr++) if (*ptr == CHAR_RIGHT_PARENTHESIS) break; - goto FAIL_EXIT; - } - - /* Handle a condition. If it is an assertion, just carry on so that it - is processed as normal. If not, skip to the closing parenthesis of the - condition (there can't be any nested parens). */ - - else if (ptr[2] == CHAR_LEFT_PARENTHESIS) - { - ptr += 2; - if (ptr[1] != CHAR_QUESTION_MARK) - { - while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++; - if (*ptr != 0) ptr++; - } - } - - /* Start with (? but not a condition. */ - - else - { - ptr += 2; - if (*ptr == CHAR_P) ptr++; /* Allow optional P */ - - /* We have to disambiguate (? for named groups */ - - if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK && - ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE) - { - int term; - const uschar *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) - return *count; - term++; - } - } - } - -/* Past any initial parenthesis handling, scan for parentheses or vertical -bars. Stop if we get to cd->end_pattern. Note that this is important for the -first-pass call when this value is temporarily adjusted to stop at the current -position. So DO NOT change this to a test for binary zero. */ - -for (; ptr < cd->end_pattern; ptr++) - { - /* Skip over backslashed characters and also entire \Q...\E */ - - if (*ptr == CHAR_BACKSLASH) - { - if (*(++ptr) == 0) goto FAIL_EXIT; - if (*ptr == CHAR_Q) for (;;) - { - while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {}; - if (*ptr == 0) goto FAIL_EXIT; - if (*(++ptr) == CHAR_E) break; - } - continue; - } - - /* Skip over character classes; this logic must be similar to the way they - are handled for real. If the first character is '^', skip it. Also, if the - first few characters (either before or after ^) are \Q\E or \E we skip them - too. This makes for compatibility with Perl. Note the use of STR macros to - encode "Q\\E" so that it works in UTF-8 on EBCDIC platforms. */ - - if (*ptr == CHAR_LEFT_SQUARE_BRACKET) - { - BOOL negate_class = FALSE; - for (;;) - { - if (ptr[1] == CHAR_BACKSLASH) - { - if (ptr[2] == CHAR_E) - ptr+= 2; - else if (strncmp((const char *)ptr+2, - STR_Q STR_BACKSLASH STR_E, 3) == 0) - ptr += 4; - else - break; - } - else if (!negate_class && ptr[1] == CHAR_CIRCUMFLEX_ACCENT) - { - negate_class = TRUE; - ptr++; - } - else break; - } - - /* If the next character is ']', it is a data character that must be - skipped, except in JavaScript compatibility mode. */ - - if (ptr[1] == CHAR_RIGHT_SQUARE_BRACKET && - (cd->external_options & PCRE_JAVASCRIPT_COMPAT) == 0) - ptr++; - - while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET) - { - if (*ptr == 0) return -1; - if (*ptr == CHAR_BACKSLASH) - { - if (*(++ptr) == 0) goto FAIL_EXIT; - if (*ptr == CHAR_Q) for (;;) - { - while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {}; - if (*ptr == 0) goto FAIL_EXIT; - if (*(++ptr) == CHAR_E) break; - } - continue; - } - } - continue; - } - - /* Skip comments in /x mode */ - - if (xmode && *ptr == CHAR_NUMBER_SIGN) - { - ptr++; - while (*ptr != 0) - { - if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; } - ptr++; -#ifdef SUPPORT_UTF8 - if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++; -#endif - } - if (*ptr == 0) goto FAIL_EXIT; - continue; - } - - /* Check for the special metacharacters */ - - if (*ptr == CHAR_LEFT_PARENTHESIS) - { - int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, count); - if (rc > 0) return rc; - if (*ptr == 0) goto FAIL_EXIT; - } - - else if (*ptr == CHAR_RIGHT_PARENTHESIS) - { - if (dup_parens && *count < hwm_count) *count = hwm_count; - goto FAIL_EXIT; - } - - else if (*ptr == CHAR_VERTICAL_LINE && dup_parens) - { - if (*count > hwm_count) hwm_count = *count; - *count = start_count; - } - } - -FAIL_EXIT: -*ptrptr = ptr; -return -1; -} - - - - -/************************************************* -* Find forward referenced subpattern * -*************************************************/ - -/* This function scans along a pattern's text looking for capturing -subpatterns, and counting them. If it finds a named pattern that matches the -name it is given, it returns its number. Alternatively, if the name is NULL, it -returns when it reaches a given numbered subpattern. This is used for forward -references to subpatterns. We used to be able to start this scan from the -current compiling point, using the current count value from cd->bracount, and -do it all in a single loop, but the addition of the possibility of duplicate -subpattern numbers means that we have to scan from the very start, in order to -take account of such duplicates, and to use a recursive function to keep track -of the different types of group. - -Arguments: - cd compile background data - 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 - -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) -{ -uschar *ptr = (uschar *)cd->start_pattern; -int count = 0; -int rc; - -/* If the pattern does not start with an opening parenthesis, the first call -to find_parens_sub() will scan right to the end (if necessary). However, if it -does start with a parenthesis, find_parens_sub() will return when it hits the -matching closing parens. That is why we have to have a loop. */ - -for (;;) - { - rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, &count); - if (rc > 0 || *ptr++ == 0) break; - } - -return rc; -} - - - - -/************************************************* * Find first significant op code * *************************************************/ @@ -1513,8 +1640,8 @@ 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 +1652,7 @@ 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: @@ -1535,11 +1662,11 @@ for (;;) case OP_CALLOUT: case OP_CREF: - case OP_NCREF: + case OP_DNCREF: case OP_RREF: - case OP_NRREF: + case OP_DNRREF: case OP_DEF: - code += _pcre_OP_lengths[*code]; + code += PRIV(OP_lengths)[*code]; break; default: @@ -1551,7 +1678,6 @@ for (;;) - /************************************************* * Find the fixed length of a branch * *************************************************/ @@ -1569,7 +1695,7 @@ 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 +1707,12 @@ 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 +1720,9 @@ 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 +1735,7 @@ 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 +1766,10 @@ 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 +1782,8 @@ 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 +1791,7 @@ 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: @@ -1673,13 +1801,13 @@ for (;;) case OP_COMMIT: case OP_CREF: case OP_DEF: + case OP_DNCREF: + case OP_DNRREF: case OP_DOLL: case OP_DOLLM: case OP_EOD: case OP_EODN: case OP_FAIL: - case OP_NCREF: - case OP_NRREF: case OP_NOT_WORD_BOUNDARY: case OP_PRUNE: case OP_REVERSE: @@ -1690,7 +1818,7 @@ 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 +1829,8 @@ 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 +1841,18 @@ 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 */ @@ -1757,31 +1886,39 @@ 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: - case OP_CRMINPLUS: case OP_CRSTAR: case OP_CRMINSTAR: + case OP_CRPLUS: + case OP_CRMINPLUS: case OP_CRQUERY: case OP_CRMINQUERY: + case OP_CRPOSSTAR: + case OP_CRPOSPLUS: + case OP_CRPOSQUERY: return -1; case OP_CRRANGE: case OP_CRMINRANGE: - if (GET2(cc,1) != GET2(cc,3)) return -1; - branchlength += GET2(cc,1); - cc += 5; + case OP_CRPOSRANGE: + if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1; + branchlength += (int)GET2(cc,1); + cc += 1 + 2 * IMM2_SIZE; break; default: @@ -1847,6 +1984,8 @@ for (;;) case OP_QUERYI: case OP_REF: case OP_REFI: + case OP_DNREF: + case OP_DNREFI: case OP_SBRA: case OP_SBRAPOS: case OP_SCBRA: @@ -1883,7 +2022,6 @@ for (;;) - /************************************************* * Scan compiled regex for specific bracket * *************************************************/ @@ -1896,18 +2034,18 @@ 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 +2059,8 @@ 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 +2068,9 @@ 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 +2098,13 @@ 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 +2112,14 @@ 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 +2149,11 @@ 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 +2170,17 @@ 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 +2215,13 @@ 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 +2229,76 @@ 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 +2321,30 @@ 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. + recurses chain of recurse_check to catch mutual recursion Returns: TRUE if what is matched could be empty */ +typedef struct recurse_check { + struct recurse_check *prev; + const pcre_uchar *group; +} recurse_check; + 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, recurse_check *recurses) { -register int c; -for (code = first_significant_code(code + _pcre_OP_lengths[*code], TRUE); +register pcre_uchar c; +recurse_check this_recurse; + +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; @@ -2197,25 +2367,50 @@ for (code = first_significant_code(code + _pcre_OP_len if (c == OP_RECURSE) { - const uschar *scode; + const pcre_uchar *scode = cd->start_code + GET(code, 1); BOOL empty_branch; - /* Test for forward reference */ + /* Test for forward reference or uncompleted reference. This is disabled + when called to scan a completed pattern by setting cd->start_workspace to + NULL. */ - for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE) - if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE; + if (cd->start_workspace != NULL) + { + const pcre_uchar *tcode; + for (tcode = cd->start_workspace; tcode < cd->hwm; tcode += LINK_SIZE) + if ((int)GET(tcode, 0) == (int)(code + 1 - cd->start_code)) return TRUE; + if (GET(scode, 1) == 0) return TRUE; /* Unclosed */ + } - /* Not a forward reference, test for completed backward reference */ + /* If we are scanning a completed pattern, there are no forward references + and all groups are complete. We need to detect whether this is a recursive + call, as otherwise there will be an infinite loop. If it is a recursion, + just skip over it. Simple recursions are easily detected. For mutual + recursions we keep a chain on the stack. */ - empty_branch = FALSE; - scode = cd->start_code + GET(code, 1); - if (GET(scode, 1) == 0) return TRUE; /* Unclosed */ + else + { + recurse_check *r = recurses; + const pcre_uchar *endgroup = scode; - /* Completed backwards reference */ + do endgroup += GET(endgroup, 1); while (*endgroup == OP_ALT); + if (code >= scode && code <= endgroup) continue; /* Simple recursion */ + for (r = recurses; r != NULL; r = r->prev) + if (r->group == scode) break; + if (r != NULL) continue; /* Mutual recursion */ + } + + /* Completed reference; scan the referenced group, remembering it on the + stack chain to detect mutual recursions. */ + + empty_branch = FALSE; + this_recurse.prev = recurses; + this_recurse.group = scode; + do { - if (could_be_empty_branch(scode, endcode, utf8, cd)) + if (could_be_empty_branch(scode, endcode, utf, cd, &this_recurse)) { empty_branch = TRUE; break; @@ -2233,7 +2428,7 @@ 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 +2466,7 @@ 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, NULL)) empty_branch = TRUE; code += GET(code, 1); } @@ -2289,11 +2484,11 @@ 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 +2496,9 @@ 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 @@ -2313,15 +2508,19 @@ for (code = first_significant_code(code + _pcre_OP_len case OP_CRMINSTAR: case OP_CRQUERY: case OP_CRMINQUERY: + case OP_CRPOSSTAR: + case OP_CRPOSQUERY: break; default: /* Non-repeat => class must match */ case OP_CRPLUS: /* These repeats aren't empty */ case OP_CRMINPLUS: + case OP_CRPOSPLUS: return FALSE; case OP_CRRANGE: case OP_CRMINRANGE: + case OP_CRPOSRANGE: if (GET2(ccode, 1) > 0) return FALSE; /* Minimum > 0 */ break; } @@ -2329,34 +2528,57 @@ for (code = first_significant_code(code + _pcre_OP_len /* Opcodes that must match a character */ + case OP_ANY: + case OP_ALLANY: + case OP_ANYBYTE: + case OP_PROP: case OP_NOTPROP: + case OP_ANYNL: + + case OP_NOT_HSPACE: + case OP_HSPACE: + case OP_NOT_VSPACE: + case OP_VSPACE: case OP_EXTUNI: + case OP_NOT_DIGIT: case OP_DIGIT: case OP_NOT_WHITESPACE: case OP_WHITESPACE: case OP_NOT_WORDCHAR: case OP_WORDCHAR: - case OP_ANY: - case OP_ALLANY: - case OP_ANYBYTE: + case OP_CHAR: case OP_CHARI: case OP_NOT: case OP_NOTI: + case OP_PLUS: + case OP_PLUSI: case OP_MINPLUS: - case OP_POSPLUS: - case OP_EXACT: + case OP_MINPLUSI: + case OP_NOTPLUS: + case OP_NOTPLUSI: case OP_NOTMINPLUS: + case OP_NOTMINPLUSI: + + case OP_POSPLUS: + case OP_POSPLUSI: case OP_NOTPOSPLUS: + case OP_NOTPOSPLUSI: + + case OP_EXACT: + case OP_EXACTI: case OP_NOTEXACT: + case OP_NOTEXACTI: + case OP_TYPEPLUS: case OP_TYPEMINPLUS: case OP_TYPEPOSPLUS: case OP_TYPEEXACT: + return FALSE; /* These are going to continue, as they may be empty, but we have to @@ -2376,7 +2598,8 @@ 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 */ @@ -2389,31 +2612,59 @@ for (code = first_significant_code(code + _pcre_OP_len return TRUE; /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO, - MINUPTO, and POSUPTO may be followed by a multibyte character */ + MINUPTO, and POSUPTO and their caseless and negative versions 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_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_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 (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f]; + case OP_NOTPOSQUERY: + case OP_NOTPOSQUERYI: + + if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]); break; 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: - if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f]; + case OP_NOTPOSUPTO: + case OP_NOTPOSUPTOI: + + if (utf && HAS_EXTRALEN(code[1 + IMM2_SIZE])) code += GET_EXTRALEN(code[1 + IMM2_SIZE]); break; #endif @@ -2423,9 +2674,6 @@ 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 +2705,19 @@ 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, NULL)) return FALSE; bcptr = bcptr->outer; } @@ -2479,6 +2727,1072 @@ return TRUE; /************************************************* +* Base opcode of repeated opcodes * +*************************************************/ + +/* Returns the base opcode for repeated single character type opcodes. If the +opcode is not a repeated character type, it returns with the original value. + +Arguments: c opcode +Returns: base opcode for the type +*/ + +static pcre_uchar +get_repeat_base(pcre_uchar c) +{ +return (c > OP_TYPEPOSUPTO)? c : + (c >= OP_TYPESTAR)? OP_TYPESTAR : + (c >= OP_NOTSTARI)? OP_NOTSTARI : + (c >= OP_NOTSTAR)? OP_NOTSTAR : + (c >= OP_STARI)? OP_STARI : + OP_STAR; +} + + + +#ifdef SUPPORT_UCP +/************************************************* +* Check a character and a property * +*************************************************/ + +/* This function is called by check_auto_possessive() when a property item +is adjacent to a fixed character. + +Arguments: + c the character + ptype the property type + pdata the data for the type + negated TRUE if it's a negated property (\P or \p{^) + +Returns: TRUE if auto-possessifying is OK +*/ + +static BOOL +check_char_prop(pcre_uint32 c, unsigned int ptype, unsigned int pdata, + BOOL negated) +{ +const pcre_uint32 *p; +const ucd_record *prop = GET_UCD(c); + +switch(ptype) + { + case PT_LAMP: + return (prop->chartype == ucp_Lu || + prop->chartype == ucp_Ll || + prop->chartype == ucp_Lt) == negated; + + case PT_GC: + return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated; + + case PT_PC: + return (pdata == prop->chartype) == negated; + + case PT_SC: + return (pdata == prop->script) == negated; + + /* These are specials */ + + case PT_ALNUM: + return (PRIV(ucp_gentype)[prop->chartype] == ucp_L || + PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated; + + /* Perl space used to exclude VT, but from Perl 5.18 it is included, which + means that Perl space and POSIX space are now identical. PCRE was changed + at release 8.34. */ + + case PT_SPACE: /* Perl space */ + case PT_PXSPACE: /* POSIX space */ + switch(c) + { + HSPACE_CASES: + VSPACE_CASES: + return negated; + + default: + return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z) == negated; + } + break; /* Control never reaches here */ + + case PT_WORD: + return (PRIV(ucp_gentype)[prop->chartype] == ucp_L || + PRIV(ucp_gentype)[prop->chartype] == ucp_N || + c == CHAR_UNDERSCORE) == negated; + + 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 */ + } + +return FALSE; +} +#endif /* SUPPORT_UCP */ + + + +/************************************************* +* Fill the character property list * +*************************************************/ + +/* Checks whether the code points to an opcode that can take part in auto- +possessification, and if so, fills a list with its properties. + +Arguments: + code points to start of expression + utf TRUE if in UTF-8 / UTF-16 / UTF-32 mode + fcc points to case-flipping table + list points to output list + list[0] will be filled with the opcode + list[1] will be non-zero if this opcode + can match an empty character string + list[2..7] depends on the opcode + +Returns: points to the start of the next opcode if *code is accepted + NULL if *code is not accepted +*/ + +static const pcre_uchar * +get_chr_property_list(const pcre_uchar *code, BOOL utf, + const pcre_uint8 *fcc, pcre_uint32 *list) +{ +pcre_uchar c = *code; +pcre_uchar base; +const pcre_uchar *end; +pcre_uint32 chr; + +#ifdef SUPPORT_UCP +pcre_uint32 *clist_dest; +const pcre_uint32 *clist_src; +#else +utf = utf; /* Suppress "unused parameter" compiler warning */ +#endif + +list[0] = c; +list[1] = FALSE; +code++; + +if (c >= OP_STAR && c <= OP_TYPEPOSUPTO) + { + base = get_repeat_base(c); + c -= (base - OP_STAR); + + if (c == OP_UPTO || c == OP_MINUPTO || c == OP_EXACT || c == OP_POSUPTO) + code += IMM2_SIZE; + + list[1] = (c != OP_PLUS && c != OP_MINPLUS && c != OP_EXACT && c != OP_POSPLUS); + + switch(base) + { + case OP_STAR: + list[0] = OP_CHAR; + break; + + case OP_STARI: + list[0] = OP_CHARI; + break; + + case OP_NOTSTAR: + list[0] = OP_NOT; + break; + + case OP_NOTSTARI: + list[0] = OP_NOTI; + break; + + case OP_TYPESTAR: + list[0] = *code; + code++; + break; + } + c = list[0]; + } + +switch(c) + { + case OP_NOT_DIGIT: + case OP_DIGIT: + case OP_NOT_WHITESPACE: + case OP_WHITESPACE: + case OP_NOT_WORDCHAR: + case OP_WORDCHAR: + case OP_ANY: + case OP_ALLANY: + case OP_ANYNL: + case OP_NOT_HSPACE: + case OP_HSPACE: + case OP_NOT_VSPACE: + case OP_VSPACE: + case OP_EXTUNI: + case OP_EODN: + case OP_EOD: + case OP_DOLL: + case OP_DOLLM: + return code; + + case OP_CHAR: + case OP_NOT: + GETCHARINCTEST(chr, code); + list[2] = chr; + list[3] = NOTACHAR; + return code; + + case OP_CHARI: + case OP_NOTI: + list[0] = (c == OP_CHARI) ? OP_CHAR : OP_NOT; + GETCHARINCTEST(chr, code); + list[2] = chr; + +#ifdef SUPPORT_UCP + if (chr < 128 || (chr < 256 && !utf)) + list[3] = fcc[chr]; + else + list[3] = UCD_OTHERCASE(chr); +#elif defined SUPPORT_UTF || !defined COMPILE_PCRE8 + list[3] = (chr < 256) ? fcc[chr] : chr; +#else + list[3] = fcc[chr]; +#endif + + /* The othercase might be the same value. */ + + if (chr == list[3]) + list[3] = NOTACHAR; + else + list[4] = NOTACHAR; + return code; + +#ifdef SUPPORT_UCP + case OP_PROP: + case OP_NOTPROP: + if (code[0] != PT_CLIST) + { + list[2] = code[0]; + list[3] = code[1]; + return code + 2; + } + + /* Convert only if we have enough space. */ + + clist_src = PRIV(ucd_caseless_sets) + code[1]; + clist_dest = list + 2; + code += 2; + + do { + if (clist_dest >= list + 8) + { + /* Early return if there is not enough space. This should never + happen, since all clists are shorter than 5 character now. */ + list[2] = code[0]; + list[3] = code[1]; + return code; + } + *clist_dest++ = *clist_src; + } + while(*clist_src++ != NOTACHAR); + + /* All characters are stored. The terminating NOTACHAR + is copied form the clist itself. */ + + list[0] = (c == OP_PROP) ? OP_CHAR : OP_NOT; + return code; +#endif + + case OP_NCLASS: + case OP_CLASS: +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + case OP_XCLASS: + if (c == OP_XCLASS) + end = code + GET(code, 0) - 1; + else +#endif + end = code + 32 / sizeof(pcre_uchar); + + switch(*end) + { + case OP_CRSTAR: + case OP_CRMINSTAR: + case OP_CRQUERY: + case OP_CRMINQUERY: + case OP_CRPOSSTAR: + case OP_CRPOSQUERY: + list[1] = TRUE; + end++; + break; + + case OP_CRPLUS: + case OP_CRMINPLUS: + case OP_CRPOSPLUS: + end++; + break; + + case OP_CRRANGE: + case OP_CRMINRANGE: + case OP_CRPOSRANGE: + list[1] = (GET2(end, 1) == 0); + end += 1 + 2 * IMM2_SIZE; + break; + } + list[2] = end - code; + return end; + } +return NULL; /* Opcode not accepted */ +} + + + +/************************************************* +* Scan further character sets for match * +*************************************************/ + +/* Checks whether the base and the current opcode have a common character, in +which case the base cannot be possessified. + +Arguments: + code points to the byte code + utf TRUE in UTF-8 / UTF-16 / UTF-32 mode + cd static compile data + base_list the data list of the base opcode + +Returns: TRUE if the auto-possessification is possible +*/ + +static BOOL +compare_opcodes(const pcre_uchar *code, BOOL utf, const compile_data *cd, + const pcre_uint32 *base_list, const pcre_uchar *base_end) +{ +pcre_uchar c; +pcre_uint32 list[8]; +const pcre_uint32 *chr_ptr; +const pcre_uint32 *ochr_ptr; +const pcre_uint32 *list_ptr; +const pcre_uchar *next_code; +const pcre_uint8 *class_bitset; +const pcre_uint32 *set1, *set2, *set_end; +pcre_uint32 chr; +BOOL accepted, invert_bits; + +/* Note: the base_list[1] contains whether the current opcode has greedy +(represented by a non-zero value) quantifier. This is a different from +other character type lists, which stores here that the character iterator +matches to an empty string (also represented by a non-zero value). */ + +for(;;) + { + /* All operations move the code pointer forward. + Therefore infinite recursions are not possible. */ + + c = *code; + + /* Skip over callouts */ + + if (c == OP_CALLOUT) + { + code += PRIV(OP_lengths)[c]; + continue; + } + + if (c == OP_ALT) + { + do code += GET(code, 1); while (*code == OP_ALT); + c = *code; + } + + switch(c) + { + case OP_END: + case OP_KETRPOS: + /* TRUE only in greedy case. The non-greedy case could be replaced by + an OP_EXACT, but it is probably not worth it. (And note that OP_EXACT + uses more memory, which we cannot get at this stage.) */ + + return base_list[1] != 0; + + case OP_KET: + /* If the bracket is capturing, and referenced by an OP_RECURSE, or + it is an atomic sub-pattern (assert, once, etc.) the non-greedy case + cannot be converted to a possessive form. */ + + if (base_list[1] == 0) return FALSE; + + switch(*(code - GET(code, 1))) + { + case OP_ASSERT: + case OP_ASSERT_NOT: + case OP_ASSERTBACK: + case OP_ASSERTBACK_NOT: + case OP_ONCE: + case OP_ONCE_NC: + /* Atomic sub-patterns and assertions can always auto-possessify their + last iterator. */ + return TRUE; + } + + code += PRIV(OP_lengths)[c]; + continue; + + case OP_ONCE: + case OP_ONCE_NC: + case OP_BRA: + case OP_CBRA: + next_code = code + GET(code, 1); + code += PRIV(OP_lengths)[c]; + + while (*next_code == OP_ALT) + { + if (!compare_opcodes(code, utf, cd, base_list, base_end)) return FALSE; + code = next_code + 1 + LINK_SIZE; + next_code += GET(next_code, 1); + } + continue; + + case OP_BRAZERO: + case OP_BRAMINZERO: + + next_code = code + 1; + if (*next_code != OP_BRA && *next_code != OP_CBRA + && *next_code != OP_ONCE && *next_code != OP_ONCE_NC) return FALSE; + + do next_code += GET(next_code, 1); while (*next_code == OP_ALT); + + /* The bracket content will be checked by the + OP_BRA/OP_CBRA case above. */ + next_code += 1 + LINK_SIZE; + if (!compare_opcodes(next_code, utf, cd, base_list, base_end)) + return FALSE; + + code += PRIV(OP_lengths)[c]; + continue; + } + + /* Check for a supported opcode, and load its properties. */ + + code = get_chr_property_list(code, utf, cd->fcc, list); + if (code == NULL) return FALSE; /* Unsupported */ + + /* If either opcode is a small character list, set pointers for comparing + characters from that list with another list, or with a property. */ + + if (base_list[0] == OP_CHAR) + { + chr_ptr = base_list + 2; + list_ptr = list; + } + else if (list[0] == OP_CHAR) + { + chr_ptr = list + 2; + list_ptr = base_list; + } + + /* Character bitsets can also be compared to certain opcodes. */ + + else if (base_list[0] == OP_CLASS || list[0] == OP_CLASS +#ifdef COMPILE_PCRE8 + /* In 8 bit, non-UTF mode, OP_CLASS and OP_NCLASS are the same. */ + || (!utf && (base_list[0] == OP_NCLASS || list[0] == OP_NCLASS)) +#endif + ) + { +#ifdef COMPILE_PCRE8 + if (base_list[0] == OP_CLASS || (!utf && base_list[0] == OP_NCLASS)) +#else + if (base_list[0] == OP_CLASS) +#endif + { + set1 = (pcre_uint32 *)(base_end - base_list[2]); + list_ptr = list; + } + else + { + set1 = (pcre_uint32 *)(code - list[2]); + list_ptr = base_list; + } + + invert_bits = FALSE; + switch(list_ptr[0]) + { + case OP_CLASS: + case OP_NCLASS: + set2 = (pcre_uint32 *) + ((list_ptr == list ? code : base_end) - list_ptr[2]); + break; + + /* OP_XCLASS cannot be supported here, because its bitset + is not necessarily complete. E.g: [a-\0x{200}] is stored + as a character range, and the appropriate bits are not set. */ + + case OP_NOT_DIGIT: + invert_bits = TRUE; + /* Fall through */ + case OP_DIGIT: + set2 = (pcre_uint32 *)(cd->cbits + cbit_digit); + break; + + case OP_NOT_WHITESPACE: + invert_bits = TRUE; + /* Fall through */ + case OP_WHITESPACE: + set2 = (pcre_uint32 *)(cd->cbits + cbit_space); + break; + + case OP_NOT_WORDCHAR: + invert_bits = TRUE; + /* Fall through */ + case OP_WORDCHAR: + set2 = (pcre_uint32 *)(cd->cbits + cbit_word); + break; + + default: + return FALSE; + } + + /* Compare 4 bytes to improve speed. */ + set_end = set1 + (32 / 4); + if (invert_bits) + { + do + { + if ((*set1++ & ~(*set2++)) != 0) return FALSE; + } + while (set1 < set_end); + } + else + { + do + { + if ((*set1++ & *set2++) != 0) return FALSE; + } + while (set1 < set_end); + } + + if (list[1] == 0) return TRUE; + /* Might be an empty repeat. */ + continue; + } + + /* Some property combinations also acceptable. Unicode property opcodes are + processed specially; the rest can be handled with a lookup table. */ + + else + { + pcre_uint32 leftop, rightop; + + leftop = base_list[0]; + rightop = list[0]; + +#ifdef SUPPORT_UCP + accepted = FALSE; /* Always set in non-unicode case. */ + if (leftop == OP_PROP || leftop == OP_NOTPROP) + { + if (rightop == OP_EOD) + accepted = TRUE; + else if (rightop == OP_PROP || rightop == OP_NOTPROP) + { + int n; + const pcre_uint8 *p; + BOOL same = leftop == rightop; + BOOL lisprop = leftop == OP_PROP; + BOOL risprop = rightop == OP_PROP; + BOOL bothprop = lisprop && risprop; + + /* There's a table that specifies how each combination is to be + processed: + 0 Always return FALSE (never auto-possessify) + 1 Character groups are distinct (possessify if both are OP_PROP) + 2 Check character categories in the same group (general or particular) + 3 Return TRUE if the two opcodes are not the same + ... see comments below + */ + + n = propposstab[base_list[2]][list[2]]; + switch(n) + { + case 0: break; + case 1: accepted = bothprop; break; + case 2: accepted = (base_list[3] == list[3]) != same; break; + case 3: accepted = !same; break; + + case 4: /* Left general category, right particular category */ + accepted = risprop && catposstab[base_list[3]][list[3]] == same; + break; + + case 5: /* Right general category, left particular category */ + accepted = lisprop && catposstab[list[3]][base_list[3]] == same; + break; + + /* This code is logically tricky. Think hard before fiddling with it. + The posspropstab table has four entries per row. Each row relates to + one of PCRE's special properties such as ALNUM or SPACE or WORD. + Only WORD actually needs all four entries, but using repeats for the + others means they can all use the same code below. + + The first two entries in each row are Unicode general categories, and + apply always, because all the characters they include are part of the + PCRE character set. The third and fourth entries are a general and a + particular category, respectively, that include one or more relevant + characters. One or the other is used, depending on whether the check + is for a general or a particular category. However, in both cases the + category contains more characters than the specials that are defined + for the property being tested against. Therefore, it cannot be used + in a NOTPROP case. + + Example: the row for WORD contains ucp_L, ucp_N, ucp_P, ucp_Po. + Underscore is covered by ucp_P or ucp_Po. */ + + case 6: /* Left alphanum vs right general category */ + case 7: /* Left space vs right general category */ + case 8: /* Left word vs right general category */ + p = posspropstab[n-6]; + accepted = risprop && lisprop == + (list[3] != p[0] && + list[3] != p[1] && + (list[3] != p[2] || !lisprop)); + break; + + case 9: /* Right alphanum vs left general category */ + case 10: /* Right space vs left general category */ + case 11: /* Right word vs left general category */ + p = posspropstab[n-9]; + accepted = lisprop && risprop == + (base_list[3] != p[0] && + base_list[3] != p[1] && + (base_list[3] != p[2] || !risprop)); + break; + + case 12: /* Left alphanum vs right particular category */ + case 13: /* Left space vs right particular category */ + case 14: /* Left word vs right particular category */ + p = posspropstab[n-12]; + accepted = risprop && lisprop == + (catposstab[p[0]][list[3]] && + catposstab[p[1]][list[3]] && + (list[3] != p[3] || !lisprop)); + break; + + case 15: /* Right alphanum vs left particular category */ + case 16: /* Right space vs left particular category */ + case 17: /* Right word vs left particular category */ + p = posspropstab[n-15]; + accepted = lisprop && risprop == + (catposstab[p[0]][base_list[3]] && + catposstab[p[1]][base_list[3]] && + (base_list[3] != p[3] || !risprop)); + break; + } + } + } + + else +#endif /* SUPPORT_UCP */ + + accepted = leftop >= FIRST_AUTOTAB_OP && leftop <= LAST_AUTOTAB_LEFT_OP && + rightop >= FIRST_AUTOTAB_OP && rightop <= LAST_AUTOTAB_RIGHT_OP && + autoposstab[leftop - FIRST_AUTOTAB_OP][rightop - FIRST_AUTOTAB_OP]; + + if (!accepted) + return FALSE; + + if (list[1] == 0) return TRUE; + /* Might be an empty repeat. */ + continue; + } + + /* Control reaches here only if one of the items is a small character list. + All characters are checked against the other side. */ + + do + { + chr = *chr_ptr; + + switch(list_ptr[0]) + { + case OP_CHAR: + ochr_ptr = list_ptr + 2; + do + { + if (chr == *ochr_ptr) return FALSE; + ochr_ptr++; + } + while(*ochr_ptr != NOTACHAR); + break; + + case OP_NOT: + ochr_ptr = list_ptr + 2; + do + { + if (chr == *ochr_ptr) + break; + ochr_ptr++; + } + while(*ochr_ptr != NOTACHAR); + if (*ochr_ptr == NOTACHAR) return FALSE; /* Not found */ + break; + + /* 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: + if (chr < 256 && (cd->ctypes[chr] & ctype_digit) != 0) return FALSE; + break; + + case OP_NOT_DIGIT: + if (chr > 255 || (cd->ctypes[chr] & ctype_digit) == 0) return FALSE; + break; + + case OP_WHITESPACE: + if (chr < 256 && (cd->ctypes[chr] & ctype_space) != 0) return FALSE; + break; + + case OP_NOT_WHITESPACE: + if (chr > 255 || (cd->ctypes[chr] & ctype_space) == 0) return FALSE; + break; + + case OP_WORDCHAR: + if (chr < 255 && (cd->ctypes[chr] & ctype_word) != 0) return FALSE; + break; + + case OP_NOT_WORDCHAR: + if (chr > 255 || (cd->ctypes[chr] & ctype_word) == 0) return FALSE; + break; + + case OP_HSPACE: + switch(chr) + { + HSPACE_CASES: return FALSE; + default: break; + } + break; + + case OP_NOT_HSPACE: + switch(chr) + { + HSPACE_CASES: break; + default: return FALSE; + } + break; + + case OP_ANYNL: + case OP_VSPACE: + switch(chr) + { + VSPACE_CASES: return FALSE; + default: break; + } + break; + + case OP_NOT_VSPACE: + switch(chr) + { + VSPACE_CASES: break; + default: return FALSE; + } + break; + + case OP_DOLL: + case OP_EODN: + switch (chr) + { + case CHAR_CR: + case CHAR_LF: + case CHAR_VT: + case CHAR_FF: + case CHAR_NEL: +#ifndef EBCDIC + case 0x2028: + case 0x2029: +#endif /* Not EBCDIC */ + return FALSE; + } + break; + + case OP_EOD: /* Can always possessify before \z */ + break; + +#ifdef SUPPORT_UCP + case OP_PROP: + case OP_NOTPROP: + if (!check_char_prop(chr, list_ptr[2], list_ptr[3], + list_ptr[0] == OP_NOTPROP)) + return FALSE; + break; +#endif + + case OP_NCLASS: + if (chr > 255) return FALSE; + /* Fall through */ + + case OP_CLASS: + if (chr > 255) break; + class_bitset = (pcre_uint8 *) + ((list_ptr == list ? code : base_end) - list_ptr[2]); + if ((class_bitset[chr >> 3] & (1 << (chr & 7))) != 0) return FALSE; + break; + +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + case OP_XCLASS: + if (PRIV(xclass)(chr, (list_ptr == list ? code : base_end) - + list_ptr[2] + LINK_SIZE, utf)) return FALSE; + break; +#endif + + default: + return FALSE; + } + + chr_ptr++; + } + while(*chr_ptr != NOTACHAR); + + /* At least one character must be matched from this opcode. */ + + if (list[1] == 0) return TRUE; + } + +return FALSE; +} + + + +/************************************************* +* Scan compiled regex for auto-possession * +*************************************************/ + +/* Replaces single character iterations with their possessive alternatives +if appropriate. This function modifies the compiled opcode! + +Arguments: + code points to start of the byte code + utf TRUE in UTF-8 / UTF-16 / UTF-32 mode + cd static compile data + +Returns: nothing +*/ + +static void +auto_possessify(pcre_uchar *code, BOOL utf, const compile_data *cd) +{ +register pcre_uchar c; +const pcre_uchar *end; +pcre_uchar *repeat_opcode; +pcre_uint32 list[8]; + +for (;;) + { + c = *code; + + if (c >= OP_STAR && c <= OP_TYPEPOSUPTO) + { + c -= get_repeat_base(c) - OP_STAR; + end = (c <= OP_MINUPTO) ? + get_chr_property_list(code, utf, cd->fcc, list) : NULL; + list[1] = c == OP_STAR || c == OP_PLUS || c == OP_QUERY || c == OP_UPTO; + + if (end != NULL && compare_opcodes(end, utf, cd, list, end)) + { + switch(c) + { + case OP_STAR: + *code += OP_POSSTAR - OP_STAR; + break; + + case OP_MINSTAR: + *code += OP_POSSTAR - OP_MINSTAR; + break; + + case OP_PLUS: + *code += OP_POSPLUS - OP_PLUS; + break; + + case OP_MINPLUS: + *code += OP_POSPLUS - OP_MINPLUS; + break; + + case OP_QUERY: + *code += OP_POSQUERY - OP_QUERY; + break; + + case OP_MINQUERY: + *code += OP_POSQUERY - OP_MINQUERY; + break; + + case OP_UPTO: + *code += OP_POSUPTO - OP_UPTO; + break; + + case OP_MINUPTO: + *code += OP_MINUPTO - OP_UPTO; + break; + } + } + c = *code; + } + else if (c == OP_CLASS || c == OP_NCLASS || c == OP_XCLASS) + { +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + if (c == OP_XCLASS) + repeat_opcode = code + GET(code, 1); + else +#endif + repeat_opcode = code + 1 + (32 / sizeof(pcre_uchar)); + + c = *repeat_opcode; + if (c >= OP_CRSTAR && c <= OP_CRMINRANGE) + { + /* end must not be NULL. */ + end = get_chr_property_list(code, utf, cd->fcc, list); + + list[1] = (c & 1) == 0; + + if (compare_opcodes(end, utf, cd, list, end)) + { + switch (c) + { + case OP_CRSTAR: + case OP_CRMINSTAR: + *repeat_opcode = OP_CRPOSSTAR; + break; + + case OP_CRPLUS: + case OP_CRMINPLUS: + *repeat_opcode = OP_CRPOSPLUS; + break; + + case OP_CRQUERY: + case OP_CRMINQUERY: + *repeat_opcode = OP_CRPOSQUERY; + break; + + case OP_CRRANGE: + case OP_CRMINRANGE: + *repeat_opcode = OP_CRPOSRANGE; + break; + } + } + } + c = *code; + } + + switch(c) + { + case OP_END: + return; + + case OP_TYPESTAR: + case OP_TYPEMINSTAR: + case OP_TYPEPLUS: + case OP_TYPEMINPLUS: + case OP_TYPEQUERY: + case OP_TYPEMINQUERY: + case OP_TYPEPOSSTAR: + case OP_TYPEPOSPLUS: + case OP_TYPEPOSQUERY: + if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2; + break; + + case OP_TYPEUPTO: + case OP_TYPEMINUPTO: + case OP_TYPEEXACT: + case OP_TYPEPOSUPTO: + if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP) + code += 2; + break; + +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + case OP_XCLASS: + code += GET(code, 1); + break; +#endif + + case OP_MARK: + case OP_PRUNE_ARG: + case OP_SKIP_ARG: + case OP_THEN_ARG: + code += code[1]; + break; + } + + /* Add in the fixed length from the table */ + + 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. */ + +#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_STAR: + case OP_MINSTAR: + case OP_PLUS: + case OP_MINPLUS: + case OP_QUERY: + case OP_MINQUERY: + case OP_UPTO: + case OP_MINUPTO: + case OP_EXACT: + case OP_POSSTAR: + case OP_POSPLUS: + case OP_POSQUERY: + case OP_POSUPTO: + case OP_STARI: + case OP_MINSTARI: + case OP_PLUSI: + case OP_MINPLUSI: + case OP_QUERYI: + case OP_MINQUERYI: + case OP_UPTOI: + case OP_MINUPTOI: + case OP_EXACTI: + case OP_POSSTARI: + case OP_POSPLUSI: + case OP_POSQUERYI: + case OP_POSUPTOI: + case OP_NOTSTAR: + case OP_NOTMINSTAR: + case OP_NOTPLUS: + case OP_NOTMINPLUS: + case OP_NOTQUERY: + case OP_NOTMINQUERY: + case OP_NOTUPTO: + case OP_NOTMINUPTO: + case OP_NOTEXACT: + case OP_NOTPOSSTAR: + case OP_NOTPOSPLUS: + case OP_NOTPOSQUERY: + case OP_NOTPOSUPTO: + case OP_NOTSTARI: + case OP_NOTMINSTARI: + case OP_NOTPLUSI: + case OP_NOTMINPLUSI: + case OP_NOTQUERYI: + case OP_NOTMINQUERYI: + case OP_NOTUPTOI: + case OP_NOTMINUPTOI: + case OP_NOTEXACTI: + case OP_NOTPOSSTARI: + case OP_NOTPOSPLUSI: + case OP_NOTPOSQUERYI: + case OP_NOTPOSUPTOI: + if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]); + break; + } +#else + (void)(utf); /* Keep compiler happy by referencing function argument */ +#endif + } +} + + + +/************************************************* * Check for POSIX class syntax * *************************************************/ @@ -2499,7 +3813,7 @@ class, but [abc[:x\]pqr:]] is (so that an error can be below handles the special case of \], but does not try to do any other escape processing. This makes it different from Perl for cases such as [:l\ower:] where Perl recognizes it as the POSIX class "lower" but PCRE does not recognize -"l\ower". This is a lesser evil that not diagnosing bad classes when Perl does, +"l\ower". This is a lesser evil than not diagnosing bad classes when Perl does, I think. A user pointed out that PCRE was rejecting [:a[:digit:]] whereas Perl was not. @@ -2521,11 +3835,11 @@ 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 +3879,14 @@ 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 +3918,7 @@ 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 +3926,22 @@ 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 +3954,7 @@ 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 +3979,14 @@ 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 +4008,7 @@ 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 +4022,10 @@ 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 +4033,34 @@ 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; @@ -2741,494 +4071,248 @@ 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; } +#endif /* SUPPORT_UCP */ /************************************************* -* Check a character and a property * +* Add a character or range to a class * *************************************************/ -/* This function is called by check_auto_possessive() when a property item -is adjacent to a fixed character. +/* 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: - c the character - ptype the property type - pdata the data for the type - negated TRUE if it's a negated property (\P or \p{^) + 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: TRUE if auto-possessifying is OK +Returns: the number of < 256 characters added + the pointer to extra data is updated */ -static BOOL -check_char_prop(int c, int ptype, int pdata, BOOL negated) +static int +add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options, + compile_data *cd, pcre_uint32 start, pcre_uint32 end) { -const ucd_record *prop = GET_UCD(c); -switch(ptype) - { - case PT_LAMP: - return (prop->chartype == ucp_Lu || - prop->chartype == ucp_Ll || - prop->chartype == ucp_Lt) == negated; +pcre_uint32 c; +int n8 = 0; - case PT_GC: - return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated; +/* 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. */ - case PT_PC: - return (pdata == prop->chartype) == negated; +if ((options & PCRE_CASELESS) != 0) + { +#ifdef SUPPORT_UCP + if ((options & PCRE_UTF8) != 0) + { + int rc; + pcre_uint32 oc, od; - case PT_SC: - return (pdata == prop->script) == negated; + options &= ~PCRE_CASELESS; /* Remove for recursive calls */ + c = start; - /* These are specials */ + while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0) + { + /* Handle a single character that has more than one other case. */ - case PT_ALNUM: - return (_pcre_ucp_gentype[prop->chartype] == ucp_L || - _pcre_ucp_gentype[prop->chartype] == ucp_N) == negated; + if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd, + PRIV(ucd_caseless_sets) + rc, oc); - case PT_SPACE: /* Perl space */ - return (_pcre_ucp_gentype[prop->chartype] == ucp_Z || - c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR) - == negated; + /* Do nothing if the other case range is within the original range. */ - case PT_PXSPACE: /* POSIX space */ - return (_pcre_ucp_gentype[prop->chartype] == ucp_Z || - c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || - c == CHAR_FF || c == CHAR_CR) - == negated; + else if (oc >= start && od <= end) continue; - case PT_WORD: - return (_pcre_ucp_gentype[prop->chartype] == ucp_L || - _pcre_ucp_gentype[prop->chartype] == ucp_N || - c == CHAR_UNDERSCORE) == negated; - } -return FALSE; -} + /* 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 */ - -/************************************************* -* Check if auto-possessifying is possible * -*************************************************/ - -/* This function is called for unlimited repeats of certain items, to see -whether the next thing could possibly match the repeated item. If not, it makes -sense to automatically possessify the repeated item. - -Arguments: - previous pointer to the repeated opcode - utf8 TRUE in UTF-8 mode - ptr next character in pattern - options options bits - cd contains pointers to tables etc. - -Returns: TRUE if possessifying is wanted -*/ - -static BOOL -check_auto_possessive(const uschar *previous, BOOL utf8, const uschar *ptr, - int options, compile_data *cd) -{ -int c, next; -int op_code = *previous++; - -/* Skip whitespace and comments in extended mode */ - -if ((options & PCRE_EXTENDED) != 0) - { - for (;;) + for (c = start; c <= end && c < 256; c++) { - while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++; - if (*ptr == CHAR_NUMBER_SIGN) - { - ptr++; - while (*ptr != 0) - { - if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; } - ptr++; -#ifdef SUPPORT_UTF8 - if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++; -#endif - } - } - else break; + SETBIT(classbits, cd->fcc[c]); + n8++; } } -/* If the next item is one that we can handle, get its value. A non-negative -value is a character, a negative value is an escape value. */ +/* 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 (*ptr == CHAR_BACKSLASH) - { - int temperrorcode = 0; - next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE); - if (temperrorcode != 0) return FALSE; - ptr++; /* Point after the escape sequence */ - } +#if defined COMPILE_PCRE8 +#ifdef SUPPORT_UTF + if ((options & PCRE_UTF8) == 0) +#endif + if (end > 0xff) end = 0xff; -else if ((cd->ctypes[*ptr] & ctype_meta) == 0) - { -#ifdef SUPPORT_UTF8 - if (utf8) { GETCHARINC(next, ptr); } else +#elif defined COMPILE_PCRE16 +#ifdef SUPPORT_UTF + if ((options & PCRE_UTF16) == 0) #endif - next = *ptr++; - } + if (end > 0xffff) end = 0xffff; -else return FALSE; +#endif /* COMPILE_PCRE[8|16] */ -/* Skip whitespace and comments in extended mode */ +/* If all characters are less than 256, use the bit map. Otherwise use extra +data. */ -if ((options & PCRE_EXTENDED) != 0) +if (end < 0x100) { - for (;;) + for (c = start; c <= end; c++) { - while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++; - if (*ptr == CHAR_NUMBER_SIGN) - { - ptr++; - while (*ptr != 0) - { - if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; } - ptr++; -#ifdef SUPPORT_UTF8 - if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++; -#endif - } - } - else break; + n8++; + SETBIT(classbits, c); } } -/* 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) - return FALSE; - -/* Now compare the next item with the previous opcode. First, handle cases when -the next item is a character. */ - -if (next >= 0) switch(op_code) +else { - case OP_CHAR: -#ifdef SUPPORT_UTF8 - GETCHARTEST(c, previous); -#else - c = *previous; -#endif - return c != next; + pcre_uchar *uchardata = *uchardptr; - /* 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. */ - - case OP_CHARI: -#ifdef SUPPORT_UTF8 - GETCHARTEST(c, previous); -#else - c = *previous; -#endif - if (c == next) return FALSE; -#ifdef SUPPORT_UTF8 - if (utf8) +#ifdef SUPPORT_UTF + if ((options & PCRE_UTF8) != 0) /* All UTFs use the same flag bit */ { - unsigned int othercase; - if (next < 128) othercase = cd->fcc[next]; else -#ifdef SUPPORT_UCP - othercase = UCD_OTHERCASE((unsigned int)next); -#else - othercase = NOTACHAR; -#endif - return (unsigned int)c != othercase; + 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_UTF8 */ - return (c != cd->fcc[next]); /* Non-UTF-8 mode */ +#endif /* SUPPORT_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. */ + /* Without UTF support, character values are constrained by the bit length, + and can only be > 256 for 16-bit and 32-bit libraries. */ - 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 -#ifdef SUPPORT_UCP - othercase = UCD_OTHERCASE(next); +#ifdef COMPILE_PCRE8 + {} #else - othercase = NOTACHAR; -#endif - return (unsigned int)c == othercase; - } - else -#endif /* SUPPORT_UTF8 */ - return (c == cd->fcc[next]); /* Non-UTF-8 mode */ - - /* 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_NOT_DIGIT: - return next <= 127 && (cd->ctypes[next] & ctype_digit) != 0; - - case OP_WHITESPACE: - return next > 127 || (cd->ctypes[next] & ctype_space) == 0; - - case OP_NOT_WHITESPACE: - return next <= 127 && (cd->ctypes[next] & ctype_space) != 0; - - case OP_WORDCHAR: - return next > 127 || (cd->ctypes[next] & ctype_word) == 0; - - case OP_NOT_WORDCHAR: - return next <= 127 && (cd->ctypes[next] & ctype_word) != 0; - - case OP_HSPACE: - case OP_NOT_HSPACE: - switch(next) + if (start < end) { - 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; + *uchardata++ = XCL_RANGE; + *uchardata++ = start; + *uchardata++ = end; } - - case OP_ANYNL: - case OP_VSPACE: - case OP_NOT_VSPACE: - switch(next) + else if (start == end) { - 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; + *uchardata++ = XCL_SINGLE; + *uchardata++ = start; } - -#ifdef SUPPORT_UCP - 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); #endif - default: - return FALSE; + *uchardptr = uchardata; /* Updata extra data pointer */ } +return n8; /* Number of 8-bit characters */ +} -/* 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 -characteristics are recognized. Similarly, the opcodes OP_DIGIT etc. are -replaced by OP_PROP codes when PCRE_UCP is set. */ -switch(op_code) - { - case OP_CHAR: - case OP_CHARI: -#ifdef SUPPORT_UTF8 - GETCHARTEST(c, previous); -#else - c = *previous; -#endif - switch(-next) - { - case ESC_d: - return c > 127 || (cd->ctypes[c] & ctype_digit) == 0; - case ESC_D: - return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0; - case ESC_s: - return c > 127 || (cd->ctypes[c] & ctype_space) == 0; +/************************************************* +* Add a list of characters to a class * +*************************************************/ - case ESC_S: - return c <= 127 && (cd->ctypes[c] & ctype_space) != 0; +/* 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. - case ESC_w: - return c > 127 || (cd->ctypes[c] & ctype_word) == 0; +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 - case ESC_W: - return c <= 127 && (cd->ctypes[c] & ctype_word) != 0; +Returns: the number of < 256 characters added + the pointer to extra data is updated +*/ - 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; - default: - return -next == 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; - default: - return -next == 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. */ - -#ifdef SUPPORT_UCP - case ESC_du: - case ESC_DU: - case ESC_wu: - case ESC_WU: - case ESC_su: - case ESC_SU: - { - int temperrorcode = 0; - ptr = substitutes[-next - ESC_DU]; - next = check_escape(&ptr, &temperrorcode, 0, options, FALSE); - if (temperrorcode != 0) return FALSE; - ptr++; /* For compatibility */ - } - /* Fall through */ - - case ESC_p: - case ESC_P: - { - int ptype, pdata, errorcodeptr; - BOOL negated; - - ptr--; /* Make ptr point at the p or P */ - ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr); - if (ptype < 0) return FALSE; - ptr++; /* Point past the final curly ket */ - - /* If the property item is optional, we have to give up. (When generated - from \d etc by PCRE_UCP, this test will have been applied much earlier, - 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) - return FALSE; - - /* Do the property check. */ - - return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated); - } -#endif - - default: - return FALSE; +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; +} - /* In principle, support for Unicode properties should be integrated here as - well. It means re-organizing the above code so as to get hold of the property - values before switching on the op-code. However, I wonder how many patterns - combine ASCII \d etc with Unicode properties? (Note that if PCRE_UCP is set, - 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; - case OP_NOT_DIGIT: - return next == -ESC_d; +/************************************************* +* Add characters not in a list to a class * +*************************************************/ - case OP_WHITESPACE: - return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R; +/* 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. - case OP_NOT_WHITESPACE: - return next == -ESC_s || next == -ESC_h || next == -ESC_v; +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 - case OP_HSPACE: - return next == -ESC_S || next == -ESC_H || next == -ESC_d || - next == -ESC_w || next == -ESC_v || next == -ESC_R; +Returns: the number of < 256 characters added + the pointer to extra data is updated +*/ - case OP_NOT_HSPACE: - return next == -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; - - case OP_NOT_VSPACE: - return next == -ESC_v || next == -ESC_R; - - case OP_WORDCHAR: - return next == -ESC_W || next == -ESC_s || next == -ESC_h || - next == -ESC_v || next == -ESC_R; - - case OP_NOT_WORDCHAR: - return next == -ESC_w || next == -ESC_d; - - default: - return FALSE; +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++; } - -/* Control does not reach here */ +return n8; } @@ -3244,66 +4328,85 @@ to find out the amount of memory needed, as well as du phase. The value of lengthptr distinguishes the two phases. Arguments: - optionsptr pointer to the option bits - codeptr points to the pointer to the current code point - ptrptr points to the current pattern pointer - errorcodeptr points to error code variable - firstbyteptr set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE) - reqbyteptr set to the last literal character required, else < 0 - bcptr points to current branch chain - cond_depth conditional nesting depth - cd contains pointers to tables etc. - lengthptr NULL during the real compile phase - points to length accumulator during pre-compile phase + optionsptr pointer to the option bits + codeptr points to the pointer to the current code point + ptrptr points to the current pattern pointer + errorcodeptr points to error code variable + 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. + lengthptr NULL during the real compile phase + points to length accumulator during pre-compile phase -Returns: TRUE on success - FALSE, with *errorcodeptr set non-zero on error +Returns: TRUE on success + FALSE, with *errorcodeptr set non-zero on error */ 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 +4418,24 @@ 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 +4447,28 @@ 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 +4507,8 @@ 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 +4518,7 @@ 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 +4543,7 @@ 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) { @@ -3460,62 +4566,73 @@ for (;; ptr++) } goto NORMAL_CHAR; } + /* Control does not reach here. */ } - /* Fill in length of a previous callout, except when the next thing is - a quantifier. */ + /* In extended mode, skip white space and comments. We need a loop in order + to check for more white space and more comments after a comment. */ - is_quantifier = - c == CHAR_ASTERISK || c == CHAR_PLUS || c == CHAR_QUESTION_MARK || - (c == CHAR_LEFT_CURLY_BRACKET && is_counted_repeat(ptr+1)); - - if (!is_quantifier && previous_callout != NULL && - after_manual_callout-- <= 0) - { - if (lengthptr == NULL) /* Don't attempt in pre-compile phase */ - complete_callout(previous_callout, ptr, cd); - previous_callout = NULL; - } - - /* In extended mode, skip white space and comments. */ - if ((options & PCRE_EXTENDED) != 0) { - if ((cd->ctypes[c] & ctype_space) != 0) continue; - if (c == CHAR_NUMBER_SIGN) + for (;;) { + while (MAX_255(c) && (cd->ctypes[c] & ctype_space) != 0) c = *(++ptr); + if (c != CHAR_NUMBER_SIGN) break; ptr++; - while (*ptr != 0) + while (*ptr != CHAR_NULL) { - if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; } + if (IS_NEWLINE(ptr)) /* For non-fixed-length newline cases, */ + { /* IS_NEWLINE sets cd->nllen. */ + ptr += cd->nllen; + break; + } ptr++; -#ifdef SUPPORT_UTF8 - if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++; +#ifdef SUPPORT_UTF + if (utf) FORWARDCHAR(ptr); #endif } - if (*ptr != 0) continue; - - /* Else fall through to handle end of string */ - c = 0; + c = *ptr; /* Either NULL or the char after a newline */ } } - /* No auto callout for quantifiers. */ + /* See if the next thing is a quantifier. */ - if ((options & PCRE_AUTO_CALLOUT) != 0 && !is_quantifier) + is_quantifier = + c == CHAR_ASTERISK || c == CHAR_PLUS || c == CHAR_QUESTION_MARK || + (c == CHAR_LEFT_CURLY_BRACKET && is_counted_repeat(ptr+1)); + + /* Fill in length of a previous callout, except when the next thing is a + quantifier or when processing a property substitution string in UCP mode. */ + + if (!is_quantifier && previous_callout != NULL && nestptr == NULL && + after_manual_callout-- <= 0) { + if (lengthptr == NULL) /* Don't attempt in pre-compile phase */ + complete_callout(previous_callout, ptr, cd); + previous_callout = NULL; + } + + /* Create auto callout, except for quantifiers, or while processing property + strings that are substituted for \w etc in UCP mode. */ + + if ((options & PCRE_AUTO_CALLOUT) != 0 && !is_quantifier && nestptr == NULL) + { previous_callout = code; code = auto_callout(code, ptr, cd); } + /* Process the next pattern item. */ + switch(c) { /* ===================================================================*/ - case 0: /* The branch terminates at string end */ + case CHAR_NULL: /* 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 +4656,7 @@ 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 +4668,14 @@ 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; @@ -3585,7 +4704,29 @@ for (;; ptr++) } goto NORMAL_CHAR; + /* In another (POSIX) regex library, the ugly syntax [[:<:]] and [[:>:]] is + used for "start of word" and "end of word". As these are otherwise illegal + sequences, we don't break anything by recognizing them. They are replaced + by \b(?=\w) and \b(?<=\w) respectively. Sequences like [a[:<:]] are + erroneous and are handled by the normal code below. */ + case CHAR_LEFT_SQUARE_BRACKET: + if (STRNCMP_UC_C8(ptr+1, STRING_WEIRD_STARTWORD, 6) == 0) + { + nestptr = ptr + 7; + ptr = sub_start_of_word - 1; + continue; + } + + if (STRNCMP_UC_C8(ptr+1, STRING_WEIRD_ENDWORD, 6) == 0) + { + nestptr = ptr + 7; + ptr = sub_end_of_word - 1; + continue; + } + + /* Handle a real character class. */ + previous = code; /* PCRE supports POSIX class stuff inside a class. Perl gives an error if @@ -3611,8 +4752,7 @@ 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 +4771,9 @@ 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 +4783,55 @@ 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 +4859,8 @@ 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) { @@ -3746,31 +4891,65 @@ for (;; ptr++) posix_class = 0; /* When PCRE_UCP is set, some of the POSIX classes are converted to - different escape sequences that use Unicode properties. */ + different escape sequences that use Unicode properties \p or \P. Others + that are not available via \p or \P generate XCL_PROP/XCL_NOTPROP + directly. */ #ifdef SUPPORT_UCP if ((options & PCRE_UCP) != 0) { + unsigned int ptype = 0; int pc = posix_class + ((local_negate)? POSIX_SUBSIZE/2 : 0); + + /* The posix_substitutes table specifies which POSIX classes can be + converted to \p or \P items. */ + if (posix_substitutes[pc] != NULL) { nestptr = tempptr + 1; ptr = posix_substitutes[pc] - 1; continue; } + + /* There are three other classes that generate special property calls + that are recognized only in an XCLASS. */ + + else switch(posix_class) + { + case PC_GRAPH: + ptype = PT_PXGRAPH; + /* Fall through */ + case PC_PRINT: + if (ptype == 0) ptype = PT_PXPRINT; + /* Fall through */ + case PC_PUNCT: + if (ptype == 0) ptype = PT_PXPUNCT; + *class_uchardata++ = local_negate? XCL_NOTPROP : XCL_PROP; + *class_uchardata++ = ptype; + *class_uchardata++ = 0; + ptr = tempptr + 1; + continue; + + /* For all other POSIX classes, no special action is taken in UCP + mode. Fall through to the non_UCP case. */ + + default: + break; + } } #endif - /* In the non-UCP case, we build the bit map for the POSIX class in a - chunk of local store because we may be adding and subtracting from it, - and we don't want to subtract bits that may be in the main map already. - At the end we or the result into the bit map that is being built. */ + /* In the non-UCP case, or when UCP makes no difference, we build the + bit map for the POSIX class in a chunk of local store because we may be + adding and subtracting from it, and we don't want to subtract bits that + may be in the main map already. At the end we or the result into the + bit map that is being built. */ posix_class *= 3; /* Copy in the first table (always present) */ memcpy(pbits, cbits + posix_class_maps[posix_class], - 32 * sizeof(uschar)); + 32 * sizeof(pcre_uint8)); /* If there is a second table, add or remove it as required. */ @@ -3785,7 +4964,7 @@ 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 +4980,35 @@ 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 +5017,17 @@ 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 +5037,8 @@ 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: @@ -3872,136 +5059,42 @@ for (;; ptr++) for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word]; continue; - /* 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. */ + /* Perl 5.004 onwards omitted VT from \s, but restored it at Perl + 5.18. Before PCRE 8.34, we had to preserve the VT bit if it was + previously set by something earlier in the character class. + Luckily, the value of CHAR_VT is 0x0b in both ASCII and EBCDIC, so + we could just adjust the appropriate bit. From PCRE 8.34 we no + longer treat \s and \S specially. */ case ESC_s: - classbits[0] |= cbits[cbit_space]; - classbits[1] |= cbits[cbit_space+1] & ~0x08; - for (c = 2; c < 32; c++) classbits[c] |= cbits[c+cbit_space]; + for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_space]; continue; case ESC_S: should_flip_negation = TRUE; for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space]; - 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 +5102,14 @@ 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 +5123,24 @@ 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 +5148,9 @@ 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 +5158,7 @@ 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 +5174,19 @@ 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 */ } @@ -4095,263 +5194,196 @@ for (;; ptr++) #endif d = *ptr; /* Not UTF-8 mode */ - /* The second part of a range can be a single-character escape, but - not any of the other escapes. Perl 5.6 treats a hyphen as a literal - in such circumstances. */ + /* The second part of a range can be a single-character escape + sequence, but not any of the other escapes. Perl treats a hyphen as a + literal in such circumstances. However, in Perl's warning mode, a + warning is given, so PCRE now faults it as it is almost certainly a + mistake on the user's part. */ - if (!inescq && d == CHAR_BACKSLASH) + if (!inescq) { - d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE); - if (*errorcodeptr != 0) goto FAILED; + if (d == CHAR_BACKSLASH) + { + 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 */ + /* 0 means a character was put into d; \b is backspace; any other + special causes an error. */ - if (d < 0) - { - if (d == -ESC_b) d = CHAR_BS; else + if (descape != 0) { - ptr = oldptr; - goto LONE_SINGLE_CHARACTER; /* A few lines below */ + if (descape == ESC_b) d = CHAR_BS; else + { + *errorcodeptr = ERR83; + goto FAILED; + } } } + + /* A hyphen followed by a POSIX class is treated in the same way. */ + + else if (d == CHAR_LEFT_SQUARE_BRACKET && + (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT || + ptr[1] == CHAR_EQUALS_SIGN) && + check_posix_syntax(ptr, &tempptr)) + { + *errorcodeptr = ERR83; + goto FAILED; + } } /* 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 +5392,30 @@ 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 +5431,15 @@ 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 +5476,10 @@ 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 */ @@ -4456,6 +5494,34 @@ for (;; ptr++) tempcode = previous; + /* Before checking for a possessive quantifier, we must skip over + whitespace and comments in extended mode because Perl allows white space at + this point. */ + + if ((options & PCRE_EXTENDED) != 0) + { + const pcre_uchar *p = ptr + 1; + for (;;) + { + while (MAX_255(*p) && (cd->ctypes[*p] & ctype_space) != 0) p++; + if (*p != CHAR_NUMBER_SIGN) break; + p++; + while (*p != CHAR_NULL) + { + if (IS_NEWLINE(p)) /* For non-fixed-length newline cases, */ + { /* IS_NEWLINE sets cd->nllen. */ + p += cd->nllen; + break; + } + p++; +#ifdef SUPPORT_UTF + if (utf) FORWARDCHAR(p); +#endif + } /* Loop for comment characters */ + } /* Loop for multiple comments */ + ptr = p - 1; /* Character before the next significant one. */ + } + /* If the next character is '+', we have a possessive quantifier. This implies greediness, whatever the setting of the PCRE_UNGREEDY option. If the next character is '?' this is a minimizing repeat, by default, @@ -4483,7 +5549,7 @@ 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; @@ -4504,77 +5570,55 @@ 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 - the line is something that cannot possibly match this character. If so, - automatically possessifying this item gains some performance in the case - where the match fails. */ - - 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; /* 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 @@ -4584,19 +5628,11 @@ 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)) - { - repeat_type = 0; /* Force greedy */ - possessive_quantifier = TRUE; - } - OUTPUT_SINGLE_REPEAT: if (*previous == OP_PROP || *previous == OP_NOTPROP) { @@ -4613,16 +5649,6 @@ 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 +5697,14 @@ 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 +5726,10 @@ 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 +5756,10 @@ 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 @@ -4755,13 +5781,12 @@ for (;; ptr++) /* If previous was a character class or a back reference, we put the repeat stuff after it, but just skip the item if the repeat was {0,0}. */ - else if (*previous == OP_CLASS || - *previous == OP_NCLASS || -#ifdef SUPPORT_UTF8 + else if (*previous == OP_CLASS || *previous == OP_NCLASS || +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 *previous == OP_XCLASS || #endif - *previous == OP_REF || - *previous == OP_REFI) + *previous == OP_REF || *previous == OP_REFI || + *previous == OP_DNREF || *previous == OP_DNREFI) { if (repeat_max == 0) { @@ -4769,16 +5794,6 @@ 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) @@ -4799,15 +5814,16 @@ for (;; ptr++) opcodes such as BRA and CBRA, as this is the place where they get converted into the more special varieties such as BRAPOS and SBRA. A test for >= OP_ASSERT and <= OP_COND includes ASSERT, ASSERT_NOT, ASSERTBACK, - ASSERTBACK_NOT, ONCE, BRA, CBRA, and COND. Originally, PCRE did not allow - repetition of assertions, but now it does, for Perl compatibility. */ + ASSERTBACK_NOT, ONCE, ONCE_NC, BRA, BRAPOS, CBRA, CBRAPOS, and COND. + Originally, PCRE did not allow repetition of assertions, but now it does, + for Perl compatibility. */ else if (*previous >= OP_ASSERT && *previous <= OP_COND) { 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. */ @@ -4818,7 +5834,7 @@ for (;; ptr++) /* There is no sense in actually repeating assertions. The only potential use of repetition is in cases when the assertion is optional. Therefore, if the minimum is greater than zero, just ignore the repeat. If the - maximum is not not zero or one, set it to 1. */ + maximum is not zero or one, set it to 1. */ if (*previous < OP_ONCE) /* Assertion */ { @@ -4860,8 +5876,8 @@ 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 +5900,8 @@ 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 +5954,17 @@ 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 +5973,8 @@ 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 +6026,8 @@ 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 +6043,7 @@ 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 +6055,8 @@ 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 +6075,7 @@ 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 +6111,8 @@ 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 +6134,10 @@ 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, NULL)) { *bracode += OP_SBRA - OP_BRA; break; @@ -5140,8 +6160,8 @@ 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; @@ -5187,42 +6207,105 @@ for (;; ptr++) goto FAILED; } - /* If the character following a repeat is '+', or if certain optimization - tests above succeeded, possessive_quantifier is TRUE. For some opcodes, - there are special alternative opcodes for this case. For anything else, we - wrap the entire repeated item inside OP_ONCE brackets. Logically, the '+' - notation is just syntactic sugar, taken from Sun's Java package, but the - special opcodes can optimize it. + /* If the character following a repeat is '+', possessive_quantifier is + TRUE. For some opcodes, there are special alternative opcodes for this + case. For anything else, we wrap the entire repeated item inside OP_ONCE + brackets. Logically, the '+' notation is just syntactic sugar, taken from + Sun's Java package, but the special opcodes can optimize it. Some (but not all) possessively repeated subpatterns have already been completely handled in the code just above. For them, possessive_quantifier - is always FALSE at this stage. + is always FALSE at this stage. Note that the repeated item starts at + tempcode, not at previous, which might be the first part of a string whose + (former) last char we repeated. */ - Note that the repeated item starts at tempcode, not at previous, which - might be the first part of a string whose (former) last char we repeated. - - Possessifying an 'exact' quantifier has no effect, so we can ignore it. But - an 'upto' may follow. We skip over an 'exact' item, and then test the - length of what remains before proceeding. */ - if (possessive_quantifier) { int len; - if (*tempcode == OP_TYPEEXACT) - tempcode += _pcre_OP_lengths[*tempcode] + - ((tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP)? 2 : 0); + /* Possessifying an EXACT quantifier has no effect, so we can ignore it. + However, QUERY, STAR, or UPTO may follow (for quantifiers such as {5,6}, + {5,}, or {5,10}). We skip over an EXACT item; if the length of what + remains is greater than zero, there's a further opcode that can be + handled. If not, do nothing, leaving the EXACT alone. */ - else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT) + switch(*tempcode) { - tempcode += _pcre_OP_lengths[*tempcode]; -#ifdef SUPPORT_UTF8 - if (utf8 && tempcode[-1] >= 0xc0) - tempcode += _pcre_utf8_table4[tempcode[-1] & 0x3f]; + case OP_TYPEEXACT: + tempcode += PRIV(OP_lengths)[*tempcode] + + ((tempcode[1 + IMM2_SIZE] == OP_PROP + || tempcode[1 + IMM2_SIZE] == OP_NOTPROP)? 2 : 0); + break; + + /* CHAR opcodes are used for exacts whose count is 1. */ + + case OP_CHAR: + case OP_CHARI: + case OP_NOT: + case OP_NOTI: + case OP_EXACT: + case OP_EXACTI: + case OP_NOTEXACT: + case OP_NOTEXACTI: + tempcode += PRIV(OP_lengths)[*tempcode]; +#ifdef SUPPORT_UTF + if (utf && HAS_EXTRALEN(tempcode[-1])) + tempcode += GET_EXTRALEN(tempcode[-1]); #endif + break; + + /* For the class opcodes, the repeat operator appears at the end; + adjust tempcode to point to it. */ + + case OP_CLASS: + case OP_NCLASS: + tempcode += 1 + 32/sizeof(pcre_uchar); + break; + +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + case OP_XCLASS: + tempcode += GET(tempcode, 1); + break; +#endif } + /* If tempcode is equal to code (which points to the end of the repeated + item), it means we have skipped an EXACT item but there is no following + QUERY, STAR, or UPTO; the value of len will be 0, and we do nothing. In + all other cases, tempcode will be pointing to the repeat opcode, and will + be less than code, so the value of len will be greater than 0. */ + len = (int)(code - tempcode); + if (len > 0) + { + unsigned int repcode = *tempcode; + + /* There is a table for possessifying opcodes, all of which are less + than OP_CALLOUT. A zero entry means there is no possessified version. + */ + + if (repcode < OP_CALLOUT && opcode_possessify[repcode] > 0) + *tempcode = opcode_possessify[repcode]; + + /* For opcode without a special possessified version, wrap the item in + ONCE brackets. Because we are moving code along, we must ensure that any + pending recursive references are updated. */ + + else + { + *code = OP_END; + 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; + *code++ = OP_KET; + PUTINC(code, 0, len); + PUT(tempcode, 1, len); + } + } + +#ifdef NEVER if (len > 0) switch (*tempcode) { case OP_STAR: *tempcode = OP_POSSTAR; break; @@ -5250,13 +6333,18 @@ for (;; ptr++) case OP_TYPEQUERY: *tempcode = OP_TYPEPOSQUERY; break; case OP_TYPEUPTO: *tempcode = OP_TYPEPOSUPTO; break; + case OP_CRSTAR: *tempcode = OP_CRPOSSTAR; break; + case OP_CRPLUS: *tempcode = OP_CRPOSPLUS; break; + case OP_CRQUERY: *tempcode = OP_CRPOSQUERY; break; + case OP_CRRANGE: *tempcode = OP_CRPOSRANGE; break; + /* Because we are moving code along, we must ensure that any pending recursive references are updated. */ 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; @@ -5265,10 +6353,11 @@ for (;; ptr++) PUT(tempcode, 1, len); break; } +#endif } /* 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 +6380,18 @@ 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 +6401,13 @@ 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 +6421,10 @@ 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 +6442,11 @@ 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 +6458,7 @@ 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 +6468,28 @@ 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 +6508,15 @@ 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; @@ -5427,38 +6539,53 @@ 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 - recursion. R and R&name are also permitted for recursion tests. + group's having been set), a name (referring to a named group), or 'R', + referring to recursion. R and R&name are also permitted for + recursion tests. - There are several syntaxes for testing a named group: (?(name)) is used - by Python; Perl 5.10 onwards uses (?() or (?('name')). + There are ways of testing a named group: (?(name)) is used by Python; + Perl 5.10 onwards uses (?() or (?('name')). - There are two unfortunate ambiguities, caused by history. (a) 'R' can - 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. */ + There is one unfortunate ambiguity, caused by history. 'R' can be the + recursive thing or the name 'R' (and similarly for 'R' followed by + digits). We look for a name first; if not found, we try the other case. + 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. */ + /* Other conditions use OP_CREF/OP_DNCREF/OP_RREF/OP_DNRREF, and all + need to skip at least 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. */ - if (ptr[1] == CHAR_R && ptr[2] == CHAR_AMPERSAND) + ptr++; + if (*ptr == CHAR_R && ptr[1] == CHAR_AMPERSAND) { terminator = -1; ptr += 2; @@ -5466,51 +6593,71 @@ for (;; ptr++) } /* Check for a test for a named group's having been set, using the Perl - syntax (?() or (?('name') */ + syntax (?() or (?('name'), and also allow for the original PCRE + syntax of (?(name) or for (?(+n), (?(-n), and just (?(n). */ - else if (ptr[1] == CHAR_LESS_THAN_SIGN) + else if (*ptr == CHAR_LESS_THAN_SIGN) { terminator = CHAR_GREATER_THAN_SIGN; ptr++; } - else if (ptr[1] == CHAR_APOSTROPHE) + else if (*ptr == CHAR_APOSTROPHE) { terminator = CHAR_APOSTROPHE; ptr++; } else { - terminator = 0; - if (ptr[1] == CHAR_MINUS || ptr[1] == CHAR_PLUS) refsign = *(++ptr); + terminator = CHAR_NULL; + if (*ptr == CHAR_MINUS || *ptr == CHAR_PLUS) refsign = *ptr++; + else if (IS_DIGIT(*ptr)) refsign = 0; } - /* We now expect to read a name; any thing else is an error */ + /* Handle a number */ - if ((cd->ctypes[ptr[1]] & ctype_word) == 0) + if (refsign >= 0) { - ptr += 1; /* To get the right offset */ - *errorcodeptr = ERR28; - goto FAILED; + recno = 0; + while (IS_DIGIT(*ptr)) + { + recno = recno * 10 + (int)(*ptr - CHAR_0); + ptr++; + } } - /* Read the name, but also get it as a number if it's all digits */ + /* Otherwise we expect to read a name; anything else is an error. When + a name is one of a number of duplicates, a different opcode is used and + it needs more memory. Unfortunately we cannot tell whether a name is a + duplicate in the first pass, so we have to allow for more memory. */ - recno = 0; - name = ++ptr; - while ((cd->ctypes[*ptr] & ctype_word) != 0) + else { - if (recno >= 0) - recno = ((digitab[*ptr] & ctype_digit) != 0)? - recno * 10 + *ptr - CHAR_0 : -1; - ptr++; + if (IS_DIGIT(*ptr)) + { + *errorcodeptr = ERR84; + goto FAILED; + } + if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_word) == 0) + { + *errorcodeptr = ERR28; /* Assertion expected */ + goto FAILED; + } + name = ptr++; + while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) + { + ptr++; + } + namelen = (int)(ptr - name); + if (lengthptr != NULL) *lengthptr += IMM2_SIZE; } - namelen = (int)(ptr - name); - if ((terminator > 0 && *ptr++ != terminator) || + /* Check the terminator */ + + if ((terminator > 0 && *ptr++ != (pcre_uchar)terminator) || *ptr++ != CHAR_RIGHT_PARENTHESIS) { - ptr--; /* Error offset */ - *errorcodeptr = ERR26; + ptr--; /* Error offset */ + *errorcodeptr = ERR26; /* Malformed number or name */ goto FAILED; } @@ -5519,18 +6666,18 @@ for (;; ptr++) if (lengthptr != NULL) break; /* In the real compile we do the work of looking for the actual - reference. If the string started with "+" or "-" we require the rest to - be digits, in which case recno will be set. */ + reference. If refsign is not negative, it means we have a number in + recno. */ - if (refsign > 0) + if (refsign >= 0) { if (recno <= 0) { - *errorcodeptr = ERR58; + *errorcodeptr = ERR35; goto FAILED; } - recno = (refsign == CHAR_MINUS)? - cd->bracount - recno + 1 : recno +cd->bracount; + if (refsign != 0) recno = (refsign == CHAR_MINUS)? + cd->bracount - recno + 1 : recno + cd->bracount; if (recno <= 0 || recno > cd->final_bracount) { *errorcodeptr = ERR15; @@ -5540,45 +6687,52 @@ for (;; ptr++) break; } - /* Otherwise (did not start with "+" or "-"), start by looking for the - name. If we find a name, add one to the opcode to change OP_CREF or - OP_RREF into OP_NCREF or OP_NRREF. These behave exactly the same, - except they record that the reference was originally to a name. The - information is used to check duplicate names. */ + /* Otherwise look for the name. */ 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; } - /* Found a previous named subpattern */ + /* Found the named subpattern. If the name is duplicated, add one to + the opcode to change CREF/RREF into DNCREF/DNRREF and insert + appropriate data values. Otherwise, just insert the unique subpattern + number. */ if (i < cd->names_found) { - recno = GET2(slot, 0); - PUT2(code, 2+LINK_SIZE, recno); - code[1+LINK_SIZE]++; + int offset = i++; + int count = 1; + recno = GET2(slot, 0); /* Number from first found */ + for (; i < cd->names_found; i++) + { + slot += cd->name_entry_size; + if (STRNCMP_UC_UC(name, slot+IMM2_SIZE, namelen) != 0) break; + count++; + } + if (count > 1) + { + PUT2(code, 2+LINK_SIZE, offset); + PUT2(code, 2+LINK_SIZE+IMM2_SIZE, count); + skipbytes += IMM2_SIZE; + code[1+LINK_SIZE]++; + } + else /* Not a duplicated name */ + { + PUT2(code, 2+LINK_SIZE, recno); + } } - /* Search the pattern for a forward reference */ + /* 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 != + CHAR_NULL [things like (?(... or (?('name')... or (?(R&name)... ] + we have now checked all the possibilities, so give an error. */ - else if ((i = find_parens(cd, name, namelen, - (options & PCRE_EXTENDED) != 0, utf8)) > 0) + else if (terminator != CHAR_NULL) { - 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 - now checked all the possibilities, so give an error. */ - - else if (terminator != 0) - { *errorcodeptr = ERR15; goto FAILED; } @@ -5591,7 +6745,7 @@ 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; @@ -5606,25 +6760,17 @@ 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; } - /* Check for the "name" actually being a subpattern number. We are - in the second pass here, so final_bracount is set. */ + /* Reference to an unidentified subpattern. */ - else if (recno > 0 && recno <= cd->final_bracount) - { - PUT2(code, 2+LINK_SIZE, recno); - } - - /* Either an unidentified subpattern, or a reference to (?(0) */ - else { - *errorcodeptr = (recno == 0)? ERR35: ERR15; + *errorcodeptr = ERR15; goto FAILED; } break; @@ -5637,11 +6783,18 @@ for (;; ptr++) ptr++; break; + /* Optimize (?!) to (*FAIL) unless it is quantified - which is a weird + thing to do, but Perl allows all assertions to be quantified, and when + they contain capturing parentheses there may be a potential use for + this feature. Not that that applies to a quantified (?!) but we allow + it for uniformity. */ /* ------------------------------------------------------------ */ case CHAR_EXCLAMATION_MARK: /* Negative lookahead */ ptr++; - if (*ptr == CHAR_RIGHT_PARENTHESIS) /* Optimize (?!) */ + if (*ptr == CHAR_RIGHT_PARENTHESIS && ptr[1] != CHAR_ASTERISK && + ptr[1] != CHAR_PLUS && ptr[1] != CHAR_QUESTION_MARK && + (ptr[1] != CHAR_LEFT_CURLY_BRACKET || !is_counted_repeat(ptr+2))) { *code++ = OP_FAIL; previous = NULL; @@ -5669,7 +6822,8 @@ 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 +6845,9 @@ 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; @@ -5732,124 +6887,110 @@ for (;; ptr++) /* ------------------------------------------------------------ */ DEFINE_NAME: /* Come here from (?< handling */ case CHAR_APOSTROPHE: + terminator = (*ptr == CHAR_LESS_THAN_SIGN)? + CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE; + name = ++ptr; + if (IS_DIGIT(*ptr)) { - terminator = (*ptr == CHAR_LESS_THAN_SIGN)? - CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE; - name = ++ptr; + *errorcodeptr = ERR84; /* Group name must start with non-digit */ + goto FAILED; + } + while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++; + namelen = (int)(ptr - name); - while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++; - namelen = (int)(ptr - name); + /* In the pre-compile phase, do a syntax check, remember the longest + name, and then remember the group in a vector, expanding it if + necessary. Duplicates for the same number are skipped; other duplicates + are checked for validity. In the actual compile, there is nothing to + do. */ - /* In the pre-compile phase, just do a syntax check. */ + if (lengthptr != NULL) + { + named_group *ng; + pcre_uint32 number = cd->bracount + 1; - if (lengthptr != NULL) + if (*ptr != (pcre_uchar)terminator) { - if (*ptr != terminator) + *errorcodeptr = ERR42; + goto FAILED; + } + + if (cd->names_found >= MAX_NAME_COUNT) + { + *errorcodeptr = ERR49; + goto FAILED; + } + + if (namelen + IMM2_SIZE + 1 > cd->name_entry_size) + { + cd->name_entry_size = namelen + IMM2_SIZE + 1; + if (namelen > MAX_NAME_SIZE) { - *errorcodeptr = ERR42; + *errorcodeptr = ERR48; goto FAILED; } - if (cd->names_found >= MAX_NAME_COUNT) + } + + /* Scan the list to check for duplicates. For duplicate names, if the + number is the same, break the loop, which causes the name to be + discarded; otherwise, if DUPNAMES is not set, give an error. + If it is set, allow the name with a different number, but continue + scanning in case this is a duplicate with the same number. For + non-duplicate names, give an error if the number is duplicated. */ + + ng = cd->named_groups; + for (i = 0; i < cd->names_found; i++, ng++) + { + if (namelen == ng->length && + STRNCMP_UC_UC(name, ng->name, namelen) == 0) { - *errorcodeptr = ERR49; - goto FAILED; - } - if (namelen + 3 > cd->name_entry_size) - { - cd->name_entry_size = namelen + 3; - if (namelen > MAX_NAME_SIZE) + if (ng->number == number) break; + if ((options & PCRE_DUPNAMES) == 0) { - *errorcodeptr = ERR48; + *errorcodeptr = ERR43; goto FAILED; } + cd->dupnames = TRUE; /* Duplicate names exist */ } + else if (ng->number == number) + { + *errorcodeptr = ERR65; + goto FAILED; + } } - /* In the real compile, create the entry in the table, maintaining - alphabetical order. Duplicate names for different numbers are - permitted only if PCRE_DUPNAMES is set. Duplicate names for the same - number are always OK. (An existing number can be re-used if (?| - appears in the pattern.) In either event, a duplicate name results in - a duplicate entry in the table, even if the number is the same. This - is because the number of names, and hence the table size, is computed - in the pre-compile, and it affects various numbers and pointers which - would all have to be modified, and the compiled code moved down, if - duplicates with the same number were omitted from the table. This - doesn't seem worth the hassle. However, *different* names for the - same number are not permitted. */ - - else + if (i >= cd->names_found) /* Not a duplicate with same number */ { - BOOL dupname = FALSE; - slot = cd->name_table; + /* Increase the list size if necessary */ - for (i = 0; i < cd->names_found; i++) + if (cd->names_found >= cd->named_group_list_size) { - int crc = memcmp(name, slot+2, namelen); - if (crc == 0) - { - if (slot[2+namelen] == 0) - { - if (GET2(slot, 0) != cd->bracount + 1 && - (options & PCRE_DUPNAMES) == 0) - { - *errorcodeptr = ERR43; - goto FAILED; - } - else dupname = TRUE; - } - else crc = -1; /* Current name is a substring */ - } + int newsize = cd->named_group_list_size * 2; + named_group *newspace = (PUBL(malloc)) + (newsize * sizeof(named_group)); - /* Make space in the table and break the loop for an earlier - name. For a duplicate or later name, carry on. We do this for - duplicates so that in the simple case (when ?(| is not used) they - are in order of their numbers. */ - - if (crc < 0) + if (newspace == NULL) { - memmove(slot + cd->name_entry_size, slot, - (cd->names_found - i) * cd->name_entry_size); - break; + *errorcodeptr = ERR21; + goto FAILED; } - /* Continue the loop for a later or duplicate name */ - - slot += cd->name_entry_size; + memcpy(newspace, cd->named_groups, + cd->named_group_list_size * sizeof(named_group)); + if (cd->named_group_list_size > NAMED_GROUP_LIST_SIZE) + (PUBL(free))((void *)cd->named_groups); + cd->named_groups = newspace; + cd->named_group_list_size = newsize; } - /* For non-duplicate names, check for a duplicate number before - adding the new name. */ - - if (!dupname) - { - uschar *cslot = cd->name_table; - for (i = 0; i < cd->names_found; i++) - { - if (cslot != slot) - { - if (GET2(cslot, 0) == cd->bracount + 1) - { - *errorcodeptr = ERR65; - goto FAILED; - } - } - else i--; - cslot += cd->name_entry_size; - } - } - - PUT2(slot, 0, cd->bracount + 1); - memcpy(slot + 2, name, namelen); - slot[2+namelen] = 0; + cd->named_groups[cd->names_found].name = name; + cd->named_groups[cd->names_found].length = namelen; + cd->named_groups[cd->names_found].number = number; + cd->names_found++; } } - /* In both pre-compile and compile, count the number of names we've - encountered. */ - - cd->names_found++; - ptr++; /* Move past > or ' */ + ptr++; /* Move past > or ' in both passes. */ goto NUMBERED_GROUP; @@ -5867,7 +7008,12 @@ for (;; ptr++) NAMED_REF_OR_RECURSE: name = ++ptr; - while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++; + if (IS_DIGIT(*ptr)) + { + *errorcodeptr = ERR84; /* Group name must start with non-digit */ + goto FAILED; + } + 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 +7025,14 @@ for (;; ptr++) if (lengthptr != NULL) { - const uschar *temp; + named_group *ng; if (namelen == 0) { *errorcodeptr = ERR62; goto FAILED; } - if (*ptr != terminator) + if (*ptr != (pcre_uchar)terminator) { *errorcodeptr = ERR42; goto FAILED; @@ -5897,59 +7043,113 @@ for (;; ptr++) goto FAILED; } - /* The name table does not exist in the first pass, so we cannot - do a simple search as in the code below. Instead, we have to scan the - pattern to find the number. It is important that we scan it only as - far as we have got because the syntax of named subpatterns has not - been checked for the rest of the pattern, and find_parens() assumes - correct syntax. In any case, it's a waste of resources to scan - further. We stop the scan at the current point by temporarily - adjusting the value of cd->endpattern. */ + /* The name table does not exist in the first pass; instead we must + scan the list of names encountered so far in order to get the + number. If the name is not found, set the value to 0 for a forward + reference. */ - temp = cd->end_pattern; - cd->end_pattern = ptr; - recno = find_parens(cd, name, namelen, - (options & PCRE_EXTENDED) != 0, utf8); - cd->end_pattern = temp; - if (recno < 0) recno = 0; /* Forward ref; set dummy number */ + ng = cd->named_groups; + for (i = 0; i < cd->names_found; i++, ng++) + { + if (namelen == ng->length && + STRNCMP_UC_UC(name, ng->name, namelen) == 0) + break; + } + recno = (i < cd->names_found)? ng->number : 0; + + /* Count named back references. */ + + if (!is_recurse) cd->namedrefcount++; } - /* In the real compile, seek the name in the table. We check the name + /* In the real compile, search the name table. We check the name first, and then check that we have reached the end of the name in the - table. That way, if the name that is longer than any in the table, - the comparison will fail without reading beyond the table entry. */ + table. That way, if the name is longer than any in the table, the + comparison will fail without reading beyond the table entry. */ else { 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; } - if (i < cd->names_found) /* Back reference */ + if (i < cd->names_found) { recno = GET2(slot, 0); } - else if ((recno = /* Forward back reference */ - find_parens(cd, name, namelen, - (options & PCRE_EXTENDED) != 0, utf8)) <= 0) + else { *errorcodeptr = ERR15; goto FAILED; } } - /* In both phases, we can now go to the code than handles numerical - recursion or backreferences. */ + /* In both phases, for recursions, we can now go to the code than + handles numerical recursion. */ if (is_recurse) goto HANDLE_RECURSION; - else goto HANDLE_REFERENCE; + /* In the second pass we must see if the name is duplicated. If so, we + generate a different opcode. */ + if (lengthptr == NULL && cd->dupnames) + { + int count = 1; + unsigned int index = i; + pcre_uchar *cslot = slot + cd->name_entry_size; + + for (i++; i < cd->names_found; i++) + { + if (STRCMP_UC_UC(slot + IMM2_SIZE, cslot + IMM2_SIZE) != 0) break; + count++; + cslot += cd->name_entry_size; + } + + if (count > 1) + { + if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; + previous = code; + *code++ = ((options & PCRE_CASELESS) != 0)? OP_DNREFI : OP_DNREF; + PUT2INC(code, 0, index); + PUT2INC(code, 0, count); + + /* Process each potentially referenced group. */ + + for (; slot < cslot; slot += cd->name_entry_size) + { + open_capitem *oc; + recno = GET2(slot, 0); + cd->backref_map |= (recno < 32)? (1 << recno) : 1; + if (recno > cd->top_backref) cd->top_backref = recno; + + /* Check to see if this back reference is recursive, that it, it + is inside the group that it references. A flag is set so that the + group can be made atomic. */ + + for (oc = cd->open_caps; oc != NULL; oc = oc->next) + { + if (oc->number == recno) + { + oc->flag = TRUE; + break; + } + } + } + + continue; /* End of back ref handling */ + } + } + + /* First pass, or a non-duplicated name. */ + + goto HANDLE_REFERENCE; + + /* ------------------------------------------------------------ */ case CHAR_R: /* Recursion */ ptr++; /* Same as (?0) */ @@ -5961,7 +7161,7 @@ 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 +7175,7 @@ for (;; ptr++) if ((refsign = *ptr) == CHAR_PLUS) { ptr++; - if ((digitab[*ptr] & ctype_digit) == 0) + if (!IS_DIGIT(*ptr)) { *errorcodeptr = ERR63; goto FAILED; @@ -5983,16 +7183,16 @@ 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 +7240,13 @@ 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) + if (recno > cd->final_bracount) { *errorcodeptr = ERR15; goto FAILED; @@ -6077,7 +7276,7 @@ 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 +7284,18 @@ 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 +7352,7 @@ 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 +7365,7 @@ 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 @@ -6203,14 +7402,23 @@ 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, - but this was changed for Perl compatibility, so all kinds can now be - repeated. We copy code into a non-register variable (tempcode) in order to - be able to pass its address because some compilers complain otherwise. */ + /* Process nested bracketed regex. First check for parentheses nested too + deeply. */ + if ((cd->parens_depth += 1) > PARENS_NEST_LIMIT) + { + *errorcodeptr = ERR82; + goto FAILED; + } + + /* Assertions used not to be repeatable, but this was changed for Perl + compatibility, so all kinds can now be repeated. We copy code into a + non-register variable (tempcode) in order to be able to pass its address + because some compilers complain otherwise. */ + previous = code; /* For handling repetition */ *code = bravalue; tempcode = code; @@ -6229,8 +7437,10 @@ 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 */ @@ -6238,6 +7448,8 @@ for (;; ptr++) )) goto FAILED; + cd->parens_depth -= 1; + /* If this was an atomic group and there are no capturing groups within it, generate OP_ONCE_NC instead of OP_ONCE. */ @@ -6258,7 +7470,7 @@ for (;; ptr++) if (bravalue == OP_COND && lengthptr == NULL) { - uschar *tc = code; + pcre_uchar *tc = code; int condcount = 0; do { @@ -6281,7 +7493,7 @@ 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 +7503,7 @@ for (;; ptr++) *errorcodeptr = ERR27; goto FAILED; } - if (condcount == 1) subfirstbyte = subreqbyte = REQ_NONE; + if (condcount == 1) subfirstcharflags = subreqcharflags = REQ_NONE; } } @@ -6335,55 +7547,69 @@ 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 +7617,21 @@ 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 */ @@ -6411,70 +7639,60 @@ 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; + pcre_uint32 cf; + save_hwm = cd->hwm; /* Normally this is set when '(' is read */ terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)? CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE; /* These two statements stop the compiler for warning about possibly unset variables caused by the jump to HANDLE_NUMERICAL_RECURSION. In - fact, because we actually check for a number below, the paths that + fact, because we do the check for a number below, the paths that would actually be in error are never taken. */ skipbytes = 0; reset_bracount = FALSE; - /* Test for a name */ + /* If it's not a signed or unsigned number, treat it as a name. */ - if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS) + cf = ptr[1]; + if (cf != CHAR_PLUS && cf != CHAR_MINUS && !IS_DIGIT(cf)) { - BOOL isnumber = TRUE; - for (p = ptr + 1; *p != 0 && *p != terminator; p++) - { - if ((cd->ctypes[*p] & ctype_digit) == 0) isnumber = FALSE; - if ((cd->ctypes[*p] & ctype_word) == 0) break; - } - if (*p != terminator) - { - *errorcodeptr = ERR57; - break; - } - if (isnumber) - { - ptr++; - goto HANDLE_NUMERICAL_RECURSION; - } is_recurse = TRUE; goto NAMED_REF_OR_RECURSE; } - /* Test a signed number in angle brackets or quotes. */ + /* Signed or unsigned number (cf = ptr[1]) is known to be plus or minus + or a digit. */ 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 +7704,7 @@ 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 +7719,20 @@ 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; + /* Come here from named backref handling when the reference is to a + single group (i.e. not to a duplicated name. */ + + HANDLE_REFERENCE: + if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; previous = code; *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF; PUT2INC(code, 0, recno); @@ -6535,14 +7756,14 @@ 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 +7772,7 @@ 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 +7782,19 @@ 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 +7802,8 @@ 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 +7813,9 @@ 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 @@ -6603,20 +7828,17 @@ for (;; ptr++) /* ===================================================================*/ /* Handle a literal character. It is guaranteed not to be whitespace or # - when the extended flag is set. If we are in UTF-8 mode, it may be a - multi-byte literal character. */ + when the extended flag is set. If we are in a UTF mode, it may be a + multi-unit literal character. */ default: NORMAL_CHAR: 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 +7846,29 @@ 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 +7879,47 @@ 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 +7938,6 @@ return FALSE; - /************************************************* * Compile sequence of alternatives * *************************************************/ @@ -6693,48 +7950,55 @@ out the amount of memory needed, as well as during the value of lengthptr distinguishes the two phases. Arguments: - options option bits, including any changes for this subpattern - codeptr -> the address of the current code pointer - ptrptr -> the address of the current pattern pointer - errorcodeptr -> pointer to error code variable - lookbehind TRUE if this is a lookbehind assertion - 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 - 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 - points to length accumulator during pre-compile phase + options option bits, including any changes for this subpattern + codeptr -> the address of the current code pointer + ptrptr -> the address of the current pattern pointer + errorcodeptr -> pointer to error code variable + lookbehind TRUE if this is a lookbehind assertion + 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 + 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 + points to length accumulator during pre-compile phase -Returns: TRUE on success +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 +8057,9 @@ 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 +8074,58 @@ 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 +8154,12 @@ 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 +8200,7 @@ 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 +8220,10 @@ 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 +8293,27 @@ 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 +8321,7 @@ 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 +8331,40 @@ 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 +8382,28 @@ 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 @@ -7111,18 +8414,18 @@ 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: - case OP_NCREF: + case OP_DNCREF: case OP_RREF: - case OP_NRREF: + case OP_DNRREF: case OP_DEF: 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 +8439,7 @@ 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 +8449,40 @@ 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; @@ -7186,32 +8504,40 @@ return TRUE; discarded, because they can cause conflicts with actual literals that follow. However, if we end up without a first char setting for an unanchored pattern, it is worth scanning the regex to see if there is an initial asserted first -char. If all branches start with the same asserted char, or with a bracket all -of whose alternatives start with the same asserted char (recurse ad lib), then -we return that char, otherwise -1. +char. If all branches start with the same asserted char, or with a +non-conditional bracket all of whose alternatives start with the same asserted +char (recurse ad lib), then we return that char, with the flags set to zero or +REQ_CASELESS; otherwise return zero with REQ_NONE in the flags. 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: @@ -7222,48 +8548,105 @@ do { case OP_ASSERT: 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; } /************************************************* +* Add an entry to the name/number table * +*************************************************/ + +/* This function is called between compiling passes to add an entry to the +name/number table, maintaining alphabetical order. Checking for permitted +and forbidden duplicates has already been done. + +Arguments: + cd the compile data block + name the name to add + length the length of the name + groupno the group number + +Returns: nothing +*/ + +static void +add_name(compile_data *cd, const pcre_uchar *name, int length, + unsigned int groupno) +{ +int i; +pcre_uchar *slot = cd->name_table; + +for (i = 0; i < cd->names_found; i++) + { + int crc = memcmp(name, slot+IMM2_SIZE, IN_UCHARS(length)); + if (crc == 0 && slot[IMM2_SIZE+length] != 0) + crc = -1; /* Current name is a substring */ + + /* Make space in the table and break the loop for an earlier name. For a + duplicate or later name, carry on. We do this for duplicates so that in the + simple case (when ?(| is not used) they are in order of their numbers. In all + cases they are in the order in which they appear in the pattern. */ + + if (crc < 0) + { + memmove(slot + cd->name_entry_size, slot, + IN_UCHARS((cd->names_found - i) * cd->name_entry_size)); + break; + } + + /* Continue the loop for a later or duplicate name */ + + slot += cd->name_entry_size; + } + +PUT2(slot, 0, groupno); +memcpy(slot + IMM2_SIZE, name, IN_UCHARS(length)); +slot[IMM2_SIZE + length] = 0; +cd->names_found++; +} + + + +/************************************************* * Compile a Regular Expression * *************************************************/ @@ -7285,28 +8668,59 @@ 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 +8731,16 @@ 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]; +/* This vector is used for remembering name groups during the pre-compile. In a +similar way to cworkspace, it can be expanded using malloc() if necessary. */ + +named_group named_groups[NAMED_GROUP_LIST_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 +8767,7 @@ 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 +8781,99 @@ 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_AUTO_POSSESS_RIGHTPAR, 16) == 0) + { skipatstart += 18; options |= PCRE_NO_AUTO_POSSESS; continue; } + 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 +8883,34 @@ 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 +8986,10 @@ 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 @@ -7505,15 +9002,21 @@ cd->bracount = cd->final_bracount = 0; cd->names_found = 0; cd->name_entry_size = 0; cd->name_table = NULL; +cd->dupnames = FALSE; +cd->namedrefcount = 0; 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->named_groups = named_groups; +cd->named_group_list_size = NAMED_GROUP_LIST_SIZE; +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->parens_depth = 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 @@ -7525,12 +9028,14 @@ outside can help speed up starting point checks. */ 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) { @@ -7538,14 +9043,23 @@ if (length > MAX_PATTERN_SIZE) goto PCRE_EARLY_ERROR_RETURN; } -/* Compute the size of data block needed and get it, either from malloc or -externally provided function. Integer overflow should no longer be possible -because nowadays we limit the maximum value of cd->names_found and -cd->name_entry_size. */ +/* If there are groups with duplicate names and there are also references by +name, we must allow for the possibility of named references to duplicated +groups. These require an extra data item each. */ -size = length + sizeof(real_pcre) + cd->names_found * cd->name_entry_size; -re = (real_pcre *)(pcre_malloc)(size); +if (cd->dupnames && cd->namedrefcount > 0) + length += cd->namedrefcount * IMM2_SIZE * sizeof(pcre_uchar); +/* Compute the size of the data block for storing the compiled pattern. Integer +overflow should no longer be possible because nowadays we limit the maximum +value of cd->names_found and cd->name_entry_size. */ + +size = sizeof(REAL_PCRE) + + (length + cd->names_found * cd->name_entry_size) * sizeof(pcre_uchar); + +/* Get the memory. */ + +re = (REAL_PCRE *)(PUBL(malloc))(size); if (re == NULL) { errorcode = ERR21; @@ -7562,15 +9076,21 @@ 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 @@ -7580,36 +9100,57 @@ field; this time it's used for remembering forward ref */ cd->final_bracount = cd->bracount; /* Save for checking forward references */ +cd->parens_depth = 0; cd->assert_depth = 0; cd->bracount = 0; -cd->names_found = 0; -cd->name_table = (uschar *)re + re->name_table_offset; +cd->max_lookbehind = 0; +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; +/* If any named groups were found, create the name/number table from the list +created in the first pass. */ + +if (cd->names_found > 0) + { + int i = cd->names_found; + named_group *ng = cd->named_groups; + cd->names_found = 0; + for (; i > 0; i--, ng++) + add_name(cd, ng->name, ng->length, ng->number); + if (cd->named_group_list_size > NAMED_GROUP_LIST_SIZE) + (PUBL(free))((void *)cd->named_groups); + } + /* Set up a starting, non-extracting bracket, then compile the expression. On error, errorcode will be set non-zero, so we don't need to look at the result of the function here. */ -ptr = (const uschar *)pattern + 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 +9161,20 @@ 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; @@ -7635,27 +9183,35 @@ 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 the workspace had to be expanded, free the new memory. Set the pointer to +NULL to indicate that forward references have been filled in. */ if (cd->workspace_size > COMPILE_WORK_SIZE) - (pcre_free)((void *)cd->start_workspace); + (PUBL(free))((void *)cd->start_workspace); +cd->start_workspace = NULL; /* Give an error if there's back reference to a non-existent capturing subpattern. */ if (errorcode == 0 && re->top_backref > re->top_bracket) errorcode = ERR15; +/* Unless disabled, check whether single character iterators can be +auto-possessified. The function overwrites the appropriate opcode values. */ + +if ((options & PCRE_NO_AUTO_POSSESS) == 0) + auto_possessify((pcre_uchar *)codestart, utf, cd); + /* 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 +9219,21 @@ 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 +9246,7 @@ 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 +9257,9 @@ 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 +9267,57 @@ 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 +9325,36 @@ 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 +9369,65 @@ 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 */ +/* Check for a pattern than can match an empty string, so that this information +can be provided to applications. */ + +do + { + if (could_be_empty_branch(codestart, code, utf, cd, NULL)) + { + re->flags |= PCRE_MATCH_EMPTY; + break; + } + codestart += GET(codestart, 1); + } +while (*codestart == OP_ALT); + +#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 */ +