src/utf.c

   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   int 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     char zBuf[100];
 219     sqlite3VdbeMemPrettyPrint(pMem, zBuf);
 220     fprintf(stderr, "INPUT:  %s\n", zBuf);
 221   }
 222 #endif
 223
 224   /* If the translation is between UTF-16 little and big endian, then
 225   ** all that is required is to swap the byte order. This case is handled
 226   ** differently from the others.
 227   */
 228   if( pMem->enc!=SQLITE_UTF8 && desiredEnc!=SQLITE_UTF8 ){
 229     u8 temp;
 230     int rc;
 231     rc = sqlite3VdbeMemMakeWriteable(pMem);
 232     if( rc!=SQLITE_OK ){
 233       assert( rc==SQLITE_NOMEM );
 234       return SQLITE_NOMEM_BKPT;
 235     }
 236     zIn = (u8*)pMem->z;
 237     zTerm = &zIn[pMem->n&~1];
 238     while( zIn<zTerm ){
 239       temp = *zIn;
 240       *zIn = *(zIn+1);
 241       zIn++;
 242       *zIn++ = temp;
 243     }
 244     pMem->enc = desiredEnc;
 245     goto translate_out;
 246   }
 247
 248   /* Set len to the maximum number of bytes required in the output buffer. */
 249   if( desiredEnc==SQLITE_UTF8 ){
 250     /* When converting from UTF-16, the maximum growth results from
 251     ** translating a 2-byte character to a 4-byte UTF-8 character.
 252     ** A single byte is required for the output string
 253     ** nul-terminator.
 254     */
 255     pMem->n &= ~1;
 256     len = pMem->n * 2 + 1;
 257   }else{
 258     /* When converting from UTF-8 to UTF-16 the maximum growth is caused
 259     ** when a 1-byte UTF-8 character is translated into a 2-byte UTF-16
 260     ** character. Two bytes are required in the output buffer for the
 261     ** nul-terminator.
 262     */
 263     len = pMem->n * 2 + 2;
 264   }
 265
 266   /* Set zIn to point at the start of the input buffer and zTerm to point 1
 267   ** byte past the end.
 268   **
 269   ** Variable zOut is set to point at the output buffer, space obtained
 270   ** from sqlite3_malloc().
 271   */
 272   zIn = (u8*)pMem->z;
 273   zTerm = &zIn[pMem->n];
 274   zOut = sqlite3DbMallocRaw(pMem->db, len);
 275   if( !zOut ){
 276     return SQLITE_NOMEM_BKPT;
 277   }
 278   z = zOut;
 279
 280   if( pMem->enc==SQLITE_UTF8 ){
 281     if( desiredEnc==SQLITE_UTF16LE ){
 282       /* UTF-8 -> UTF-16 Little-endian */
 283       while( zIn<zTerm ){
 284         READ_UTF8(zIn, zTerm, c);
 285         WRITE_UTF16LE(z, c);
 286       }
 287     }else{
 288       assert( desiredEnc==SQLITE_UTF16BE );
 289       /* UTF-8 -> UTF-16 Big-endian */
 290       while( zIn<zTerm ){
 291         READ_UTF8(zIn, zTerm, c);
 292         WRITE_UTF16BE(z, c);
 293       }
 294     }
 295     pMem->n = (int)(z - zOut);
 296     *z++ = 0;
 297   }else{
 298     assert( desiredEnc==SQLITE_UTF8 );
 299     if( pMem->enc==SQLITE_UTF16LE ){
 300       /* UTF-16 Little-endian -> UTF-8 */
 301       while( zIn<zTerm ){
 302         READ_UTF16LE(zIn, zIn<zTerm, c);
 303         WRITE_UTF8(z, c);
 304       }
 305     }else{
 306       /* UTF-16 Big-endian -> UTF-8 */
 307       while( zIn<zTerm ){
 308         READ_UTF16BE(zIn, zIn<zTerm, c);
 309         WRITE_UTF8(z, c);
 310       }
 311     }
 312     pMem->n = (int)(z - zOut);
 313   }
 314   *z = 0;
 315   assert( (pMem->n+(desiredEnc==SQLITE_UTF8?1:2))<=len );
 316
 317   c = pMem->flags;
 318   sqlite3VdbeMemRelease(pMem);
 319   pMem->flags = MEM_Str|MEM_Term|(c&(MEM_AffMask|MEM_Subtype));
 320   pMem->enc = desiredEnc;
 321   pMem->z = (char*)zOut;
 322   pMem->zMalloc = pMem->z;
 323   pMem->szMalloc = sqlite3DbMallocSize(pMem->db, pMem->z);
 324
 325 translate_out:
 326 #if defined(TRANSLATE_TRACE) && defined(SQLITE_DEBUG)
 327   {
 328     char zBuf[100];
 329     sqlite3VdbeMemPrettyPrint(pMem, zBuf);
 330     fprintf(stderr, "OUTPUT: %s\n", zBuf);
 331   }
 332 #endif
 333   return SQLITE_OK;
 334 }
 335 #endif /* SQLITE_OMIT_UTF16 */
 336
 337 #ifndef SQLITE_OMIT_UTF16
 338 /*
 339 ** This routine checks for a byte-order mark at the beginning of the
 340 ** UTF-16 string stored in *pMem. If one is present, it is removed and
 341 ** the encoding of the Mem adjusted. This routine does not do any
 342 ** byte-swapping, it just sets Mem.enc appropriately.
 343 **
 344 ** The allocation (static, dynamic etc.) and encoding of the Mem may be
 345 ** changed by this function.
 346 */
 347 int sqlite3VdbeMemHandleBom(Mem *pMem){
 348   int rc = SQLITE_OK;
 349   u8 bom = 0;
 350
 351   assert( pMem->n>=0 );
 352   if( pMem->n>1 ){
 353     u8 b1 = *(u8 *)pMem->z;
 354     u8 b2 = *(((u8 *)pMem->z) + 1);
 355     if( b1==0xFE && b2==0xFF ){
 356       bom = SQLITE_UTF16BE;
 357     }
 358     if( b1==0xFF && b2==0xFE ){
 359       bom = SQLITE_UTF16LE;
 360     }
 361   }
 362
 363   if( bom ){
 364     rc = sqlite3VdbeMemMakeWriteable(pMem);
 365     if( rc==SQLITE_OK ){
 366       pMem->n -= 2;
 367       memmove(pMem->z, &pMem->z[2], pMem->n);
 368       pMem->z[pMem->n] = '\0';
 369       pMem->z[pMem->n+1] = '\0';
 370       pMem->flags |= MEM_Term;
 371       pMem->enc = bom;
 372     }
 373   }
 374   return rc;
 375 }
 376 #endif /* SQLITE_OMIT_UTF16 */
 377
 378 /*
 379 ** pZ is a UTF-8 encoded unicode string. If nByte is less than zero,
 380 ** return the number of unicode characters in pZ up to (but not including)
 381 ** the first 0x00 byte. If nByte is not less than zero, return the
 382 ** number of unicode characters in the first nByte of pZ (or up to
 383 ** the first 0x00, whichever comes first).
 384 */
 385 int sqlite3Utf8CharLen(const char *zIn, int nByte){
 386   int r = 0;
 387   const u8 *z = (const u8*)zIn;
 388   const u8 *zTerm;
 389   if( nByte>=0 ){
 390     zTerm = &z[nByte];
 391   }else{
 392     zTerm = (const u8*)(-1);
 393   }
 394   assert( z<=zTerm );
 395   while( *z!=0 && z<zTerm ){
 396     SQLITE_SKIP_UTF8(z);
 397     r++;
 398   }
 399   return r;
 400 }
 401
 402 /* This test function is not currently used by the automated test-suite.
 403 ** Hence it is only available in debug builds.
 404 */
 405 #if defined(SQLITE_TEST) && defined(SQLITE_DEBUG)
 406 /*
 407 ** Translate UTF-8 to UTF-8.
 408 **
 409 ** This has the effect of making sure that the string is well-formed
 410 ** UTF-8.  Miscoded characters are removed.
 411 **
 412 ** The translation is done in-place and aborted if the output
 413 ** overruns the input.
 414 */
 415 int sqlite3Utf8To8(unsigned char *zIn){
 416   unsigned char *zOut = zIn;
 417   unsigned char *zStart = zIn;
 418   u32 c;
 419
 420   while( zIn[0] && zOut<=zIn ){
 421     c = sqlite3Utf8Read((const u8**)&zIn);
 422     if( c!=0xfffd ){
 423       WRITE_UTF8(zOut, c);
 424     }
 425   }
 426   *zOut = 0;
 427   return (int)(zOut - zStart);
 428 }
 429 #endif
 430
 431 #ifndef SQLITE_OMIT_UTF16
 432 /*
 433 ** Convert a UTF-16 string in the native encoding into a UTF-8 string.
 434 ** Memory to hold the UTF-8 string is obtained from sqlite3_malloc and must
 435 ** be freed by the calling function.
 436 **
 437 ** NULL is returned if there is an allocation error.
 438 */
 439 char *sqlite3Utf16to8(sqlite3 *db, const void *z, int nByte, u8 enc){
 440   Mem m;
 441   memset(&m, 0, sizeof(m));
 442   m.db = db;
 443   sqlite3VdbeMemSetStr(&m, z, nByte, enc, SQLITE_STATIC);
 444   sqlite3VdbeChangeEncoding(&m, SQLITE_UTF8);
 445   if( db->mallocFailed ){
 446     sqlite3VdbeMemRelease(&m);
 447     m.z = 0;
 448   }
 449   assert( (m.flags & MEM_Term)!=0 || db->mallocFailed );
 450   assert( (m.flags & MEM_Str)!=0 || db->mallocFailed );
 451   assert( m.z || db->mallocFailed );
 452   return m.z;
 453 }
 454
 455 /*
 456 ** zIn is a UTF-16 encoded unicode string at least nChar characters long.
 457 ** Return the number of bytes in the first nChar unicode characters
 458 ** in pZ.  nChar must be non-negative.
 459 */
 460 int sqlite3Utf16ByteLen(const void *zIn, int nChar){
 461   int c;
 462   unsigned char const *z = zIn;
 463   int n = 0;
 464
 465   if( SQLITE_UTF16NATIVE==SQLITE_UTF16BE ){
 466     while( n<nChar ){
 467       READ_UTF16BE(z, 1, c);
 468       n++;
 469     }
 470   }else{
 471     while( n<nChar ){
 472       READ_UTF16LE(z, 1, c);
 473       n++;
 474     }
 475   }
 476   return (int)(z-(unsigned char const *)zIn);
 477 }
 478
 479 #if defined(SQLITE_TEST)
 480 /*
 481 ** This routine is called from the TCL test function "translate_selftest".
 482 ** It checks that the primitives for serializing and deserializing
 483 ** characters in each encoding are inverses of each other.
 484 */
 485 void sqlite3UtfSelfTest(void){
 486   unsigned int i, t;
 487   unsigned char zBuf[20];
 488   unsigned char *z;
 489   int n;
 490   unsigned int c;
 491
 492   for(i=0; i<0x00110000; i++){
 493     z = zBuf;
 494     WRITE_UTF8(z, i);
 495     n = (int)(z-zBuf);
 496     assert( n>0 && n<=4 );
 497     z[0] = 0;
 498     z = zBuf;
 499     c = sqlite3Utf8Read((const u8**)&z);
 500     t = i;
 501     if( i>=0xD800 && i<=0xDFFF ) t = 0xFFFD;
 502     if( (i&0xFFFFFFFE)==0xFFFE ) t = 0xFFFD;
 503     assert( c==t );
 504     assert( (z-zBuf)==n );
 505   }
 506   for(i=0; i<0x00110000; i++){
 507     if( i>=0xD800 && i<0xE000 ) continue;
 508     z = zBuf;
 509     WRITE_UTF16LE(z, i);
 510     n = (int)(z-zBuf);
 511     assert( n>0 && n<=4 );
 512     z[0] = 0;
 513     z = zBuf;
 514     READ_UTF16LE(z, 1, c);
 515     assert( c==i );
 516     assert( (z-zBuf)==n );
 517   }
 518   for(i=0; i<0x00110000; i++){
 519     if( i>=0xD800 && i<0xE000 ) continue;
 520     z = zBuf;
 521     WRITE_UTF16BE(z, i);
 522     n = (int)(z-zBuf);
 523     assert( n>0 && n<=4 );
 524     z[0] = 0;
 525     z = zBuf;
 526     READ_UTF16BE(z, 1, c);
 527     assert( c==i );
 528     assert( (z-zBuf)==n );
 529   }
 530 }
 531 #endif /* SQLITE_TEST */
 532 #endif /* SQLITE_OMIT_UTF16 */