1 /* 2 ** 2004 April 13 3 ** 4 ** The author disclaims copyright to this source code. In place of 5 ** a legal notice, here is a blessing: 6 ** 7 ** May you do good and not evil. 8 ** May you find forgiveness for yourself and forgive others. 9 ** May you share freely, never taking more than you give. 10 ** 11 ************************************************************************* 12 ** This file contains routines used to translate between UTF-8, 13 ** UTF-16, UTF-16BE, and UTF-16LE. 14 ** 15 ** Notes on UTF-8: 16 ** 17 ** Byte-0 Byte-1 Byte-2 Byte-3 Value 18 ** 0xxxxxxx 00000000 00000000 0xxxxxxx 19 ** 110yyyyy 10xxxxxx 00000000 00000yyy yyxxxxxx 20 ** 1110zzzz 10yyyyyy 10xxxxxx 00000000 zzzzyyyy yyxxxxxx 21 ** 11110uuu 10uuzzzz 10yyyyyy 10xxxxxx 000uuuuu zzzzyyyy yyxxxxxx 22 ** 23 ** 24 ** Notes on UTF-16: (with wwww+1==uuuuu) 25 ** 26 ** Word-0 Word-1 Value 27 ** 110110ww wwzzzzyy 110111yy yyxxxxxx 000uuuuu zzzzyyyy yyxxxxxx 28 ** zzzzyyyy yyxxxxxx 00000000 zzzzyyyy yyxxxxxx 29 ** 30 ** 31 ** BOM or Byte Order Mark: 32 ** 0xff 0xfe little-endian utf-16 follows 33 ** 0xfe 0xff big-endian utf-16 follows 34 ** 35 */ 36 #include "sqliteInt.h" 37 #include <assert.h> 38 #include "vdbeInt.h" 39 40 #if !defined(SQLITE_AMALGAMATION) && SQLITE_BYTEORDER==0 41 /* 42 ** The following constant value is used by the SQLITE_BIGENDIAN and 43 ** SQLITE_LITTLEENDIAN macros. 44 */ 45 const int sqlite3one = 1; 46 #endif /* SQLITE_AMALGAMATION && SQLITE_BYTEORDER==0 */ 47 48 /* 49 ** This lookup table is used to help decode the first byte of 50 ** a multi-byte UTF8 character. 51 */ 52 static const unsigned char sqlite3Utf8Trans1[] = { 53 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 54 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 55 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, 56 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 57 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 58 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 59 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 60 0x00, 0x01, 0x02, 0x03, 0x00, 0x01, 0x00, 0x00, 61 }; 62 63 64 #define WRITE_UTF8(zOut, c) { \ 65 if( c<0x00080 ){ \ 66 *zOut++ = (u8)(c&0xFF); \ 67 } \ 68 else if( c<0x00800 ){ \ 69 *zOut++ = 0xC0 + (u8)((c>>6)&0x1F); \ 70 *zOut++ = 0x80 + (u8)(c & 0x3F); \ 71 } \ 72 else if( c<0x10000 ){ \ 73 *zOut++ = 0xE0 + (u8)((c>>12)&0x0F); \ 74 *zOut++ = 0x80 + (u8)((c>>6) & 0x3F); \ 75 *zOut++ = 0x80 + (u8)(c & 0x3F); \ 76 }else{ \ 77 *zOut++ = 0xF0 + (u8)((c>>18) & 0x07); \ 78 *zOut++ = 0x80 + (u8)((c>>12) & 0x3F); \ 79 *zOut++ = 0x80 + (u8)((c>>6) & 0x3F); \ 80 *zOut++ = 0x80 + (u8)(c & 0x3F); \ 81 } \ 82 } 83 84 #define WRITE_UTF16LE(zOut, c) { \ 85 if( c<=0xFFFF ){ \ 86 *zOut++ = (u8)(c&0x00FF); \ 87 *zOut++ = (u8)((c>>8)&0x00FF); \ 88 }else{ \ 89 *zOut++ = (u8)(((c>>10)&0x003F) + (((c-0x10000)>>10)&0x00C0)); \ 90 *zOut++ = (u8)(0x00D8 + (((c-0x10000)>>18)&0x03)); \ 91 *zOut++ = (u8)(c&0x00FF); \ 92 *zOut++ = (u8)(0x00DC + ((c>>8)&0x03)); \ 93 } \ 94 } 95 96 #define WRITE_UTF16BE(zOut, c) { \ 97 if( c<=0xFFFF ){ \ 98 *zOut++ = (u8)((c>>8)&0x00FF); \ 99 *zOut++ = (u8)(c&0x00FF); \ 100 }else{ \ 101 *zOut++ = (u8)(0x00D8 + (((c-0x10000)>>18)&0x03)); \ 102 *zOut++ = (u8)(((c>>10)&0x003F) + (((c-0x10000)>>10)&0x00C0)); \ 103 *zOut++ = (u8)(0x00DC + ((c>>8)&0x03)); \ 104 *zOut++ = (u8)(c&0x00FF); \ 105 } \ 106 } 107 108 #define READ_UTF16LE(zIn, TERM, c){ \ 109 c = (*zIn++); \ 110 c += ((*zIn++)<<8); \ 111 if( c>=0xD800 && c<0xE000 && TERM ){ \ 112 int c2 = (*zIn++); \ 113 c2 += ((*zIn++)<<8); \ 114 c = (c2&0x03FF) + ((c&0x003F)<<10) + (((c&0x03C0)+0x0040)<<10); \ 115 } \ 116 } 117 118 #define READ_UTF16BE(zIn, TERM, c){ \ 119 c = ((*zIn++)<<8); \ 120 c += (*zIn++); \ 121 if( c>=0xD800 && c<0xE000 && TERM ){ \ 122 int c2 = ((*zIn++)<<8); \ 123 c2 += (*zIn++); \ 124 c = (c2&0x03FF) + ((c&0x003F)<<10) + (((c&0x03C0)+0x0040)<<10); \ 125 } \ 126 } 127 128 /* 129 ** Translate a single UTF-8 character. Return the unicode value. 130 ** 131 ** During translation, assume that the byte that zTerm points 132 ** is a 0x00. 133 ** 134 ** Write a pointer to the next unread byte back into *pzNext. 135 ** 136 ** Notes On Invalid UTF-8: 137 ** 138 ** * This routine never allows a 7-bit character (0x00 through 0x7f) to 139 ** be encoded as a multi-byte character. Any multi-byte character that 140 ** attempts to encode a value between 0x00 and 0x7f is rendered as 0xfffd. 141 ** 142 ** * This routine never allows a UTF16 surrogate value to be encoded. 143 ** If a multi-byte character attempts to encode a value between 144 ** 0xd800 and 0xe000 then it is rendered as 0xfffd. 145 ** 146 ** * Bytes in the range of 0x80 through 0xbf which occur as the first 147 ** byte of a character are interpreted as single-byte characters 148 ** and rendered as themselves even though they are technically 149 ** invalid characters. 150 ** 151 ** * This routine accepts over-length UTF8 encodings 152 ** for unicode values 0x80 and greater. It does not change over-length 153 ** encodings to 0xfffd as some systems recommend. 154 */ 155 #define READ_UTF8(zIn, zTerm, c) \ 156 c = *(zIn++); \ 157 if( c>=0xc0 ){ \ 158 c = sqlite3Utf8Trans1[c-0xc0]; \ 159 while( zIn!=zTerm && (*zIn & 0xc0)==0x80 ){ \ 160 c = (c<<6) + (0x3f & *(zIn++)); \ 161 } \ 162 if( c<0x80 \ 163 || (c&0xFFFFF800)==0xD800 \ 164 || (c&0xFFFFFFFE)==0xFFFE ){ c = 0xFFFD; } \ 165 } 166 u32 sqlite3Utf8Read( 167 const unsigned char **pz /* Pointer to string from which to read char */ 168 ){ 169 unsigned int c; 170 171 /* Same as READ_UTF8() above but without the zTerm parameter. 172 ** For this routine, we assume the UTF8 string is always zero-terminated. 173 */ 174 c = *((*pz)++); 175 if( c>=0xc0 ){ 176 c = sqlite3Utf8Trans1[c-0xc0]; 177 while( (*(*pz) & 0xc0)==0x80 ){ 178 c = (c<<6) + (0x3f & *((*pz)++)); 179 } 180 if( c<0x80 181 || (c&0xFFFFF800)==0xD800 182 || (c&0xFFFFFFFE)==0xFFFE ){ c = 0xFFFD; } 183 } 184 return c; 185 } 186 187 188 189 190 /* 191 ** If the TRANSLATE_TRACE macro is defined, the value of each Mem is 192 ** printed on stderr on the way into and out of sqlite3VdbeMemTranslate(). 193 */ 194 /* #define TRANSLATE_TRACE 1 */ 195 196 #ifndef SQLITE_OMIT_UTF16 197 /* 198 ** This routine transforms the internal text encoding used by pMem to 199 ** desiredEnc. It is an error if the string is already of the desired 200 ** encoding, or if *pMem does not contain a string value. 201 */ 202 SQLITE_NOINLINE int sqlite3VdbeMemTranslate(Mem *pMem, u8 desiredEnc){ 203 sqlite3_int64 len; /* Maximum length of output string in bytes */ 204 unsigned char *zOut; /* Output buffer */ 205 unsigned char *zIn; /* Input iterator */ 206 unsigned char *zTerm; /* End of input */ 207 unsigned char *z; /* Output iterator */ 208 unsigned int c; 209 210 assert( pMem->db==0 || sqlite3_mutex_held(pMem->db->mutex) ); 211 assert( pMem->flags&MEM_Str ); 212 assert( pMem->enc!=desiredEnc ); 213 assert( pMem->enc!=0 ); 214 assert( pMem->n>=0 ); 215 216 #if defined(TRANSLATE_TRACE) && defined(SQLITE_DEBUG) 217 { 218 StrAccum acc; 219 char zBuf[1000]; 220 sqlite3StrAccumInit(&acc, 0, zBuf, sizeof(zBuf), 0); 221 sqlite3VdbeMemPrettyPrint(pMem, &acc); 222 fprintf(stderr, "INPUT: %s\n", sqlite3StrAccumFinish(&acc)); 223 } 224 #endif 225 226 /* If the translation is between UTF-16 little and big endian, then 227 ** all that is required is to swap the byte order. This case is handled 228 ** differently from the others. 229 */ 230 if( pMem->enc!=SQLITE_UTF8 && desiredEnc!=SQLITE_UTF8 ){ 231 u8 temp; 232 int rc; 233 rc = sqlite3VdbeMemMakeWriteable(pMem); 234 if( rc!=SQLITE_OK ){ 235 assert( rc==SQLITE_NOMEM ); 236 return SQLITE_NOMEM_BKPT; 237 } 238 zIn = (u8*)pMem->z; 239 zTerm = &zIn[pMem->n&~1]; 240 while( zIn<zTerm ){ 241 temp = *zIn; 242 *zIn = *(zIn+1); 243 zIn++; 244 *zIn++ = temp; 245 } 246 pMem->enc = desiredEnc; 247 goto translate_out; 248 } 249 250 /* Set len to the maximum number of bytes required in the output buffer. */ 251 if( desiredEnc==SQLITE_UTF8 ){ 252 /* When converting from UTF-16, the maximum growth results from 253 ** translating a 2-byte character to a 4-byte UTF-8 character. 254 ** A single byte is required for the output string 255 ** nul-terminator. 256 */ 257 pMem->n &= ~1; 258 len = 2 * (sqlite3_int64)pMem->n + 1; 259 }else{ 260 /* When converting from UTF-8 to UTF-16 the maximum growth is caused 261 ** when a 1-byte UTF-8 character is translated into a 2-byte UTF-16 262 ** character. Two bytes are required in the output buffer for the 263 ** nul-terminator. 264 */ 265 len = 2 * (sqlite3_int64)pMem->n + 2; 266 } 267 268 /* Set zIn to point at the start of the input buffer and zTerm to point 1 269 ** byte past the end. 270 ** 271 ** Variable zOut is set to point at the output buffer, space obtained 272 ** from sqlite3_malloc(). 273 */ 274 zIn = (u8*)pMem->z; 275 zTerm = &zIn[pMem->n]; 276 zOut = sqlite3DbMallocRaw(pMem->db, len); 277 if( !zOut ){ 278 return SQLITE_NOMEM_BKPT; 279 } 280 z = zOut; 281 282 if( pMem->enc==SQLITE_UTF8 ){ 283 if( desiredEnc==SQLITE_UTF16LE ){ 284 /* UTF-8 -> UTF-16 Little-endian */ 285 while( zIn<zTerm ){ 286 READ_UTF8(zIn, zTerm, c); 287 WRITE_UTF16LE(z, c); 288 } 289 }else{ 290 assert( desiredEnc==SQLITE_UTF16BE ); 291 /* UTF-8 -> UTF-16 Big-endian */ 292 while( zIn<zTerm ){ 293 READ_UTF8(zIn, zTerm, c); 294 WRITE_UTF16BE(z, c); 295 } 296 } 297 pMem->n = (int)(z - zOut); 298 *z++ = 0; 299 }else{ 300 assert( desiredEnc==SQLITE_UTF8 ); 301 if( pMem->enc==SQLITE_UTF16LE ){ 302 /* UTF-16 Little-endian -> UTF-8 */ 303 while( zIn<zTerm ){ 304 READ_UTF16LE(zIn, zIn<zTerm, c); 305 WRITE_UTF8(z, c); 306 } 307 }else{ 308 /* UTF-16 Big-endian -> UTF-8 */ 309 while( zIn<zTerm ){ 310 READ_UTF16BE(zIn, zIn<zTerm, c); 311 WRITE_UTF8(z, c); 312 } 313 } 314 pMem->n = (int)(z - zOut); 315 } 316 *z = 0; 317 assert( (pMem->n+(desiredEnc==SQLITE_UTF8?1:2))<=len ); 318 319 c = pMem->flags; 320 sqlite3VdbeMemRelease(pMem); 321 pMem->flags = MEM_Str|MEM_Term|(c&(MEM_AffMask|MEM_Subtype)); 322 pMem->enc = desiredEnc; 323 pMem->z = (char*)zOut; 324 pMem->zMalloc = pMem->z; 325 pMem->szMalloc = sqlite3DbMallocSize(pMem->db, pMem->z); 326 327 translate_out: 328 #if defined(TRANSLATE_TRACE) && defined(SQLITE_DEBUG) 329 { 330 StrAccum acc; 331 char zBuf[1000]; 332 sqlite3StrAccumInit(&acc, 0, zBuf, sizeof(zBuf), 0); 333 sqlite3VdbeMemPrettyPrint(pMem, &acc); 334 fprintf(stderr, "OUTPUT: %s\n", sqlite3StrAccumFinish(&acc)); 335 } 336 #endif 337 return SQLITE_OK; 338 } 339 #endif /* SQLITE_OMIT_UTF16 */ 340 341 #ifndef SQLITE_OMIT_UTF16 342 /* 343 ** This routine checks for a byte-order mark at the beginning of the 344 ** UTF-16 string stored in *pMem. If one is present, it is removed and 345 ** the encoding of the Mem adjusted. This routine does not do any 346 ** byte-swapping, it just sets Mem.enc appropriately. 347 ** 348 ** The allocation (static, dynamic etc.) and encoding of the Mem may be 349 ** changed by this function. 350 */ 351 int sqlite3VdbeMemHandleBom(Mem *pMem){ 352 int rc = SQLITE_OK; 353 u8 bom = 0; 354 355 assert( pMem->n>=0 ); 356 if( pMem->n>1 ){ 357 u8 b1 = *(u8 *)pMem->z; 358 u8 b2 = *(((u8 *)pMem->z) + 1); 359 if( b1==0xFE && b2==0xFF ){ 360 bom = SQLITE_UTF16BE; 361 } 362 if( b1==0xFF && b2==0xFE ){ 363 bom = SQLITE_UTF16LE; 364 } 365 } 366 367 if( bom ){ 368 rc = sqlite3VdbeMemMakeWriteable(pMem); 369 if( rc==SQLITE_OK ){ 370 pMem->n -= 2; 371 memmove(pMem->z, &pMem->z[2], pMem->n); 372 pMem->z[pMem->n] = '\0'; 373 pMem->z[pMem->n+1] = '\0'; 374 pMem->flags |= MEM_Term; 375 pMem->enc = bom; 376 } 377 } 378 return rc; 379 } 380 #endif /* SQLITE_OMIT_UTF16 */ 381 382 /* 383 ** pZ is a UTF-8 encoded unicode string. If nByte is less than zero, 384 ** return the number of unicode characters in pZ up to (but not including) 385 ** the first 0x00 byte. If nByte is not less than zero, return the 386 ** number of unicode characters in the first nByte of pZ (or up to 387 ** the first 0x00, whichever comes first). 388 */ 389 int sqlite3Utf8CharLen(const char *zIn, int nByte){ 390 int r = 0; 391 const u8 *z = (const u8*)zIn; 392 const u8 *zTerm; 393 if( nByte>=0 ){ 394 zTerm = &z[nByte]; 395 }else{ 396 zTerm = (const u8*)(-1); 397 } 398 assert( z<=zTerm ); 399 while( *z!=0 && z<zTerm ){ 400 SQLITE_SKIP_UTF8(z); 401 r++; 402 } 403 return r; 404 } 405 406 /* This test function is not currently used by the automated test-suite. 407 ** Hence it is only available in debug builds. 408 */ 409 #if defined(SQLITE_TEST) && defined(SQLITE_DEBUG) 410 /* 411 ** Translate UTF-8 to UTF-8. 412 ** 413 ** This has the effect of making sure that the string is well-formed 414 ** UTF-8. Miscoded characters are removed. 415 ** 416 ** The translation is done in-place and aborted if the output 417 ** overruns the input. 418 */ 419 int sqlite3Utf8To8(unsigned char *zIn){ 420 unsigned char *zOut = zIn; 421 unsigned char *zStart = zIn; 422 u32 c; 423 424 while( zIn[0] && zOut<=zIn ){ 425 c = sqlite3Utf8Read((const u8**)&zIn); 426 if( c!=0xfffd ){ 427 WRITE_UTF8(zOut, c); 428 } 429 } 430 *zOut = 0; 431 return (int)(zOut - zStart); 432 } 433 #endif 434 435 #ifndef SQLITE_OMIT_UTF16 436 /* 437 ** Convert a UTF-16 string in the native encoding into a UTF-8 string. 438 ** Memory to hold the UTF-8 string is obtained from sqlite3_malloc and must 439 ** be freed by the calling function. 440 ** 441 ** NULL is returned if there is an allocation error. 442 */ 443 char *sqlite3Utf16to8(sqlite3 *db, const void *z, int nByte, u8 enc){ 444 Mem m; 445 memset(&m, 0, sizeof(m)); 446 m.db = db; 447 sqlite3VdbeMemSetStr(&m, z, nByte, enc, SQLITE_STATIC); 448 sqlite3VdbeChangeEncoding(&m, SQLITE_UTF8); 449 if( db->mallocFailed ){ 450 sqlite3VdbeMemRelease(&m); 451 m.z = 0; 452 } 453 assert( (m.flags & MEM_Term)!=0 || db->mallocFailed ); 454 assert( (m.flags & MEM_Str)!=0 || db->mallocFailed ); 455 assert( m.z || db->mallocFailed ); 456 return m.z; 457 } 458 459 /* 460 ** zIn is a UTF-16 encoded unicode string at least nChar characters long. 461 ** Return the number of bytes in the first nChar unicode characters 462 ** in pZ. nChar must be non-negative. 463 */ 464 int sqlite3Utf16ByteLen(const void *zIn, int nChar){ 465 int c; 466 unsigned char const *z = zIn; 467 int n = 0; 468 469 if( SQLITE_UTF16NATIVE==SQLITE_UTF16BE ){ 470 while( n<nChar ){ 471 READ_UTF16BE(z, 1, c); 472 n++; 473 } 474 }else{ 475 while( n<nChar ){ 476 READ_UTF16LE(z, 1, c); 477 n++; 478 } 479 } 480 return (int)(z-(unsigned char const *)zIn); 481 } 482 483 #if defined(SQLITE_TEST) 484 /* 485 ** This routine is called from the TCL test function "translate_selftest". 486 ** It checks that the primitives for serializing and deserializing 487 ** characters in each encoding are inverses of each other. 488 */ 489 void sqlite3UtfSelfTest(void){ 490 unsigned int i, t; 491 unsigned char zBuf[20]; 492 unsigned char *z; 493 int n; 494 unsigned int c; 495 496 for(i=0; i<0x00110000; i++){ 497 z = zBuf; 498 WRITE_UTF8(z, i); 499 n = (int)(z-zBuf); 500 assert( n>0 && n<=4 ); 501 z[0] = 0; 502 z = zBuf; 503 c = sqlite3Utf8Read((const u8**)&z); 504 t = i; 505 if( i>=0xD800 && i<=0xDFFF ) t = 0xFFFD; 506 if( (i&0xFFFFFFFE)==0xFFFE ) t = 0xFFFD; 507 assert( c==t ); 508 assert( (z-zBuf)==n ); 509 } 510 for(i=0; i<0x00110000; i++){ 511 if( i>=0xD800 && i<0xE000 ) continue; 512 z = zBuf; 513 WRITE_UTF16LE(z, i); 514 n = (int)(z-zBuf); 515 assert( n>0 && n<=4 ); 516 z[0] = 0; 517 z = zBuf; 518 READ_UTF16LE(z, 1, c); 519 assert( c==i ); 520 assert( (z-zBuf)==n ); 521 } 522 for(i=0; i<0x00110000; i++){ 523 if( i>=0xD800 && i<0xE000 ) continue; 524 z = zBuf; 525 WRITE_UTF16BE(z, i); 526 n = (int)(z-zBuf); 527 assert( n>0 && n<=4 ); 528 z[0] = 0; 529 z = zBuf; 530 READ_UTF16BE(z, 1, c); 531 assert( c==i ); 532 assert( (z-zBuf)==n ); 533 } 534 } 535 #endif /* SQLITE_TEST */ 536 #endif /* SQLITE_OMIT_UTF16 */ 537