uhc.c (4875B)
1 2 #include "wc.h" 3 #include "uhc.h" 4 #include "wtf.h" 5 #ifdef USE_UNICODE 6 #include "ucs.h" 7 #endif 8 9 #define C0 WC_UHC_MAP_C0 10 #define GL WC_UHC_MAP_GL 11 #define C1 WC_UHC_MAP_C1 12 #define LB WC_UHC_MAP_LB 13 #define UB WC_UHC_MAP_UB 14 15 wc_uint8 WC_UHC_MAP[ 0x100 ] = { 16 C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, 17 C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, 18 GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, 19 GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, 20 GL, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, 21 LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, GL, GL, GL, GL, GL, 22 GL, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, 23 LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, GL, GL, GL, GL, C0, 24 25 C1, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, 26 UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, 27 UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, 28 UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, 29 UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, 30 UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, 31 UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, 32 UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, UB, C1, 33 }; 34 35 wc_wchar_t 36 wc_uhc_to_cs128w(wc_wchar_t cc) 37 { 38 cc.code = WC_UHC_N(cc.code); 39 if (cc.code < 0x4000) 40 cc.ccs = WC_CCS_UHC_1; 41 else { 42 cc.ccs = WC_CCS_UHC_2; 43 cc.code -= 0x4000; 44 } 45 cc.code = WC_N_CS128W(cc.code); 46 return cc; 47 } 48 49 wc_wchar_t 50 wc_cs128w_to_uhc(wc_wchar_t cc) 51 { 52 cc.code = WC_CS128W_N(cc.code); 53 if (cc.ccs == WC_CCS_UHC_2) 54 cc.code += 0x4000; 55 cc.ccs = WC_CCS_UHC; 56 cc.code = WC_N_UHC(cc.code); 57 return cc; 58 } 59 60 wc_uint32 61 wc_uhc_to_N(wc_uint32 c) 62 { 63 if (c <= 0xA1A0) /* 0x8141 - 0xA1A0 */ 64 return WC_UHC_N(c); 65 if (c <= 0xA2A0) /* 0xA240 - 0xA2A0 */ 66 return WC_UHC_N(c) - 0x5E; 67 if (c <= 0xA2E7) /* 0xA2E6 - 0xA2E7 */ 68 return WC_UHC_N(0xA2A0) - 0x5E + c - 0xA2E5; 69 /* 0xA340 - 0xFEA0 */ 70 return WC_UHC_N(c) - ((c >> 8) - 0xA1) * 0x5E + 2; 71 } 72 73 Str 74 wc_conv_from_uhc(Str is, wc_ces ces) 75 { 76 Str os; 77 wc_uchar *sp = (wc_uchar *)is->ptr; 78 wc_uchar *ep = sp + is->length; 79 wc_uchar *p; 80 int state = WC_UHC_NOSTATE; 81 wc_uint32 uhc; 82 83 for (p = sp; p < ep && *p < 0x80; p++) 84 ; 85 if (p == ep) 86 return is; 87 os = Strnew_size(is->length); 88 if (p > sp) 89 Strcat_charp_n(os, (char *)is->ptr, (int)(p - sp)); 90 91 for (; p < ep; p++) { 92 switch (state) { 93 case WC_UHC_NOSTATE: 94 switch (WC_UHC_MAP[*p]) { 95 case UB: 96 state = WC_UHC_MBYTE1; 97 break; 98 case C1: 99 wtf_push_unknown(os, p, 1); 100 break; 101 default: 102 Strcat_char(os, (char)*p); 103 break; 104 } 105 break; 106 case WC_UHC_MBYTE1: 107 if (WC_UHC_MAP[*p] & LB) { 108 uhc = ((wc_uint32)*(p-1) << 8) | *p; 109 if (*(p-1) >= 0xA1 && *p >= 0xA1 && 110 uhc != 0xA2E6 && uhc != 0xA2E7) 111 wtf_push(os, WC_CCS_KS_X_1001, uhc); 112 else 113 wtf_push(os, WC_CCS_UHC, uhc); 114 } else 115 wtf_push_unknown(os, p-1, 2); 116 state = WC_UHC_NOSTATE; 117 break; 118 } 119 } 120 switch (state) { 121 case WC_UHC_MBYTE1: 122 wtf_push_unknown(os, p-1, 1); 123 break; 124 } 125 return os; 126 } 127 128 void 129 wc_push_to_uhc(Str os, wc_wchar_t cc, wc_status *st) 130 { 131 while (1) { 132 switch (cc.ccs) { 133 case WC_CCS_US_ASCII: 134 Strcat_char(os, (char)cc.code); 135 return; 136 case WC_CCS_KS_X_1001: 137 Strcat_char(os, (char)((cc.code >> 8) | 0x80)); 138 Strcat_char(os, (char)((cc.code & 0xff) | 0x80)); 139 return; 140 case WC_CCS_UHC_1: 141 case WC_CCS_UHC_2: 142 cc = wc_cs128w_to_uhc(cc); 143 case WC_CCS_UHC: 144 Strcat_char(os, (char)(cc.code >> 8)); 145 Strcat_char(os, (char)(cc.code & 0xff)); 146 return; 147 case WC_CCS_UNKNOWN_W: 148 if (!WcOption.no_replace) 149 Strcat_charp(os, WC_REPLACE_W); 150 return; 151 case WC_CCS_UNKNOWN: 152 if (!WcOption.no_replace) 153 Strcat_charp(os, WC_REPLACE); 154 return; 155 default: 156 #ifdef USE_UNICODE 157 if (WcOption.ucs_conv) 158 cc = wc_any_to_any_ces(cc, st); 159 else 160 #endif 161 cc.ccs = WC_CCS_IS_WIDE(cc.ccs) ? WC_CCS_UNKNOWN_W : WC_CCS_UNKNOWN; 162 continue; 163 } 164 } 165 } 166 167 Str 168 wc_char_conv_from_uhc(wc_uchar c, wc_status *st) 169 { 170 static Str os; 171 static wc_uchar uhcu; 172 wc_uint32 uhc; 173 174 if (st->state == -1) { 175 st->state = WC_UHC_NOSTATE; 176 os = Strnew_size(8); 177 } 178 179 switch (st->state) { 180 case WC_UHC_NOSTATE: 181 switch (WC_UHC_MAP[c]) { 182 case UB: 183 uhcu = c; 184 st->state = WC_UHC_MBYTE1; 185 return NULL; 186 case C1: 187 break; 188 default: 189 Strcat_char(os, (char)c); 190 break; 191 } 192 break; 193 case WC_UHC_MBYTE1: 194 if (WC_UHC_MAP[c] & LB) { 195 uhc = ((wc_uint32)uhcu << 8) | c; 196 if (uhcu >= 0xA1 && c >= 0xA1 && 197 uhc != 0xA2E6 && uhc != 0xA2E7) 198 wtf_push(os, WC_CCS_KS_X_1001, uhc); 199 else 200 wtf_push(os, WC_CCS_UHC, uhc); 201 } 202 break; 203 } 204 st->state = -1; 205 return os; 206 }