diff options
Diffstat (limited to 'regint.h')
| -rw-r--r-- | regint.h | 566 |
1 files changed, 360 insertions, 206 deletions
@@ -1,10 +1,11 @@ -#ifndef REGINT_H -#define REGINT_H +#ifndef ONIGMO_REGINT_H +#define ONIGMO_REGINT_H /********************************************************************** - regint.h - Oniguruma (regular expression library) + regint.h - Onigmo (Oniguruma-mod) (regular expression library) **********************************************************************/ /*- - * Copyright (c) 2002-2007 K.Kosako <sndgk393 AT ybb DOT ne DOT jp> + * Copyright (c) 2002-2013 K.Kosako <sndgk393 AT ybb DOT ne DOT jp> + * Copyright (c) 2011-2019 K.Takata <kentkt AT csc DOT jp> * All rights reserved. * * Redistribution and use in source and binary forms, with or without @@ -34,198 +35,260 @@ /* #define ONIG_DEBUG_COMPILE */ /* #define ONIG_DEBUG_SEARCH */ /* #define ONIG_DEBUG_MATCH */ +/* #define ONIG_DEBUG_MATCH_CACHE */ +/* #define ONIG_DEBUG_MEMLEAK */ /* #define ONIG_DONT_OPTIMIZE */ /* for byte-code statistical data. */ /* #define ONIG_DEBUG_STATISTICS */ +/* enable the match optimization by using a cache. */ +#define USE_MATCH_CACHE + #if defined(ONIG_DEBUG_PARSE_TREE) || defined(ONIG_DEBUG_MATCH) || \ defined(ONIG_DEBUG_SEARCH) || defined(ONIG_DEBUG_COMPILE) || \ - defined(ONIG_DEBUG_STATISTICS) -#ifndef ONIG_DEBUG -#define ONIG_DEBUG + defined(ONIG_DEBUG_STATISTICS) || defined(ONIG_DEBUG_MEMLEAK) +# ifndef ONIG_DEBUG +# define ONIG_DEBUG +# endif #endif + +/* __POWERPC__ added to accommodate Darwin case. */ +#ifndef UNALIGNED_WORD_ACCESS +# if defined(__i386) || defined(__i386__) || defined(_M_IX86) || \ + defined(__x86_64) || defined(__x86_64__) || defined(_M_AMD64) || \ + defined(__powerpc64__) || defined(__POWERPC__) || defined(__aarch64__) || \ + defined(__mc68020__) +# define UNALIGNED_WORD_ACCESS 1 +# else +# define UNALIGNED_WORD_ACCESS 0 +# endif #endif -#if defined(__i386) || defined(__i386__) || defined(_M_IX86) || \ - (defined(__ppc__) && defined(__APPLE__)) || \ - defined(__x86_64) || defined(__x86_64__) || defined(_M_AMD86) || \ - defined(__mc68020__) -/* #define PLATFORM_UNALIGNED_WORD_ACCESS */ +#if UNALIGNED_WORD_ACCESS +# define PLATFORM_UNALIGNED_WORD_ACCESS #endif /* config */ /* spec. config */ #define USE_NAMED_GROUP #define USE_SUBEXP_CALL -#define USE_BACKREF_AT_LEVEL /* \k<name+n>, \k<name-n> */ -#define USE_INFINITE_REPEAT_MONOMANIAC_MEM_STATUS_CHECK /* /(?:()|())*\2/ */ +#define USE_PERL_SUBEXP_CALL +#define USE_CAPITAL_P_NAMED_GROUP +#define USE_BACKREF_WITH_LEVEL /* \k<name+n>, \k<name-n> */ +#define USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT /* /(?:()|())*\2/ */ #define USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE /* /\n$/ =~ "\n" */ #define USE_WARNING_REDUNDANT_NESTED_REPEAT_OPERATOR -/* #define USE_RECOMPILE_API */ -/* #define USE_CRNL_AS_LINE_TERMINATOR */ /* moved to regenc.h. */ +/* !!! moved to regenc.h. */ /* #define USE_CRNL_AS_LINE_TERMINATOR */ +#define USE_NO_INVALID_QUANTIFIER /* internal config */ -#define USE_PARSE_TREE_NODE_RECYCLE -#define USE_OP_PUSH_OR_JUMP_EXACT +/* #define USE_OP_PUSH_OR_JUMP_EXACT */ #define USE_QTFR_PEEK_NEXT #define USE_ST_LIBRARY -#define USE_SHARED_CCLASS_TABLE #define INIT_MATCH_STACK_SIZE 160 #define DEFAULT_MATCH_STACK_LIMIT_SIZE 0 /* unlimited */ +#define DEFAULT_PARSE_DEPTH_LIMIT 4096 + +#define OPT_EXACT_MAXLEN 24 /* This must be smaller than ONIG_CHAR_TABLE_SIZE. */ + +/* check config */ +#if defined(USE_PERL_SUBEXP_CALL) || defined(USE_CAPITAL_P_NAMED_GROUP) +# if !defined(USE_NAMED_GROUP) || !defined(USE_SUBEXP_CALL) +# error USE_NAMED_GROUP and USE_SUBEXP_CALL must be defined. +# endif +#endif + +#if defined(__GNUC__) +# define ARG_UNUSED __attribute__ ((unused)) +#else +# define ARG_UNUSED +#endif + +#if !defined(RUBY) && defined(RUBY_EXPORT) +# define RUBY +#endif +#ifdef RUBY +# ifndef RUBY_DEFINES_H +# include "ruby/ruby.h" +# undef xmalloc +# undef xrealloc +# undef xcalloc +# undef xfree +# endif +#else /* RUBY */ +# include "config.h" +# if SIZEOF_LONG_LONG > 0 +# define LONG_LONG long long +# endif +#endif /* RUBY */ + +#include <stdarg.h> /* */ /* escape other system UChar definition */ -#ifndef RUBY_DEFINES_H -#include "ruby/config.h" -#include "ruby/defines.h" -#endif #ifdef ONIG_ESCAPE_UCHAR_COLLISION -#undef ONIG_ESCAPE_UCHAR_COLLISION +# undef ONIG_ESCAPE_UCHAR_COLLISION #endif -#undef USE_MATCH_RANGE_IS_COMPLETE_RANGE -#define USE_CAPTURE_HISTORY -#define USE_VARIABLE_META_CHARS + #define USE_WORD_BEGIN_END /* "\<": word-begin, "\>": word-end */ -#define USE_POSIX_REGION_OPTION /* needed for POSIX API support */ +#ifdef RUBY +# undef USE_CAPTURE_HISTORY +#else +# define USE_CAPTURE_HISTORY +#endif +#define USE_VARIABLE_META_CHARS #define USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE /* #define USE_COMBINATION_EXPLOSION_CHECK */ /* (X*)* */ -/* #define USE_MULTI_THREAD_SYSTEM */ -#define THREAD_SYSTEM_INIT /* depend on thread system */ -#define THREAD_SYSTEM_END /* depend on thread system */ -#define THREAD_ATOMIC_START /* depend on thread system */ -#define THREAD_ATOMIC_END /* depend on thread system */ -#define THREAD_PASS /* depend on thread system */ + + #ifndef xmalloc -#define xmalloc malloc -#define xrealloc realloc -#define xcalloc calloc -#define xfree free +# define xmalloc malloc +# define xrealloc realloc +# define xcalloc calloc +# define xfree free #endif -#define CHECK_INTERRUPT_IN_MATCH_AT - #ifdef RUBY -#define onig_st_init_table st_init_table -#define onig_st_init_table_with_size st_init_table_with_size -#define onig_st_init_numtable st_init_numtable -#define onig_st_init_numtable_with_size st_init_numtable_with_size -#define onig_st_init_strtable st_init_strtable -#define onig_st_init_strtable_with_size st_init_strtable_with_size -#define onig_st_delete st_delete -#define onig_st_delete_safe st_delete_safe -#define onig_st_insert st_insert -#define onig_st_lookup st_lookup -#define onig_st_foreach st_foreach -#define onig_st_add_direct st_add_direct -#define onig_st_free_table st_free_table -#define onig_st_cleanup_safe st_cleanup_safe -#define onig_st_copy st_copy -#define onig_st_nothing_key_clone st_nothing_key_clone -#define onig_st_nothing_key_free st_nothing_key_free -#define onig_st_is_member st_is_member - -#define USE_UPPER_CASE_TABLE -#else - -#define st_init_table onig_st_init_table -#define st_init_table_with_size onig_st_init_table_with_size -#define st_init_numtable onig_st_init_numtable -#define st_init_numtable_with_size onig_st_init_numtable_with_size -#define st_init_strtable onig_st_init_strtable -#define st_init_strtable_with_size onig_st_init_strtable_with_size -#define st_delete onig_st_delete -#define st_delete_safe onig_st_delete_safe -#define st_insert onig_st_insert -#define st_lookup onig_st_lookup -#define st_foreach onig_st_foreach -#define st_add_direct onig_st_add_direct -#define st_free_table onig_st_free_table -#define st_cleanup_safe onig_st_cleanup_safe -#define st_copy onig_st_copy -#define st_nothing_key_clone onig_st_nothing_key_clone -#define st_nothing_key_free onig_st_nothing_key_free +# define CHECK_INTERRUPT_IN_MATCH_AT do { \ + msa->counter++; \ + if (msa->counter >= 128) { \ + msa->counter = 0; \ + if (rb_reg_timeout_p(reg, &msa->end_time)) { \ + goto timeout; \ + } \ + rb_thread_check_ints(); \ + } \ +} while(0) +# define onig_st_init_table st_init_table +# define onig_st_init_table_with_size st_init_table_with_size +# define onig_st_init_numtable st_init_numtable +# define onig_st_init_numtable_with_size st_init_numtable_with_size +# define onig_st_init_strtable st_init_strtable +# define onig_st_init_strtable_with_size st_init_strtable_with_size +# define onig_st_delete st_delete +# define onig_st_delete_safe st_delete_safe +# define onig_st_insert st_insert +# define onig_st_lookup st_lookup +# define onig_st_foreach st_foreach +# define onig_st_add_direct st_add_direct +# define onig_st_free_table st_free_table +# define onig_st_cleanup_safe st_cleanup_safe +# define onig_st_copy st_copy +# define onig_st_nothing_key_clone st_nothing_key_clone +# define onig_st_nothing_key_free st_nothing_key_free +# define onig_st_is_member st_is_member + +# define USE_UPPER_CASE_TABLE +#else /* RUBY */ + +# define CHECK_INTERRUPT_IN_MATCH_AT + +# define st_init_table onig_st_init_table +# define st_init_table_with_size onig_st_init_table_with_size +# define st_init_numtable onig_st_init_numtable +# define st_init_numtable_with_size onig_st_init_numtable_with_size +# define st_init_strtable onig_st_init_strtable +# define st_init_strtable_with_size onig_st_init_strtable_with_size +# define st_delete onig_st_delete +# define st_delete_safe onig_st_delete_safe +# define st_insert onig_st_insert +# define st_lookup onig_st_lookup +# define st_foreach onig_st_foreach +# define st_add_direct onig_st_add_direct +# define st_free_table onig_st_free_table +# define st_cleanup_safe onig_st_cleanup_safe +# define st_copy onig_st_copy +# define st_nothing_key_clone onig_st_nothing_key_clone +# define st_nothing_key_free onig_st_nothing_key_free /* */ -#define onig_st_is_member st_is_member +# define onig_st_is_member st_is_member -#endif +#endif /* RUBY */ #define STATE_CHECK_STRING_THRESHOLD_LEN 7 #define STATE_CHECK_BUFF_MAX_SIZE 0x4000 -#define THREAD_PASS_LIMIT_COUNT 8 #define xmemset memset #define xmemcpy memcpy #define xmemmove memmove #if defined(_WIN32) && !defined(__GNUC__) -#define xalloca _alloca -#define xvsnprintf _vsnprintf +# define xalloca _alloca +# define xvsnprintf(buf,size,fmt,args) _vsnprintf_s(buf,size,_TRUNCATE,fmt,args) +# define xsnprintf sprintf_s +# define xstrcat(dest,src,size) strcat_s(dest,size,src) #else -#define xalloca alloca -#define xvsnprintf vsnprintf +# define xalloca alloca +# define xvsnprintf vsnprintf +# define xsnprintf snprintf +# define xstrcat(dest,src,size) strcat(dest,src) #endif +#if defined(ONIG_DEBUG_MEMLEAK) && defined(_MSC_VER) +# define _CRTDBG_MAP_ALLOC +# include <malloc.h> +# include <crtdbg.h> +#endif -#if defined(USE_RECOMPILE_API) && defined(USE_MULTI_THREAD_SYSTEM) -#define ONIG_STATE_INC(reg) (reg)->state++ -#define ONIG_STATE_DEC(reg) (reg)->state-- - -#define ONIG_STATE_INC_THREAD(reg) do {\ - THREAD_ATOMIC_START;\ - (reg)->state++;\ - THREAD_ATOMIC_END;\ -} while(0) -#define ONIG_STATE_DEC_THREAD(reg) do {\ - THREAD_ATOMIC_START;\ - (reg)->state--;\ - THREAD_ATOMIC_END;\ -} while(0) -#else -#define ONIG_STATE_INC(reg) /* Nothing */ -#define ONIG_STATE_DEC(reg) /* Nothing */ -#define ONIG_STATE_INC_THREAD(reg) /* Nothing */ -#define ONIG_STATE_DEC_THREAD(reg) /* Nothing */ -#endif /* USE_RECOMPILE_API && USE_MULTI_THREAD_SYSTEM */ - -#ifdef HAVE_STDLIB_H #include <stdlib.h> -#endif #if defined(HAVE_ALLOCA_H) && (defined(_AIX) || !defined(__GNUC__)) -#include <alloca.h> +# include <alloca.h> #endif -#ifdef HAVE_STRING_H -# include <string.h> -#else -# include <strings.h> -#endif +#include <string.h> #include <ctype.h> #ifdef HAVE_SYS_TYPES_H -#ifndef __BORLANDC__ -#include <sys/types.h> +# include <sys/types.h> +#endif + +#ifdef HAVE_STDINT_H +# include <stdint.h> #endif + +#ifdef HAVE_INTTYPES_H +# include <inttypes.h> #endif -#ifdef __BORLANDC__ -#include <malloc.h> +#include <stddef.h> + +#ifdef _WIN32 +# include <malloc.h> /* for alloca() */ #endif #ifdef ONIG_DEBUG # include <stdio.h> #endif +#ifndef PRIdPTR +# ifdef _WIN64 +# define PRIdPTR "I64d" +# define PRIuPTR "I64u" +# define PRIxPTR "I64x" +# else +# define PRIdPTR "ld" +# define PRIuPTR "lu" +# define PRIxPTR "lx" +# endif +#endif + +#ifndef PRIdPTRDIFF +# define PRIdPTRDIFF PRIdPTR +#endif + #include "regenc.h" +RUBY_SYMBOL_EXPORT_BEGIN + #ifdef MIN -#undef MIN +# undef MIN #endif #ifdef MAX -#undef MAX +# undef MAX #endif #define MIN(a,b) (((a)>(b))?(b):(a)) #define MAX(a,b) (((a)<(b))?(b):(a)) @@ -236,33 +299,41 @@ #define CHECK_NULL_RETURN_MEMERR(p) if (IS_NULL(p)) return ONIGERR_MEMORY #define NULL_UCHARP ((UChar* )0) +#define ONIG_LAST_CODE_POINT (~((OnigCodePoint )0)) + +#define PLATFORM_GET_INC_ARGUMENTS_ASSERT(val, type) \ + ((void)sizeof(char[2 * (sizeof(val) == sizeof(type)) - 1])) #ifdef PLATFORM_UNALIGNED_WORD_ACCESS -#define PLATFORM_GET_INC(val,p,type) do{\ +# define PLATFORM_GET_INC(val,p,type) do{\ + PLATFORM_GET_INC_ARGUMENTS_ASSERT(val, type);\ val = *(type* )p;\ (p) += sizeof(type);\ } while(0) #else -#define PLATFORM_GET_INC(val,p,type) do{\ - xmemcpy(&val, (p), sizeof(type));\ +# define PLATFORM_GET_INC(val,p,type) do{\ + PLATFORM_GET_INC_ARGUMENTS_ASSERT(val, type);\ + type platform_get_value;\ + xmemcpy(&platform_get_value, (p), sizeof(type));\ + val = platform_get_value;\ (p) += sizeof(type);\ } while(0) /* sizeof(OnigCodePoint) */ -#define WORD_ALIGNMENT_SIZE SIZEOF_LONG +# define WORD_ALIGNMENT_SIZE SIZEOF_LONG -#define GET_ALIGNMENT_PAD_SIZE(addr,pad_size) do {\ +# define GET_ALIGNMENT_PAD_SIZE(addr,pad_size) do {\ (pad_size) = WORD_ALIGNMENT_SIZE \ - - ((unsigned int )(addr) % WORD_ALIGNMENT_SIZE);\ + - ((uintptr_t )(addr) % WORD_ALIGNMENT_SIZE);\ if ((pad_size) == WORD_ALIGNMENT_SIZE) (pad_size) = 0;\ } while (0) -#define ALIGNMENT_RIGHT(addr) do {\ +# define ALIGNMENT_RIGHT(addr) do {\ (addr) += (WORD_ALIGNMENT_SIZE - 1);\ - (addr) -= ((unsigned int )(addr) % WORD_ALIGNMENT_SIZE);\ + (addr) -= ((uintptr_t )(addr) % WORD_ALIGNMENT_SIZE);\ } while (0) #endif /* PLATFORM_UNALIGNED_WORD_ACCESS */ @@ -276,9 +347,11 @@ #define ONIG_OPTIMIZE_NONE 0 #define ONIG_OPTIMIZE_EXACT 1 /* Slow Search */ #define ONIG_OPTIMIZE_EXACT_BM 2 /* Boyer Moore Search */ -#define ONIG_OPTIMIZE_EXACT_BM_NOT_REV 3 /* BM (but not simple match) */ +#define ONIG_OPTIMIZE_EXACT_BM_NOT_REV 3 /* BM (applied to a multibyte string) */ #define ONIG_OPTIMIZE_EXACT_IC 4 /* Slow Search (ignore case) */ #define ONIG_OPTIMIZE_MAP 5 /* char map */ +#define ONIG_OPTIMIZE_EXACT_BM_IC 6 /* BM (ignore case) */ +#define ONIG_OPTIMIZE_EXACT_BM_NOT_REV_IC 7 /* BM (applied to a multibyte string) (ignore case) */ /* bit status */ typedef unsigned int BitStatusType; @@ -287,22 +360,23 @@ typedef unsigned int BitStatusType; #define BIT_STATUS_CLEAR(stats) (stats) = 0 #define BIT_STATUS_ON_ALL(stats) (stats) = ~((BitStatusType )0) #define BIT_STATUS_AT(stats,n) \ - ((n) < BIT_STATUS_BITS_NUM ? ((stats) & (1 << n)) : ((stats) & 1)) + ((n) < (int )BIT_STATUS_BITS_NUM ? ((stats) & ((BitStatusType )1 << n)) : ((stats) & 1)) #define BIT_STATUS_ON_AT(stats,n) do {\ - if ((n) < BIT_STATUS_BITS_NUM)\ + if ((n) < (int )BIT_STATUS_BITS_NUM)\ (stats) |= (1 << (n));\ else\ (stats) |= 1;\ } while (0) #define BIT_STATUS_ON_AT_SIMPLE(stats,n) do {\ - if ((n) < BIT_STATUS_BITS_NUM)\ + if ((n) < (int )BIT_STATUS_BITS_NUM)\ (stats) |= (1 << (n));\ } while (0) #define INT_MAX_LIMIT ((1UL << (SIZEOF_INT * 8 - 1)) - 1) +#define LONG_MAX_LIMIT ((1UL << (SIZEOF_LONG * 8 - 1)) - 1) #define DIGITVAL(code) ((code) - '0') #define ODIGITVAL(code) DIGITVAL(code) @@ -320,7 +394,12 @@ typedef unsigned int BitStatusType; (ONIG_OPTION_FIND_LONGEST | ONIG_OPTION_FIND_NOT_EMPTY)) #define IS_NOTBOL(option) ((option) & ONIG_OPTION_NOTBOL) #define IS_NOTEOL(option) ((option) & ONIG_OPTION_NOTEOL) -#define IS_POSIX_REGION(option) ((option) & ONIG_OPTION_POSIX_REGION) +#define IS_NOTBOS(option) ((option) & ONIG_OPTION_NOTBOS) +#define IS_NOTEOS(option) ((option) & ONIG_OPTION_NOTEOS) +#define IS_ASCII_RANGE(option) ((option) & ONIG_OPTION_ASCII_RANGE) +#define IS_POSIX_BRACKET_ALL_RANGE(option) ((option) & ONIG_OPTION_POSIX_BRACKET_ALL_RANGE) +#define IS_WORD_BOUND_ALL_RANGE(option) ((option) & ONIG_OPTION_WORD_BOUND_ALL_RANGE) +#define IS_NEWLINE_CRLF(option) ((option) & ONIG_OPTION_NEWLINE_CRLF) /* OP_SET_OPTION is required for these options. #define IS_DYNAMIC_OPTION(option) \ @@ -338,7 +417,7 @@ typedef unsigned int BitStatusType; /* bitset */ #define BITS_PER_BYTE 8 #define SINGLE_BYTE_SIZE (1 << BITS_PER_BYTE) -#define BITS_IN_ROOM (sizeof(Bits) * BITS_PER_BYTE) +#define BITS_IN_ROOM ((int )sizeof(Bits) * BITS_PER_BYTE) #define BITSET_SIZE (SINGLE_BYTE_SIZE / BITS_IN_ROOM) #ifdef PLATFORM_UNALIGNED_WORD_ACCESS @@ -349,15 +428,15 @@ typedef unsigned char Bits; typedef Bits BitSet[BITSET_SIZE]; typedef Bits* BitSetRef; -#define SIZE_BITSET sizeof(BitSet) +#define SIZE_BITSET (int )sizeof(BitSet) #define BITSET_CLEAR(bs) do {\ int i;\ - for (i = 0; i < BITSET_SIZE; i++) { (bs)[i] = 0; }\ + for (i = 0; i < BITSET_SIZE; i++) { (bs)[i] = 0; } \ } while (0) -#define BS_ROOM(bs,pos) (bs)[pos / BITS_IN_ROOM] -#define BS_BIT(pos) (1 << (pos % BITS_IN_ROOM)) +#define BS_ROOM(bs,pos) (bs)[(int )(pos) / BITS_IN_ROOM] +#define BS_BIT(pos) (1U << ((int )(pos) % BITS_IN_ROOM)) #define BITSET_AT(bs, pos) (BS_ROOM(bs,pos) & BS_BIT(pos)) #define BITSET_SET_BIT(bs, pos) BS_ROOM(bs,pos) |= BS_BIT(pos) @@ -374,29 +453,35 @@ typedef struct _BBuf { #define BBUF_INIT(buf,size) onig_bbuf_init((BBuf* )(buf), (size)) #define BBUF_SIZE_INC(buf,inc) do{\ + UChar *tmp;\ (buf)->alloc += (inc);\ - (buf)->p = (UChar* )xrealloc((buf)->p, (buf)->alloc);\ - if (IS_NULL((buf)->p)) return(ONIGERR_MEMORY);\ + tmp = (UChar* )xrealloc((buf)->p, (buf)->alloc);\ + if (IS_NULL(tmp)) return(ONIGERR_MEMORY);\ + (buf)->p = tmp;\ } while (0) #define BBUF_EXPAND(buf,low) do{\ + UChar *tmp;\ do { (buf)->alloc *= 2; } while ((buf)->alloc < (unsigned int )low);\ - (buf)->p = (UChar* )xrealloc((buf)->p, (buf)->alloc);\ - if (IS_NULL((buf)->p)) return(ONIGERR_MEMORY);\ + tmp = (UChar* )xrealloc((buf)->p, (buf)->alloc);\ + if (IS_NULL(tmp)) return(ONIGERR_MEMORY);\ + (buf)->p = tmp;\ } while (0) #define BBUF_ENSURE_SIZE(buf,size) do{\ unsigned int new_alloc = (buf)->alloc;\ while (new_alloc < (unsigned int )(size)) { new_alloc *= 2; }\ if ((buf)->alloc != new_alloc) {\ - (buf)->p = (UChar* )xrealloc((buf)->p, new_alloc);\ - if (IS_NULL((buf)->p)) return(ONIGERR_MEMORY);\ + UChar *tmp;\ + tmp = (UChar* )xrealloc((buf)->p, new_alloc);\ + if (IS_NULL(tmp)) return(ONIGERR_MEMORY);\ + (buf)->p = tmp;\ (buf)->alloc = new_alloc;\ }\ } while (0) #define BBUF_WRITE(buf,pos,bytes,n) do{\ - int used = (pos) + (n);\ + int used = (pos) + (int )(n);\ if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\ xmemcpy((buf)->p + (pos), (bytes), (n));\ if ((buf)->used < (unsigned int )used) (buf)->used = used;\ @@ -405,7 +490,7 @@ typedef struct _BBuf { #define BBUF_WRITE1(buf,pos,byte) do{\ int used = (pos) + 1;\ if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\ - (buf)->p[(pos)] = (byte);\ + (buf)->p[(pos)] = (UChar )(byte);\ if ((buf)->used < (unsigned int )used) (buf)->used = used;\ } while (0) @@ -464,6 +549,8 @@ typedef struct _BBuf { #define ANCHOR_ANYCHAR_STAR (1<<14) /* ".*" optimize info */ #define ANCHOR_ANYCHAR_STAR_ML (1<<15) /* ".*" optimize info (multi-line) */ +#define ANCHOR_KEEP (1<<16) + /* operation code */ enum OpCode { OP_FINISH = 0, /* matching process terminator (no more alternative) */ @@ -491,7 +578,6 @@ enum OpCode { OP_CCLASS_NOT, OP_CCLASS_MB_NOT, OP_CCLASS_MIX_NOT, - OP_CCLASS_NODE, /* pointer to CClassNode node */ OP_ANYCHAR, /* "." */ OP_ANYCHAR_ML, /* "." multi-line */ @@ -507,6 +593,13 @@ enum OpCode { OP_WORD_BEGIN, OP_WORD_END, + OP_ASCII_WORD, + OP_NOT_ASCII_WORD, + OP_ASCII_WORD_BOUND, + OP_NOT_ASCII_WORD_BOUND, + OP_ASCII_WORD_BEGIN, + OP_ASCII_WORD_END, + OP_BEGIN_BUF, OP_END_BUF, OP_BEGIN_LINE, @@ -520,7 +613,7 @@ enum OpCode { OP_BACKREFN_IC, OP_BACKREF_MULTI, OP_BACKREF_MULTI_IC, - OP_BACKREF_AT_LEVEL, /* \k<xxx+n>, \k<xxx-n> */ + OP_BACKREF_WITH_LEVEL, /* \k<xxx+n>, \k<xxx-n> */ OP_MEMORY_START, OP_MEMORY_START_PUSH, /* push back-tracker to stack */ @@ -529,6 +622,8 @@ enum OpCode { OP_MEMORY_END, OP_MEMORY_END_REC, /* push marker to stack */ + OP_KEEP, + OP_FAIL, /* pop stack and move */ OP_JUMP, OP_PUSH, @@ -555,10 +650,15 @@ enum OpCode { OP_LOOK_BEHIND, /* (?<=...) start (no needs end opcode) */ OP_PUSH_LOOK_BEHIND_NOT, /* (?<!...) start */ OP_FAIL_LOOK_BEHIND_NOT, /* (?<!...) end */ + OP_PUSH_ABSENT_POS, /* (?~...) start */ + OP_ABSENT, /* (?~...) start of inner loop */ + OP_ABSENT_END, /* (?~...) end */ OP_CALL, /* \g<name> */ OP_RETURN, + OP_CONDITION, + OP_STATE_CHECK_PUSH, /* combination explosion check and push */ OP_STATE_CHECK_PUSH_OR_JUMP, /* check ok -> push, else jump */ OP_STATE_CHECK, /* check only */ @@ -579,15 +679,15 @@ typedef short int StateCheckNumType; typedef void* PointerType; #define SIZE_OPCODE 1 -#define SIZE_RELADDR sizeof(RelAddrType) -#define SIZE_ABSADDR sizeof(AbsAddrType) -#define SIZE_LENGTH sizeof(LengthType) -#define SIZE_MEMNUM sizeof(MemNumType) -#define SIZE_STATE_CHECK_NUM sizeof(StateCheckNumType) -#define SIZE_REPEATNUM sizeof(RepeatNumType) -#define SIZE_OPTION sizeof(OnigOptionType) -#define SIZE_CODE_POINT sizeof(OnigCodePoint) -#define SIZE_POINTER sizeof(PointerType) +#define SIZE_RELADDR (int )sizeof(RelAddrType) +#define SIZE_ABSADDR (int )sizeof(AbsAddrType) +#define SIZE_LENGTH (int )sizeof(LengthType) +#define SIZE_MEMNUM (int )sizeof(MemNumType) +#define SIZE_STATE_CHECK_NUM (int )sizeof(StateCheckNumType) +#define SIZE_REPEATNUM (int )sizeof(RepeatNumType) +#define SIZE_OPTION (int )sizeof(OnigOptionType) +#define SIZE_CODE_POINT (int )sizeof(OnigCodePoint) +#define SIZE_POINTER (int )sizeof(PointerType) #define GET_RELADDR_INC(addr,p) PLATFORM_GET_INC(addr, p, RelAddrType) @@ -639,12 +739,16 @@ typedef void* PointerType; #define SIZE_OP_FAIL_LOOK_BEHIND_NOT SIZE_OPCODE #define SIZE_OP_CALL (SIZE_OPCODE + SIZE_ABSADDR) #define SIZE_OP_RETURN SIZE_OPCODE +#define SIZE_OP_CONDITION (SIZE_OPCODE + SIZE_MEMNUM + SIZE_RELADDR) +#define SIZE_OP_PUSH_ABSENT_POS SIZE_OPCODE +#define SIZE_OP_ABSENT (SIZE_OPCODE + SIZE_RELADDR) +#define SIZE_OP_ABSENT_END SIZE_OPCODE #ifdef USE_COMBINATION_EXPLOSION_CHECK -#define SIZE_OP_STATE_CHECK (SIZE_OPCODE + SIZE_STATE_CHECK_NUM) -#define SIZE_OP_STATE_CHECK_PUSH (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR) -#define SIZE_OP_STATE_CHECK_PUSH_OR_JUMP (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR) -#define SIZE_OP_STATE_CHECK_ANYCHAR_STAR (SIZE_OPCODE + SIZE_STATE_CHECK_NUM) +# define SIZE_OP_STATE_CHECK (SIZE_OPCODE + SIZE_STATE_CHECK_NUM) +# define SIZE_OP_STATE_CHECK_PUSH (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR) +# define SIZE_OP_STATE_CHECK_PUSH_OR_JUMP (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR) +# define SIZE_OP_STATE_CHECK_ANYCHAR_STAR (SIZE_OPCODE + SIZE_STATE_CHECK_NUM) #endif #define MC_ESC(syn) (syn)->meta_char_table.esc @@ -692,13 +796,10 @@ typedef void* PointerType; /* cclass node */ #define FLAG_NCCLASS_NOT (1<<0) -#define FLAG_NCCLASS_SHARE (1<<1) #define NCCLASS_SET_NOT(nd) NCCLASS_FLAG_SET(nd, FLAG_NCCLASS_NOT) -#define NCCLASS_SET_SHARE(nd) NCCLASS_FLAG_SET(nd, FLAG_NCCLASS_SHARE) #define NCCLASS_CLEAR_NOT(nd) NCCLASS_FLAG_CLEAR(nd, FLAG_NCCLASS_NOT) #define IS_NCCLASS_NOT(nd) IS_NCCLASS_FLAG_ON(nd, FLAG_NCCLASS_NOT) -#define IS_NCCLASS_SHARE(nd) IS_NCCLASS_FLAG_ON(nd, FLAG_NCCLASS_SHARE) typedef struct { int type; @@ -713,10 +814,11 @@ typedef struct { BBuf* mbuf; /* multi-byte info or NULL */ } CClassNode; -typedef long OnigStackIndex; +typedef intptr_t OnigStackIndex; typedef struct _OnigStackType { unsigned int type; + OnigStackIndex null_check; union { struct { UChar *pcode; /* byte code position */ @@ -725,6 +827,7 @@ typedef struct _OnigStackType { #ifdef USE_COMBINATION_EXPLOSION_CHECK unsigned int state_check; #endif + UChar *pkeep; /* keep pattern position */ } state; struct { int count; /* for OP_REPEAT_INC, OP_REPEAT_INC_NG */ @@ -737,7 +840,7 @@ typedef struct _OnigStackType { struct { int num; /* memory num */ UChar *pstr; /* start/end position */ - /* Following information is setted, if this stack type is MEM-START */ + /* Following information is set, if this stack type is MEM-START */ OnigStackIndex start; /* prev. info (for backtrack "(...)*" ) */ OnigStackIndex end; /* prev. info (for backtrack "(...)*" ) */ } mem; @@ -752,85 +855,136 @@ typedef struct _OnigStackType { UChar *pstr; /* string position */ } call_frame; #endif + struct { + UChar *abs_pstr; /* absent start position */ + const UChar *end_pstr; /* end position */ + } absent_pos; +#ifdef USE_MATCH_CACHE + struct { + long index; /* index of the match cache buffer */ + uint8_t mask; /* bit-mask for the match cache buffer */ + } match_cache_point; +#endif } u; } OnigStackType; +#ifdef USE_MATCH_CACHE +typedef struct { + UChar *addr; + long cache_point; + int outer_repeat_mem; + long num_cache_points_at_outer_repeat; + long num_cache_points_in_outer_repeat; + int lookaround_nesting; + UChar *match_addr; +} OnigCacheOpcode; +#endif + typedef struct { void* stack_p; - int stack_n; + size_t stack_n; OnigOptionType options; OnigRegion* region; - const UChar* start; /* search start position (for \G: BEGIN_POSITION) */ + const UChar* start; /* search start position */ + const UChar* gpos; /* global position (for \G: BEGIN_POSITION) */ #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE - int best_len; /* for ONIG_OPTION_FIND_LONGEST */ + OnigPosition best_len; /* for ONIG_OPTION_FIND_LONGEST */ UChar* best_s; #endif #ifdef USE_COMBINATION_EXPLOSION_CHECK void* state_check_buff; int state_check_buff_size; #endif + int counter; + /* rb_hrtime_t from hrtime.h */ +#ifdef MY_RUBY_BUILD_MAY_TIME_TRAVEL + int128_t end_time; +#else + uint64_t end_time; +#endif +#ifdef USE_MATCH_CACHE + int match_cache_status; + long num_fails; + long num_cache_opcodes; + OnigCacheOpcode* cache_opcodes; + long num_cache_points; + uint8_t* match_cache_buf; +#endif } OnigMatchArg; +#define NUM_CACHE_OPCODES_UNINIT 1 +#define NUM_CACHE_OPCODES_IMPOSSIBLE -1 + +#define MATCH_CACHE_STATUS_UNINIT 1 +#define MATCH_CACHE_STATUS_INIT 2 +#define MATCH_CACHE_STATUS_DISABLED -1 +#define MATCH_CACHE_STATUS_ENABLED 0 #define IS_CODE_SB_WORD(enc,code) \ (ONIGENC_IS_CODE_ASCII(code) && ONIGENC_IS_CODE_WORD(enc,code)) +typedef struct OnigEndCallListItem { + struct OnigEndCallListItem* next; + void (*func)(void); +} OnigEndCallListItemType; + +extern void onig_add_end_call(void (*func)(void)); + + #ifdef ONIG_DEBUG typedef struct { short int opcode; - char* name; + const char* name; short int arg_type; } OnigOpInfoType; extern OnigOpInfoType OnigOpInfo[]; -extern void onig_print_compiled_byte_code P_((FILE* f, UChar* bp, UChar** nextp, OnigEncoding enc)); -#ifdef ONIG_DEBUG_STATISTICS -extern void onig_statistics_init P_((void)); -extern void onig_print_statistics P_((FILE* f)); +extern void onig_print_compiled_byte_code(FILE* f, UChar* bp, UChar* bpend, UChar** nextp, OnigEncoding enc); + +# ifdef ONIG_DEBUG_STATISTICS +extern void onig_statistics_init(void); +extern void onig_print_statistics(FILE* f); +# endif #endif + +#ifndef PRINTF_ARGS +#define PRINTF_ARGS(func, fmt, vargs) func #endif -extern UChar* onig_error_code_to_format P_((int code)); -extern void onig_snprintf_with_pattern PV_((UChar buf[], int bufsize, OnigEncoding enc, UChar* pat, UChar* pat_end, const UChar *fmt, ...)); -extern int onig_bbuf_init P_((BBuf* buf, int size)); -extern int onig_alloc_init P_((regex_t** reg, OnigOptionType option, OnigCaseFoldType case_fold_flag, OnigEncoding enc, OnigSyntaxType* syntax)); -extern int onig_compile P_((regex_t* reg, const UChar* pattern, const UChar* pattern_end, OnigErrorInfo* einfo)); -extern void onig_chain_reduce P_((regex_t* reg)); -extern void onig_chain_link_add P_((regex_t* to, regex_t* add)); -extern void onig_transfer P_((regex_t* to, regex_t* from)); -extern int onig_is_code_in_cc P_((OnigEncoding enc, OnigCodePoint code, CClassNode* cc)); -extern int onig_is_code_in_cc_len P_((int enclen, OnigCodePoint code, CClassNode* cc)); +extern UChar* onig_error_code_to_format(OnigPosition code); +PRINTF_ARGS(extern void onig_vsnprintf_with_pattern(UChar buf[], int bufsize, OnigEncoding enc, UChar* pat, UChar* pat_end, const char *fmt, va_list args), 6, 0); +PRINTF_ARGS(extern void onig_snprintf_with_pattern(UChar buf[], int bufsize, OnigEncoding enc, UChar* pat, UChar* pat_end, const char *fmt, ...), 6, 7); +extern int onig_bbuf_init(BBuf* buf, OnigDistance size); +extern int onig_compile(regex_t* reg, const UChar* pattern, const UChar* pattern_end, OnigErrorInfo* einfo); +#ifdef RUBY +extern int onig_compile_ruby(regex_t* reg, const UChar* pattern, const UChar* pattern_end, OnigErrorInfo* einfo, const char *sourcefile, int sourceline); +#endif +extern void onig_transfer(regex_t* to, regex_t* from); +extern int onig_is_code_in_cc(OnigEncoding enc, OnigCodePoint code, CClassNode* cc); +extern int onig_is_code_in_cc_len(int enclen, OnigCodePoint code, CClassNode* cc); /* strend hash */ typedef void hash_table_type; #ifdef RUBY -#include <ruby/st.h> -typedef st_data_t hash_data_type; +# include "ruby/st.h" #else -typedef unsigned long hash_data_type; +# include "st.h" #endif +typedef st_data_t hash_data_type; -extern hash_table_type* onig_st_init_strend_table_with_size P_((int size)); -extern int onig_st_lookup_strend P_((hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type *value)); -extern int onig_st_insert_strend P_((hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type value)); - -/* encoding property management */ -#define PROPERTY_LIST_ADD_PROP(Name, CR) \ - r = onigenc_property_list_add_property((UChar* )Name, CR,\ - &PropertyNameTable, &PropertyList, &PropertyListNum,\ - &PropertyListSize);\ - if (r != 0) goto end +extern hash_table_type* onig_st_init_strend_table_with_size(st_index_t size); +extern int onig_st_lookup_strend(hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type *value); +extern int onig_st_insert_strend(hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type value); -#define PROPERTY_LIST_INIT_CHECK \ - if (PropertyInited == 0) {\ - int r = onigenc_property_list_init(init_property_list);\ - if (r != 0) return r;\ - } +#ifdef RUBY +extern size_t onig_memsize(const regex_t *reg); +extern size_t onig_region_memsize(const struct re_registers *regs); +bool rb_reg_timeout_p(regex_t *reg, void *end_time); +#endif -extern int onigenc_property_list_add_property P_((UChar* name, const OnigCodePoint* prop, hash_table_type **table, const OnigCodePoint*** plist, int *pnum, int *psize)); -extern int onigenc_property_list_init P_((int (*f)())); +RUBY_SYMBOL_EXPORT_END -#endif /* REGINT_H */ +#endif /* ONIGMO_REGINT_H */ |