| 1 | n/a | /* |
|---|
| 2 | n/a | * _codecs_iso2022.c: Codecs collection for ISO-2022 encodings. |
|---|
| 3 | n/a | * |
|---|
| 4 | n/a | * Written by Hye-Shik Chang <perky@FreeBSD.org> |
|---|
| 5 | n/a | */ |
|---|
| 6 | n/a | |
|---|
| 7 | n/a | #define USING_IMPORTED_MAPS |
|---|
| 8 | n/a | #define USING_BINARY_PAIR_SEARCH |
|---|
| 9 | n/a | #define EXTERN_JISX0213_PAIR |
|---|
| 10 | n/a | #define EMULATE_JISX0213_2000_ENCODE_INVALID MAP_UNMAPPABLE |
|---|
| 11 | n/a | #define EMULATE_JISX0213_2000_DECODE_INVALID MAP_UNMAPPABLE |
|---|
| 12 | n/a | |
|---|
| 13 | n/a | #include "cjkcodecs.h" |
|---|
| 14 | n/a | #include "alg_jisx0201.h" |
|---|
| 15 | n/a | #include "emu_jisx0213_2000.h" |
|---|
| 16 | n/a | #include "mappings_jisx0213_pair.h" |
|---|
| 17 | n/a | |
|---|
| 18 | n/a | /* STATE |
|---|
| 19 | n/a | |
|---|
| 20 | n/a | state->c[0-3] |
|---|
| 21 | n/a | |
|---|
| 22 | n/a | 00000000 |
|---|
| 23 | n/a | ||^^^^^| |
|---|
| 24 | n/a | |+-----+---- G0-3 Character Set |
|---|
| 25 | n/a | +----------- Is G0-3 double byte? |
|---|
| 26 | n/a | |
|---|
| 27 | n/a | state->c[4] |
|---|
| 28 | n/a | |
|---|
| 29 | n/a | 00000000 |
|---|
| 30 | n/a | || |
|---|
| 31 | n/a | |+---- Locked-Shift? |
|---|
| 32 | n/a | +----- ESC Throughout |
|---|
| 33 | n/a | */ |
|---|
| 34 | n/a | |
|---|
| 35 | n/a | #define ESC 0x1B |
|---|
| 36 | n/a | #define SO 0x0E |
|---|
| 37 | n/a | #define SI 0x0F |
|---|
| 38 | n/a | #define LF 0x0A |
|---|
| 39 | n/a | |
|---|
| 40 | n/a | #define MAX_ESCSEQLEN 16 |
|---|
| 41 | n/a | |
|---|
| 42 | n/a | #define CHARSET_ISO8859_1 'A' |
|---|
| 43 | n/a | #define CHARSET_ASCII 'B' |
|---|
| 44 | n/a | #define CHARSET_ISO8859_7 'F' |
|---|
| 45 | n/a | #define CHARSET_JISX0201_K 'I' |
|---|
| 46 | n/a | #define CHARSET_JISX0201_R 'J' |
|---|
| 47 | n/a | |
|---|
| 48 | n/a | #define CHARSET_GB2312 ('A'|CHARSET_DBCS) |
|---|
| 49 | n/a | #define CHARSET_JISX0208 ('B'|CHARSET_DBCS) |
|---|
| 50 | n/a | #define CHARSET_KSX1001 ('C'|CHARSET_DBCS) |
|---|
| 51 | n/a | #define CHARSET_JISX0212 ('D'|CHARSET_DBCS) |
|---|
| 52 | n/a | #define CHARSET_GB2312_8565 ('E'|CHARSET_DBCS) |
|---|
| 53 | n/a | #define CHARSET_CNS11643_1 ('G'|CHARSET_DBCS) |
|---|
| 54 | n/a | #define CHARSET_CNS11643_2 ('H'|CHARSET_DBCS) |
|---|
| 55 | n/a | #define CHARSET_JISX0213_2000_1 ('O'|CHARSET_DBCS) |
|---|
| 56 | n/a | #define CHARSET_JISX0213_2 ('P'|CHARSET_DBCS) |
|---|
| 57 | n/a | #define CHARSET_JISX0213_2004_1 ('Q'|CHARSET_DBCS) |
|---|
| 58 | n/a | #define CHARSET_JISX0208_O ('@'|CHARSET_DBCS) |
|---|
| 59 | n/a | |
|---|
| 60 | n/a | #define CHARSET_DBCS 0x80 |
|---|
| 61 | n/a | #define ESCMARK(mark) ((mark) & 0x7f) |
|---|
| 62 | n/a | |
|---|
| 63 | n/a | #define IS_ESCEND(c) (((c) >= 'A' && (c) <= 'Z') || (c) == '@') |
|---|
| 64 | n/a | #define IS_ISO2022ESC(c2) \ |
|---|
| 65 | n/a | ((c2) == '(' || (c2) == ')' || (c2) == '$' || \ |
|---|
| 66 | n/a | (c2) == '.' || (c2) == '&') |
|---|
| 67 | n/a | /* this is not a complete list of ISO-2022 escape sequence headers. |
|---|
| 68 | n/a | * but, it's enough to implement CJK instances of iso-2022. */ |
|---|
| 69 | n/a | |
|---|
| 70 | n/a | #define MAP_UNMAPPABLE 0xFFFF |
|---|
| 71 | n/a | #define MAP_MULTIPLE_AVAIL 0xFFFE /* for JIS X 0213 */ |
|---|
| 72 | n/a | |
|---|
| 73 | n/a | #define F_SHIFTED 0x01 |
|---|
| 74 | n/a | #define F_ESCTHROUGHOUT 0x02 |
|---|
| 75 | n/a | |
|---|
| 76 | n/a | #define STATE_SETG(dn, v) do { ((state)->c[dn]) = (v); } while (0) |
|---|
| 77 | n/a | #define STATE_GETG(dn) ((state)->c[dn]) |
|---|
| 78 | n/a | |
|---|
| 79 | n/a | #define STATE_G0 STATE_GETG(0) |
|---|
| 80 | n/a | #define STATE_G1 STATE_GETG(1) |
|---|
| 81 | n/a | #define STATE_G2 STATE_GETG(2) |
|---|
| 82 | n/a | #define STATE_G3 STATE_GETG(3) |
|---|
| 83 | n/a | #define STATE_SETG0(v) STATE_SETG(0, v) |
|---|
| 84 | n/a | #define STATE_SETG1(v) STATE_SETG(1, v) |
|---|
| 85 | n/a | #define STATE_SETG2(v) STATE_SETG(2, v) |
|---|
| 86 | n/a | #define STATE_SETG3(v) STATE_SETG(3, v) |
|---|
| 87 | n/a | |
|---|
| 88 | n/a | #define STATE_SETFLAG(f) do { ((state)->c[4]) |= (f); } while (0) |
|---|
| 89 | n/a | #define STATE_GETFLAG(f) ((state)->c[4] & (f)) |
|---|
| 90 | n/a | #define STATE_CLEARFLAG(f) do { ((state)->c[4]) &= ~(f); } while (0) |
|---|
| 91 | n/a | #define STATE_CLEARFLAGS() do { ((state)->c[4]) = 0; } while (0) |
|---|
| 92 | n/a | |
|---|
| 93 | n/a | #define ISO2022_CONFIG ((const struct iso2022_config *)config) |
|---|
| 94 | n/a | #define CONFIG_ISSET(flag) (ISO2022_CONFIG->flags & (flag)) |
|---|
| 95 | n/a | #define CONFIG_DESIGNATIONS (ISO2022_CONFIG->designations) |
|---|
| 96 | n/a | |
|---|
| 97 | n/a | /* iso2022_config.flags */ |
|---|
| 98 | n/a | #define NO_SHIFT 0x01 |
|---|
| 99 | n/a | #define USE_G2 0x02 |
|---|
| 100 | n/a | #define USE_JISX0208_EXT 0x04 |
|---|
| 101 | n/a | |
|---|
| 102 | n/a | /*-*- internal data structures -*-*/ |
|---|
| 103 | n/a | |
|---|
| 104 | n/a | typedef int (*iso2022_init_func)(void); |
|---|
| 105 | n/a | typedef Py_UCS4 (*iso2022_decode_func)(const unsigned char *data); |
|---|
| 106 | n/a | typedef DBCHAR (*iso2022_encode_func)(const Py_UCS4 *data, Py_ssize_t *length); |
|---|
| 107 | n/a | |
|---|
| 108 | n/a | struct iso2022_designation { |
|---|
| 109 | n/a | unsigned char mark; |
|---|
| 110 | n/a | unsigned char plane; |
|---|
| 111 | n/a | unsigned char width; |
|---|
| 112 | n/a | iso2022_init_func initializer; |
|---|
| 113 | n/a | iso2022_decode_func decoder; |
|---|
| 114 | n/a | iso2022_encode_func encoder; |
|---|
| 115 | n/a | }; |
|---|
| 116 | n/a | |
|---|
| 117 | n/a | struct iso2022_config { |
|---|
| 118 | n/a | int flags; |
|---|
| 119 | n/a | const struct iso2022_designation *designations; /* non-ascii desigs */ |
|---|
| 120 | n/a | }; |
|---|
| 121 | n/a | |
|---|
| 122 | n/a | /*-*- iso-2022 codec implementation -*-*/ |
|---|
| 123 | n/a | |
|---|
| 124 | n/a | CODEC_INIT(iso2022) |
|---|
| 125 | n/a | { |
|---|
| 126 | n/a | const struct iso2022_designation *desig; |
|---|
| 127 | n/a | for (desig = CONFIG_DESIGNATIONS; desig->mark; desig++) |
|---|
| 128 | n/a | if (desig->initializer != NULL && desig->initializer() != 0) |
|---|
| 129 | n/a | return -1; |
|---|
| 130 | n/a | return 0; |
|---|
| 131 | n/a | } |
|---|
| 132 | n/a | |
|---|
| 133 | n/a | ENCODER_INIT(iso2022) |
|---|
| 134 | n/a | { |
|---|
| 135 | n/a | STATE_CLEARFLAGS(); |
|---|
| 136 | n/a | STATE_SETG0(CHARSET_ASCII); |
|---|
| 137 | n/a | STATE_SETG1(CHARSET_ASCII); |
|---|
| 138 | n/a | return 0; |
|---|
| 139 | n/a | } |
|---|
| 140 | n/a | |
|---|
| 141 | n/a | ENCODER_RESET(iso2022) |
|---|
| 142 | n/a | { |
|---|
| 143 | n/a | if (STATE_GETFLAG(F_SHIFTED)) { |
|---|
| 144 | n/a | WRITEBYTE1(SI); |
|---|
| 145 | n/a | NEXT_OUT(1); |
|---|
| 146 | n/a | STATE_CLEARFLAG(F_SHIFTED); |
|---|
| 147 | n/a | } |
|---|
| 148 | n/a | if (STATE_G0 != CHARSET_ASCII) { |
|---|
| 149 | n/a | WRITEBYTE3(ESC, '(', 'B'); |
|---|
| 150 | n/a | NEXT_OUT(3); |
|---|
| 151 | n/a | STATE_SETG0(CHARSET_ASCII); |
|---|
| 152 | n/a | } |
|---|
| 153 | n/a | return 0; |
|---|
| 154 | n/a | } |
|---|
| 155 | n/a | |
|---|
| 156 | n/a | ENCODER(iso2022) |
|---|
| 157 | n/a | { |
|---|
| 158 | n/a | while (*inpos < inlen) { |
|---|
| 159 | n/a | const struct iso2022_designation *dsg; |
|---|
| 160 | n/a | DBCHAR encoded; |
|---|
| 161 | n/a | Py_UCS4 c = INCHAR1; |
|---|
| 162 | n/a | Py_ssize_t insize; |
|---|
| 163 | n/a | |
|---|
| 164 | n/a | if (c < 0x80) { |
|---|
| 165 | n/a | if (STATE_G0 != CHARSET_ASCII) { |
|---|
| 166 | n/a | WRITEBYTE3(ESC, '(', 'B'); |
|---|
| 167 | n/a | STATE_SETG0(CHARSET_ASCII); |
|---|
| 168 | n/a | NEXT_OUT(3); |
|---|
| 169 | n/a | } |
|---|
| 170 | n/a | if (STATE_GETFLAG(F_SHIFTED)) { |
|---|
| 171 | n/a | WRITEBYTE1(SI); |
|---|
| 172 | n/a | STATE_CLEARFLAG(F_SHIFTED); |
|---|
| 173 | n/a | NEXT_OUT(1); |
|---|
| 174 | n/a | } |
|---|
| 175 | n/a | WRITEBYTE1((unsigned char)c); |
|---|
| 176 | n/a | NEXT(1, 1); |
|---|
| 177 | n/a | continue; |
|---|
| 178 | n/a | } |
|---|
| 179 | n/a | |
|---|
| 180 | n/a | insize = 1; |
|---|
| 181 | n/a | |
|---|
| 182 | n/a | encoded = MAP_UNMAPPABLE; |
|---|
| 183 | n/a | for (dsg = CONFIG_DESIGNATIONS; dsg->mark; dsg++) { |
|---|
| 184 | n/a | Py_ssize_t length = 1; |
|---|
| 185 | n/a | encoded = dsg->encoder(&c, &length); |
|---|
| 186 | n/a | if (encoded == MAP_MULTIPLE_AVAIL) { |
|---|
| 187 | n/a | /* this implementation won't work for pair |
|---|
| 188 | n/a | * of non-bmp characters. */ |
|---|
| 189 | n/a | if (inlen - *inpos < 2) { |
|---|
| 190 | n/a | if (!(flags & MBENC_FLUSH)) |
|---|
| 191 | n/a | return MBERR_TOOFEW; |
|---|
| 192 | n/a | length = -1; |
|---|
| 193 | n/a | } |
|---|
| 194 | n/a | else |
|---|
| 195 | n/a | length = 2; |
|---|
| 196 | n/a | encoded = dsg->encoder(&c, &length); |
|---|
| 197 | n/a | if (encoded != MAP_UNMAPPABLE) { |
|---|
| 198 | n/a | insize = length; |
|---|
| 199 | n/a | break; |
|---|
| 200 | n/a | } |
|---|
| 201 | n/a | } |
|---|
| 202 | n/a | else if (encoded != MAP_UNMAPPABLE) |
|---|
| 203 | n/a | break; |
|---|
| 204 | n/a | } |
|---|
| 205 | n/a | |
|---|
| 206 | n/a | if (!dsg->mark) |
|---|
| 207 | n/a | return 1; |
|---|
| 208 | n/a | assert(dsg->width == 1 || dsg->width == 2); |
|---|
| 209 | n/a | |
|---|
| 210 | n/a | switch (dsg->plane) { |
|---|
| 211 | n/a | case 0: /* G0 */ |
|---|
| 212 | n/a | if (STATE_GETFLAG(F_SHIFTED)) { |
|---|
| 213 | n/a | WRITEBYTE1(SI); |
|---|
| 214 | n/a | STATE_CLEARFLAG(F_SHIFTED); |
|---|
| 215 | n/a | NEXT_OUT(1); |
|---|
| 216 | n/a | } |
|---|
| 217 | n/a | if (STATE_G0 != dsg->mark) { |
|---|
| 218 | n/a | if (dsg->width == 1) { |
|---|
| 219 | n/a | WRITEBYTE3(ESC, '(', ESCMARK(dsg->mark)); |
|---|
| 220 | n/a | STATE_SETG0(dsg->mark); |
|---|
| 221 | n/a | NEXT_OUT(3); |
|---|
| 222 | n/a | } |
|---|
| 223 | n/a | else if (dsg->mark == CHARSET_JISX0208) { |
|---|
| 224 | n/a | WRITEBYTE3(ESC, '$', ESCMARK(dsg->mark)); |
|---|
| 225 | n/a | STATE_SETG0(dsg->mark); |
|---|
| 226 | n/a | NEXT_OUT(3); |
|---|
| 227 | n/a | } |
|---|
| 228 | n/a | else { |
|---|
| 229 | n/a | WRITEBYTE4(ESC, '$', '(', |
|---|
| 230 | n/a | ESCMARK(dsg->mark)); |
|---|
| 231 | n/a | STATE_SETG0(dsg->mark); |
|---|
| 232 | n/a | NEXT_OUT(4); |
|---|
| 233 | n/a | } |
|---|
| 234 | n/a | } |
|---|
| 235 | n/a | break; |
|---|
| 236 | n/a | case 1: /* G1 */ |
|---|
| 237 | n/a | if (STATE_G1 != dsg->mark) { |
|---|
| 238 | n/a | if (dsg->width == 1) { |
|---|
| 239 | n/a | WRITEBYTE3(ESC, ')', ESCMARK(dsg->mark)); |
|---|
| 240 | n/a | STATE_SETG1(dsg->mark); |
|---|
| 241 | n/a | NEXT_OUT(3); |
|---|
| 242 | n/a | } |
|---|
| 243 | n/a | else { |
|---|
| 244 | n/a | WRITEBYTE4(ESC, '$', ')', ESCMARK(dsg->mark)); |
|---|
| 245 | n/a | STATE_SETG1(dsg->mark); |
|---|
| 246 | n/a | NEXT_OUT(4); |
|---|
| 247 | n/a | } |
|---|
| 248 | n/a | } |
|---|
| 249 | n/a | if (!STATE_GETFLAG(F_SHIFTED)) { |
|---|
| 250 | n/a | WRITEBYTE1(SO); |
|---|
| 251 | n/a | STATE_SETFLAG(F_SHIFTED); |
|---|
| 252 | n/a | NEXT_OUT(1); |
|---|
| 253 | n/a | } |
|---|
| 254 | n/a | break; |
|---|
| 255 | n/a | default: /* G2 and G3 is not supported: no encoding in |
|---|
| 256 | n/a | * CJKCodecs are using them yet */ |
|---|
| 257 | n/a | return MBERR_INTERNAL; |
|---|
| 258 | n/a | } |
|---|
| 259 | n/a | |
|---|
| 260 | n/a | if (dsg->width == 1) { |
|---|
| 261 | n/a | WRITEBYTE1((unsigned char)encoded); |
|---|
| 262 | n/a | NEXT_OUT(1); |
|---|
| 263 | n/a | } |
|---|
| 264 | n/a | else { |
|---|
| 265 | n/a | WRITEBYTE2(encoded >> 8, encoded & 0xff); |
|---|
| 266 | n/a | NEXT_OUT(2); |
|---|
| 267 | n/a | } |
|---|
| 268 | n/a | NEXT_INCHAR(insize); |
|---|
| 269 | n/a | } |
|---|
| 270 | n/a | |
|---|
| 271 | n/a | return 0; |
|---|
| 272 | n/a | } |
|---|
| 273 | n/a | |
|---|
| 274 | n/a | DECODER_INIT(iso2022) |
|---|
| 275 | n/a | { |
|---|
| 276 | n/a | STATE_CLEARFLAGS(); |
|---|
| 277 | n/a | STATE_SETG0(CHARSET_ASCII); |
|---|
| 278 | n/a | STATE_SETG1(CHARSET_ASCII); |
|---|
| 279 | n/a | STATE_SETG2(CHARSET_ASCII); |
|---|
| 280 | n/a | return 0; |
|---|
| 281 | n/a | } |
|---|
| 282 | n/a | |
|---|
| 283 | n/a | DECODER_RESET(iso2022) |
|---|
| 284 | n/a | { |
|---|
| 285 | n/a | STATE_SETG0(CHARSET_ASCII); |
|---|
| 286 | n/a | STATE_CLEARFLAG(F_SHIFTED); |
|---|
| 287 | n/a | return 0; |
|---|
| 288 | n/a | } |
|---|
| 289 | n/a | |
|---|
| 290 | n/a | static Py_ssize_t |
|---|
| 291 | n/a | iso2022processesc(const void *config, MultibyteCodec_State *state, |
|---|
| 292 | n/a | const unsigned char **inbuf, Py_ssize_t *inleft) |
|---|
| 293 | n/a | { |
|---|
| 294 | n/a | unsigned char charset, designation; |
|---|
| 295 | n/a | Py_ssize_t i, esclen = 0; |
|---|
| 296 | n/a | |
|---|
| 297 | n/a | for (i = 1;i < MAX_ESCSEQLEN;i++) { |
|---|
| 298 | n/a | if (i >= *inleft) |
|---|
| 299 | n/a | return MBERR_TOOFEW; |
|---|
| 300 | n/a | if (IS_ESCEND((*inbuf)[i])) { |
|---|
| 301 | n/a | esclen = i + 1; |
|---|
| 302 | n/a | break; |
|---|
| 303 | n/a | } |
|---|
| 304 | n/a | else if (CONFIG_ISSET(USE_JISX0208_EXT) && i+1 < *inleft && |
|---|
| 305 | n/a | (*inbuf)[i] == '&' && (*inbuf)[i+1] == '@') { |
|---|
| 306 | n/a | i += 2; |
|---|
| 307 | n/a | } |
|---|
| 308 | n/a | } |
|---|
| 309 | n/a | |
|---|
| 310 | n/a | switch (esclen) { |
|---|
| 311 | n/a | case 0: |
|---|
| 312 | n/a | return 1; /* unterminated escape sequence */ |
|---|
| 313 | n/a | case 3: |
|---|
| 314 | n/a | if (INBYTE2 == '$') { |
|---|
| 315 | n/a | charset = INBYTE3 | CHARSET_DBCS; |
|---|
| 316 | n/a | designation = 0; |
|---|
| 317 | n/a | } |
|---|
| 318 | n/a | else { |
|---|
| 319 | n/a | charset = INBYTE3; |
|---|
| 320 | n/a | if (INBYTE2 == '(') |
|---|
| 321 | n/a | designation = 0; |
|---|
| 322 | n/a | else if (INBYTE2 == ')') |
|---|
| 323 | n/a | designation = 1; |
|---|
| 324 | n/a | else if (CONFIG_ISSET(USE_G2) && INBYTE2 == '.') |
|---|
| 325 | n/a | designation = 2; |
|---|
| 326 | n/a | else |
|---|
| 327 | n/a | return 3; |
|---|
| 328 | n/a | } |
|---|
| 329 | n/a | break; |
|---|
| 330 | n/a | case 4: |
|---|
| 331 | n/a | if (INBYTE2 != '$') |
|---|
| 332 | n/a | return 4; |
|---|
| 333 | n/a | |
|---|
| 334 | n/a | charset = INBYTE4 | CHARSET_DBCS; |
|---|
| 335 | n/a | if (INBYTE3 == '(') |
|---|
| 336 | n/a | designation = 0; |
|---|
| 337 | n/a | else if (INBYTE3 == ')') |
|---|
| 338 | n/a | designation = 1; |
|---|
| 339 | n/a | else |
|---|
| 340 | n/a | return 4; |
|---|
| 341 | n/a | break; |
|---|
| 342 | n/a | case 6: /* designation with prefix */ |
|---|
| 343 | n/a | if (CONFIG_ISSET(USE_JISX0208_EXT) && |
|---|
| 344 | n/a | (*inbuf)[3] == ESC && (*inbuf)[4] == '$' && |
|---|
| 345 | n/a | (*inbuf)[5] == 'B') { |
|---|
| 346 | n/a | charset = 'B' | CHARSET_DBCS; |
|---|
| 347 | n/a | designation = 0; |
|---|
| 348 | n/a | } |
|---|
| 349 | n/a | else |
|---|
| 350 | n/a | return 6; |
|---|
| 351 | n/a | break; |
|---|
| 352 | n/a | default: |
|---|
| 353 | n/a | return esclen; |
|---|
| 354 | n/a | } |
|---|
| 355 | n/a | |
|---|
| 356 | n/a | /* raise error when the charset is not designated for this encoding */ |
|---|
| 357 | n/a | if (charset != CHARSET_ASCII) { |
|---|
| 358 | n/a | const struct iso2022_designation *dsg; |
|---|
| 359 | n/a | |
|---|
| 360 | n/a | for (dsg = CONFIG_DESIGNATIONS; dsg->mark; dsg++) { |
|---|
| 361 | n/a | if (dsg->mark == charset) |
|---|
| 362 | n/a | break; |
|---|
| 363 | n/a | } |
|---|
| 364 | n/a | if (!dsg->mark) |
|---|
| 365 | n/a | return esclen; |
|---|
| 366 | n/a | } |
|---|
| 367 | n/a | |
|---|
| 368 | n/a | STATE_SETG(designation, charset); |
|---|
| 369 | n/a | *inleft -= esclen; |
|---|
| 370 | n/a | (*inbuf) += esclen; |
|---|
| 371 | n/a | return 0; |
|---|
| 372 | n/a | } |
|---|
| 373 | n/a | |
|---|
| 374 | n/a | #define ISO8859_7_DECODE(c, writer) \ |
|---|
| 375 | n/a | if ((c) < 0xa0) { \ |
|---|
| 376 | n/a | OUTCHAR(c); \ |
|---|
| 377 | n/a | } else if ((c) < 0xc0 && (0x288f3bc9L & (1L << ((c)-0xa0)))) { \ |
|---|
| 378 | n/a | OUTCHAR(c); \ |
|---|
| 379 | n/a | } else if ((c) >= 0xb4 && (c) <= 0xfe && ((c) >= 0xd4 || \ |
|---|
| 380 | n/a | (0xbffffd77L & (1L << ((c)-0xb4))))) { \ |
|---|
| 381 | n/a | OUTCHAR(0x02d0 + (c)); \ |
|---|
| 382 | n/a | } else if ((c) == 0xa1) { \ |
|---|
| 383 | n/a | OUTCHAR(0x2018); \ |
|---|
| 384 | n/a | } else if ((c) == 0xa2) { \ |
|---|
| 385 | n/a | OUTCHAR(0x2019); \ |
|---|
| 386 | n/a | } else if ((c) == 0xaf) { \ |
|---|
| 387 | n/a | OUTCHAR(0x2015); \ |
|---|
| 388 | n/a | } |
|---|
| 389 | n/a | |
|---|
| 390 | n/a | static Py_ssize_t |
|---|
| 391 | n/a | iso2022processg2(const void *config, MultibyteCodec_State *state, |
|---|
| 392 | n/a | const unsigned char **inbuf, Py_ssize_t *inleft, |
|---|
| 393 | n/a | _PyUnicodeWriter *writer) |
|---|
| 394 | n/a | { |
|---|
| 395 | n/a | /* not written to use encoder, decoder functions because only few |
|---|
| 396 | n/a | * encodings use G2 designations in CJKCodecs */ |
|---|
| 397 | n/a | if (STATE_G2 == CHARSET_ISO8859_1) { |
|---|
| 398 | n/a | if (INBYTE3 < 0x80) |
|---|
| 399 | n/a | OUTCHAR(INBYTE3 + 0x80); |
|---|
| 400 | n/a | else |
|---|
| 401 | n/a | return 3; |
|---|
| 402 | n/a | } |
|---|
| 403 | n/a | else if (STATE_G2 == CHARSET_ISO8859_7) { |
|---|
| 404 | n/a | ISO8859_7_DECODE(INBYTE3 ^ 0x80, writer) |
|---|
| 405 | n/a | else |
|---|
| 406 | n/a | return 3; |
|---|
| 407 | n/a | } |
|---|
| 408 | n/a | else if (STATE_G2 == CHARSET_ASCII) { |
|---|
| 409 | n/a | if (INBYTE3 & 0x80) |
|---|
| 410 | n/a | return 3; |
|---|
| 411 | n/a | else |
|---|
| 412 | n/a | OUTCHAR(INBYTE3); |
|---|
| 413 | n/a | } |
|---|
| 414 | n/a | else |
|---|
| 415 | n/a | return MBERR_INTERNAL; |
|---|
| 416 | n/a | |
|---|
| 417 | n/a | (*inbuf) += 3; |
|---|
| 418 | n/a | *inleft -= 3; |
|---|
| 419 | n/a | return 0; |
|---|
| 420 | n/a | } |
|---|
| 421 | n/a | |
|---|
| 422 | n/a | DECODER(iso2022) |
|---|
| 423 | n/a | { |
|---|
| 424 | n/a | const struct iso2022_designation *dsgcache = NULL; |
|---|
| 425 | n/a | |
|---|
| 426 | n/a | while (inleft > 0) { |
|---|
| 427 | n/a | unsigned char c = INBYTE1; |
|---|
| 428 | n/a | Py_ssize_t err; |
|---|
| 429 | n/a | |
|---|
| 430 | n/a | if (STATE_GETFLAG(F_ESCTHROUGHOUT)) { |
|---|
| 431 | n/a | /* ESC throughout mode: |
|---|
| 432 | n/a | * for non-iso2022 escape sequences */ |
|---|
| 433 | n/a | OUTCHAR(c); /* assume as ISO-8859-1 */ |
|---|
| 434 | n/a | NEXT_IN(1); |
|---|
| 435 | n/a | if (IS_ESCEND(c)) { |
|---|
| 436 | n/a | STATE_CLEARFLAG(F_ESCTHROUGHOUT); |
|---|
| 437 | n/a | } |
|---|
| 438 | n/a | continue; |
|---|
| 439 | n/a | } |
|---|
| 440 | n/a | |
|---|
| 441 | n/a | switch (c) { |
|---|
| 442 | n/a | case ESC: |
|---|
| 443 | n/a | REQUIRE_INBUF(2); |
|---|
| 444 | n/a | if (IS_ISO2022ESC(INBYTE2)) { |
|---|
| 445 | n/a | err = iso2022processesc(config, state, |
|---|
| 446 | n/a | inbuf, &inleft); |
|---|
| 447 | n/a | if (err != 0) |
|---|
| 448 | n/a | return err; |
|---|
| 449 | n/a | } |
|---|
| 450 | n/a | else if (CONFIG_ISSET(USE_G2) && INBYTE2 == 'N') {/* SS2 */ |
|---|
| 451 | n/a | REQUIRE_INBUF(3); |
|---|
| 452 | n/a | err = iso2022processg2(config, state, |
|---|
| 453 | n/a | inbuf, &inleft, writer); |
|---|
| 454 | n/a | if (err != 0) |
|---|
| 455 | n/a | return err; |
|---|
| 456 | n/a | } |
|---|
| 457 | n/a | else { |
|---|
| 458 | n/a | OUTCHAR(ESC); |
|---|
| 459 | n/a | STATE_SETFLAG(F_ESCTHROUGHOUT); |
|---|
| 460 | n/a | NEXT_IN(1); |
|---|
| 461 | n/a | } |
|---|
| 462 | n/a | break; |
|---|
| 463 | n/a | case SI: |
|---|
| 464 | n/a | if (CONFIG_ISSET(NO_SHIFT)) |
|---|
| 465 | n/a | goto bypass; |
|---|
| 466 | n/a | STATE_CLEARFLAG(F_SHIFTED); |
|---|
| 467 | n/a | NEXT_IN(1); |
|---|
| 468 | n/a | break; |
|---|
| 469 | n/a | case SO: |
|---|
| 470 | n/a | if (CONFIG_ISSET(NO_SHIFT)) |
|---|
| 471 | n/a | goto bypass; |
|---|
| 472 | n/a | STATE_SETFLAG(F_SHIFTED); |
|---|
| 473 | n/a | NEXT_IN(1); |
|---|
| 474 | n/a | break; |
|---|
| 475 | n/a | case LF: |
|---|
| 476 | n/a | STATE_CLEARFLAG(F_SHIFTED); |
|---|
| 477 | n/a | OUTCHAR(LF); |
|---|
| 478 | n/a | NEXT_IN(1); |
|---|
| 479 | n/a | break; |
|---|
| 480 | n/a | default: |
|---|
| 481 | n/a | if (c < 0x20) /* C0 */ |
|---|
| 482 | n/a | goto bypass; |
|---|
| 483 | n/a | else if (c >= 0x80) |
|---|
| 484 | n/a | return 1; |
|---|
| 485 | n/a | else { |
|---|
| 486 | n/a | const struct iso2022_designation *dsg; |
|---|
| 487 | n/a | unsigned char charset; |
|---|
| 488 | n/a | Py_UCS4 decoded; |
|---|
| 489 | n/a | |
|---|
| 490 | n/a | if (STATE_GETFLAG(F_SHIFTED)) |
|---|
| 491 | n/a | charset = STATE_G1; |
|---|
| 492 | n/a | else |
|---|
| 493 | n/a | charset = STATE_G0; |
|---|
| 494 | n/a | |
|---|
| 495 | n/a | if (charset == CHARSET_ASCII) { |
|---|
| 496 | n/a | bypass: |
|---|
| 497 | n/a | OUTCHAR(c); |
|---|
| 498 | n/a | NEXT_IN(1); |
|---|
| 499 | n/a | break; |
|---|
| 500 | n/a | } |
|---|
| 501 | n/a | |
|---|
| 502 | n/a | if (dsgcache != NULL && |
|---|
| 503 | n/a | dsgcache->mark == charset) |
|---|
| 504 | n/a | dsg = dsgcache; |
|---|
| 505 | n/a | else { |
|---|
| 506 | n/a | for (dsg = CONFIG_DESIGNATIONS; |
|---|
| 507 | n/a | dsg->mark != charset |
|---|
| 508 | n/a | #ifdef Py_DEBUG |
|---|
| 509 | n/a | && dsg->mark != '\0' |
|---|
| 510 | n/a | #endif |
|---|
| 511 | n/a | ; dsg++) |
|---|
| 512 | n/a | { |
|---|
| 513 | n/a | /* noop */ |
|---|
| 514 | n/a | } |
|---|
| 515 | n/a | assert(dsg->mark != '\0'); |
|---|
| 516 | n/a | dsgcache = dsg; |
|---|
| 517 | n/a | } |
|---|
| 518 | n/a | |
|---|
| 519 | n/a | REQUIRE_INBUF(dsg->width); |
|---|
| 520 | n/a | decoded = dsg->decoder(*inbuf); |
|---|
| 521 | n/a | if (decoded == MAP_UNMAPPABLE) |
|---|
| 522 | n/a | return dsg->width; |
|---|
| 523 | n/a | |
|---|
| 524 | n/a | if (decoded < 0x10000) { |
|---|
| 525 | n/a | OUTCHAR(decoded); |
|---|
| 526 | n/a | } |
|---|
| 527 | n/a | else if (decoded < 0x30000) { |
|---|
| 528 | n/a | OUTCHAR(decoded); |
|---|
| 529 | n/a | } |
|---|
| 530 | n/a | else { /* JIS X 0213 pairs */ |
|---|
| 531 | n/a | OUTCHAR2(decoded >> 16, decoded & 0xffff); |
|---|
| 532 | n/a | } |
|---|
| 533 | n/a | NEXT_IN(dsg->width); |
|---|
| 534 | n/a | } |
|---|
| 535 | n/a | break; |
|---|
| 536 | n/a | } |
|---|
| 537 | n/a | } |
|---|
| 538 | n/a | return 0; |
|---|
| 539 | n/a | } |
|---|
| 540 | n/a | |
|---|
| 541 | n/a | /*-*- mapping table holders -*-*/ |
|---|
| 542 | n/a | |
|---|
| 543 | n/a | #define ENCMAP(enc) static const encode_map *enc##_encmap = NULL; |
|---|
| 544 | n/a | #define DECMAP(enc) static const decode_map *enc##_decmap = NULL; |
|---|
| 545 | n/a | |
|---|
| 546 | n/a | /* kr */ |
|---|
| 547 | n/a | ENCMAP(cp949) |
|---|
| 548 | n/a | DECMAP(ksx1001) |
|---|
| 549 | n/a | |
|---|
| 550 | n/a | /* jp */ |
|---|
| 551 | n/a | ENCMAP(jisxcommon) |
|---|
| 552 | n/a | DECMAP(jisx0208) |
|---|
| 553 | n/a | DECMAP(jisx0212) |
|---|
| 554 | n/a | ENCMAP(jisx0213_bmp) |
|---|
| 555 | n/a | DECMAP(jisx0213_1_bmp) |
|---|
| 556 | n/a | DECMAP(jisx0213_2_bmp) |
|---|
| 557 | n/a | ENCMAP(jisx0213_emp) |
|---|
| 558 | n/a | DECMAP(jisx0213_1_emp) |
|---|
| 559 | n/a | DECMAP(jisx0213_2_emp) |
|---|
| 560 | n/a | |
|---|
| 561 | n/a | /* cn */ |
|---|
| 562 | n/a | ENCMAP(gbcommon) |
|---|
| 563 | n/a | DECMAP(gb2312) |
|---|
| 564 | n/a | |
|---|
| 565 | n/a | /* tw */ |
|---|
| 566 | n/a | |
|---|
| 567 | n/a | /*-*- mapping access functions -*-*/ |
|---|
| 568 | n/a | |
|---|
| 569 | n/a | static int |
|---|
| 570 | n/a | ksx1001_init(void) |
|---|
| 571 | n/a | { |
|---|
| 572 | n/a | static int initialized = 0; |
|---|
| 573 | n/a | |
|---|
| 574 | n/a | if (!initialized && ( |
|---|
| 575 | n/a | IMPORT_MAP(kr, cp949, &cp949_encmap, NULL) || |
|---|
| 576 | n/a | IMPORT_MAP(kr, ksx1001, NULL, &ksx1001_decmap))) |
|---|
| 577 | n/a | return -1; |
|---|
| 578 | n/a | initialized = 1; |
|---|
| 579 | n/a | return 0; |
|---|
| 580 | n/a | } |
|---|
| 581 | n/a | |
|---|
| 582 | n/a | static Py_UCS4 |
|---|
| 583 | n/a | ksx1001_decoder(const unsigned char *data) |
|---|
| 584 | n/a | { |
|---|
| 585 | n/a | Py_UCS4 u; |
|---|
| 586 | n/a | if (TRYMAP_DEC(ksx1001, u, data[0], data[1])) |
|---|
| 587 | n/a | return u; |
|---|
| 588 | n/a | else |
|---|
| 589 | n/a | return MAP_UNMAPPABLE; |
|---|
| 590 | n/a | } |
|---|
| 591 | n/a | |
|---|
| 592 | n/a | static DBCHAR |
|---|
| 593 | n/a | ksx1001_encoder(const Py_UCS4 *data, Py_ssize_t *length) |
|---|
| 594 | n/a | { |
|---|
| 595 | n/a | DBCHAR coded; |
|---|
| 596 | n/a | assert(*length == 1); |
|---|
| 597 | n/a | if (*data < 0x10000) { |
|---|
| 598 | n/a | if (TRYMAP_ENC(cp949, coded, *data)) { |
|---|
| 599 | n/a | if (!(coded & 0x8000)) |
|---|
| 600 | n/a | return coded; |
|---|
| 601 | n/a | } |
|---|
| 602 | n/a | } |
|---|
| 603 | n/a | return MAP_UNMAPPABLE; |
|---|
| 604 | n/a | } |
|---|
| 605 | n/a | |
|---|
| 606 | n/a | static int |
|---|
| 607 | n/a | jisx0208_init(void) |
|---|
| 608 | n/a | { |
|---|
| 609 | n/a | static int initialized = 0; |
|---|
| 610 | n/a | |
|---|
| 611 | n/a | if (!initialized && ( |
|---|
| 612 | n/a | IMPORT_MAP(jp, jisxcommon, &jisxcommon_encmap, NULL) || |
|---|
| 613 | n/a | IMPORT_MAP(jp, jisx0208, NULL, &jisx0208_decmap))) |
|---|
| 614 | n/a | return -1; |
|---|
| 615 | n/a | initialized = 1; |
|---|
| 616 | n/a | return 0; |
|---|
| 617 | n/a | } |
|---|
| 618 | n/a | |
|---|
| 619 | n/a | static Py_UCS4 |
|---|
| 620 | n/a | jisx0208_decoder(const unsigned char *data) |
|---|
| 621 | n/a | { |
|---|
| 622 | n/a | Py_UCS4 u; |
|---|
| 623 | n/a | if (data[0] == 0x21 && data[1] == 0x40) /* F/W REVERSE SOLIDUS */ |
|---|
| 624 | n/a | return 0xff3c; |
|---|
| 625 | n/a | else if (TRYMAP_DEC(jisx0208, u, data[0], data[1])) |
|---|
| 626 | n/a | return u; |
|---|
| 627 | n/a | else |
|---|
| 628 | n/a | return MAP_UNMAPPABLE; |
|---|
| 629 | n/a | } |
|---|
| 630 | n/a | |
|---|
| 631 | n/a | static DBCHAR |
|---|
| 632 | n/a | jisx0208_encoder(const Py_UCS4 *data, Py_ssize_t *length) |
|---|
| 633 | n/a | { |
|---|
| 634 | n/a | DBCHAR coded; |
|---|
| 635 | n/a | assert(*length == 1); |
|---|
| 636 | n/a | if (*data < 0x10000) { |
|---|
| 637 | n/a | if (*data == 0xff3c) /* F/W REVERSE SOLIDUS */ |
|---|
| 638 | n/a | return 0x2140; |
|---|
| 639 | n/a | else if (TRYMAP_ENC(jisxcommon, coded, *data)) { |
|---|
| 640 | n/a | if (!(coded & 0x8000)) |
|---|
| 641 | n/a | return coded; |
|---|
| 642 | n/a | } |
|---|
| 643 | n/a | } |
|---|
| 644 | n/a | return MAP_UNMAPPABLE; |
|---|
| 645 | n/a | } |
|---|
| 646 | n/a | |
|---|
| 647 | n/a | static int |
|---|
| 648 | n/a | jisx0212_init(void) |
|---|
| 649 | n/a | { |
|---|
| 650 | n/a | static int initialized = 0; |
|---|
| 651 | n/a | |
|---|
| 652 | n/a | if (!initialized && ( |
|---|
| 653 | n/a | IMPORT_MAP(jp, jisxcommon, &jisxcommon_encmap, NULL) || |
|---|
| 654 | n/a | IMPORT_MAP(jp, jisx0212, NULL, &jisx0212_decmap))) |
|---|
| 655 | n/a | return -1; |
|---|
| 656 | n/a | initialized = 1; |
|---|
| 657 | n/a | return 0; |
|---|
| 658 | n/a | } |
|---|
| 659 | n/a | |
|---|
| 660 | n/a | static Py_UCS4 |
|---|
| 661 | n/a | jisx0212_decoder(const unsigned char *data) |
|---|
| 662 | n/a | { |
|---|
| 663 | n/a | Py_UCS4 u; |
|---|
| 664 | n/a | if (TRYMAP_DEC(jisx0212, u, data[0], data[1])) |
|---|
| 665 | n/a | return u; |
|---|
| 666 | n/a | else |
|---|
| 667 | n/a | return MAP_UNMAPPABLE; |
|---|
| 668 | n/a | } |
|---|
| 669 | n/a | |
|---|
| 670 | n/a | static DBCHAR |
|---|
| 671 | n/a | jisx0212_encoder(const Py_UCS4 *data, Py_ssize_t *length) |
|---|
| 672 | n/a | { |
|---|
| 673 | n/a | DBCHAR coded; |
|---|
| 674 | n/a | assert(*length == 1); |
|---|
| 675 | n/a | if (*data < 0x10000) { |
|---|
| 676 | n/a | if (TRYMAP_ENC(jisxcommon, coded, *data)) { |
|---|
| 677 | n/a | if (coded & 0x8000) |
|---|
| 678 | n/a | return coded & 0x7fff; |
|---|
| 679 | n/a | } |
|---|
| 680 | n/a | } |
|---|
| 681 | n/a | return MAP_UNMAPPABLE; |
|---|
| 682 | n/a | } |
|---|
| 683 | n/a | |
|---|
| 684 | n/a | static int |
|---|
| 685 | n/a | jisx0213_init(void) |
|---|
| 686 | n/a | { |
|---|
| 687 | n/a | static int initialized = 0; |
|---|
| 688 | n/a | |
|---|
| 689 | n/a | if (!initialized && ( |
|---|
| 690 | n/a | jisx0208_init() || |
|---|
| 691 | n/a | IMPORT_MAP(jp, jisx0213_bmp, |
|---|
| 692 | n/a | &jisx0213_bmp_encmap, NULL) || |
|---|
| 693 | n/a | IMPORT_MAP(jp, jisx0213_1_bmp, |
|---|
| 694 | n/a | NULL, &jisx0213_1_bmp_decmap) || |
|---|
| 695 | n/a | IMPORT_MAP(jp, jisx0213_2_bmp, |
|---|
| 696 | n/a | NULL, &jisx0213_2_bmp_decmap) || |
|---|
| 697 | n/a | IMPORT_MAP(jp, jisx0213_emp, |
|---|
| 698 | n/a | &jisx0213_emp_encmap, NULL) || |
|---|
| 699 | n/a | IMPORT_MAP(jp, jisx0213_1_emp, |
|---|
| 700 | n/a | NULL, &jisx0213_1_emp_decmap) || |
|---|
| 701 | n/a | IMPORT_MAP(jp, jisx0213_2_emp, |
|---|
| 702 | n/a | NULL, &jisx0213_2_emp_decmap) || |
|---|
| 703 | n/a | IMPORT_MAP(jp, jisx0213_pair, &jisx0213_pair_encmap, |
|---|
| 704 | n/a | &jisx0213_pair_decmap))) |
|---|
| 705 | n/a | return -1; |
|---|
| 706 | n/a | initialized = 1; |
|---|
| 707 | n/a | return 0; |
|---|
| 708 | n/a | } |
|---|
| 709 | n/a | |
|---|
| 710 | n/a | #define config ((void *)2000) |
|---|
| 711 | n/a | static Py_UCS4 |
|---|
| 712 | n/a | jisx0213_2000_1_decoder(const unsigned char *data) |
|---|
| 713 | n/a | { |
|---|
| 714 | n/a | Py_UCS4 u; |
|---|
| 715 | n/a | EMULATE_JISX0213_2000_DECODE_PLANE1(u, data[0], data[1]) |
|---|
| 716 | n/a | else if (data[0] == 0x21 && data[1] == 0x40) /* F/W REVERSE SOLIDUS */ |
|---|
| 717 | n/a | return 0xff3c; |
|---|
| 718 | n/a | else if (TRYMAP_DEC(jisx0208, u, data[0], data[1])) |
|---|
| 719 | n/a | ; |
|---|
| 720 | n/a | else if (TRYMAP_DEC(jisx0213_1_bmp, u, data[0], data[1])) |
|---|
| 721 | n/a | ; |
|---|
| 722 | n/a | else if (TRYMAP_DEC(jisx0213_1_emp, u, data[0], data[1])) |
|---|
| 723 | n/a | u |= 0x20000; |
|---|
| 724 | n/a | else if (TRYMAP_DEC(jisx0213_pair, u, data[0], data[1])) |
|---|
| 725 | n/a | ; |
|---|
| 726 | n/a | else |
|---|
| 727 | n/a | return MAP_UNMAPPABLE; |
|---|
| 728 | n/a | return u; |
|---|
| 729 | n/a | } |
|---|
| 730 | n/a | |
|---|
| 731 | n/a | static Py_UCS4 |
|---|
| 732 | n/a | jisx0213_2000_2_decoder(const unsigned char *data) |
|---|
| 733 | n/a | { |
|---|
| 734 | n/a | Py_UCS4 u; |
|---|
| 735 | n/a | EMULATE_JISX0213_2000_DECODE_PLANE2_CHAR(u, data[0], data[1]) |
|---|
| 736 | n/a | if (TRYMAP_DEC(jisx0213_2_bmp, u, data[0], data[1])) |
|---|
| 737 | n/a | ; |
|---|
| 738 | n/a | else if (TRYMAP_DEC(jisx0213_2_emp, u, data[0], data[1])) |
|---|
| 739 | n/a | u |= 0x20000; |
|---|
| 740 | n/a | else |
|---|
| 741 | n/a | return MAP_UNMAPPABLE; |
|---|
| 742 | n/a | return u; |
|---|
| 743 | n/a | } |
|---|
| 744 | n/a | #undef config |
|---|
| 745 | n/a | |
|---|
| 746 | n/a | static Py_UCS4 |
|---|
| 747 | n/a | jisx0213_2004_1_decoder(const unsigned char *data) |
|---|
| 748 | n/a | { |
|---|
| 749 | n/a | Py_UCS4 u; |
|---|
| 750 | n/a | if (data[0] == 0x21 && data[1] == 0x40) /* F/W REVERSE SOLIDUS */ |
|---|
| 751 | n/a | return 0xff3c; |
|---|
| 752 | n/a | else if (TRYMAP_DEC(jisx0208, u, data[0], data[1])) |
|---|
| 753 | n/a | ; |
|---|
| 754 | n/a | else if (TRYMAP_DEC(jisx0213_1_bmp, u, data[0], data[1])) |
|---|
| 755 | n/a | ; |
|---|
| 756 | n/a | else if (TRYMAP_DEC(jisx0213_1_emp, u, data[0], data[1])) |
|---|
| 757 | n/a | u |= 0x20000; |
|---|
| 758 | n/a | else if (TRYMAP_DEC(jisx0213_pair, u, data[0], data[1])) |
|---|
| 759 | n/a | ; |
|---|
| 760 | n/a | else |
|---|
| 761 | n/a | return MAP_UNMAPPABLE; |
|---|
| 762 | n/a | return u; |
|---|
| 763 | n/a | } |
|---|
| 764 | n/a | |
|---|
| 765 | n/a | static Py_UCS4 |
|---|
| 766 | n/a | jisx0213_2004_2_decoder(const unsigned char *data) |
|---|
| 767 | n/a | { |
|---|
| 768 | n/a | Py_UCS4 u; |
|---|
| 769 | n/a | if (TRYMAP_DEC(jisx0213_2_bmp, u, data[0], data[1])) |
|---|
| 770 | n/a | ; |
|---|
| 771 | n/a | else if (TRYMAP_DEC(jisx0213_2_emp, u, data[0], data[1])) |
|---|
| 772 | n/a | u |= 0x20000; |
|---|
| 773 | n/a | else |
|---|
| 774 | n/a | return MAP_UNMAPPABLE; |
|---|
| 775 | n/a | return u; |
|---|
| 776 | n/a | } |
|---|
| 777 | n/a | |
|---|
| 778 | n/a | static DBCHAR |
|---|
| 779 | n/a | jisx0213_encoder(const Py_UCS4 *data, Py_ssize_t *length, void *config) |
|---|
| 780 | n/a | { |
|---|
| 781 | n/a | DBCHAR coded; |
|---|
| 782 | n/a | |
|---|
| 783 | n/a | switch (*length) { |
|---|
| 784 | n/a | case 1: /* first character */ |
|---|
| 785 | n/a | if (*data >= 0x10000) { |
|---|
| 786 | n/a | if ((*data) >> 16 == 0x20000 >> 16) { |
|---|
| 787 | n/a | EMULATE_JISX0213_2000_ENCODE_EMP(coded, *data) |
|---|
| 788 | n/a | else if (TRYMAP_ENC(jisx0213_emp, coded, (*data) & 0xffff)) |
|---|
| 789 | n/a | return coded; |
|---|
| 790 | n/a | } |
|---|
| 791 | n/a | return MAP_UNMAPPABLE; |
|---|
| 792 | n/a | } |
|---|
| 793 | n/a | |
|---|
| 794 | n/a | EMULATE_JISX0213_2000_ENCODE_BMP(coded, *data) |
|---|
| 795 | n/a | else if (TRYMAP_ENC(jisx0213_bmp, coded, *data)) { |
|---|
| 796 | n/a | if (coded == MULTIC) |
|---|
| 797 | n/a | return MAP_MULTIPLE_AVAIL; |
|---|
| 798 | n/a | } |
|---|
| 799 | n/a | else if (TRYMAP_ENC(jisxcommon, coded, *data)) { |
|---|
| 800 | n/a | if (coded & 0x8000) |
|---|
| 801 | n/a | return MAP_UNMAPPABLE; |
|---|
| 802 | n/a | } |
|---|
| 803 | n/a | else |
|---|
| 804 | n/a | return MAP_UNMAPPABLE; |
|---|
| 805 | n/a | return coded; |
|---|
| 806 | n/a | |
|---|
| 807 | n/a | case 2: /* second character of unicode pair */ |
|---|
| 808 | n/a | coded = find_pairencmap((ucs2_t)data[0], (ucs2_t)data[1], |
|---|
| 809 | n/a | jisx0213_pair_encmap, JISX0213_ENCPAIRS); |
|---|
| 810 | n/a | if (coded == DBCINV) { |
|---|
| 811 | n/a | *length = 1; |
|---|
| 812 | n/a | coded = find_pairencmap((ucs2_t)data[0], 0, |
|---|
| 813 | n/a | jisx0213_pair_encmap, JISX0213_ENCPAIRS); |
|---|
| 814 | n/a | if (coded == DBCINV) |
|---|
| 815 | n/a | return MAP_UNMAPPABLE; |
|---|
| 816 | n/a | } |
|---|
| 817 | n/a | else |
|---|
| 818 | n/a | return coded; |
|---|
| 819 | n/a | |
|---|
| 820 | n/a | case -1: /* flush unterminated */ |
|---|
| 821 | n/a | *length = 1; |
|---|
| 822 | n/a | coded = find_pairencmap((ucs2_t)data[0], 0, |
|---|
| 823 | n/a | jisx0213_pair_encmap, JISX0213_ENCPAIRS); |
|---|
| 824 | n/a | if (coded == DBCINV) |
|---|
| 825 | n/a | return MAP_UNMAPPABLE; |
|---|
| 826 | n/a | else |
|---|
| 827 | n/a | return coded; |
|---|
| 828 | n/a | break; |
|---|
| 829 | n/a | |
|---|
| 830 | n/a | default: |
|---|
| 831 | n/a | return MAP_UNMAPPABLE; |
|---|
| 832 | n/a | } |
|---|
| 833 | n/a | } |
|---|
| 834 | n/a | |
|---|
| 835 | n/a | static DBCHAR |
|---|
| 836 | n/a | jisx0213_2000_1_encoder(const Py_UCS4 *data, Py_ssize_t *length) |
|---|
| 837 | n/a | { |
|---|
| 838 | n/a | DBCHAR coded = jisx0213_encoder(data, length, (void *)2000); |
|---|
| 839 | n/a | if (coded == MAP_UNMAPPABLE || coded == MAP_MULTIPLE_AVAIL) |
|---|
| 840 | n/a | return coded; |
|---|
| 841 | n/a | else if (coded & 0x8000) |
|---|
| 842 | n/a | return MAP_UNMAPPABLE; |
|---|
| 843 | n/a | else |
|---|
| 844 | n/a | return coded; |
|---|
| 845 | n/a | } |
|---|
| 846 | n/a | |
|---|
| 847 | n/a | static DBCHAR |
|---|
| 848 | n/a | jisx0213_2000_1_encoder_paironly(const Py_UCS4 *data, Py_ssize_t *length) |
|---|
| 849 | n/a | { |
|---|
| 850 | n/a | DBCHAR coded; |
|---|
| 851 | n/a | Py_ssize_t ilength = *length; |
|---|
| 852 | n/a | |
|---|
| 853 | n/a | coded = jisx0213_encoder(data, length, (void *)2000); |
|---|
| 854 | n/a | switch (ilength) { |
|---|
| 855 | n/a | case 1: |
|---|
| 856 | n/a | if (coded == MAP_MULTIPLE_AVAIL) |
|---|
| 857 | n/a | return MAP_MULTIPLE_AVAIL; |
|---|
| 858 | n/a | else |
|---|
| 859 | n/a | return MAP_UNMAPPABLE; |
|---|
| 860 | n/a | case 2: |
|---|
| 861 | n/a | if (*length != 2) |
|---|
| 862 | n/a | return MAP_UNMAPPABLE; |
|---|
| 863 | n/a | else |
|---|
| 864 | n/a | return coded; |
|---|
| 865 | n/a | default: |
|---|
| 866 | n/a | return MAP_UNMAPPABLE; |
|---|
| 867 | n/a | } |
|---|
| 868 | n/a | } |
|---|
| 869 | n/a | |
|---|
| 870 | n/a | static DBCHAR |
|---|
| 871 | n/a | jisx0213_2000_2_encoder(const Py_UCS4 *data, Py_ssize_t *length) |
|---|
| 872 | n/a | { |
|---|
| 873 | n/a | DBCHAR coded = jisx0213_encoder(data, length, (void *)2000); |
|---|
| 874 | n/a | if (coded == MAP_UNMAPPABLE || coded == MAP_MULTIPLE_AVAIL) |
|---|
| 875 | n/a | return coded; |
|---|
| 876 | n/a | else if (coded & 0x8000) |
|---|
| 877 | n/a | return coded & 0x7fff; |
|---|
| 878 | n/a | else |
|---|
| 879 | n/a | return MAP_UNMAPPABLE; |
|---|
| 880 | n/a | } |
|---|
| 881 | n/a | |
|---|
| 882 | n/a | static DBCHAR |
|---|
| 883 | n/a | jisx0213_2004_1_encoder(const Py_UCS4 *data, Py_ssize_t *length) |
|---|
| 884 | n/a | { |
|---|
| 885 | n/a | DBCHAR coded = jisx0213_encoder(data, length, NULL); |
|---|
| 886 | n/a | if (coded == MAP_UNMAPPABLE || coded == MAP_MULTIPLE_AVAIL) |
|---|
| 887 | n/a | return coded; |
|---|
| 888 | n/a | else if (coded & 0x8000) |
|---|
| 889 | n/a | return MAP_UNMAPPABLE; |
|---|
| 890 | n/a | else |
|---|
| 891 | n/a | return coded; |
|---|
| 892 | n/a | } |
|---|
| 893 | n/a | |
|---|
| 894 | n/a | static DBCHAR |
|---|
| 895 | n/a | jisx0213_2004_1_encoder_paironly(const Py_UCS4 *data, Py_ssize_t *length) |
|---|
| 896 | n/a | { |
|---|
| 897 | n/a | DBCHAR coded; |
|---|
| 898 | n/a | Py_ssize_t ilength = *length; |
|---|
| 899 | n/a | |
|---|
| 900 | n/a | coded = jisx0213_encoder(data, length, NULL); |
|---|
| 901 | n/a | switch (ilength) { |
|---|
| 902 | n/a | case 1: |
|---|
| 903 | n/a | if (coded == MAP_MULTIPLE_AVAIL) |
|---|
| 904 | n/a | return MAP_MULTIPLE_AVAIL; |
|---|
| 905 | n/a | else |
|---|
| 906 | n/a | return MAP_UNMAPPABLE; |
|---|
| 907 | n/a | case 2: |
|---|
| 908 | n/a | if (*length != 2) |
|---|
| 909 | n/a | return MAP_UNMAPPABLE; |
|---|
| 910 | n/a | else |
|---|
| 911 | n/a | return coded; |
|---|
| 912 | n/a | default: |
|---|
| 913 | n/a | return MAP_UNMAPPABLE; |
|---|
| 914 | n/a | } |
|---|
| 915 | n/a | } |
|---|
| 916 | n/a | |
|---|
| 917 | n/a | static DBCHAR |
|---|
| 918 | n/a | jisx0213_2004_2_encoder(const Py_UCS4 *data, Py_ssize_t *length) |
|---|
| 919 | n/a | { |
|---|
| 920 | n/a | DBCHAR coded = jisx0213_encoder(data, length, NULL); |
|---|
| 921 | n/a | if (coded == MAP_UNMAPPABLE || coded == MAP_MULTIPLE_AVAIL) |
|---|
| 922 | n/a | return coded; |
|---|
| 923 | n/a | else if (coded & 0x8000) |
|---|
| 924 | n/a | return coded & 0x7fff; |
|---|
| 925 | n/a | else |
|---|
| 926 | n/a | return MAP_UNMAPPABLE; |
|---|
| 927 | n/a | } |
|---|
| 928 | n/a | |
|---|
| 929 | n/a | static Py_UCS4 |
|---|
| 930 | n/a | jisx0201_r_decoder(const unsigned char *data) |
|---|
| 931 | n/a | { |
|---|
| 932 | n/a | Py_UCS4 u; |
|---|
| 933 | n/a | JISX0201_R_DECODE_CHAR(*data, u) |
|---|
| 934 | n/a | else |
|---|
| 935 | n/a | return MAP_UNMAPPABLE; |
|---|
| 936 | n/a | return u; |
|---|
| 937 | n/a | } |
|---|
| 938 | n/a | |
|---|
| 939 | n/a | static DBCHAR |
|---|
| 940 | n/a | jisx0201_r_encoder(const Py_UCS4 *data, Py_ssize_t *length) |
|---|
| 941 | n/a | { |
|---|
| 942 | n/a | DBCHAR coded; |
|---|
| 943 | n/a | JISX0201_R_ENCODE(*data, coded) |
|---|
| 944 | n/a | else |
|---|
| 945 | n/a | return MAP_UNMAPPABLE; |
|---|
| 946 | n/a | return coded; |
|---|
| 947 | n/a | } |
|---|
| 948 | n/a | |
|---|
| 949 | n/a | static Py_UCS4 |
|---|
| 950 | n/a | jisx0201_k_decoder(const unsigned char *data) |
|---|
| 951 | n/a | { |
|---|
| 952 | n/a | Py_UCS4 u; |
|---|
| 953 | n/a | JISX0201_K_DECODE_CHAR(*data ^ 0x80, u) |
|---|
| 954 | n/a | else |
|---|
| 955 | n/a | return MAP_UNMAPPABLE; |
|---|
| 956 | n/a | return u; |
|---|
| 957 | n/a | } |
|---|
| 958 | n/a | |
|---|
| 959 | n/a | static DBCHAR |
|---|
| 960 | n/a | jisx0201_k_encoder(const Py_UCS4 *data, Py_ssize_t *length) |
|---|
| 961 | n/a | { |
|---|
| 962 | n/a | DBCHAR coded; |
|---|
| 963 | n/a | JISX0201_K_ENCODE(*data, coded) |
|---|
| 964 | n/a | else |
|---|
| 965 | n/a | return MAP_UNMAPPABLE; |
|---|
| 966 | n/a | return coded - 0x80; |
|---|
| 967 | n/a | } |
|---|
| 968 | n/a | |
|---|
| 969 | n/a | static int |
|---|
| 970 | n/a | gb2312_init(void) |
|---|
| 971 | n/a | { |
|---|
| 972 | n/a | static int initialized = 0; |
|---|
| 973 | n/a | |
|---|
| 974 | n/a | if (!initialized && ( |
|---|
| 975 | n/a | IMPORT_MAP(cn, gbcommon, &gbcommon_encmap, NULL) || |
|---|
| 976 | n/a | IMPORT_MAP(cn, gb2312, NULL, &gb2312_decmap))) |
|---|
| 977 | n/a | return -1; |
|---|
| 978 | n/a | initialized = 1; |
|---|
| 979 | n/a | return 0; |
|---|
| 980 | n/a | } |
|---|
| 981 | n/a | |
|---|
| 982 | n/a | static Py_UCS4 |
|---|
| 983 | n/a | gb2312_decoder(const unsigned char *data) |
|---|
| 984 | n/a | { |
|---|
| 985 | n/a | Py_UCS4 u; |
|---|
| 986 | n/a | if (TRYMAP_DEC(gb2312, u, data[0], data[1])) |
|---|
| 987 | n/a | return u; |
|---|
| 988 | n/a | else |
|---|
| 989 | n/a | return MAP_UNMAPPABLE; |
|---|
| 990 | n/a | } |
|---|
| 991 | n/a | |
|---|
| 992 | n/a | static DBCHAR |
|---|
| 993 | n/a | gb2312_encoder(const Py_UCS4 *data, Py_ssize_t *length) |
|---|
| 994 | n/a | { |
|---|
| 995 | n/a | DBCHAR coded; |
|---|
| 996 | n/a | assert(*length == 1); |
|---|
| 997 | n/a | if (*data < 0x10000) { |
|---|
| 998 | n/a | if (TRYMAP_ENC(gbcommon, coded, *data)) { |
|---|
| 999 | n/a | if (!(coded & 0x8000)) |
|---|
| 1000 | n/a | return coded; |
|---|
| 1001 | n/a | } |
|---|
| 1002 | n/a | } |
|---|
| 1003 | n/a | return MAP_UNMAPPABLE; |
|---|
| 1004 | n/a | } |
|---|
| 1005 | n/a | |
|---|
| 1006 | n/a | |
|---|
| 1007 | n/a | static Py_UCS4 |
|---|
| 1008 | n/a | dummy_decoder(const unsigned char *data) |
|---|
| 1009 | n/a | { |
|---|
| 1010 | n/a | return MAP_UNMAPPABLE; |
|---|
| 1011 | n/a | } |
|---|
| 1012 | n/a | |
|---|
| 1013 | n/a | static DBCHAR |
|---|
| 1014 | n/a | dummy_encoder(const Py_UCS4 *data, Py_ssize_t *length) |
|---|
| 1015 | n/a | { |
|---|
| 1016 | n/a | return MAP_UNMAPPABLE; |
|---|
| 1017 | n/a | } |
|---|
| 1018 | n/a | |
|---|
| 1019 | n/a | /*-*- registry tables -*-*/ |
|---|
| 1020 | n/a | |
|---|
| 1021 | n/a | #define REGISTRY_KSX1001_G0 { CHARSET_KSX1001, 0, 2, \ |
|---|
| 1022 | n/a | ksx1001_init, \ |
|---|
| 1023 | n/a | ksx1001_decoder, ksx1001_encoder } |
|---|
| 1024 | n/a | #define REGISTRY_KSX1001_G1 { CHARSET_KSX1001, 1, 2, \ |
|---|
| 1025 | n/a | ksx1001_init, \ |
|---|
| 1026 | n/a | ksx1001_decoder, ksx1001_encoder } |
|---|
| 1027 | n/a | #define REGISTRY_JISX0201_R { CHARSET_JISX0201_R, 0, 1, \ |
|---|
| 1028 | n/a | NULL, \ |
|---|
| 1029 | n/a | jisx0201_r_decoder, jisx0201_r_encoder } |
|---|
| 1030 | n/a | #define REGISTRY_JISX0201_K { CHARSET_JISX0201_K, 0, 1, \ |
|---|
| 1031 | n/a | NULL, \ |
|---|
| 1032 | n/a | jisx0201_k_decoder, jisx0201_k_encoder } |
|---|
| 1033 | n/a | #define REGISTRY_JISX0208 { CHARSET_JISX0208, 0, 2, \ |
|---|
| 1034 | n/a | jisx0208_init, \ |
|---|
| 1035 | n/a | jisx0208_decoder, jisx0208_encoder } |
|---|
| 1036 | n/a | #define REGISTRY_JISX0208_O { CHARSET_JISX0208_O, 0, 2, \ |
|---|
| 1037 | n/a | jisx0208_init, \ |
|---|
| 1038 | n/a | jisx0208_decoder, jisx0208_encoder } |
|---|
| 1039 | n/a | #define REGISTRY_JISX0212 { CHARSET_JISX0212, 0, 2, \ |
|---|
| 1040 | n/a | jisx0212_init, \ |
|---|
| 1041 | n/a | jisx0212_decoder, jisx0212_encoder } |
|---|
| 1042 | n/a | #define REGISTRY_JISX0213_2000_1 { CHARSET_JISX0213_2000_1, 0, 2, \ |
|---|
| 1043 | n/a | jisx0213_init, \ |
|---|
| 1044 | n/a | jisx0213_2000_1_decoder, \ |
|---|
| 1045 | n/a | jisx0213_2000_1_encoder } |
|---|
| 1046 | n/a | #define REGISTRY_JISX0213_2000_1_PAIRONLY { CHARSET_JISX0213_2000_1, 0, 2, \ |
|---|
| 1047 | n/a | jisx0213_init, \ |
|---|
| 1048 | n/a | jisx0213_2000_1_decoder, \ |
|---|
| 1049 | n/a | jisx0213_2000_1_encoder_paironly } |
|---|
| 1050 | n/a | #define REGISTRY_JISX0213_2000_2 { CHARSET_JISX0213_2, 0, 2, \ |
|---|
| 1051 | n/a | jisx0213_init, \ |
|---|
| 1052 | n/a | jisx0213_2000_2_decoder, \ |
|---|
| 1053 | n/a | jisx0213_2000_2_encoder } |
|---|
| 1054 | n/a | #define REGISTRY_JISX0213_2004_1 { CHARSET_JISX0213_2004_1, 0, 2, \ |
|---|
| 1055 | n/a | jisx0213_init, \ |
|---|
| 1056 | n/a | jisx0213_2004_1_decoder, \ |
|---|
| 1057 | n/a | jisx0213_2004_1_encoder } |
|---|
| 1058 | n/a | #define REGISTRY_JISX0213_2004_1_PAIRONLY { CHARSET_JISX0213_2004_1, 0, 2, \ |
|---|
| 1059 | n/a | jisx0213_init, \ |
|---|
| 1060 | n/a | jisx0213_2004_1_decoder, \ |
|---|
| 1061 | n/a | jisx0213_2004_1_encoder_paironly } |
|---|
| 1062 | n/a | #define REGISTRY_JISX0213_2004_2 { CHARSET_JISX0213_2, 0, 2, \ |
|---|
| 1063 | n/a | jisx0213_init, \ |
|---|
| 1064 | n/a | jisx0213_2004_2_decoder, \ |
|---|
| 1065 | n/a | jisx0213_2004_2_encoder } |
|---|
| 1066 | n/a | #define REGISTRY_GB2312 { CHARSET_GB2312, 0, 2, \ |
|---|
| 1067 | n/a | gb2312_init, \ |
|---|
| 1068 | n/a | gb2312_decoder, gb2312_encoder } |
|---|
| 1069 | n/a | #define REGISTRY_CNS11643_1 { CHARSET_CNS11643_1, 1, 2, \ |
|---|
| 1070 | n/a | cns11643_init, \ |
|---|
| 1071 | n/a | cns11643_1_decoder, cns11643_1_encoder } |
|---|
| 1072 | n/a | #define REGISTRY_CNS11643_2 { CHARSET_CNS11643_2, 2, 2, \ |
|---|
| 1073 | n/a | cns11643_init, \ |
|---|
| 1074 | n/a | cns11643_2_decoder, cns11643_2_encoder } |
|---|
| 1075 | n/a | #define REGISTRY_ISO8859_1 { CHARSET_ISO8859_1, 2, 1, \ |
|---|
| 1076 | n/a | NULL, dummy_decoder, dummy_encoder } |
|---|
| 1077 | n/a | #define REGISTRY_ISO8859_7 { CHARSET_ISO8859_7, 2, 1, \ |
|---|
| 1078 | n/a | NULL, dummy_decoder, dummy_encoder } |
|---|
| 1079 | n/a | #define REGISTRY_SENTINEL { 0, } |
|---|
| 1080 | n/a | #define CONFIGDEF(var, attrs) \ |
|---|
| 1081 | n/a | static const struct iso2022_config iso2022_##var##_config = { \ |
|---|
| 1082 | n/a | attrs, iso2022_##var##_designations \ |
|---|
| 1083 | n/a | }; |
|---|
| 1084 | n/a | |
|---|
| 1085 | n/a | static const struct iso2022_designation iso2022_kr_designations[] = { |
|---|
| 1086 | n/a | REGISTRY_KSX1001_G1, REGISTRY_SENTINEL |
|---|
| 1087 | n/a | }; |
|---|
| 1088 | n/a | CONFIGDEF(kr, 0) |
|---|
| 1089 | n/a | |
|---|
| 1090 | n/a | static const struct iso2022_designation iso2022_jp_designations[] = { |
|---|
| 1091 | n/a | REGISTRY_JISX0208, REGISTRY_JISX0201_R, REGISTRY_JISX0208_O, |
|---|
| 1092 | n/a | REGISTRY_SENTINEL |
|---|
| 1093 | n/a | }; |
|---|
| 1094 | n/a | CONFIGDEF(jp, NO_SHIFT | USE_JISX0208_EXT) |
|---|
| 1095 | n/a | |
|---|
| 1096 | n/a | static const struct iso2022_designation iso2022_jp_1_designations[] = { |
|---|
| 1097 | n/a | REGISTRY_JISX0208, REGISTRY_JISX0212, REGISTRY_JISX0201_R, |
|---|
| 1098 | n/a | REGISTRY_JISX0208_O, REGISTRY_SENTINEL |
|---|
| 1099 | n/a | }; |
|---|
| 1100 | n/a | CONFIGDEF(jp_1, NO_SHIFT | USE_JISX0208_EXT) |
|---|
| 1101 | n/a | |
|---|
| 1102 | n/a | static const struct iso2022_designation iso2022_jp_2_designations[] = { |
|---|
| 1103 | n/a | REGISTRY_JISX0208, REGISTRY_JISX0212, REGISTRY_KSX1001_G0, |
|---|
| 1104 | n/a | REGISTRY_GB2312, REGISTRY_JISX0201_R, REGISTRY_JISX0208_O, |
|---|
| 1105 | n/a | REGISTRY_ISO8859_1, REGISTRY_ISO8859_7, REGISTRY_SENTINEL |
|---|
| 1106 | n/a | }; |
|---|
| 1107 | n/a | CONFIGDEF(jp_2, NO_SHIFT | USE_G2 | USE_JISX0208_EXT) |
|---|
| 1108 | n/a | |
|---|
| 1109 | n/a | static const struct iso2022_designation iso2022_jp_2004_designations[] = { |
|---|
| 1110 | n/a | REGISTRY_JISX0213_2004_1_PAIRONLY, REGISTRY_JISX0208, |
|---|
| 1111 | n/a | REGISTRY_JISX0213_2004_1, REGISTRY_JISX0213_2004_2, REGISTRY_SENTINEL |
|---|
| 1112 | n/a | }; |
|---|
| 1113 | n/a | CONFIGDEF(jp_2004, NO_SHIFT | USE_JISX0208_EXT) |
|---|
| 1114 | n/a | |
|---|
| 1115 | n/a | static const struct iso2022_designation iso2022_jp_3_designations[] = { |
|---|
| 1116 | n/a | REGISTRY_JISX0213_2000_1_PAIRONLY, REGISTRY_JISX0208, |
|---|
| 1117 | n/a | REGISTRY_JISX0213_2000_1, REGISTRY_JISX0213_2000_2, REGISTRY_SENTINEL |
|---|
| 1118 | n/a | }; |
|---|
| 1119 | n/a | CONFIGDEF(jp_3, NO_SHIFT | USE_JISX0208_EXT) |
|---|
| 1120 | n/a | |
|---|
| 1121 | n/a | static const struct iso2022_designation iso2022_jp_ext_designations[] = { |
|---|
| 1122 | n/a | REGISTRY_JISX0208, REGISTRY_JISX0212, REGISTRY_JISX0201_R, |
|---|
| 1123 | n/a | REGISTRY_JISX0201_K, REGISTRY_JISX0208_O, REGISTRY_SENTINEL |
|---|
| 1124 | n/a | }; |
|---|
| 1125 | n/a | CONFIGDEF(jp_ext, NO_SHIFT | USE_JISX0208_EXT) |
|---|
| 1126 | n/a | |
|---|
| 1127 | n/a | |
|---|
| 1128 | n/a | BEGIN_MAPPINGS_LIST |
|---|
| 1129 | n/a | /* no mapping table here */ |
|---|
| 1130 | n/a | END_MAPPINGS_LIST |
|---|
| 1131 | n/a | |
|---|
| 1132 | n/a | #define ISO2022_CODEC(variation) { \ |
|---|
| 1133 | n/a | "iso2022_" #variation, \ |
|---|
| 1134 | n/a | &iso2022_##variation##_config, \ |
|---|
| 1135 | n/a | iso2022_codec_init, \ |
|---|
| 1136 | n/a | _STATEFUL_METHODS(iso2022) \ |
|---|
| 1137 | n/a | }, |
|---|
| 1138 | n/a | |
|---|
| 1139 | n/a | BEGIN_CODECS_LIST |
|---|
| 1140 | n/a | ISO2022_CODEC(kr) |
|---|
| 1141 | n/a | ISO2022_CODEC(jp) |
|---|
| 1142 | n/a | ISO2022_CODEC(jp_1) |
|---|
| 1143 | n/a | ISO2022_CODEC(jp_2) |
|---|
| 1144 | n/a | ISO2022_CODEC(jp_2004) |
|---|
| 1145 | n/a | ISO2022_CODEC(jp_3) |
|---|
| 1146 | n/a | ISO2022_CODEC(jp_ext) |
|---|
| 1147 | n/a | END_CODECS_LIST |
|---|
| 1148 | n/a | |
|---|
| 1149 | n/a | I_AM_A_MODULE_FOR(iso2022) |
|---|