1/********************************************************************** 2 3 transcode_data.h - 4 5 $Author: yugui $ 6 created at: Mon 10 Dec 2007 14:01:47 JST 2007 7 8 Copyright (C) 2007 Martin Duerst 9 10**********************************************************************/ 11 12#include "ruby/ruby.h" 13 14#ifndef RUBY_TRANSCODE_DATA_H 15#define RUBY_TRANSCODE_DATA_H 1 16 17#if defined __GNUC__ && __GNUC__ >= 4 18#pragma GCC visibility push(default) 19#endif 20 21#define WORDINDEX_SHIFT_BITS 2 22#define WORDINDEX2INFO(widx) ((widx) << WORDINDEX_SHIFT_BITS) 23#define INFO2WORDINDEX(info) ((info) >> WORDINDEX_SHIFT_BITS) 24#define BYTE_LOOKUP_BASE(bl) ((bl)[0]) 25#define BYTE_LOOKUP_INFO(bl) ((bl)[1]) 26 27#define PType (unsigned int) 28 29#define NOMAP (PType 0x01) /* direct map */ 30#define ONEbt (0x02) /* one byte payload */ 31#define TWObt (0x03) /* two bytes payload */ 32#define THREEbt (0x05) /* three bytes payload */ 33#define FOURbt (0x06) /* four bytes payload, UTF-8 only, macros start at getBT0 */ 34#define INVALID (PType 0x07) /* invalid byte sequence */ 35#define UNDEF (PType 0x09) /* legal but undefined */ 36#define ZERObt (PType 0x0A) /* zero bytes of payload, i.e. remove */ 37#define FUNii (PType 0x0B) /* function from info to info */ 38#define FUNsi (PType 0x0D) /* function from start to info */ 39#define FUNio (PType 0x0E) /* function from info to output */ 40#define FUNso (PType 0x0F) /* function from start to output */ 41#define STR1 (PType 0x11) /* string 4 <= len <= 259 bytes: 1byte length + content */ 42#define GB4bt (PType 0x12) /* GB18030 four bytes payload */ 43#define FUNsio (PType 0x13) /* function from start and info to output */ 44 45#define STR1_LENGTH(byte_addr) (unsigned int)(*(byte_addr) + 4) 46#define STR1_BYTEINDEX(w) ((w) >> 6) 47#define makeSTR1(bi) (((bi) << 6) | STR1) 48#define makeSTR1LEN(len) ((len)-4) 49 50#define o1(b1) (PType((((unsigned char)(b1))<<8)|ONEbt)) 51#define o2(b1,b2) (PType((((unsigned char)(b1))<<8)|(((unsigned char)(b2))<<16)|TWObt)) 52#define o3(b1,b2,b3) (PType(((((unsigned char)(b1))<<8)|(((unsigned char)(b2))<<16)|(((unsigned int)(unsigned char)(b3))<<24)|THREEbt)&0xffffffffU)) 53#define o4(b0,b1,b2,b3) (PType(((((unsigned char)(b1))<<8)|(((unsigned char)(b2))<<16)|(((unsigned char)(b3))<<24)|((((unsigned char)(b0))&0x07)<<5)|FOURbt)&0xffffffffU)) 54#define g4(b0,b1,b2,b3) (PType(((((unsigned char)(b0))<<8)|(((unsigned char)(b2))<<16)|((((unsigned char)(b1))&0x0f)<<24)|((((unsigned int)(unsigned char)(b3))&0x0f)<<28)|GB4bt)&0xffffffffU)) 55#define funsio(diff) (PType((((unsigned int)(diff))<<8)|FUNsio)) 56 57#define getBT1(a) ((unsigned char)((a)>> 8)) 58#define getBT2(a) ((unsigned char)((a)>>16)) 59#define getBT3(a) ((unsigned char)((a)>>24)) 60#define getBT0(a) (((unsigned char)((a)>> 5)&0x07)|0xF0) /* for UTF-8 only!!! */ 61 62#define getGB4bt0(a) ((unsigned char)((a)>> 8)) 63#define getGB4bt1(a) (((unsigned char)((a)>>24)&0x0F)|0x30) 64#define getGB4bt2(a) ((unsigned char)((a)>>16)) 65#define getGB4bt3(a) (((unsigned char)((a)>>28)&0x0F)|0x30) 66 67#define o2FUNii(b1,b2) (PType((((unsigned char)(b1))<<8)|(((unsigned char)(b2))<<16)|FUNii)) 68 69/* do we need these??? maybe not, can be done with simple tables */ 70#define ONETRAIL /* legal but undefined if one more trailing UTF-8 */ 71#define TWOTRAIL /* legal but undefined if two more trailing UTF-8 */ 72#define THREETRAIL /* legal but undefined if three more trailing UTF-8 */ 73 74typedef enum { 75 asciicompat_converter, /* ASCII-compatible -> ASCII-compatible */ 76 asciicompat_decoder, /* ASCII-incompatible -> ASCII-compatible */ 77 asciicompat_encoder /* ASCII-compatible -> ASCII-incompatible */ 78 /* ASCII-incompatible -> ASCII-incompatible is intentionally omitted. */ 79} rb_transcoder_asciicompat_type_t; 80 81typedef struct rb_transcoder rb_transcoder; 82 83/* static structure, one per supported encoding pair */ 84struct rb_transcoder { 85 const char *src_encoding; 86 const char *dst_encoding; 87 unsigned int conv_tree_start; 88 const unsigned char *byte_array; 89 unsigned int byte_array_length; 90 const unsigned int *word_array; 91 unsigned int word_array_length; 92 int word_size; 93 int input_unit_length; 94 int max_input; 95 int max_output; 96 rb_transcoder_asciicompat_type_t asciicompat_type; 97 size_t state_size; 98 int (*state_init_func)(void*); /* ret==0:success ret!=0:failure(errno) */ 99 int (*state_fini_func)(void*); /* ret==0:success ret!=0:failure(errno) */ 100 VALUE (*func_ii)(void*, VALUE); /* info -> info */ 101 VALUE (*func_si)(void*, const unsigned char*, size_t); /* start -> info */ 102 ssize_t (*func_io)(void*, VALUE, const unsigned char*, size_t); /* info -> output */ 103 ssize_t (*func_so)(void*, const unsigned char*, size_t, unsigned char*, size_t); /* start -> output */ 104 ssize_t (*finish_func)(void*, unsigned char*, size_t); /* -> output */ 105 ssize_t (*resetsize_func)(void*); /* -> len */ 106 ssize_t (*resetstate_func)(void*, unsigned char*, size_t); /* -> output */ 107 ssize_t (*func_sio)(void*, const unsigned char*, size_t, VALUE, unsigned char*, size_t); /* start -> output */ 108}; 109 110void rb_declare_transcoder(const char *enc1, const char *enc2, const char *lib); 111void rb_register_transcoder(const rb_transcoder *); 112 113/* 114 * To get rid of collision of initializer symbols in statically-linked encodings 115 * and transcoders 116 */ 117#if defined(EXTSTATIC) && EXTSTATIC 118# define TRANS_INIT(name) void Init_trans_ ## name(void) 119#else 120# define TRANS_INIT(name) void Init_ ## name(void) 121#endif 122 123#if defined __GNUC__ && __GNUC__ >= 4 124#pragma GCC visibility pop 125#endif 126 127#endif /* RUBY_TRANSCODE_DATA_H */ 128