| 1 | n/a | /* |
|---|
| 2 | n/a | * _codecs_kr.c: Codecs collection for Korean encodings |
|---|
| 3 | n/a | * |
|---|
| 4 | n/a | * Written by Hye-Shik Chang <perky@FreeBSD.org> |
|---|
| 5 | n/a | */ |
|---|
| 6 | n/a | |
|---|
| 7 | n/a | #include "cjkcodecs.h" |
|---|
| 8 | n/a | #include "mappings_kr.h" |
|---|
| 9 | n/a | |
|---|
| 10 | n/a | /* |
|---|
| 11 | n/a | * EUC-KR codec |
|---|
| 12 | n/a | */ |
|---|
| 13 | n/a | |
|---|
| 14 | n/a | #define EUCKR_JAMO_FIRSTBYTE 0xA4 |
|---|
| 15 | n/a | #define EUCKR_JAMO_FILLER 0xD4 |
|---|
| 16 | n/a | |
|---|
| 17 | n/a | static const unsigned char u2cgk_choseong[19] = { |
|---|
| 18 | n/a | 0xa1, 0xa2, 0xa4, 0xa7, 0xa8, 0xa9, 0xb1, 0xb2, |
|---|
| 19 | n/a | 0xb3, 0xb5, 0xb6, 0xb7, 0xb8, 0xb9, 0xba, 0xbb, |
|---|
| 20 | n/a | 0xbc, 0xbd, 0xbe |
|---|
| 21 | n/a | }; |
|---|
| 22 | n/a | static const unsigned char u2cgk_jungseong[21] = { |
|---|
| 23 | n/a | 0xbf, 0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5, 0xc6, |
|---|
| 24 | n/a | 0xc7, 0xc8, 0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce, |
|---|
| 25 | n/a | 0xcf, 0xd0, 0xd1, 0xd2, 0xd3 |
|---|
| 26 | n/a | }; |
|---|
| 27 | n/a | static const unsigned char u2cgk_jongseong[28] = { |
|---|
| 28 | n/a | 0xd4, 0xa1, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7, |
|---|
| 29 | n/a | 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf, 0xb0, |
|---|
| 30 | n/a | 0xb1, 0xb2, 0xb4, 0xb5, 0xb6, 0xb7, 0xb8, 0xba, |
|---|
| 31 | n/a | 0xbb, 0xbc, 0xbd, 0xbe |
|---|
| 32 | n/a | }; |
|---|
| 33 | n/a | |
|---|
| 34 | n/a | ENCODER(euc_kr) |
|---|
| 35 | n/a | { |
|---|
| 36 | n/a | while (*inpos < inlen) { |
|---|
| 37 | n/a | Py_UCS4 c = INCHAR1; |
|---|
| 38 | n/a | DBCHAR code; |
|---|
| 39 | n/a | |
|---|
| 40 | n/a | if (c < 0x80) { |
|---|
| 41 | n/a | WRITEBYTE1((unsigned char)c); |
|---|
| 42 | n/a | NEXT(1, 1); |
|---|
| 43 | n/a | continue; |
|---|
| 44 | n/a | } |
|---|
| 45 | n/a | |
|---|
| 46 | n/a | if (c > 0xFFFF) |
|---|
| 47 | n/a | return 1; |
|---|
| 48 | n/a | |
|---|
| 49 | n/a | REQUIRE_OUTBUF(2); |
|---|
| 50 | n/a | if (TRYMAP_ENC(cp949, code, c)) |
|---|
| 51 | n/a | ; |
|---|
| 52 | n/a | else |
|---|
| 53 | n/a | return 1; |
|---|
| 54 | n/a | |
|---|
| 55 | n/a | if ((code & 0x8000) == 0) { |
|---|
| 56 | n/a | /* KS X 1001 coded character */ |
|---|
| 57 | n/a | OUTBYTE1((code >> 8) | 0x80); |
|---|
| 58 | n/a | OUTBYTE2((code & 0xFF) | 0x80); |
|---|
| 59 | n/a | NEXT(1, 2); |
|---|
| 60 | n/a | } |
|---|
| 61 | n/a | else { |
|---|
| 62 | n/a | /* Mapping is found in CP949 extension, |
|---|
| 63 | n/a | but we encode it in KS X 1001:1998 Annex 3, |
|---|
| 64 | n/a | make-up sequence for EUC-KR. */ |
|---|
| 65 | n/a | |
|---|
| 66 | n/a | REQUIRE_OUTBUF(8); |
|---|
| 67 | n/a | |
|---|
| 68 | n/a | /* syllable composition precedence */ |
|---|
| 69 | n/a | OUTBYTE1(EUCKR_JAMO_FIRSTBYTE); |
|---|
| 70 | n/a | OUTBYTE2(EUCKR_JAMO_FILLER); |
|---|
| 71 | n/a | |
|---|
| 72 | n/a | /* All code points in CP949 extension are in unicode |
|---|
| 73 | n/a | * Hangul Syllable area. */ |
|---|
| 74 | n/a | assert(0xac00 <= c && c <= 0xd7a3); |
|---|
| 75 | n/a | c -= 0xac00; |
|---|
| 76 | n/a | |
|---|
| 77 | n/a | OUTBYTE3(EUCKR_JAMO_FIRSTBYTE); |
|---|
| 78 | n/a | OUTBYTE4(u2cgk_choseong[c / 588]); |
|---|
| 79 | n/a | NEXT_OUT(4); |
|---|
| 80 | n/a | |
|---|
| 81 | n/a | OUTBYTE1(EUCKR_JAMO_FIRSTBYTE); |
|---|
| 82 | n/a | OUTBYTE2(u2cgk_jungseong[(c / 28) % 21]); |
|---|
| 83 | n/a | OUTBYTE3(EUCKR_JAMO_FIRSTBYTE); |
|---|
| 84 | n/a | OUTBYTE4(u2cgk_jongseong[c % 28]); |
|---|
| 85 | n/a | NEXT(1, 4); |
|---|
| 86 | n/a | } |
|---|
| 87 | n/a | } |
|---|
| 88 | n/a | |
|---|
| 89 | n/a | return 0; |
|---|
| 90 | n/a | } |
|---|
| 91 | n/a | |
|---|
| 92 | n/a | #define NONE 127 |
|---|
| 93 | n/a | |
|---|
| 94 | n/a | static const unsigned char cgk2u_choseong[] = { /* [A1, BE] */ |
|---|
| 95 | n/a | 0, 1, NONE, 2, NONE, NONE, 3, 4, |
|---|
| 96 | n/a | 5, NONE, NONE, NONE, NONE, NONE, NONE, NONE, |
|---|
| 97 | n/a | 6, 7, 8, NONE, 9, 10, 11, 12, |
|---|
| 98 | n/a | 13, 14, 15, 16, 17, 18 |
|---|
| 99 | n/a | }; |
|---|
| 100 | n/a | static const unsigned char cgk2u_jongseong[] = { /* [A1, BE] */ |
|---|
| 101 | n/a | 1, 2, 3, 4, 5, 6, 7, NONE, |
|---|
| 102 | n/a | 8, 9, 10, 11, 12, 13, 14, 15, |
|---|
| 103 | n/a | 16, 17, NONE, 18, 19, 20, 21, 22, |
|---|
| 104 | n/a | NONE, 23, 24, 25, 26, 27 |
|---|
| 105 | n/a | }; |
|---|
| 106 | n/a | |
|---|
| 107 | n/a | DECODER(euc_kr) |
|---|
| 108 | n/a | { |
|---|
| 109 | n/a | while (inleft > 0) { |
|---|
| 110 | n/a | unsigned char c = INBYTE1; |
|---|
| 111 | n/a | Py_UCS4 decoded; |
|---|
| 112 | n/a | |
|---|
| 113 | n/a | if (c < 0x80) { |
|---|
| 114 | n/a | OUTCHAR(c); |
|---|
| 115 | n/a | NEXT_IN(1); |
|---|
| 116 | n/a | continue; |
|---|
| 117 | n/a | } |
|---|
| 118 | n/a | |
|---|
| 119 | n/a | REQUIRE_INBUF(2); |
|---|
| 120 | n/a | |
|---|
| 121 | n/a | if (c == EUCKR_JAMO_FIRSTBYTE && |
|---|
| 122 | n/a | INBYTE2 == EUCKR_JAMO_FILLER) { |
|---|
| 123 | n/a | /* KS X 1001:1998 Annex 3 make-up sequence */ |
|---|
| 124 | n/a | DBCHAR cho, jung, jong; |
|---|
| 125 | n/a | |
|---|
| 126 | n/a | REQUIRE_INBUF(8); |
|---|
| 127 | n/a | if ((*inbuf)[2] != EUCKR_JAMO_FIRSTBYTE || |
|---|
| 128 | n/a | (*inbuf)[4] != EUCKR_JAMO_FIRSTBYTE || |
|---|
| 129 | n/a | (*inbuf)[6] != EUCKR_JAMO_FIRSTBYTE) |
|---|
| 130 | n/a | return 1; |
|---|
| 131 | n/a | |
|---|
| 132 | n/a | c = (*inbuf)[3]; |
|---|
| 133 | n/a | if (0xa1 <= c && c <= 0xbe) |
|---|
| 134 | n/a | cho = cgk2u_choseong[c - 0xa1]; |
|---|
| 135 | n/a | else |
|---|
| 136 | n/a | cho = NONE; |
|---|
| 137 | n/a | |
|---|
| 138 | n/a | c = (*inbuf)[5]; |
|---|
| 139 | n/a | jung = (0xbf <= c && c <= 0xd3) ? c - 0xbf : NONE; |
|---|
| 140 | n/a | |
|---|
| 141 | n/a | c = (*inbuf)[7]; |
|---|
| 142 | n/a | if (c == EUCKR_JAMO_FILLER) |
|---|
| 143 | n/a | jong = 0; |
|---|
| 144 | n/a | else if (0xa1 <= c && c <= 0xbe) |
|---|
| 145 | n/a | jong = cgk2u_jongseong[c - 0xa1]; |
|---|
| 146 | n/a | else |
|---|
| 147 | n/a | jong = NONE; |
|---|
| 148 | n/a | |
|---|
| 149 | n/a | if (cho == NONE || jung == NONE || jong == NONE) |
|---|
| 150 | n/a | return 1; |
|---|
| 151 | n/a | |
|---|
| 152 | n/a | OUTCHAR(0xac00 + cho*588 + jung*28 + jong); |
|---|
| 153 | n/a | NEXT_IN(8); |
|---|
| 154 | n/a | } |
|---|
| 155 | n/a | else if (TRYMAP_DEC(ksx1001, decoded, c ^ 0x80, INBYTE2 ^ 0x80)) { |
|---|
| 156 | n/a | OUTCHAR(decoded); |
|---|
| 157 | n/a | NEXT_IN(2); |
|---|
| 158 | n/a | } |
|---|
| 159 | n/a | else |
|---|
| 160 | n/a | return 1; |
|---|
| 161 | n/a | } |
|---|
| 162 | n/a | |
|---|
| 163 | n/a | return 0; |
|---|
| 164 | n/a | } |
|---|
| 165 | n/a | #undef NONE |
|---|
| 166 | n/a | |
|---|
| 167 | n/a | |
|---|
| 168 | n/a | /* |
|---|
| 169 | n/a | * CP949 codec |
|---|
| 170 | n/a | */ |
|---|
| 171 | n/a | |
|---|
| 172 | n/a | ENCODER(cp949) |
|---|
| 173 | n/a | { |
|---|
| 174 | n/a | while (*inpos < inlen) { |
|---|
| 175 | n/a | Py_UCS4 c = INCHAR1; |
|---|
| 176 | n/a | DBCHAR code; |
|---|
| 177 | n/a | |
|---|
| 178 | n/a | if (c < 0x80) { |
|---|
| 179 | n/a | WRITEBYTE1((unsigned char)c); |
|---|
| 180 | n/a | NEXT(1, 1); |
|---|
| 181 | n/a | continue; |
|---|
| 182 | n/a | } |
|---|
| 183 | n/a | |
|---|
| 184 | n/a | if (c > 0xFFFF) |
|---|
| 185 | n/a | return 1; |
|---|
| 186 | n/a | |
|---|
| 187 | n/a | REQUIRE_OUTBUF(2); |
|---|
| 188 | n/a | if (TRYMAP_ENC(cp949, code, c)) |
|---|
| 189 | n/a | ; |
|---|
| 190 | n/a | else |
|---|
| 191 | n/a | return 1; |
|---|
| 192 | n/a | |
|---|
| 193 | n/a | OUTBYTE1((code >> 8) | 0x80); |
|---|
| 194 | n/a | if (code & 0x8000) |
|---|
| 195 | n/a | OUTBYTE2(code & 0xFF); /* MSB set: CP949 */ |
|---|
| 196 | n/a | else |
|---|
| 197 | n/a | OUTBYTE2((code & 0xFF) | 0x80); /* MSB unset: ks x 1001 */ |
|---|
| 198 | n/a | NEXT(1, 2); |
|---|
| 199 | n/a | } |
|---|
| 200 | n/a | |
|---|
| 201 | n/a | return 0; |
|---|
| 202 | n/a | } |
|---|
| 203 | n/a | |
|---|
| 204 | n/a | DECODER(cp949) |
|---|
| 205 | n/a | { |
|---|
| 206 | n/a | while (inleft > 0) { |
|---|
| 207 | n/a | unsigned char c = INBYTE1; |
|---|
| 208 | n/a | Py_UCS4 decoded; |
|---|
| 209 | n/a | |
|---|
| 210 | n/a | if (c < 0x80) { |
|---|
| 211 | n/a | OUTCHAR(c); |
|---|
| 212 | n/a | NEXT_IN(1); |
|---|
| 213 | n/a | continue; |
|---|
| 214 | n/a | } |
|---|
| 215 | n/a | |
|---|
| 216 | n/a | REQUIRE_INBUF(2); |
|---|
| 217 | n/a | if (TRYMAP_DEC(ksx1001, decoded, c ^ 0x80, INBYTE2 ^ 0x80)) |
|---|
| 218 | n/a | OUTCHAR(decoded); |
|---|
| 219 | n/a | else if (TRYMAP_DEC(cp949ext, decoded, c, INBYTE2)) |
|---|
| 220 | n/a | OUTCHAR(decoded); |
|---|
| 221 | n/a | else |
|---|
| 222 | n/a | return 1; |
|---|
| 223 | n/a | |
|---|
| 224 | n/a | NEXT_IN(2); |
|---|
| 225 | n/a | } |
|---|
| 226 | n/a | |
|---|
| 227 | n/a | return 0; |
|---|
| 228 | n/a | } |
|---|
| 229 | n/a | |
|---|
| 230 | n/a | |
|---|
| 231 | n/a | /* |
|---|
| 232 | n/a | * JOHAB codec |
|---|
| 233 | n/a | */ |
|---|
| 234 | n/a | |
|---|
| 235 | n/a | static const unsigned char u2johabidx_choseong[32] = { |
|---|
| 236 | n/a | 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, |
|---|
| 237 | n/a | 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, |
|---|
| 238 | n/a | 0x10, 0x11, 0x12, 0x13, 0x14, |
|---|
| 239 | n/a | }; |
|---|
| 240 | n/a | static const unsigned char u2johabidx_jungseong[32] = { |
|---|
| 241 | n/a | 0x03, 0x04, 0x05, 0x06, 0x07, |
|---|
| 242 | n/a | 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, |
|---|
| 243 | n/a | 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, |
|---|
| 244 | n/a | 0x1a, 0x1b, 0x1c, 0x1d, |
|---|
| 245 | n/a | }; |
|---|
| 246 | n/a | static const unsigned char u2johabidx_jongseong[32] = { |
|---|
| 247 | n/a | 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, |
|---|
| 248 | n/a | 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, |
|---|
| 249 | n/a | 0x10, 0x11, 0x13, 0x14, 0x15, 0x16, 0x17, |
|---|
| 250 | n/a | 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, |
|---|
| 251 | n/a | }; |
|---|
| 252 | n/a | static const DBCHAR u2johabjamo[] = { |
|---|
| 253 | n/a | 0x8841, 0x8c41, 0x8444, 0x9041, 0x8446, 0x8447, 0x9441, |
|---|
| 254 | n/a | 0x9841, 0x9c41, 0x844a, 0x844b, 0x844c, 0x844d, 0x844e, 0x844f, |
|---|
| 255 | n/a | 0x8450, 0xa041, 0xa441, 0xa841, 0x8454, 0xac41, 0xb041, 0xb441, |
|---|
| 256 | n/a | 0xb841, 0xbc41, 0xc041, 0xc441, 0xc841, 0xcc41, 0xd041, 0x8461, |
|---|
| 257 | n/a | 0x8481, 0x84a1, 0x84c1, 0x84e1, 0x8541, 0x8561, 0x8581, 0x85a1, |
|---|
| 258 | n/a | 0x85c1, 0x85e1, 0x8641, 0x8661, 0x8681, 0x86a1, 0x86c1, 0x86e1, |
|---|
| 259 | n/a | 0x8741, 0x8761, 0x8781, 0x87a1, |
|---|
| 260 | n/a | }; |
|---|
| 261 | n/a | |
|---|
| 262 | n/a | ENCODER(johab) |
|---|
| 263 | n/a | { |
|---|
| 264 | n/a | while (*inpos < inlen) { |
|---|
| 265 | n/a | Py_UCS4 c = INCHAR1; |
|---|
| 266 | n/a | DBCHAR code; |
|---|
| 267 | n/a | |
|---|
| 268 | n/a | if (c < 0x80) { |
|---|
| 269 | n/a | WRITEBYTE1((unsigned char)c); |
|---|
| 270 | n/a | NEXT(1, 1); |
|---|
| 271 | n/a | continue; |
|---|
| 272 | n/a | } |
|---|
| 273 | n/a | |
|---|
| 274 | n/a | if (c > 0xFFFF) |
|---|
| 275 | n/a | return 1; |
|---|
| 276 | n/a | |
|---|
| 277 | n/a | REQUIRE_OUTBUF(2); |
|---|
| 278 | n/a | |
|---|
| 279 | n/a | if (c >= 0xac00 && c <= 0xd7a3) { |
|---|
| 280 | n/a | c -= 0xac00; |
|---|
| 281 | n/a | code = 0x8000 | |
|---|
| 282 | n/a | (u2johabidx_choseong[c / 588] << 10) | |
|---|
| 283 | n/a | (u2johabidx_jungseong[(c / 28) % 21] << 5) | |
|---|
| 284 | n/a | u2johabidx_jongseong[c % 28]; |
|---|
| 285 | n/a | } |
|---|
| 286 | n/a | else if (c >= 0x3131 && c <= 0x3163) |
|---|
| 287 | n/a | code = u2johabjamo[c - 0x3131]; |
|---|
| 288 | n/a | else if (TRYMAP_ENC(cp949, code, c)) { |
|---|
| 289 | n/a | unsigned char c1, c2, t2; |
|---|
| 290 | n/a | unsigned short t1; |
|---|
| 291 | n/a | |
|---|
| 292 | n/a | assert((code & 0x8000) == 0); |
|---|
| 293 | n/a | c1 = code >> 8; |
|---|
| 294 | n/a | c2 = code & 0xff; |
|---|
| 295 | n/a | if (((c1 >= 0x21 && c1 <= 0x2c) || |
|---|
| 296 | n/a | (c1 >= 0x4a && c1 <= 0x7d)) && |
|---|
| 297 | n/a | (c2 >= 0x21 && c2 <= 0x7e)) { |
|---|
| 298 | n/a | t1 = (c1 < 0x4a ? (c1 - 0x21 + 0x1b2) : |
|---|
| 299 | n/a | (c1 - 0x21 + 0x197)); |
|---|
| 300 | n/a | t2 = ((t1 & 1) ? 0x5e : 0) + (c2 - 0x21); |
|---|
| 301 | n/a | OUTBYTE1(t1 >> 1); |
|---|
| 302 | n/a | OUTBYTE2(t2 < 0x4e ? t2 + 0x31 : t2 + 0x43); |
|---|
| 303 | n/a | NEXT(1, 2); |
|---|
| 304 | n/a | continue; |
|---|
| 305 | n/a | } |
|---|
| 306 | n/a | else |
|---|
| 307 | n/a | return 1; |
|---|
| 308 | n/a | } |
|---|
| 309 | n/a | else |
|---|
| 310 | n/a | return 1; |
|---|
| 311 | n/a | |
|---|
| 312 | n/a | OUTBYTE1(code >> 8); |
|---|
| 313 | n/a | OUTBYTE2(code & 0xff); |
|---|
| 314 | n/a | NEXT(1, 2); |
|---|
| 315 | n/a | } |
|---|
| 316 | n/a | |
|---|
| 317 | n/a | return 0; |
|---|
| 318 | n/a | } |
|---|
| 319 | n/a | |
|---|
| 320 | n/a | #define FILL 0xfd |
|---|
| 321 | n/a | #define NONE 0xff |
|---|
| 322 | n/a | |
|---|
| 323 | n/a | static const unsigned char johabidx_choseong[32] = { |
|---|
| 324 | n/a | NONE, FILL, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, |
|---|
| 325 | n/a | 0x06, 0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, |
|---|
| 326 | n/a | 0x0e, 0x0f, 0x10, 0x11, 0x12, NONE, NONE, NONE, |
|---|
| 327 | n/a | NONE, NONE, NONE, NONE, NONE, NONE, NONE, NONE, |
|---|
| 328 | n/a | }; |
|---|
| 329 | n/a | static const unsigned char johabidx_jungseong[32] = { |
|---|
| 330 | n/a | NONE, NONE, FILL, 0x00, 0x01, 0x02, 0x03, 0x04, |
|---|
| 331 | n/a | NONE, NONE, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0a, |
|---|
| 332 | n/a | NONE, NONE, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x10, |
|---|
| 333 | n/a | NONE, NONE, 0x11, 0x12, 0x13, 0x14, NONE, NONE, |
|---|
| 334 | n/a | }; |
|---|
| 335 | n/a | static const unsigned char johabidx_jongseong[32] = { |
|---|
| 336 | n/a | NONE, FILL, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, |
|---|
| 337 | n/a | 0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, |
|---|
| 338 | n/a | 0x0f, 0x10, NONE, 0x11, 0x12, 0x13, 0x14, 0x15, |
|---|
| 339 | n/a | 0x16, 0x17, 0x18, 0x19, 0x1a, 0x1b, NONE, NONE, |
|---|
| 340 | n/a | }; |
|---|
| 341 | n/a | |
|---|
| 342 | n/a | static const unsigned char johabjamo_choseong[32] = { |
|---|
| 343 | n/a | NONE, FILL, 0x31, 0x32, 0x34, 0x37, 0x38, 0x39, |
|---|
| 344 | n/a | 0x41, 0x42, 0x43, 0x45, 0x46, 0x47, 0x48, 0x49, |
|---|
| 345 | n/a | 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, NONE, NONE, NONE, |
|---|
| 346 | n/a | NONE, NONE, NONE, NONE, NONE, NONE, NONE, NONE, |
|---|
| 347 | n/a | }; |
|---|
| 348 | n/a | static const unsigned char johabjamo_jungseong[32] = { |
|---|
| 349 | n/a | NONE, NONE, FILL, 0x4f, 0x50, 0x51, 0x52, 0x53, |
|---|
| 350 | n/a | NONE, NONE, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59, |
|---|
| 351 | n/a | NONE, NONE, 0x5a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f, |
|---|
| 352 | n/a | NONE, NONE, 0x60, 0x61, 0x62, 0x63, NONE, NONE, |
|---|
| 353 | n/a | }; |
|---|
| 354 | n/a | static const unsigned char johabjamo_jongseong[32] = { |
|---|
| 355 | n/a | NONE, FILL, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, |
|---|
| 356 | n/a | 0x37, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f, |
|---|
| 357 | n/a | 0x40, 0x41, NONE, 0x42, 0x44, 0x45, 0x46, 0x47, |
|---|
| 358 | n/a | 0x48, 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, NONE, NONE, |
|---|
| 359 | n/a | }; |
|---|
| 360 | n/a | |
|---|
| 361 | n/a | DECODER(johab) |
|---|
| 362 | n/a | { |
|---|
| 363 | n/a | while (inleft > 0) { |
|---|
| 364 | n/a | unsigned char c = INBYTE1, c2; |
|---|
| 365 | n/a | Py_UCS4 decoded; |
|---|
| 366 | n/a | |
|---|
| 367 | n/a | if (c < 0x80) { |
|---|
| 368 | n/a | OUTCHAR(c); |
|---|
| 369 | n/a | NEXT_IN(1); |
|---|
| 370 | n/a | continue; |
|---|
| 371 | n/a | } |
|---|
| 372 | n/a | |
|---|
| 373 | n/a | REQUIRE_INBUF(2); |
|---|
| 374 | n/a | c2 = INBYTE2; |
|---|
| 375 | n/a | |
|---|
| 376 | n/a | if (c < 0xd8) { |
|---|
| 377 | n/a | /* johab hangul */ |
|---|
| 378 | n/a | unsigned char c_cho, c_jung, c_jong; |
|---|
| 379 | n/a | unsigned char i_cho, i_jung, i_jong; |
|---|
| 380 | n/a | |
|---|
| 381 | n/a | c_cho = (c >> 2) & 0x1f; |
|---|
| 382 | n/a | c_jung = ((c << 3) | c2 >> 5) & 0x1f; |
|---|
| 383 | n/a | c_jong = c2 & 0x1f; |
|---|
| 384 | n/a | |
|---|
| 385 | n/a | i_cho = johabidx_choseong[c_cho]; |
|---|
| 386 | n/a | i_jung = johabidx_jungseong[c_jung]; |
|---|
| 387 | n/a | i_jong = johabidx_jongseong[c_jong]; |
|---|
| 388 | n/a | |
|---|
| 389 | n/a | if (i_cho == NONE || i_jung == NONE || i_jong == NONE) |
|---|
| 390 | n/a | return 1; |
|---|
| 391 | n/a | |
|---|
| 392 | n/a | /* we don't use U+1100 hangul jamo yet. */ |
|---|
| 393 | n/a | if (i_cho == FILL) { |
|---|
| 394 | n/a | if (i_jung == FILL) { |
|---|
| 395 | n/a | if (i_jong == FILL) |
|---|
| 396 | n/a | OUTCHAR(0x3000); |
|---|
| 397 | n/a | else |
|---|
| 398 | n/a | OUTCHAR(0x3100 | |
|---|
| 399 | n/a | johabjamo_jongseong[c_jong]); |
|---|
| 400 | n/a | } |
|---|
| 401 | n/a | else { |
|---|
| 402 | n/a | if (i_jong == FILL) |
|---|
| 403 | n/a | OUTCHAR(0x3100 | |
|---|
| 404 | n/a | johabjamo_jungseong[c_jung]); |
|---|
| 405 | n/a | else |
|---|
| 406 | n/a | return 1; |
|---|
| 407 | n/a | } |
|---|
| 408 | n/a | } else { |
|---|
| 409 | n/a | if (i_jung == FILL) { |
|---|
| 410 | n/a | if (i_jong == FILL) |
|---|
| 411 | n/a | OUTCHAR(0x3100 | |
|---|
| 412 | n/a | johabjamo_choseong[c_cho]); |
|---|
| 413 | n/a | else |
|---|
| 414 | n/a | return 1; |
|---|
| 415 | n/a | } |
|---|
| 416 | n/a | else |
|---|
| 417 | n/a | OUTCHAR(0xac00 + |
|---|
| 418 | n/a | i_cho * 588 + |
|---|
| 419 | n/a | i_jung * 28 + |
|---|
| 420 | n/a | (i_jong == FILL ? 0 : i_jong)); |
|---|
| 421 | n/a | } |
|---|
| 422 | n/a | NEXT_IN(2); |
|---|
| 423 | n/a | } else { |
|---|
| 424 | n/a | /* KS X 1001 except hangul jamos and syllables */ |
|---|
| 425 | n/a | if (c == 0xdf || c > 0xf9 || |
|---|
| 426 | n/a | c2 < 0x31 || (c2 >= 0x80 && c2 < 0x91) || |
|---|
| 427 | n/a | (c2 & 0x7f) == 0x7f || |
|---|
| 428 | n/a | (c == 0xda && (c2 >= 0xa1 && c2 <= 0xd3))) |
|---|
| 429 | n/a | return 1; |
|---|
| 430 | n/a | else { |
|---|
| 431 | n/a | unsigned char t1, t2; |
|---|
| 432 | n/a | |
|---|
| 433 | n/a | t1 = (c < 0xe0 ? 2 * (c - 0xd9) : |
|---|
| 434 | n/a | 2 * c - 0x197); |
|---|
| 435 | n/a | t2 = (c2 < 0x91 ? c2 - 0x31 : c2 - 0x43); |
|---|
| 436 | n/a | t1 = t1 + (t2 < 0x5e ? 0 : 1) + 0x21; |
|---|
| 437 | n/a | t2 = (t2 < 0x5e ? t2 : t2 - 0x5e) + 0x21; |
|---|
| 438 | n/a | |
|---|
| 439 | n/a | if (TRYMAP_DEC(ksx1001, decoded, t1, t2)) { |
|---|
| 440 | n/a | OUTCHAR(decoded); |
|---|
| 441 | n/a | NEXT_IN(2); |
|---|
| 442 | n/a | } |
|---|
| 443 | n/a | else { |
|---|
| 444 | n/a | return 1; |
|---|
| 445 | n/a | } |
|---|
| 446 | n/a | } |
|---|
| 447 | n/a | } |
|---|
| 448 | n/a | } |
|---|
| 449 | n/a | |
|---|
| 450 | n/a | return 0; |
|---|
| 451 | n/a | } |
|---|
| 452 | n/a | #undef NONE |
|---|
| 453 | n/a | #undef FILL |
|---|
| 454 | n/a | |
|---|
| 455 | n/a | |
|---|
| 456 | n/a | BEGIN_MAPPINGS_LIST |
|---|
| 457 | n/a | MAPPING_DECONLY(ksx1001) |
|---|
| 458 | n/a | MAPPING_ENCONLY(cp949) |
|---|
| 459 | n/a | MAPPING_DECONLY(cp949ext) |
|---|
| 460 | n/a | END_MAPPINGS_LIST |
|---|
| 461 | n/a | |
|---|
| 462 | n/a | BEGIN_CODECS_LIST |
|---|
| 463 | n/a | CODEC_STATELESS(euc_kr) |
|---|
| 464 | n/a | CODEC_STATELESS(cp949) |
|---|
| 465 | n/a | CODEC_STATELESS(johab) |
|---|
| 466 | n/a | END_CODECS_LIST |
|---|
| 467 | n/a | |
|---|
| 468 | n/a | I_AM_A_MODULE_FOR(kr) |
|---|