/*
** pZ is a UTF-16 encoded unicode string. If nChar is less than zero,
** return the number of bytes up to (but not including), the first pair
** of consecutive 0x00 bytes in pZ. If nChar is not less than zero,
** then return the number of bytes in the first nChar unicode characters
** in pZ (or up until the first pair of 0x00 bytes, whichever comes first).
*/
int sqlite3utf16ByteLen(const void *zIn, int nChar){
unsigned int c = 1;
char const *z = zIn;
int n = 0;
if( SQLITE_UTF16NATIVE==SQLITE_UTF16BE ){
/* Using an "if (SQLITE_UTF16NATIVE==SQLITE_UTF16BE)" construct here
** and in other parts of this file means that at one branch will
** not be covered by coverage testing on any single host. But coverage
** will be complete if the tests are run on both a little-endian and
** big-endian host. Because both the UTF16NATIVE and SQLITE_UTF16BE
** macros are constant at compile time the compiler can determine
** which branch will be followed. It is therefore assumed that no runtime
** penalty is paid for this "if" statement.
*/
while( c && ((nChar<0) || n<nChar) ){
READ_UTF16BE(z, c);
n++;
}
}else{
while( c && ((nChar<0) || n<nChar) ){
READ_UTF16LE(z, c);
n++;
}
}
return (z-(char const *)zIn)-((c==0)?2:0);
}
/*
** This routine is called from the TCL test function "translate_selftest".
** It checks that the primitives for serializing and deserializing
** characters in each encoding are inverses of each other.
*/
void sqlite3UtfSelfTest(void){
unsigned int i, t;
unsigned char zBuf[20];
unsigned char *z;
int n;
unsigned int c;
for(i=0; i<0x00110000; i++){
z = zBuf;
WRITE_UTF8(z, i);
n = (int)(z-zBuf);
assert( n>0 && n<=4 );
z[0] = 0;
z = zBuf;
c = sqlite3Utf8Read((const u8**)&z);
t = i;
if( i>=0xD800 && i<=0xDFFF ) t = 0xFFFD;
if( (i&0xFFFFFFFE)==0xFFFE ) t = 0xFFFD;
assert( c==t );
assert( (z-zBuf)==n );
}
for(i=0; i<0x00110000; i++){
if( i>=0xD800 && i<0xE000 ) continue;
z = zBuf;
WRITE_UTF16LE(z, i);
n = (int)(z-zBuf);
assert( n>0 && n<=4 );
z[0] = 0;
z = zBuf;
READ_UTF16LE(z, 1, c);
assert( c==i );
assert( (z-zBuf)==n );
}
for(i=0; i<0x00110000; i++){
if( i>=0xD800 && i<0xE000 ) continue;
z = zBuf;
WRITE_UTF16BE(z, i);
n = (int)(z-zBuf);
assert( n>0 && n<=4 );
z[0] = 0;
z = zBuf;
READ_UTF16BE(z, 1, c);
assert( c==i );
assert( (z-zBuf)==n );
}
}
/*
** pZ is a UTF-16 encoded unicode string. If nChar is less than zero,
** return the number of bytes up to (but not including), the first pair
** of consecutive 0x00 bytes in pZ. If nChar is not less than zero,
** then return the number of bytes in the first nChar unicode characters
** in pZ (or up until the first pair of 0x00 bytes, whichever comes first).
*/
int sqlite3utf16ByteLen(const void *zIn, int nChar){
int c = 1;
char const *z = (const char*)zIn;
int n = 0;
if( SQLITE_UTF16NATIVE==SQLITE_UTF16BE ){
while( c && ((nChar<0) || n<nChar) ){
READ_UTF16BE(z, c);
n++;
}
}else{
while( c && ((nChar<0) || n<nChar) ){
READ_UTF16LE(z, c);
n++;
}
}
return (z-(char const *)zIn)-((c==0)?2:0);
}
/*
** zIn is a UTF-16 encoded unicode string at least nChar characters long.
** Return the number of bytes in the first nChar unicode characters
** in pZ. nChar must be non-negative.
*/
int sqlite3Utf16ByteLen(const void *zIn, int nChar){
int c;
unsigned char const *z = zIn;
int n = 0;
if( SQLITE_UTF16NATIVE==SQLITE_UTF16BE ){
while( n<nChar ){
READ_UTF16BE(z, 1, c);
n++;
}
}else{
while( n<nChar ){
READ_UTF16LE(z, 1, c);
n++;
}
}
return (int)(z-(unsigned char const *)zIn);
}
/*
** This routine is called from the TCL test function "translate_selftest".
** It checks that the primitives for serializing and deserializing
** characters in each encoding are inverses of each other.
*/
void sqlite3UtfSelfTest(){
unsigned int i, t;
unsigned char zBuf[20];
unsigned char *z;
int n;
unsigned int c;
for(i=0; i<0x00110000; i++){
z = zBuf;
WRITE_UTF8(z, i);
n = z-zBuf;
z[0] = 0;
z = zBuf;
SQLITE_READ_UTF8(z, c);
t = i;
if( i>=0xD800 && i<=0xDFFF ) t = 0xFFFD;
if( (i&0xFFFFFFFE)==0xFFFE ) t = 0xFFFD;
assert( c==t );
assert( (z-zBuf)==n );
}
for(i=0; i<0x00110000; i++){
if( i>=0xD800 && i<0xE000 ) continue;
z = zBuf;
WRITE_UTF16LE(z, i);
n = z-zBuf;
z[0] = 0;
z = zBuf;
READ_UTF16LE(z, c);
assert( c==i );
assert( (z-zBuf)==n );
}
for(i=0; i<0x00110000; i++){
if( i>=0xD800 && i<0xE000 ) continue;
z = zBuf;
WRITE_UTF16BE(z, i);
n = z-zBuf;
z[0] = 0;
z = zBuf;
READ_UTF16BE(z, c);
assert( c==i );
assert( (z-zBuf)==n );
}
}
/*
** This routine is called from the TCL test function "translate_selftest".
** It checks that the primitives for serializing and deserializing
** characters in each encoding are inverses of each other.
*/
void sqlite3utfSelfTest(){
int i;
unsigned char zBuf[20];
unsigned char *z;
int n;
int c;
for(i=0; i<0x00110000; i++){
z = zBuf;
WRITE_UTF8(z, i);
n = z-zBuf;
z = zBuf;
READ_UTF8(z, c);
assert( c==i );
assert( (z-zBuf)==n );
}
for(i=0; i<0x00110000; i++){
if( i>=0xD800 && i<=0xE000 ) continue;
z = zBuf;
WRITE_UTF16LE(z, i);
n = z-zBuf;
z = zBuf;
READ_UTF16LE(z, c);
assert( c==i );
assert( (z-zBuf)==n );
}
for(i=0; i<0x00110000; i++){
if( i>=0xD800 && i<=0xE000 ) continue;
z = zBuf;
WRITE_UTF16BE(z, i);
n = z-zBuf;
z = zBuf;
READ_UTF16BE(z, c);
assert( c==i );
assert( (z-zBuf)==n );
}
}
/*
** This routine transforms the internal text encoding used by pMem to
** desiredEnc. It is an error if the string is already of the desired
** encoding, or if *pMem does not contain a string value.
*/
int sqlite3VdbeMemTranslate(Mem *pMem, u8 desiredEnc){
unsigned char zShort[NBFS]; /* Temporary short output buffer */
int len; /* Maximum length of output string in bytes */
unsigned char *zOut; /* Output buffer */
unsigned char *zIn; /* Input iterator */
unsigned char *zTerm; /* End of input */
unsigned char *z; /* Output iterator */
unsigned int c;
assert( pMem->flags&MEM_Str );
assert( pMem->enc!=desiredEnc );
assert( pMem->enc!=0 );
assert( pMem->n>=0 );
#if defined(TRANSLATE_TRACE) && defined(SQLITE_DEBUG)
{
char zBuf[100];
sqlite3VdbeMemPrettyPrint(pMem, zBuf);
fprintf(stderr, "INPUT: %s\n", zBuf);
}
#endif
/* If the translation is between UTF-16 little and big endian, then
** all that is required is to swap the byte order. This case is handled
** differently from the others.
*/
if( pMem->enc!=SQLITE_UTF8 && desiredEnc!=SQLITE_UTF8 ){
u8 temp;
int rc;
rc = sqlite3VdbeMemMakeWriteable(pMem);
if( rc!=SQLITE_OK ){
assert( rc==SQLITE_NOMEM );
return SQLITE_NOMEM;
}
zIn = (u8*)pMem->z;
zTerm = &zIn[pMem->n];
while( zIn<zTerm ){
temp = *zIn;
*zIn = *(zIn+1);
zIn++;
*zIn++ = temp;
}
pMem->enc = desiredEnc;
goto translate_out;
}
/* Set len to the maximum number of bytes required in the output buffer. */
if( desiredEnc==SQLITE_UTF8 ){
/* When converting from UTF-16, the maximum growth results from
** translating a 2-byte character to a 4-byte UTF-8 character.
** A single byte is required for the output string
** nul-terminator.
*/
len = pMem->n * 2 + 1;
}else{
/* When converting from UTF-8 to UTF-16 the maximum growth is caused
** when a 1-byte UTF-8 character is translated into a 2-byte UTF-16
** character. Two bytes are required in the output buffer for the
** nul-terminator.
*/
len = pMem->n * 2 + 2;
}
/* Set zIn to point at the start of the input buffer and zTerm to point 1
** byte past the end.
**
** Variable zOut is set to point at the output buffer. This may be space
** obtained from malloc(), or Mem.zShort, if it large enough and not in
** use, or the zShort array on the stack (see above).
*/
zIn = (u8*)pMem->z;
zTerm = &zIn[pMem->n];
if( len>NBFS ){
zOut = sqliteMallocRaw(len);
if( !zOut ) return SQLITE_NOMEM;
}else{
zOut = zShort;
}
z = zOut;
if( pMem->enc==SQLITE_UTF8 ){
if( desiredEnc==SQLITE_UTF16LE ){
/* UTF-8 -> UTF-16 Little-endian */
while( zIn<zTerm ){
READ_UTF8(zIn, c);
WRITE_UTF16LE(z, c);
}
}else{
assert( desiredEnc==SQLITE_UTF16BE );
/* UTF-8 -> UTF-16 Big-endian */
while( zIn<zTerm ){
READ_UTF8(zIn, c);
WRITE_UTF16BE(z, c);
}
}
pMem->n = z - zOut;
*z++ = 0;
}else{
assert( desiredEnc==SQLITE_UTF8 );
if( pMem->enc==SQLITE_UTF16LE ){
/* UTF-16 Little-endian -> UTF-8 */
while( zIn<zTerm ){
READ_UTF16LE(zIn, c);
WRITE_UTF8(z, c);
}
}else{
/* UTF-16 Little-endian -> UTF-8 */
while( zIn<zTerm ){
READ_UTF16BE(zIn, c);
WRITE_UTF8(z, c);
}
}
pMem->n = z - zOut;
}
*z = 0;
assert( (pMem->n+(desiredEnc==SQLITE_UTF8?1:2))<=len );
sqlite3VdbeMemRelease(pMem);
pMem->flags &= ~(MEM_Static|MEM_Dyn|MEM_Ephem|MEM_Short);
pMem->enc = desiredEnc;
if( zOut==zShort ){
memcpy(pMem->zShort, zOut, len);
zOut = (u8*)pMem->zShort;
pMem->flags |= (MEM_Term|MEM_Short);
}else{
pMem->flags |= (MEM_Term|MEM_Dyn);
}
pMem->z = (char*)zOut;
translate_out:
#if defined(TRANSLATE_TRACE) && defined(SQLITE_DEBUG)
{
char zBuf[100];
sqlite3VdbeMemPrettyPrint(pMem, zBuf);
fprintf(stderr, "OUTPUT: %s\n", zBuf);
}
#endif
return SQLITE_OK;
}
/*
** This routine transforms the internal text encoding used by pMem to
** desiredEnc. It is an error if the string is already of the desired
** encoding, or if *pMem does not contain a string value.
*/
SQLITE_NOINLINE int sqlite3VdbeMemTranslate(Mem *pMem, u8 desiredEnc){
int len; /* Maximum length of output string in bytes */
unsigned char *zOut; /* Output buffer */
unsigned char *zIn; /* Input iterator */
unsigned char *zTerm; /* End of input */
unsigned char *z; /* Output iterator */
unsigned int c;
assert( pMem->db==0 || sqlite3_mutex_held(pMem->db->mutex) );
assert( pMem->flags&MEM_Str );
assert( pMem->enc!=desiredEnc );
assert( pMem->enc!=0 );
assert( pMem->n>=0 );
#if defined(TRANSLATE_TRACE) && defined(SQLITE_DEBUG)
{
char zBuf[100];
sqlite3VdbeMemPrettyPrint(pMem, zBuf);
fprintf(stderr, "INPUT: %s\n", zBuf);
}
#endif
/* If the translation is between UTF-16 little and big endian, then
** all that is required is to swap the byte order. This case is handled
** differently from the others.
*/
if( pMem->enc!=SQLITE_UTF8 && desiredEnc!=SQLITE_UTF8 ){
u8 temp;
int rc;
rc = sqlite3VdbeMemMakeWriteable(pMem);
if( rc!=SQLITE_OK ){
assert( rc==SQLITE_NOMEM );
return SQLITE_NOMEM;
}
zIn = (u8*)pMem->z;
zTerm = &zIn[pMem->n&~1];
while( zIn<zTerm ){
temp = *zIn;
*zIn = *(zIn+1);
zIn++;
*zIn++ = temp;
}
pMem->enc = desiredEnc;
goto translate_out;
}
/* Set len to the maximum number of bytes required in the output buffer. */
if( desiredEnc==SQLITE_UTF8 ){
/* When converting from UTF-16, the maximum growth results from
** translating a 2-byte character to a 4-byte UTF-8 character.
** A single byte is required for the output string
** nul-terminator.
*/
pMem->n &= ~1;
len = pMem->n * 2 + 1;
}else{
/* When converting from UTF-8 to UTF-16 the maximum growth is caused
** when a 1-byte UTF-8 character is translated into a 2-byte UTF-16
** character. Two bytes are required in the output buffer for the
** nul-terminator.
*/
len = pMem->n * 2 + 2;
}
/* Set zIn to point at the start of the input buffer and zTerm to point 1
** byte past the end.
**
** Variable zOut is set to point at the output buffer, space obtained
** from sqlite3_malloc().
*/
zIn = (u8*)pMem->z;
zTerm = &zIn[pMem->n];
zOut = sqlite3DbMallocRaw(pMem->db, len);
if( !zOut ){
return SQLITE_NOMEM;
}
z = zOut;
if( pMem->enc==SQLITE_UTF8 ){
if( desiredEnc==SQLITE_UTF16LE ){
/* UTF-8 -> UTF-16 Little-endian */
while( zIn<zTerm ){
READ_UTF8(zIn, zTerm, c);
WRITE_UTF16LE(z, c);
}
}else{
assert( desiredEnc==SQLITE_UTF16BE );
/* UTF-8 -> UTF-16 Big-endian */
while( zIn<zTerm ){
READ_UTF8(zIn, zTerm, c);
WRITE_UTF16BE(z, c);
}
}
pMem->n = (int)(z - zOut);
*z++ = 0;
}else{
assert( desiredEnc==SQLITE_UTF8 );
if( pMem->enc==SQLITE_UTF16LE ){
/* UTF-16 Little-endian -> UTF-8 */
while( zIn<zTerm ){
READ_UTF16LE(zIn, zIn<zTerm, c);
WRITE_UTF8(z, c);
}
}else{
/* UTF-16 Big-endian -> UTF-8 */
while( zIn<zTerm ){
READ_UTF16BE(zIn, zIn<zTerm, c);
WRITE_UTF8(z, c);
}
}
pMem->n = (int)(z - zOut);
}
*z = 0;
assert( (pMem->n+(desiredEnc==SQLITE_UTF8?1:2))<=len );
c = pMem->flags;
sqlite3VdbeMemRelease(pMem);
pMem->flags = MEM_Str|MEM_Term|(c&MEM_AffMask);
pMem->enc = desiredEnc;
pMem->z = (char*)zOut;
pMem->zMalloc = pMem->z;
pMem->szMalloc = sqlite3DbMallocSize(pMem->db, pMem->z);
translate_out:
#if defined(TRANSLATE_TRACE) && defined(SQLITE_DEBUG)
{
char zBuf[100];
sqlite3VdbeMemPrettyPrint(pMem, zBuf);
fprintf(stderr, "OUTPUT: %s\n", zBuf);
}
#endif
return SQLITE_OK;
}
/*
** This routine transforms the internal text encoding used by pMem to
** desiredEnc. It is an error if the string is already of the desired
** encoding, or if *pMem does not contain a string value.
*/
int sqlite3VdbeMemTranslate(Mem *pMem, u8 desiredEnc){
unsigned char zShort[NBFS]; /* Temporary short output buffer */
int len; /* Maximum length of output string in bytes */
unsigned char *zOut; /* Output buffer */
unsigned char *zIn; /* Input iterator */
unsigned char *zTerm; /* End of input */
unsigned char *z; /* Output iterator */
unsigned int c;
assert( pMem->flags&MEM_Str );
assert( pMem->enc!=desiredEnc );
assert( pMem->enc!=0 );
assert( pMem->n>=0 );
#if defined(TRANSLATE_TRACE) && defined(SQLITE_DEBUG)
{
char zBuf[100];
sqlite3VdbeMemPrettyPrint(pMem, zBuf);
fprintf(stderr, "INPUT: %s\n", zBuf);
}
#endif
/* If the translation is between UTF-16 little and big endian, then
** all that is required is to swap the byte order. This case is handled
** differently from the others.
*/
if( pMem->enc!=SQLITE_UTF8 && desiredEnc!=SQLITE_UTF8 ){
u8 temp;
int rc;
rc = sqlite3VdbeMemMakeWriteable(pMem);
if( rc!=SQLITE_OK ){
assert( rc==SQLITE_NOMEM );
return SQLITE_NOMEM;
}
zIn = (u8*)pMem->z;
zTerm = &zIn[pMem->n];
while( zIn<zTerm ){
temp = *zIn;
*zIn = *(zIn+1);
zIn++;
*zIn++ = temp;
}
pMem->enc = desiredEnc;
goto translate_out;
}
/* Set len to the maximum number of bytes required in the output buffer. */
if( desiredEnc==SQLITE_UTF8 ){
/* When converting from UTF-16, the maximum growth results from
** translating a 2-byte character to a 4-byte UTF-8 character.
** A single byte is required for the output string
** nul-terminator.
*/
len = pMem->n * 2 + 1;
}else{
/* When converting from UTF-8 to UTF-16 the maximum growth is caused
** when a 1-byte UTF-8 character is translated into a 2-byte UTF-16
** character. Two bytes are required in the output buffer for the
** nul-terminator.
*/
len = pMem->n * 2 + 2;
}
/* Set zIn to point at the start of the input buffer and zTerm to point 1
** byte past the end.
**
** Variable zOut is set to point at the output buffer. This may be space
** obtained from malloc(), or Mem.zShort, if it large enough and not in
** use, or the zShort array on the stack (see above).
*/
zIn = (u8*)pMem->z;
zTerm = &zIn[pMem->n];
if( len>NBFS ){
zOut = sqliteMallocRaw(len);
if( !zOut ) return SQLITE_NOMEM;
}else{
zOut = zShort;
}
z = zOut;
if( pMem->enc==SQLITE_UTF8 ){
unsigned int iExtra = 0xD800;
if( 0==(pMem->flags&MEM_Term) && zTerm>zIn && (zTerm[-1]&0x80) ){
/* This UTF8 string is not nul-terminated, and the last byte is
** not a character in the ascii range (codpoints 0..127). This
** means the SQLITE_READ_UTF8() macro might read past the end
** of the allocated buffer.
**
** There are four possibilities:
**
** 1. The last byte is the first byte of a non-ASCII character,
**
** 2. The final N bytes of the input string are continuation bytes
** and immediately preceding them is the first byte of a
** non-ASCII character.
**
** 3. The final N bytes of the input string are continuation bytes
** and immediately preceding them is a byte that encodes a
** character in the ASCII range.
**
** 4. The entire string consists of continuation characters.
**
** Cases (3) and (4) require no special handling. The SQLITE_READ_UTF8()
** macro will not overread the buffer in these cases.
*/
unsigned char *zExtra = &zTerm[-1];
while( zExtra>zIn && (zExtra[0]&0xC0)==0x80 ){
zExtra--;
}
if( (zExtra[0]&0xC0)==0xC0 ){
/* Make a copy of the last character encoding in the input string.
** Then make sure it is nul-terminated and use SQLITE_READ_UTF8()
** to decode the codepoint. Store the codepoint in variable iExtra,
** it will be appended to the output string later.
*/
unsigned char *zFree = 0;
unsigned char zBuf[16];
int nExtra = (pMem->n+zIn-zExtra);
zTerm = zExtra;
if( nExtra>15 ){
zExtra = sqliteMallocRaw(nExtra+1);
if( !zExtra ){
return SQLITE_NOMEM;
}
zFree = zExtra;
}else{
zExtra = zBuf;
}
memcpy(zExtra, zTerm, nExtra);
zExtra[nExtra] = '\0';
SQLITE_READ_UTF8(zExtra, iExtra);
sqliteFree(zFree);
}
}
if( desiredEnc==SQLITE_UTF16LE ){
/* UTF-8 -> UTF-16 Little-endian */
while( zIn<zTerm ){
SQLITE_READ_UTF8(zIn, c);
WRITE_UTF16LE(z, c);
}
if( iExtra!=0xD800 ){
WRITE_UTF16LE(z, iExtra);
}
}else{
assert( desiredEnc==SQLITE_UTF16BE );
/* UTF-8 -> UTF-16 Big-endian */
while( zIn<zTerm ){
SQLITE_READ_UTF8(zIn, c);
WRITE_UTF16BE(z, c);
}
if( iExtra!=0xD800 ){
WRITE_UTF16BE(z, iExtra);
}
}
pMem->n = z - zOut;
*z++ = 0;
}else{
assert( desiredEnc==SQLITE_UTF8 );
if( pMem->enc==SQLITE_UTF16LE ){
/* UTF-16 Little-endian -> UTF-8 */
while( zIn<zTerm ){
READ_UTF16LE(zIn, c);
WRITE_UTF8(z, c);
}
}else{
/* UTF-16 Little-endian -> UTF-8 */
while( zIn<zTerm ){
READ_UTF16BE(zIn, c);
WRITE_UTF8(z, c);
}
}
pMem->n = z - zOut;
}
*z = 0;
assert( (pMem->n+(desiredEnc==SQLITE_UTF8?1:2))<=len );
sqlite3VdbeMemRelease(pMem);
pMem->flags &= ~(MEM_Static|MEM_Dyn|MEM_Ephem|MEM_Short);
pMem->enc = desiredEnc;
if( zOut==zShort ){
memcpy(pMem->zShort, zOut, len);
zOut = (u8*)pMem->zShort;
pMem->flags |= (MEM_Term|MEM_Short);
}else{
pMem->flags |= (MEM_Term|MEM_Dyn);
}
pMem->z = (char*)zOut;
translate_out:
#if defined(TRANSLATE_TRACE) && defined(SQLITE_DEBUG)
{
char zBuf[100];
sqlite3VdbeMemPrettyPrint(pMem, zBuf);
fprintf(stderr, "OUTPUT: %s\n", zBuf);
}
#endif
return SQLITE_OK;
}