| 1 | /* |
| 2 | SDL - Simple DirectMedia Layer |
| 3 | Copyright (C) 1997-2009 Sam Lantinga |
| 4 | |
| 5 | This library is free software; you can redistribute it and/or |
| 6 | modify it under the terms of the GNU Lesser General Public |
| 7 | License as published by the Free Software Foundation; either |
| 8 | version 2.1 of the License, or (at your option) any later version. |
| 9 | |
| 10 | This library is distributed in the hope that it will be useful, |
| 11 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 12 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 13 | Lesser General Public License for more details. |
| 14 | |
| 15 | You should have received a copy of the GNU Lesser General Public |
| 16 | License along with this library; if not, write to the Free Software |
| 17 | Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA |
| 18 | |
| 19 | Sam Lantinga |
| 20 | slouken@libsdl.org |
| 21 | */ |
| 22 | #include "SDL_config.h" |
| 23 | |
| 24 | /* This file contains portable iconv functions for SDL */ |
| 25 | |
| 26 | #include "SDL_stdinc.h" |
| 27 | #include "SDL_endian.h" |
| 28 | |
| 29 | #ifdef HAVE_ICONV |
| 30 | |
| 31 | /* Depending on which standard the iconv() was implemented with, |
| 32 | iconv() may or may not use const char ** for the inbuf param. |
| 33 | If we get this wrong, it's just a warning, so no big deal. |
| 34 | */ |
| 35 | #if defined(_XGP6) || \ |
| 36 | defined(__GLIBC__) && ((__GLIBC__ > 2) || (__GLIBC__ == 2 && __GLIBC_MINOR__ >= 2)) |
| 37 | #define ICONV_INBUF_NONCONST |
| 38 | #endif |
| 39 | |
| 40 | #include <errno.h> |
| 41 | |
| 42 | size_t SDL_iconv(SDL_iconv_t cd, |
| 43 | const char **inbuf, size_t *inbytesleft, |
| 44 | char **outbuf, size_t *outbytesleft) |
| 45 | { |
| 46 | size_t retCode; |
| 47 | #ifdef ICONV_INBUF_NONCONST |
| 48 | retCode = iconv(cd, (char **)inbuf, inbytesleft, outbuf, outbytesleft); |
| 49 | #else |
| 50 | retCode = iconv(cd, inbuf, inbytesleft, outbuf, outbytesleft); |
| 51 | #endif |
| 52 | if ( retCode == (size_t)-1 ) { |
| 53 | switch(errno) { |
| 54 | case E2BIG: |
| 55 | return SDL_ICONV_E2BIG; |
| 56 | case EILSEQ: |
| 57 | return SDL_ICONV_EILSEQ; |
| 58 | case EINVAL: |
| 59 | return SDL_ICONV_EINVAL; |
| 60 | default: |
| 61 | return SDL_ICONV_ERROR; |
| 62 | } |
| 63 | } |
| 64 | return retCode; |
| 65 | } |
| 66 | |
| 67 | #else |
| 68 | |
| 69 | /* Lots of useful information on Unicode at: |
| 70 | http://www.cl.cam.ac.uk/~mgk25/unicode.html |
| 71 | */ |
| 72 | |
| 73 | #define UNICODE_BOM 0xFEFF |
| 74 | |
| 75 | #define UNKNOWN_ASCII '?' |
| 76 | #define UNKNOWN_UNICODE 0xFFFD |
| 77 | |
| 78 | enum { |
| 79 | ENCODING_UNKNOWN, |
| 80 | ENCODING_ASCII, |
| 81 | ENCODING_LATIN1, |
| 82 | ENCODING_UTF8, |
| 83 | ENCODING_UTF16, /* Needs byte order marker */ |
| 84 | ENCODING_UTF16BE, |
| 85 | ENCODING_UTF16LE, |
| 86 | ENCODING_UTF32, /* Needs byte order marker */ |
| 87 | ENCODING_UTF32BE, |
| 88 | ENCODING_UTF32LE, |
| 89 | ENCODING_UCS2, /* Native byte order assumed */ |
| 90 | ENCODING_UCS4, /* Native byte order assumed */ |
| 91 | }; |
| 92 | #if SDL_BYTEORDER == SDL_BIG_ENDIAN |
| 93 | #define ENCODING_UTF16NATIVE ENCODING_UTF16BE |
| 94 | #define ENCODING_UTF32NATIVE ENCODING_UTF32BE |
| 95 | #else |
| 96 | #define ENCODING_UTF16NATIVE ENCODING_UTF16LE |
| 97 | #define ENCODING_UTF32NATIVE ENCODING_UTF32LE |
| 98 | #endif |
| 99 | |
| 100 | struct _SDL_iconv_t |
| 101 | { |
| 102 | int src_fmt; |
| 103 | int dst_fmt; |
| 104 | }; |
| 105 | |
| 106 | static struct { |
| 107 | const char *name; |
| 108 | int format; |
| 109 | } encodings[] = { |
| 110 | { "ASCII", ENCODING_ASCII }, |
| 111 | { "US-ASCII", ENCODING_ASCII }, |
| 112 | { "8859-1", ENCODING_LATIN1 }, |
| 113 | { "ISO-8859-1", ENCODING_LATIN1 }, |
| 114 | { "UTF8", ENCODING_UTF8 }, |
| 115 | { "UTF-8", ENCODING_UTF8 }, |
| 116 | { "UTF16", ENCODING_UTF16 }, |
| 117 | { "UTF-16", ENCODING_UTF16 }, |
| 118 | { "UTF16BE", ENCODING_UTF16BE }, |
| 119 | { "UTF-16BE", ENCODING_UTF16BE }, |
| 120 | { "UTF16LE", ENCODING_UTF16LE }, |
| 121 | { "UTF-16LE", ENCODING_UTF16LE }, |
| 122 | { "UTF32", ENCODING_UTF32 }, |
| 123 | { "UTF-32", ENCODING_UTF32 }, |
| 124 | { "UTF32BE", ENCODING_UTF32BE }, |
| 125 | { "UTF-32BE", ENCODING_UTF32BE }, |
| 126 | { "UTF32LE", ENCODING_UTF32LE }, |
| 127 | { "UTF-32LE", ENCODING_UTF32LE }, |
| 128 | { "UCS2", ENCODING_UCS2 }, |
| 129 | { "UCS-2", ENCODING_UCS2 }, |
| 130 | { "UCS4", ENCODING_UCS4 }, |
| 131 | { "UCS-4", ENCODING_UCS4 }, |
| 132 | }; |
| 133 | |
| 134 | static const char *getlocale(char *buffer, size_t bufsize) |
| 135 | { |
| 136 | const char *lang; |
| 137 | char *ptr; |
| 138 | |
| 139 | lang = SDL_getenv("LC_ALL"); |
| 140 | if ( !lang ) { |
| 141 | lang = SDL_getenv("LC_CTYPE"); |
| 142 | } |
| 143 | if ( !lang ) { |
| 144 | lang = SDL_getenv("LC_MESSAGES"); |
| 145 | } |
| 146 | if ( !lang ) { |
| 147 | lang = SDL_getenv("LANG"); |
| 148 | } |
| 149 | if ( !lang || !*lang || SDL_strcmp(lang, "C") == 0 ) { |
| 150 | lang = "ASCII"; |
| 151 | } |
| 152 | |
| 153 | /* We need to trim down strings like "en_US.UTF-8@blah" to "UTF-8" */ |
| 154 | ptr = SDL_strchr(lang, '.'); |
| 155 | if (ptr != NULL) { |
| 156 | lang = ptr + 1; |
| 157 | } |
| 158 | |
| 159 | SDL_strlcpy(buffer, lang, bufsize); |
| 160 | ptr = SDL_strchr(buffer, '@'); |
| 161 | if (ptr != NULL) { |
| 162 | *ptr = '\0'; /* chop end of string. */ |
| 163 | } |
| 164 | |
| 165 | return buffer; |
| 166 | } |
| 167 | |
| 168 | SDL_iconv_t SDL_iconv_open(const char *tocode, const char *fromcode) |
| 169 | { |
| 170 | int src_fmt = ENCODING_UNKNOWN; |
| 171 | int dst_fmt = ENCODING_UNKNOWN; |
| 172 | int i; |
| 173 | char fromcode_buffer[64]; |
| 174 | char tocode_buffer[64]; |
| 175 | |
| 176 | if ( !fromcode || !*fromcode ) { |
| 177 | fromcode = getlocale(fromcode_buffer, sizeof(fromcode_buffer)); |
| 178 | } |
| 179 | if ( !tocode || !*tocode ) { |
| 180 | tocode = getlocale(tocode_buffer, sizeof(tocode_buffer)); |
| 181 | } |
| 182 | for ( i = 0; i < SDL_arraysize(encodings); ++i ) { |
| 183 | if ( SDL_strcasecmp(fromcode, encodings[i].name) == 0 ) { |
| 184 | src_fmt = encodings[i].format; |
| 185 | if ( dst_fmt != ENCODING_UNKNOWN ) { |
| 186 | break; |
| 187 | } |
| 188 | } |
| 189 | if ( SDL_strcasecmp(tocode, encodings[i].name) == 0 ) { |
| 190 | dst_fmt = encodings[i].format; |
| 191 | if ( src_fmt != ENCODING_UNKNOWN ) { |
| 192 | break; |
| 193 | } |
| 194 | } |
| 195 | } |
| 196 | if ( src_fmt != ENCODING_UNKNOWN && dst_fmt != ENCODING_UNKNOWN ) { |
| 197 | SDL_iconv_t cd = (SDL_iconv_t)SDL_malloc(sizeof(*cd)); |
| 198 | if ( cd ) { |
| 199 | cd->src_fmt = src_fmt; |
| 200 | cd->dst_fmt = dst_fmt; |
| 201 | return cd; |
| 202 | } |
| 203 | } |
| 204 | return (SDL_iconv_t)-1; |
| 205 | } |
| 206 | |
| 207 | size_t SDL_iconv(SDL_iconv_t cd, |
| 208 | const char **inbuf, size_t *inbytesleft, |
| 209 | char **outbuf, size_t *outbytesleft) |
| 210 | { |
| 211 | /* For simplicity, we'll convert everything to and from UCS-4 */ |
| 212 | const char *src; |
| 213 | char *dst; |
| 214 | size_t srclen, dstlen; |
| 215 | Uint32 ch = 0; |
| 216 | size_t total; |
| 217 | |
| 218 | if ( !inbuf || !*inbuf ) { |
| 219 | /* Reset the context */ |
| 220 | return 0; |
| 221 | } |
| 222 | if ( !outbuf || !*outbuf || !outbytesleft || !*outbytesleft ) { |
| 223 | return SDL_ICONV_E2BIG; |
| 224 | } |
| 225 | src = *inbuf; |
| 226 | srclen = (inbytesleft ? *inbytesleft : 0); |
| 227 | dst = *outbuf; |
| 228 | dstlen = *outbytesleft; |
| 229 | |
| 230 | switch ( cd->src_fmt ) { |
| 231 | case ENCODING_UTF16: |
| 232 | /* Scan for a byte order marker */ |
| 233 | { |
| 234 | Uint8 *p = (Uint8 *)src; |
| 235 | size_t n = srclen / 2; |
| 236 | while ( n ) { |
| 237 | if ( p[0] == 0xFF && p[1] == 0xFE ) { |
| 238 | cd->src_fmt = ENCODING_UTF16BE; |
| 239 | break; |
| 240 | } else if ( p[0] == 0xFE && p[1] == 0xFF ) { |
| 241 | cd->src_fmt = ENCODING_UTF16LE; |
| 242 | break; |
| 243 | } |
| 244 | p += 2; |
| 245 | --n; |
| 246 | } |
| 247 | if ( n == 0 ) { |
| 248 | /* We can't tell, default to host order */ |
| 249 | cd->src_fmt = ENCODING_UTF16NATIVE; |
| 250 | } |
| 251 | } |
| 252 | break; |
| 253 | case ENCODING_UTF32: |
| 254 | /* Scan for a byte order marker */ |
| 255 | { |
| 256 | Uint8 *p = (Uint8 *)src; |
| 257 | size_t n = srclen / 4; |
| 258 | while ( n ) { |
| 259 | if ( p[0] == 0xFF && p[1] == 0xFE && |
| 260 | p[2] == 0x00 && p[3] == 0x00 ) { |
| 261 | cd->src_fmt = ENCODING_UTF32BE; |
| 262 | break; |
| 263 | } else if ( p[0] == 0x00 && p[1] == 0x00 && |
| 264 | p[2] == 0xFE && p[3] == 0xFF ) { |
| 265 | cd->src_fmt = ENCODING_UTF32LE; |
| 266 | break; |
| 267 | } |
| 268 | p += 4; |
| 269 | --n; |
| 270 | } |
| 271 | if ( n == 0 ) { |
| 272 | /* We can't tell, default to host order */ |
| 273 | cd->src_fmt = ENCODING_UTF32NATIVE; |
| 274 | } |
| 275 | } |
| 276 | break; |
| 277 | } |
| 278 | |
| 279 | switch ( cd->dst_fmt ) { |
| 280 | case ENCODING_UTF16: |
| 281 | /* Default to host order, need to add byte order marker */ |
| 282 | if ( dstlen < 2 ) { |
| 283 | return SDL_ICONV_E2BIG; |
| 284 | } |
| 285 | *(Uint16 *)dst = UNICODE_BOM; |
| 286 | dst += 2; |
| 287 | dstlen -= 2; |
| 288 | cd->dst_fmt = ENCODING_UTF16NATIVE; |
| 289 | break; |
| 290 | case ENCODING_UTF32: |
| 291 | /* Default to host order, need to add byte order marker */ |
| 292 | if ( dstlen < 4 ) { |
| 293 | return SDL_ICONV_E2BIG; |
| 294 | } |
| 295 | *(Uint32 *)dst = UNICODE_BOM; |
| 296 | dst += 4; |
| 297 | dstlen -= 4; |
| 298 | cd->dst_fmt = ENCODING_UTF32NATIVE; |
| 299 | break; |
| 300 | } |
| 301 | |
| 302 | total = 0; |
| 303 | while ( srclen > 0 ) { |
| 304 | /* Decode a character */ |
| 305 | switch ( cd->src_fmt ) { |
| 306 | case ENCODING_ASCII: |
| 307 | { |
| 308 | Uint8 *p = (Uint8 *)src; |
| 309 | ch = (Uint32)(p[0] & 0x7F); |
| 310 | ++src; |
| 311 | --srclen; |
| 312 | } |
| 313 | break; |
| 314 | case ENCODING_LATIN1: |
| 315 | { |
| 316 | Uint8 *p = (Uint8 *)src; |
| 317 | ch = (Uint32)p[0]; |
| 318 | ++src; |
| 319 | --srclen; |
| 320 | } |
| 321 | break; |
| 322 | case ENCODING_UTF8: /* RFC 3629 */ |
| 323 | { |
| 324 | Uint8 *p = (Uint8 *)src; |
| 325 | size_t left = 0; |
| 326 | SDL_bool overlong = SDL_FALSE; |
| 327 | if ( p[0] >= 0xFC ) { |
| 328 | if ( (p[0] & 0xFE) != 0xFC ) { |
| 329 | /* Skip illegal sequences |
| 330 | return SDL_ICONV_EILSEQ; |
| 331 | */ |
| 332 | ch = UNKNOWN_UNICODE; |
| 333 | } else { |
| 334 | if ( p[0] == 0xFC ) { |
| 335 | overlong = SDL_TRUE; |
| 336 | } |
| 337 | ch = (Uint32)(p[0] & 0x01); |
| 338 | left = 5; |
| 339 | } |
| 340 | } else if ( p[0] >= 0xF8 ) { |
| 341 | if ( (p[0] & 0xFC) != 0xF8 ) { |
| 342 | /* Skip illegal sequences |
| 343 | return SDL_ICONV_EILSEQ; |
| 344 | */ |
| 345 | ch = UNKNOWN_UNICODE; |
| 346 | } else { |
| 347 | if ( p[0] == 0xF8 ) { |
| 348 | overlong = SDL_TRUE; |
| 349 | } |
| 350 | ch = (Uint32)(p[0] & 0x03); |
| 351 | left = 4; |
| 352 | } |
| 353 | } else if ( p[0] >= 0xF0 ) { |
| 354 | if ( (p[0] & 0xF8) != 0xF0 ) { |
| 355 | /* Skip illegal sequences |
| 356 | return SDL_ICONV_EILSEQ; |
| 357 | */ |
| 358 | ch = UNKNOWN_UNICODE; |
| 359 | } else { |
| 360 | if ( p[0] == 0xF0 ) { |
| 361 | overlong = SDL_TRUE; |
| 362 | } |
| 363 | ch = (Uint32)(p[0] & 0x07); |
| 364 | left = 3; |
| 365 | } |
| 366 | } else if ( p[0] >= 0xE0 ) { |
| 367 | if ( (p[0] & 0xF0) != 0xE0 ) { |
| 368 | /* Skip illegal sequences |
| 369 | return SDL_ICONV_EILSEQ; |
| 370 | */ |
| 371 | ch = UNKNOWN_UNICODE; |
| 372 | } else { |
| 373 | if ( p[0] == 0xE0 ) { |
| 374 | overlong = SDL_TRUE; |
| 375 | } |
| 376 | ch = (Uint32)(p[0] & 0x0F); |
| 377 | left = 2; |
| 378 | } |
| 379 | } else if ( p[0] >= 0xC0 ) { |
| 380 | if ( (p[0] & 0xE0) != 0xC0 ) { |
| 381 | /* Skip illegal sequences |
| 382 | return SDL_ICONV_EILSEQ; |
| 383 | */ |
| 384 | ch = UNKNOWN_UNICODE; |
| 385 | } else { |
| 386 | if ( (p[0] & 0xCE) == 0xC0 ) { |
| 387 | overlong = SDL_TRUE; |
| 388 | } |
| 389 | ch = (Uint32)(p[0] & 0x1F); |
| 390 | left = 1; |
| 391 | } |
| 392 | } else { |
| 393 | if ( (p[0] & 0x80) != 0x00 ) { |
| 394 | /* Skip illegal sequences |
| 395 | return SDL_ICONV_EILSEQ; |
| 396 | */ |
| 397 | ch = UNKNOWN_UNICODE; |
| 398 | } else { |
| 399 | ch = (Uint32)p[0]; |
| 400 | } |
| 401 | } |
| 402 | ++src; |
| 403 | --srclen; |
| 404 | if ( srclen < left ) { |
| 405 | return SDL_ICONV_EINVAL; |
| 406 | } |
| 407 | while ( left-- ) { |
| 408 | ++p; |
| 409 | if ( (p[0] & 0xC0) != 0x80 ) { |
| 410 | /* Skip illegal sequences |
| 411 | return SDL_ICONV_EILSEQ; |
| 412 | */ |
| 413 | ch = UNKNOWN_UNICODE; |
| 414 | break; |
| 415 | } |
| 416 | ch <<= 6; |
| 417 | ch |= (p[0] & 0x3F); |
| 418 | ++src; |
| 419 | --srclen; |
| 420 | } |
| 421 | if ( overlong ) { |
| 422 | /* Potential security risk |
| 423 | return SDL_ICONV_EILSEQ; |
| 424 | */ |
| 425 | ch = UNKNOWN_UNICODE; |
| 426 | } |
| 427 | if ( (ch >= 0xD800 && ch <= 0xDFFF) || |
| 428 | (ch == 0xFFFE || ch == 0xFFFF) || |
| 429 | ch > 0x10FFFF ) { |
| 430 | /* Skip illegal sequences |
| 431 | return SDL_ICONV_EILSEQ; |
| 432 | */ |
| 433 | ch = UNKNOWN_UNICODE; |
| 434 | } |
| 435 | } |
| 436 | break; |
| 437 | case ENCODING_UTF16BE: /* RFC 2781 */ |
| 438 | { |
| 439 | Uint8 *p = (Uint8 *)src; |
| 440 | Uint16 W1, W2; |
| 441 | if ( srclen < 2 ) { |
| 442 | return SDL_ICONV_EINVAL; |
| 443 | } |
| 444 | W1 = ((Uint16)p[0] << 8) | |
| 445 | (Uint16)p[1]; |
| 446 | src += 2; |
| 447 | srclen -= 2; |
| 448 | if ( W1 < 0xD800 || W1 > 0xDFFF ) { |
| 449 | ch = (Uint32)W1; |
| 450 | break; |
| 451 | } |
| 452 | if ( W1 > 0xDBFF ) { |
| 453 | /* Skip illegal sequences |
| 454 | return SDL_ICONV_EILSEQ; |
| 455 | */ |
| 456 | ch = UNKNOWN_UNICODE; |
| 457 | break; |
| 458 | } |
| 459 | if ( srclen < 2 ) { |
| 460 | return SDL_ICONV_EINVAL; |
| 461 | } |
| 462 | p = (Uint8 *)src; |
| 463 | W2 = ((Uint16)p[0] << 8) | |
| 464 | (Uint16)p[1]; |
| 465 | src += 2; |
| 466 | srclen -= 2; |
| 467 | if ( W2 < 0xDC00 || W2 > 0xDFFF ) { |
| 468 | /* Skip illegal sequences |
| 469 | return SDL_ICONV_EILSEQ; |
| 470 | */ |
| 471 | ch = UNKNOWN_UNICODE; |
| 472 | break; |
| 473 | } |
| 474 | ch = (((Uint32)(W1 & 0x3FF) << 10) | |
| 475 | (Uint32)(W2 & 0x3FF)) + 0x10000; |
| 476 | } |
| 477 | break; |
| 478 | case ENCODING_UTF16LE: /* RFC 2781 */ |
| 479 | { |
| 480 | Uint8 *p = (Uint8 *)src; |
| 481 | Uint16 W1, W2; |
| 482 | if ( srclen < 2 ) { |
| 483 | return SDL_ICONV_EINVAL; |
| 484 | } |
| 485 | W1 = ((Uint16)p[1] << 8) | |
| 486 | (Uint16)p[0]; |
| 487 | src += 2; |
| 488 | srclen -= 2; |
| 489 | if ( W1 < 0xD800 || W1 > 0xDFFF ) { |
| 490 | ch = (Uint32)W1; |
| 491 | break; |
| 492 | } |
| 493 | if ( W1 > 0xDBFF ) { |
| 494 | /* Skip illegal sequences |
| 495 | return SDL_ICONV_EILSEQ; |
| 496 | */ |
| 497 | ch = UNKNOWN_UNICODE; |
| 498 | break; |
| 499 | } |
| 500 | if ( srclen < 2 ) { |
| 501 | return SDL_ICONV_EINVAL; |
| 502 | } |
| 503 | p = (Uint8 *)src; |
| 504 | W2 = ((Uint16)p[1] << 8) | |
| 505 | (Uint16)p[0]; |
| 506 | src += 2; |
| 507 | srclen -= 2; |
| 508 | if ( W2 < 0xDC00 || W2 > 0xDFFF ) { |
| 509 | /* Skip illegal sequences |
| 510 | return SDL_ICONV_EILSEQ; |
| 511 | */ |
| 512 | ch = UNKNOWN_UNICODE; |
| 513 | break; |
| 514 | } |
| 515 | ch = (((Uint32)(W1 & 0x3FF) << 10) | |
| 516 | (Uint32)(W2 & 0x3FF)) + 0x10000; |
| 517 | } |
| 518 | break; |
| 519 | case ENCODING_UTF32BE: |
| 520 | { |
| 521 | Uint8 *p = (Uint8 *)src; |
| 522 | if ( srclen < 4 ) { |
| 523 | return SDL_ICONV_EINVAL; |
| 524 | } |
| 525 | ch = ((Uint32)p[0] << 24) | |
| 526 | ((Uint32)p[1] << 16) | |
| 527 | ((Uint32)p[2] << 8) | |
| 528 | (Uint32)p[3]; |
| 529 | src += 4; |
| 530 | srclen -= 4; |
| 531 | } |
| 532 | break; |
| 533 | case ENCODING_UTF32LE: |
| 534 | { |
| 535 | Uint8 *p = (Uint8 *)src; |
| 536 | if ( srclen < 4 ) { |
| 537 | return SDL_ICONV_EINVAL; |
| 538 | } |
| 539 | ch = ((Uint32)p[3] << 24) | |
| 540 | ((Uint32)p[2] << 16) | |
| 541 | ((Uint32)p[1] << 8) | |
| 542 | (Uint32)p[0]; |
| 543 | src += 4; |
| 544 | srclen -= 4; |
| 545 | } |
| 546 | break; |
| 547 | case ENCODING_UCS2: |
| 548 | { |
| 549 | Uint16 *p = (Uint16 *)src; |
| 550 | if ( srclen < 2 ) { |
| 551 | return SDL_ICONV_EINVAL; |
| 552 | } |
| 553 | ch = *p; |
| 554 | src += 2; |
| 555 | srclen -= 2; |
| 556 | } |
| 557 | break; |
| 558 | case ENCODING_UCS4: |
| 559 | { |
| 560 | Uint32 *p = (Uint32 *)src; |
| 561 | if ( srclen < 4 ) { |
| 562 | return SDL_ICONV_EINVAL; |
| 563 | } |
| 564 | ch = *p; |
| 565 | src += 4; |
| 566 | srclen -= 4; |
| 567 | } |
| 568 | break; |
| 569 | } |
| 570 | |
| 571 | /* Encode a character */ |
| 572 | switch ( cd->dst_fmt ) { |
| 573 | case ENCODING_ASCII: |
| 574 | { |
| 575 | Uint8 *p = (Uint8 *)dst; |
| 576 | if ( dstlen < 1 ) { |
| 577 | return SDL_ICONV_E2BIG; |
| 578 | } |
| 579 | if ( ch > 0x7F ) { |
| 580 | *p = UNKNOWN_ASCII; |
| 581 | } else { |
| 582 | *p = (Uint8)ch; |
| 583 | } |
| 584 | ++dst; |
| 585 | --dstlen; |
| 586 | } |
| 587 | break; |
| 588 | case ENCODING_LATIN1: |
| 589 | { |
| 590 | Uint8 *p = (Uint8 *)dst; |
| 591 | if ( dstlen < 1 ) { |
| 592 | return SDL_ICONV_E2BIG; |
| 593 | } |
| 594 | if ( ch > 0xFF ) { |
| 595 | *p = UNKNOWN_ASCII; |
| 596 | } else { |
| 597 | *p = (Uint8)ch; |
| 598 | } |
| 599 | ++dst; |
| 600 | --dstlen; |
| 601 | } |
| 602 | break; |
| 603 | case ENCODING_UTF8: /* RFC 3629 */ |
| 604 | { |
| 605 | Uint8 *p = (Uint8 *)dst; |
| 606 | if ( ch > 0x10FFFF ) { |
| 607 | ch = UNKNOWN_UNICODE; |
| 608 | } |
| 609 | if ( ch <= 0x7F ) { |
| 610 | if ( dstlen < 1 ) { |
| 611 | return SDL_ICONV_E2BIG; |
| 612 | } |
| 613 | *p = (Uint8)ch; |
| 614 | ++dst; |
| 615 | --dstlen; |
| 616 | } else if ( ch <= 0x7FF ) { |
| 617 | if ( dstlen < 2 ) { |
| 618 | return SDL_ICONV_E2BIG; |
| 619 | } |
| 620 | p[0] = 0xC0 | (Uint8)((ch >> 6) & 0x1F); |
| 621 | p[1] = 0x80 | (Uint8)(ch & 0x3F); |
| 622 | dst += 2; |
| 623 | dstlen -= 2; |
| 624 | } else if ( ch <= 0xFFFF ) { |
| 625 | if ( dstlen < 3 ) { |
| 626 | return SDL_ICONV_E2BIG; |
| 627 | } |
| 628 | p[0] = 0xE0 | (Uint8)((ch >> 12) & 0x0F); |
| 629 | p[1] = 0x80 | (Uint8)((ch >> 6) & 0x3F); |
| 630 | p[2] = 0x80 | (Uint8)(ch & 0x3F); |
| 631 | dst += 3; |
| 632 | dstlen -= 3; |
| 633 | } else if ( ch <= 0x1FFFFF ) { |
| 634 | if ( dstlen < 4 ) { |
| 635 | return SDL_ICONV_E2BIG; |
| 636 | } |
| 637 | p[0] = 0xF0 | (Uint8)((ch >> 18) & 0x07); |
| 638 | p[1] = 0x80 | (Uint8)((ch >> 12) & 0x3F); |
| 639 | p[2] = 0x80 | (Uint8)((ch >> 6) & 0x3F); |
| 640 | p[3] = 0x80 | (Uint8)(ch & 0x3F); |
| 641 | dst += 4; |
| 642 | dstlen -= 4; |
| 643 | } else if ( ch <= 0x3FFFFFF ) { |
| 644 | if ( dstlen < 5 ) { |
| 645 | return SDL_ICONV_E2BIG; |
| 646 | } |
| 647 | p[0] = 0xF8 | (Uint8)((ch >> 24) & 0x03); |
| 648 | p[1] = 0x80 | (Uint8)((ch >> 18) & 0x3F); |
| 649 | p[2] = 0x80 | (Uint8)((ch >> 12) & 0x3F); |
| 650 | p[3] = 0x80 | (Uint8)((ch >> 6) & 0x3F); |
| 651 | p[4] = 0x80 | (Uint8)(ch & 0x3F); |
| 652 | dst += 5; |
| 653 | dstlen -= 5; |
| 654 | } else { |
| 655 | if ( dstlen < 6 ) { |
| 656 | return SDL_ICONV_E2BIG; |
| 657 | } |
| 658 | p[0] = 0xFC | (Uint8)((ch >> 30) & 0x01); |
| 659 | p[1] = 0x80 | (Uint8)((ch >> 24) & 0x3F); |
| 660 | p[2] = 0x80 | (Uint8)((ch >> 18) & 0x3F); |
| 661 | p[3] = 0x80 | (Uint8)((ch >> 12) & 0x3F); |
| 662 | p[4] = 0x80 | (Uint8)((ch >> 6) & 0x3F); |
| 663 | p[5] = 0x80 | (Uint8)(ch & 0x3F); |
| 664 | dst += 6; |
| 665 | dstlen -= 6; |
| 666 | } |
| 667 | } |
| 668 | break; |
| 669 | case ENCODING_UTF16BE: /* RFC 2781 */ |
| 670 | { |
| 671 | Uint8 *p = (Uint8 *)dst; |
| 672 | if ( ch > 0x10FFFF ) { |
| 673 | ch = UNKNOWN_UNICODE; |
| 674 | } |
| 675 | if ( ch < 0x10000 ) { |
| 676 | if ( dstlen < 2 ) { |
| 677 | return SDL_ICONV_E2BIG; |
| 678 | } |
| 679 | p[0] = (Uint8)(ch >> 8); |
| 680 | p[1] = (Uint8)ch; |
| 681 | dst += 2; |
| 682 | dstlen -= 2; |
| 683 | } else { |
| 684 | Uint16 W1, W2; |
| 685 | if ( dstlen < 4 ) { |
| 686 | return SDL_ICONV_E2BIG; |
| 687 | } |
| 688 | ch = ch - 0x10000; |
| 689 | W1 = 0xD800 | (Uint16)((ch >> 10) & 0x3FF); |
| 690 | W2 = 0xDC00 | (Uint16)(ch & 0x3FF); |
| 691 | p[0] = (Uint8)(W1 >> 8); |
| 692 | p[1] = (Uint8)W1; |
| 693 | p[2] = (Uint8)(W2 >> 8); |
| 694 | p[3] = (Uint8)W2; |
| 695 | dst += 4; |
| 696 | dstlen -= 4; |
| 697 | } |
| 698 | } |
| 699 | break; |
| 700 | case ENCODING_UTF16LE: /* RFC 2781 */ |
| 701 | { |
| 702 | Uint8 *p = (Uint8 *)dst; |
| 703 | if ( ch > 0x10FFFF ) { |
| 704 | ch = UNKNOWN_UNICODE; |
| 705 | } |
| 706 | if ( ch < 0x10000 ) { |
| 707 | if ( dstlen < 2 ) { |
| 708 | return SDL_ICONV_E2BIG; |
| 709 | } |
| 710 | p[1] = (Uint8)(ch >> 8); |
| 711 | p[0] = (Uint8)ch; |
| 712 | dst += 2; |
| 713 | dstlen -= 2; |
| 714 | } else { |
| 715 | Uint16 W1, W2; |
| 716 | if ( dstlen < 4 ) { |
| 717 | return SDL_ICONV_E2BIG; |
| 718 | } |
| 719 | ch = ch - 0x10000; |
| 720 | W1 = 0xD800 | (Uint16)((ch >> 10) & 0x3FF); |
| 721 | W2 = 0xDC00 | (Uint16)(ch & 0x3FF); |
| 722 | p[1] = (Uint8)(W1 >> 8); |
| 723 | p[0] = (Uint8)W1; |
| 724 | p[3] = (Uint8)(W2 >> 8); |
| 725 | p[2] = (Uint8)W2; |
| 726 | dst += 4; |
| 727 | dstlen -= 4; |
| 728 | } |
| 729 | } |
| 730 | break; |
| 731 | case ENCODING_UTF32BE: |
| 732 | { |
| 733 | Uint8 *p = (Uint8 *)dst; |
| 734 | if ( ch > 0x10FFFF ) { |
| 735 | ch = UNKNOWN_UNICODE; |
| 736 | } |
| 737 | if ( dstlen < 4 ) { |
| 738 | return SDL_ICONV_E2BIG; |
| 739 | } |
| 740 | p[0] = (Uint8)(ch >> 24); |
| 741 | p[1] = (Uint8)(ch >> 16); |
| 742 | p[2] = (Uint8)(ch >> 8); |
| 743 | p[3] = (Uint8)ch; |
| 744 | dst += 4; |
| 745 | dstlen -= 4; |
| 746 | } |
| 747 | break; |
| 748 | case ENCODING_UTF32LE: |
| 749 | { |
| 750 | Uint8 *p = (Uint8 *)dst; |
| 751 | if ( ch > 0x10FFFF ) { |
| 752 | ch = UNKNOWN_UNICODE; |
| 753 | } |
| 754 | if ( dstlen < 4 ) { |
| 755 | return SDL_ICONV_E2BIG; |
| 756 | } |
| 757 | p[3] = (Uint8)(ch >> 24); |
| 758 | p[2] = (Uint8)(ch >> 16); |
| 759 | p[1] = (Uint8)(ch >> 8); |
| 760 | p[0] = (Uint8)ch; |
| 761 | dst += 4; |
| 762 | dstlen -= 4; |
| 763 | } |
| 764 | break; |
| 765 | case ENCODING_UCS2: |
| 766 | { |
| 767 | Uint16 *p = (Uint16 *)dst; |
| 768 | if ( ch > 0xFFFF ) { |
| 769 | ch = UNKNOWN_UNICODE; |
| 770 | } |
| 771 | if ( dstlen < 2 ) { |
| 772 | return SDL_ICONV_E2BIG; |
| 773 | } |
| 774 | *p = (Uint16)ch; |
| 775 | dst += 2; |
| 776 | dstlen -= 2; |
| 777 | } |
| 778 | break; |
| 779 | case ENCODING_UCS4: |
| 780 | { |
| 781 | Uint32 *p = (Uint32 *)dst; |
| 782 | if ( ch > 0x7FFFFFFF ) { |
| 783 | ch = UNKNOWN_UNICODE; |
| 784 | } |
| 785 | if ( dstlen < 4 ) { |
| 786 | return SDL_ICONV_E2BIG; |
| 787 | } |
| 788 | *p = ch; |
| 789 | dst += 4; |
| 790 | dstlen -= 4; |
| 791 | } |
| 792 | break; |
| 793 | } |
| 794 | |
| 795 | /* Update state */ |
| 796 | *inbuf = src; |
| 797 | *inbytesleft = srclen; |
| 798 | *outbuf = dst; |
| 799 | *outbytesleft = dstlen; |
| 800 | ++total; |
| 801 | } |
| 802 | return total; |
| 803 | } |
| 804 | |
| 805 | int SDL_iconv_close(SDL_iconv_t cd) |
| 806 | { |
| 807 | if ( cd && cd != (SDL_iconv_t)-1 ) { |
| 808 | SDL_free(cd); |
| 809 | } |
| 810 | return 0; |
| 811 | } |
| 812 | |
| 813 | #endif /* !HAVE_ICONV */ |
| 814 | |
| 815 | char *SDL_iconv_string(const char *tocode, const char *fromcode, const char *inbuf, size_t inbytesleft) |
| 816 | { |
| 817 | SDL_iconv_t cd; |
| 818 | char *string; |
| 819 | size_t stringsize; |
| 820 | char *outbuf; |
| 821 | size_t outbytesleft; |
| 822 | size_t retCode = 0; |
| 823 | |
| 824 | cd = SDL_iconv_open(tocode, fromcode); |
| 825 | if ( cd == (SDL_iconv_t)-1 ) { |
| 826 | /* See if we can recover here (fixes iconv on Solaris 11) */ |
| 827 | if ( !tocode || !*tocode ) { |
| 828 | tocode = "UTF-8"; |
| 829 | } |
| 830 | if ( !fromcode || !*fromcode ) { |
| 831 | fromcode = "UTF-8"; |
| 832 | } |
| 833 | cd = SDL_iconv_open(tocode, fromcode); |
| 834 | } |
| 835 | if ( cd == (SDL_iconv_t)-1 ) { |
| 836 | return NULL; |
| 837 | } |
| 838 | |
| 839 | stringsize = inbytesleft > 4 ? inbytesleft : 4; |
| 840 | string = SDL_malloc(stringsize); |
| 841 | if ( !string ) { |
| 842 | SDL_iconv_close(cd); |
| 843 | return NULL; |
| 844 | } |
| 845 | outbuf = string; |
| 846 | outbytesleft = stringsize; |
| 847 | SDL_memset(outbuf, 0, 4); |
| 848 | |
| 849 | while ( inbytesleft > 0 ) { |
| 850 | retCode = SDL_iconv(cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft); |
| 851 | switch (retCode) { |
| 852 | case SDL_ICONV_E2BIG: |
| 853 | { |
| 854 | char *oldstring = string; |
| 855 | stringsize *= 2; |
| 856 | string = SDL_realloc(string, stringsize); |
| 857 | if ( !string ) { |
| 858 | SDL_iconv_close(cd); |
| 859 | return NULL; |
| 860 | } |
| 861 | outbuf = string + (outbuf - oldstring); |
| 862 | outbytesleft = stringsize - (outbuf - string); |
| 863 | SDL_memset(outbuf, 0, 4); |
| 864 | } |
| 865 | break; |
| 866 | case SDL_ICONV_EILSEQ: |
| 867 | /* Try skipping some input data - not perfect, but... */ |
| 868 | ++inbuf; |
| 869 | --inbytesleft; |
| 870 | break; |
| 871 | case SDL_ICONV_EINVAL: |
| 872 | case SDL_ICONV_ERROR: |
| 873 | /* We can't continue... */ |
| 874 | inbytesleft = 0; |
| 875 | break; |
| 876 | } |
| 877 | } |
| 878 | SDL_iconv_close(cd); |
| 879 | |
| 880 | return string; |
| 881 | } |