The Tor Browser / file comparison
comparison: intl/icu/source/common/ucnvlat1.c
intl/icu/source/common/ucnvlat1.c
- branch
- TOR_BUG_3246
- changeset 4
- fc2d59ddac77
equal
deleted
inserted
replaced
| -1:000000000000 | 0:f6b0e684aa48 |
|---|---|
| 1 /* | |
| 2 ********************************************************************** | |
| 3 * Copyright (C) 2000-2012, International Business Machines | |
| 4 * Corporation and others. All Rights Reserved. | |
| 5 ********************************************************************** | |
| 6 * file name: ucnvlat1.cpp | |
| 7 * encoding: US-ASCII | |
| 8 * tab size: 8 (not used) | |
| 9 * indentation:4 | |
| 10 * | |
| 11 * created on: 2000feb07 | |
| 12 * created by: Markus W. Scherer | |
| 13 */ | |
| 14 | |
| 15 #include "unicode/utypes.h" | |
| 16 | |
| 17 #if !UCONFIG_NO_CONVERSION | |
| 18 | |
| 19 #include "unicode/ucnv.h" | |
| 20 #include "unicode/uset.h" | |
| 21 #include "unicode/utf8.h" | |
| 22 #include "ucnv_bld.h" | |
| 23 #include "ucnv_cnv.h" | |
| 24 | |
| 25 /* control optimizations according to the platform */ | |
| 26 #define LATIN1_UNROLL_FROM_UNICODE 1 | |
| 27 | |
| 28 /* ISO 8859-1 --------------------------------------------------------------- */ | |
| 29 | |
| 30 /* This is a table-less and callback-less version of ucnv_MBCSSingleToBMPWithOffsets(). */ | |
| 31 static void | |
| 32 _Latin1ToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, | |
| 33 UErrorCode *pErrorCode) { | |
| 34 const uint8_t *source; | |
| 35 UChar *target; | |
| 36 int32_t targetCapacity, length; | |
| 37 int32_t *offsets; | |
| 38 | |
| 39 int32_t sourceIndex; | |
| 40 | |
| 41 /* set up the local pointers */ | |
| 42 source=(const uint8_t *)pArgs->source; | |
| 43 target=pArgs->target; | |
| 44 targetCapacity=(int32_t)(pArgs->targetLimit-pArgs->target); | |
| 45 offsets=pArgs->offsets; | |
| 46 | |
| 47 sourceIndex=0; | |
| 48 | |
| 49 /* | |
| 50 * since the conversion here is 1:1 UChar:uint8_t, we need only one counter | |
| 51 * for the minimum of the sourceLength and targetCapacity | |
| 52 */ | |
| 53 length=(int32_t)((const uint8_t *)pArgs->sourceLimit-source); | |
| 54 if(length<=targetCapacity) { | |
| 55 targetCapacity=length; | |
| 56 } else { | |
| 57 /* target will be full */ | |
| 58 *pErrorCode=U_BUFFER_OVERFLOW_ERROR; | |
| 59 length=targetCapacity; | |
| 60 } | |
| 61 | |
| 62 if(targetCapacity>=8) { | |
| 63 /* This loop is unrolled for speed and improved pipelining. */ | |
| 64 int32_t count, loops; | |
| 65 | |
| 66 loops=count=targetCapacity>>3; | |
| 67 length=targetCapacity&=0x7; | |
| 68 do { | |
| 69 target[0]=source[0]; | |
| 70 target[1]=source[1]; | |
| 71 target[2]=source[2]; | |
| 72 target[3]=source[3]; | |
| 73 target[4]=source[4]; | |
| 74 target[5]=source[5]; | |
| 75 target[6]=source[6]; | |
| 76 target[7]=source[7]; | |
| 77 target+=8; | |
| 78 source+=8; | |
| 79 } while(--count>0); | |
| 80 | |
| 81 if(offsets!=NULL) { | |
| 82 do { | |
| 83 offsets[0]=sourceIndex++; | |
| 84 offsets[1]=sourceIndex++; | |
| 85 offsets[2]=sourceIndex++; | |
| 86 offsets[3]=sourceIndex++; | |
| 87 offsets[4]=sourceIndex++; | |
| 88 offsets[5]=sourceIndex++; | |
| 89 offsets[6]=sourceIndex++; | |
| 90 offsets[7]=sourceIndex++; | |
| 91 offsets+=8; | |
| 92 } while(--loops>0); | |
| 93 } | |
| 94 } | |
| 95 | |
| 96 /* conversion loop */ | |
| 97 while(targetCapacity>0) { | |
| 98 *target++=*source++; | |
| 99 --targetCapacity; | |
| 100 } | |
| 101 | |
| 102 /* write back the updated pointers */ | |
| 103 pArgs->source=(const char *)source; | |
| 104 pArgs->target=target; | |
| 105 | |
| 106 /* set offsets */ | |
| 107 if(offsets!=NULL) { | |
| 108 while(length>0) { | |
| 109 *offsets++=sourceIndex++; | |
| 110 --length; | |
| 111 } | |
| 112 pArgs->offsets=offsets; | |
| 113 } | |
| 114 } | |
| 115 | |
| 116 /* This is a table-less and callback-less version of ucnv_MBCSSingleGetNextUChar(). */ | |
| 117 static UChar32 | |
| 118 _Latin1GetNextUChar(UConverterToUnicodeArgs *pArgs, | |
| 119 UErrorCode *pErrorCode) { | |
| 120 const uint8_t *source=(const uint8_t *)pArgs->source; | |
| 121 if(source<(const uint8_t *)pArgs->sourceLimit) { | |
| 122 pArgs->source=(const char *)(source+1); | |
| 123 return *source; | |
| 124 } | |
| 125 | |
| 126 /* no output because of empty input */ | |
| 127 *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; | |
| 128 return 0xffff; | |
| 129 } | |
| 130 | |
| 131 /* This is a table-less version of ucnv_MBCSSingleFromBMPWithOffsets(). */ | |
| 132 static void | |
| 133 _Latin1FromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs, | |
| 134 UErrorCode *pErrorCode) { | |
| 135 UConverter *cnv; | |
| 136 const UChar *source, *sourceLimit; | |
| 137 uint8_t *target, *oldTarget; | |
| 138 int32_t targetCapacity, length; | |
| 139 int32_t *offsets; | |
| 140 | |
| 141 UChar32 cp; | |
| 142 UChar c, max; | |
| 143 | |
| 144 int32_t sourceIndex; | |
| 145 | |
| 146 /* set up the local pointers */ | |
| 147 cnv=pArgs->converter; | |
| 148 source=pArgs->source; | |
| 149 sourceLimit=pArgs->sourceLimit; | |
| 150 target=oldTarget=(uint8_t *)pArgs->target; | |
| 151 targetCapacity=(int32_t)(pArgs->targetLimit-pArgs->target); | |
| 152 offsets=pArgs->offsets; | |
| 153 | |
| 154 if(cnv->sharedData==&_Latin1Data) { | |
| 155 max=0xff; /* Latin-1 */ | |
| 156 } else { | |
| 157 max=0x7f; /* US-ASCII */ | |
| 158 } | |
| 159 | |
| 160 /* get the converter state from UConverter */ | |
| 161 cp=cnv->fromUChar32; | |
| 162 | |
| 163 /* sourceIndex=-1 if the current character began in the previous buffer */ | |
| 164 sourceIndex= cp==0 ? 0 : -1; | |
| 165 | |
| 166 /* | |
| 167 * since the conversion here is 1:1 UChar:uint8_t, we need only one counter | |
| 168 * for the minimum of the sourceLength and targetCapacity | |
| 169 */ | |
| 170 length=(int32_t)(sourceLimit-source); | |
| 171 if(length<targetCapacity) { | |
| 172 targetCapacity=length; | |
| 173 } | |
| 174 | |
| 175 /* conversion loop */ | |
| 176 if(cp!=0 && targetCapacity>0) { | |
| 177 goto getTrail; | |
| 178 } | |
| 179 | |
| 180 #if LATIN1_UNROLL_FROM_UNICODE | |
| 181 /* unroll the loop with the most common case */ | |
| 182 if(targetCapacity>=16) { | |
| 183 int32_t count, loops; | |
| 184 UChar u, oredChars; | |
| 185 | |
| 186 loops=count=targetCapacity>>4; | |
| 187 do { | |
| 188 oredChars=u=*source++; | |
| 189 *target++=(uint8_t)u; | |
| 190 oredChars|=u=*source++; | |
| 191 *target++=(uint8_t)u; | |
| 192 oredChars|=u=*source++; | |
| 193 *target++=(uint8_t)u; | |
| 194 oredChars|=u=*source++; | |
| 195 *target++=(uint8_t)u; | |
| 196 oredChars|=u=*source++; | |
| 197 *target++=(uint8_t)u; | |
| 198 oredChars|=u=*source++; | |
| 199 *target++=(uint8_t)u; | |
| 200 oredChars|=u=*source++; | |
| 201 *target++=(uint8_t)u; | |
| 202 oredChars|=u=*source++; | |
| 203 *target++=(uint8_t)u; | |
| 204 oredChars|=u=*source++; | |
| 205 *target++=(uint8_t)u; | |
| 206 oredChars|=u=*source++; | |
| 207 *target++=(uint8_t)u; | |
| 208 oredChars|=u=*source++; | |
| 209 *target++=(uint8_t)u; | |
| 210 oredChars|=u=*source++; | |
| 211 *target++=(uint8_t)u; | |
| 212 oredChars|=u=*source++; | |
| 213 *target++=(uint8_t)u; | |
| 214 oredChars|=u=*source++; | |
| 215 *target++=(uint8_t)u; | |
| 216 oredChars|=u=*source++; | |
| 217 *target++=(uint8_t)u; | |
| 218 oredChars|=u=*source++; | |
| 219 *target++=(uint8_t)u; | |
| 220 | |
| 221 /* were all 16 entries really valid? */ | |
| 222 if(oredChars>max) { | |
| 223 /* no, return to the first of these 16 */ | |
| 224 source-=16; | |
| 225 target-=16; | |
| 226 break; | |
| 227 } | |
| 228 } while(--count>0); | |
| 229 count=loops-count; | |
| 230 targetCapacity-=16*count; | |
| 231 | |
| 232 if(offsets!=NULL) { | |
| 233 oldTarget+=16*count; | |
| 234 while(count>0) { | |
| 235 *offsets++=sourceIndex++; | |
| 236 *offsets++=sourceIndex++; | |
| 237 *offsets++=sourceIndex++; | |
| 238 *offsets++=sourceIndex++; | |
| 239 *offsets++=sourceIndex++; | |
| 240 *offsets++=sourceIndex++; | |
| 241 *offsets++=sourceIndex++; | |
| 242 *offsets++=sourceIndex++; | |
| 243 *offsets++=sourceIndex++; | |
| 244 *offsets++=sourceIndex++; | |
| 245 *offsets++=sourceIndex++; | |
| 246 *offsets++=sourceIndex++; | |
| 247 *offsets++=sourceIndex++; | |
| 248 *offsets++=sourceIndex++; | |
| 249 *offsets++=sourceIndex++; | |
| 250 *offsets++=sourceIndex++; | |
| 251 --count; | |
| 252 } | |
| 253 } | |
| 254 } | |
| 255 #endif | |
| 256 | |
| 257 /* conversion loop */ | |
| 258 c=0; | |
| 259 while(targetCapacity>0 && (c=*source++)<=max) { | |
| 260 /* convert the Unicode code point */ | |
| 261 *target++=(uint8_t)c; | |
| 262 --targetCapacity; | |
| 263 } | |
| 264 | |
| 265 if(c>max) { | |
| 266 cp=c; | |
| 267 if(!U_IS_SURROGATE(cp)) { | |
| 268 /* callback(unassigned) */ | |
| 269 } else if(U_IS_SURROGATE_LEAD(cp)) { | |
| 270 getTrail: | |
| 271 if(source<sourceLimit) { | |
| 272 /* test the following code unit */ | |
| 273 UChar trail=*source; | |
| 274 if(U16_IS_TRAIL(trail)) { | |
| 275 ++source; | |
| 276 cp=U16_GET_SUPPLEMENTARY(cp, trail); | |
| 277 /* this codepage does not map supplementary code points */ | |
| 278 /* callback(unassigned) */ | |
| 279 } else { | |
| 280 /* this is an unmatched lead code unit (1st surrogate) */ | |
| 281 /* callback(illegal) */ | |
| 282 } | |
| 283 } else { | |
| 284 /* no more input */ | |
| 285 cnv->fromUChar32=cp; | |
| 286 goto noMoreInput; | |
| 287 } | |
| 288 } else { | |
| 289 /* this is an unmatched trail code unit (2nd surrogate) */ | |
| 290 /* callback(illegal) */ | |
| 291 } | |
| 292 | |
| 293 *pErrorCode= U_IS_SURROGATE(cp) ? U_ILLEGAL_CHAR_FOUND : U_INVALID_CHAR_FOUND; | |
| 294 cnv->fromUChar32=cp; | |
| 295 } | |
| 296 noMoreInput: | |
| 297 | |
| 298 /* set offsets since the start */ | |
| 299 if(offsets!=NULL) { | |
| 300 size_t count=target-oldTarget; | |
| 301 while(count>0) { | |
| 302 *offsets++=sourceIndex++; | |
| 303 --count; | |
| 304 } | |
| 305 } | |
| 306 | |
| 307 if(U_SUCCESS(*pErrorCode) && source<sourceLimit && target>=(uint8_t *)pArgs->targetLimit) { | |
| 308 /* target is full */ | |
| 309 *pErrorCode=U_BUFFER_OVERFLOW_ERROR; | |
| 310 } | |
| 311 | |
| 312 /* write back the updated pointers */ | |
| 313 pArgs->source=source; | |
| 314 pArgs->target=(char *)target; | |
| 315 pArgs->offsets=offsets; | |
| 316 } | |
| 317 | |
| 318 /* Convert UTF-8 to Latin-1. Adapted from ucnv_SBCSFromUTF8(). */ | |
| 319 static void | |
| 320 ucnv_Latin1FromUTF8(UConverterFromUnicodeArgs *pFromUArgs, | |
| 321 UConverterToUnicodeArgs *pToUArgs, | |
| 322 UErrorCode *pErrorCode) { | |
| 323 UConverter *utf8; | |
| 324 const uint8_t *source, *sourceLimit; | |
| 325 uint8_t *target; | |
| 326 int32_t targetCapacity; | |
| 327 | |
| 328 UChar32 c; | |
| 329 uint8_t b, t1; | |
| 330 | |
| 331 /* set up the local pointers */ | |
| 332 utf8=pToUArgs->converter; | |
| 333 source=(uint8_t *)pToUArgs->source; | |
| 334 sourceLimit=(uint8_t *)pToUArgs->sourceLimit; | |
| 335 target=(uint8_t *)pFromUArgs->target; | |
| 336 targetCapacity=(int32_t)(pFromUArgs->targetLimit-pFromUArgs->target); | |
| 337 | |
| 338 /* get the converter state from the UTF-8 UConverter */ | |
| 339 c=(UChar32)utf8->toUnicodeStatus; | |
| 340 if(c!=0 && source<sourceLimit) { | |
| 341 if(targetCapacity==0) { | |
| 342 *pErrorCode=U_BUFFER_OVERFLOW_ERROR; | |
| 343 return; | |
| 344 } else if(c>=0xc2 && c<=0xc3 && (t1=(uint8_t)(*source-0x80)) <= 0x3f) { | |
| 345 ++source; | |
| 346 *target++=(uint8_t)(((c&3)<<6)|t1); | |
| 347 --targetCapacity; | |
| 348 | |
| 349 utf8->toUnicodeStatus=0; | |
| 350 utf8->toULength=0; | |
| 351 } else { | |
| 352 /* complicated, illegal or unmappable input: fall back to the pivoting implementation */ | |
| 353 *pErrorCode=U_USING_DEFAULT_WARNING; | |
| 354 return; | |
| 355 } | |
| 356 } | |
| 357 | |
| 358 /* | |
| 359 * Make sure that the last byte sequence before sourceLimit is complete | |
| 360 * or runs into a lead byte. | |
| 361 * In the conversion loop compare source with sourceLimit only once | |
| 362 * per multi-byte character. | |
| 363 * For Latin-1, adjust sourceLimit only for 1 trail byte because | |
| 364 * the conversion loop handles at most 2-byte sequences. | |
| 365 */ | |
| 366 if(source<sourceLimit && U8_IS_LEAD(*(sourceLimit-1))) { | |
| 367 --sourceLimit; | |
| 368 } | |
| 369 | |
| 370 /* conversion loop */ | |
| 371 while(source<sourceLimit) { | |
| 372 if(targetCapacity>0) { | |
| 373 b=*source++; | |
| 374 if((int8_t)b>=0) { | |
| 375 /* convert ASCII */ | |
| 376 *target++=(uint8_t)b; | |
| 377 --targetCapacity; | |
| 378 } else if( /* handle U+0080..U+00FF inline */ | |
| 379 b>=0xc2 && b<=0xc3 && | |
| 380 (t1=(uint8_t)(*source-0x80)) <= 0x3f | |
| 381 ) { | |
| 382 ++source; | |
| 383 *target++=(uint8_t)(((b&3)<<6)|t1); | |
| 384 --targetCapacity; | |
| 385 } else { | |
| 386 /* complicated, illegal or unmappable input: fall back to the pivoting implementation */ | |
| 387 pToUArgs->source=(char *)(source-1); | |
| 388 pFromUArgs->target=(char *)target; | |
| 389 *pErrorCode=U_USING_DEFAULT_WARNING; | |
| 390 return; | |
| 391 } | |
| 392 } else { | |
| 393 /* target is full */ | |
| 394 *pErrorCode=U_BUFFER_OVERFLOW_ERROR; | |
| 395 break; | |
| 396 } | |
| 397 } | |
| 398 | |
| 399 /* | |
| 400 * The sourceLimit may have been adjusted before the conversion loop | |
| 401 * to stop before a truncated sequence. | |
| 402 * If so, then collect the truncated sequence now. | |
| 403 * For Latin-1, there is at most exactly one lead byte because of the | |
| 404 * smaller sourceLimit adjustment logic. | |
| 405 */ | |
| 406 if(U_SUCCESS(*pErrorCode) && source<(sourceLimit=(uint8_t *)pToUArgs->sourceLimit)) { | |
| 407 utf8->toUnicodeStatus=utf8->toUBytes[0]=b=*source++; | |
| 408 utf8->toULength=1; | |
| 409 utf8->mode=U8_COUNT_TRAIL_BYTES(b)+1; | |
| 410 } | |
| 411 | |
| 412 /* write back the updated pointers */ | |
| 413 pToUArgs->source=(char *)source; | |
| 414 pFromUArgs->target=(char *)target; | |
| 415 } | |
| 416 | |
| 417 static void | |
| 418 _Latin1GetUnicodeSet(const UConverter *cnv, | |
| 419 const USetAdder *sa, | |
| 420 UConverterUnicodeSet which, | |
| 421 UErrorCode *pErrorCode) { | |
| 422 sa->addRange(sa->set, 0, 0xff); | |
| 423 } | |
| 424 | |
| 425 static const UConverterImpl _Latin1Impl={ | |
| 426 UCNV_LATIN_1, | |
| 427 | |
| 428 NULL, | |
| 429 NULL, | |
| 430 | |
| 431 NULL, | |
| 432 NULL, | |
| 433 NULL, | |
| 434 | |
| 435 _Latin1ToUnicodeWithOffsets, | |
| 436 _Latin1ToUnicodeWithOffsets, | |
| 437 _Latin1FromUnicodeWithOffsets, | |
| 438 _Latin1FromUnicodeWithOffsets, | |
| 439 _Latin1GetNextUChar, | |
| 440 | |
| 441 NULL, | |
| 442 NULL, | |
| 443 NULL, | |
| 444 NULL, | |
| 445 _Latin1GetUnicodeSet, | |
| 446 | |
| 447 NULL, | |
| 448 ucnv_Latin1FromUTF8 | |
| 449 }; | |
| 450 | |
| 451 static const UConverterStaticData _Latin1StaticData={ | |
| 452 sizeof(UConverterStaticData), | |
| 453 "ISO-8859-1", | |
| 454 819, UCNV_IBM, UCNV_LATIN_1, 1, 1, | |
| 455 { 0x1a, 0, 0, 0 }, 1, FALSE, FALSE, | |
| 456 0, | |
| 457 0, | |
| 458 { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ | |
| 459 }; | |
| 460 | |
| 461 const UConverterSharedData _Latin1Data={ | |
| 462 sizeof(UConverterSharedData), ~((uint32_t) 0), | |
| 463 NULL, NULL, &_Latin1StaticData, FALSE, &_Latin1Impl, | |
| 464 0 | |
| 465 }; | |
| 466 | |
| 467 /* US-ASCII ----------------------------------------------------------------- */ | |
| 468 | |
| 469 /* This is a table-less version of ucnv_MBCSSingleToBMPWithOffsets(). */ | |
| 470 static void | |
| 471 _ASCIIToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, | |
| 472 UErrorCode *pErrorCode) { | |
| 473 const uint8_t *source, *sourceLimit; | |
| 474 UChar *target, *oldTarget; | |
| 475 int32_t targetCapacity, length; | |
| 476 int32_t *offsets; | |
| 477 | |
| 478 int32_t sourceIndex; | |
| 479 | |
| 480 uint8_t c; | |
| 481 | |
| 482 /* set up the local pointers */ | |
| 483 source=(const uint8_t *)pArgs->source; | |
| 484 sourceLimit=(const uint8_t *)pArgs->sourceLimit; | |
| 485 target=oldTarget=pArgs->target; | |
| 486 targetCapacity=(int32_t)(pArgs->targetLimit-pArgs->target); | |
| 487 offsets=pArgs->offsets; | |
| 488 | |
| 489 /* sourceIndex=-1 if the current character began in the previous buffer */ | |
| 490 sourceIndex=0; | |
| 491 | |
| 492 /* | |
| 493 * since the conversion here is 1:1 UChar:uint8_t, we need only one counter | |
| 494 * for the minimum of the sourceLength and targetCapacity | |
| 495 */ | |
| 496 length=(int32_t)(sourceLimit-source); | |
| 497 if(length<targetCapacity) { | |
| 498 targetCapacity=length; | |
| 499 } | |
| 500 | |
| 501 if(targetCapacity>=8) { | |
| 502 /* This loop is unrolled for speed and improved pipelining. */ | |
| 503 int32_t count, loops; | |
| 504 UChar oredChars; | |
| 505 | |
| 506 loops=count=targetCapacity>>3; | |
| 507 do { | |
| 508 oredChars=target[0]=source[0]; | |
| 509 oredChars|=target[1]=source[1]; | |
| 510 oredChars|=target[2]=source[2]; | |
| 511 oredChars|=target[3]=source[3]; | |
| 512 oredChars|=target[4]=source[4]; | |
| 513 oredChars|=target[5]=source[5]; | |
| 514 oredChars|=target[6]=source[6]; | |
| 515 oredChars|=target[7]=source[7]; | |
| 516 | |
| 517 /* were all 16 entries really valid? */ | |
| 518 if(oredChars>0x7f) { | |
| 519 /* no, return to the first of these 16 */ | |
| 520 break; | |
| 521 } | |
| 522 source+=8; | |
| 523 target+=8; | |
| 524 } while(--count>0); | |
| 525 count=loops-count; | |
| 526 targetCapacity-=count*8; | |
| 527 | |
| 528 if(offsets!=NULL) { | |
| 529 oldTarget+=count*8; | |
| 530 while(count>0) { | |
| 531 offsets[0]=sourceIndex++; | |
| 532 offsets[1]=sourceIndex++; | |
| 533 offsets[2]=sourceIndex++; | |
| 534 offsets[3]=sourceIndex++; | |
| 535 offsets[4]=sourceIndex++; | |
| 536 offsets[5]=sourceIndex++; | |
| 537 offsets[6]=sourceIndex++; | |
| 538 offsets[7]=sourceIndex++; | |
| 539 offsets+=8; | |
| 540 --count; | |
| 541 } | |
| 542 } | |
| 543 } | |
| 544 | |
| 545 /* conversion loop */ | |
| 546 c=0; | |
| 547 while(targetCapacity>0 && (c=*source++)<=0x7f) { | |
| 548 *target++=c; | |
| 549 --targetCapacity; | |
| 550 } | |
| 551 | |
| 552 if(c>0x7f) { | |
| 553 /* callback(illegal); copy the current bytes to toUBytes[] */ | |
| 554 UConverter *cnv=pArgs->converter; | |
| 555 cnv->toUBytes[0]=c; | |
| 556 cnv->toULength=1; | |
| 557 *pErrorCode=U_ILLEGAL_CHAR_FOUND; | |
| 558 } else if(source<sourceLimit && target>=pArgs->targetLimit) { | |
| 559 /* target is full */ | |
| 560 *pErrorCode=U_BUFFER_OVERFLOW_ERROR; | |
| 561 } | |
| 562 | |
| 563 /* set offsets since the start */ | |
| 564 if(offsets!=NULL) { | |
| 565 size_t count=target-oldTarget; | |
| 566 while(count>0) { | |
| 567 *offsets++=sourceIndex++; | |
| 568 --count; | |
| 569 } | |
| 570 } | |
| 571 | |
| 572 /* write back the updated pointers */ | |
| 573 pArgs->source=(const char *)source; | |
| 574 pArgs->target=target; | |
| 575 pArgs->offsets=offsets; | |
| 576 } | |
| 577 | |
| 578 /* This is a table-less version of ucnv_MBCSSingleGetNextUChar(). */ | |
| 579 static UChar32 | |
| 580 _ASCIIGetNextUChar(UConverterToUnicodeArgs *pArgs, | |
| 581 UErrorCode *pErrorCode) { | |
| 582 const uint8_t *source; | |
| 583 uint8_t b; | |
| 584 | |
| 585 source=(const uint8_t *)pArgs->source; | |
| 586 if(source<(const uint8_t *)pArgs->sourceLimit) { | |
| 587 b=*source++; | |
| 588 pArgs->source=(const char *)source; | |
| 589 if(b<=0x7f) { | |
| 590 return b; | |
| 591 } else { | |
| 592 UConverter *cnv=pArgs->converter; | |
| 593 cnv->toUBytes[0]=b; | |
| 594 cnv->toULength=1; | |
| 595 *pErrorCode=U_ILLEGAL_CHAR_FOUND; | |
| 596 return 0xffff; | |
| 597 } | |
| 598 } | |
| 599 | |
| 600 /* no output because of empty input */ | |
| 601 *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; | |
| 602 return 0xffff; | |
| 603 } | |
| 604 | |
| 605 /* "Convert" UTF-8 to US-ASCII: Validate and copy. */ | |
| 606 static void | |
| 607 ucnv_ASCIIFromUTF8(UConverterFromUnicodeArgs *pFromUArgs, | |
| 608 UConverterToUnicodeArgs *pToUArgs, | |
| 609 UErrorCode *pErrorCode) { | |
| 610 const uint8_t *source, *sourceLimit; | |
| 611 uint8_t *target; | |
| 612 int32_t targetCapacity, length; | |
| 613 | |
| 614 uint8_t c; | |
| 615 | |
| 616 if(pToUArgs->converter->toUnicodeStatus!=0) { | |
| 617 /* no handling of partial UTF-8 characters here, fall back to pivoting */ | |
| 618 *pErrorCode=U_USING_DEFAULT_WARNING; | |
| 619 return; | |
| 620 } | |
| 621 | |
| 622 /* set up the local pointers */ | |
| 623 source=(const uint8_t *)pToUArgs->source; | |
| 624 sourceLimit=(const uint8_t *)pToUArgs->sourceLimit; | |
| 625 target=(uint8_t *)pFromUArgs->target; | |
| 626 targetCapacity=(int32_t)(pFromUArgs->targetLimit-pFromUArgs->target); | |
| 627 | |
| 628 /* | |
| 629 * since the conversion here is 1:1 uint8_t:uint8_t, we need only one counter | |
| 630 * for the minimum of the sourceLength and targetCapacity | |
| 631 */ | |
| 632 length=(int32_t)(sourceLimit-source); | |
| 633 if(length<targetCapacity) { | |
| 634 targetCapacity=length; | |
| 635 } | |
| 636 | |
| 637 /* unroll the loop with the most common case */ | |
| 638 if(targetCapacity>=16) { | |
| 639 int32_t count, loops; | |
| 640 uint8_t oredChars; | |
| 641 | |
| 642 loops=count=targetCapacity>>4; | |
| 643 do { | |
| 644 oredChars=*target++=*source++; | |
| 645 oredChars|=*target++=*source++; | |
| 646 oredChars|=*target++=*source++; | |
| 647 oredChars|=*target++=*source++; | |
| 648 oredChars|=*target++=*source++; | |
| 649 oredChars|=*target++=*source++; | |
| 650 oredChars|=*target++=*source++; | |
| 651 oredChars|=*target++=*source++; | |
| 652 oredChars|=*target++=*source++; | |
| 653 oredChars|=*target++=*source++; | |
| 654 oredChars|=*target++=*source++; | |
| 655 oredChars|=*target++=*source++; | |
| 656 oredChars|=*target++=*source++; | |
| 657 oredChars|=*target++=*source++; | |
| 658 oredChars|=*target++=*source++; | |
| 659 oredChars|=*target++=*source++; | |
| 660 | |
| 661 /* were all 16 entries really valid? */ | |
| 662 if(oredChars>0x7f) { | |
| 663 /* no, return to the first of these 16 */ | |
| 664 source-=16; | |
| 665 target-=16; | |
| 666 break; | |
| 667 } | |
| 668 } while(--count>0); | |
| 669 count=loops-count; | |
| 670 targetCapacity-=16*count; | |
| 671 } | |
| 672 | |
| 673 /* conversion loop */ | |
| 674 c=0; | |
| 675 while(targetCapacity>0 && (c=*source)<=0x7f) { | |
| 676 ++source; | |
| 677 *target++=c; | |
| 678 --targetCapacity; | |
| 679 } | |
| 680 | |
| 681 if(c>0x7f) { | |
| 682 /* non-ASCII character, handle in standard converter */ | |
| 683 *pErrorCode=U_USING_DEFAULT_WARNING; | |
| 684 } else if(source<sourceLimit && target>=(const uint8_t *)pFromUArgs->targetLimit) { | |
| 685 /* target is full */ | |
| 686 *pErrorCode=U_BUFFER_OVERFLOW_ERROR; | |
| 687 } | |
| 688 | |
| 689 /* write back the updated pointers */ | |
| 690 pToUArgs->source=(const char *)source; | |
| 691 pFromUArgs->target=(char *)target; | |
| 692 } | |
| 693 | |
| 694 static void | |
| 695 _ASCIIGetUnicodeSet(const UConverter *cnv, | |
| 696 const USetAdder *sa, | |
| 697 UConverterUnicodeSet which, | |
| 698 UErrorCode *pErrorCode) { | |
| 699 sa->addRange(sa->set, 0, 0x7f); | |
| 700 } | |
| 701 | |
| 702 static const UConverterImpl _ASCIIImpl={ | |
| 703 UCNV_US_ASCII, | |
| 704 | |
| 705 NULL, | |
| 706 NULL, | |
| 707 | |
| 708 NULL, | |
| 709 NULL, | |
| 710 NULL, | |
| 711 | |
| 712 _ASCIIToUnicodeWithOffsets, | |
| 713 _ASCIIToUnicodeWithOffsets, | |
| 714 _Latin1FromUnicodeWithOffsets, | |
| 715 _Latin1FromUnicodeWithOffsets, | |
| 716 _ASCIIGetNextUChar, | |
| 717 | |
| 718 NULL, | |
| 719 NULL, | |
| 720 NULL, | |
| 721 NULL, | |
| 722 _ASCIIGetUnicodeSet, | |
| 723 | |
| 724 NULL, | |
| 725 ucnv_ASCIIFromUTF8 | |
| 726 }; | |
| 727 | |
| 728 static const UConverterStaticData _ASCIIStaticData={ | |
| 729 sizeof(UConverterStaticData), | |
| 730 "US-ASCII", | |
| 731 367, UCNV_IBM, UCNV_US_ASCII, 1, 1, | |
| 732 { 0x1a, 0, 0, 0 }, 1, FALSE, FALSE, | |
| 733 0, | |
| 734 0, | |
| 735 { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ | |
| 736 }; | |
| 737 | |
| 738 const UConverterSharedData _ASCIIData={ | |
| 739 sizeof(UConverterSharedData), ~((uint32_t) 0), | |
| 740 NULL, NULL, &_ASCIIStaticData, FALSE, &_ASCIIImpl, | |
| 741 0 | |
| 742 }; | |
| 743 | |
| 744 #endif |
