/*
** This routine checks for a byte-order mark at the beginning of the
** UTF-16 string stored in *pMem. If one is present, it is removed and
** the encoding of the Mem adjusted. This routine does not do any
** byte-swapping, it just sets Mem.enc appropriately.
**
** The allocation (static, dynamic etc.) and encoding of the Mem may be
** changed by this function.
*/
int sqlite3VdbeMemHandleBom(Mem *pMem){
int rc = SQLITE_OK;
u8 bom = 0;
if( pMem->n<0 || pMem->n>1 ){
u8 b1 = *(u8 *)pMem->z;
u8 b2 = *(((u8 *)pMem->z) + 1);
if( b1==0xFE && b2==0xFF ){
bom = SQLITE_UTF16BE;
}
if( b1==0xFF && b2==0xFE ){
bom = SQLITE_UTF16LE;
}
}
if( bom ){
rc = sqlite3VdbeMemMakeWriteable(pMem);
if( rc==SQLITE_OK ){
pMem->n -= 2;
memmove(pMem->z, &pMem->z[2], pMem->n);
pMem->z[pMem->n] = '\0';
pMem->z[pMem->n+1] = '\0';
pMem->flags |= MEM_Term;
pMem->enc = bom;
}
}
return rc;
}
/* This function is only available internally, it is not part of the
** external API. It works in a similar way to sqlite3_value_text(),
** except the data returned is in the encoding specified by the second
** parameter, which must be one of SQLITE_UTF16BE, SQLITE_UTF16LE or
** SQLITE_UTF8.
**
** (2006-02-16:) The enc value can be or-ed with SQLITE_UTF16_ALIGNED.
** If that is the case, then the result must be aligned on an even byte
** boundary.
*/
const void *sqlite3ValueText(sqlite3_value* pVal, u8 enc){
if( !pVal ) return 0;
assert( pVal->db==0 || sqlite3_mutex_held(pVal->db->mutex) );
assert( (enc&3)==(enc&~SQLITE_UTF16_ALIGNED) );
if( pVal->flags&MEM_Null ){
return 0;
}
assert( (MEM_Blob>>3) == MEM_Str );
pVal->flags |= (pVal->flags & MEM_Blob)>>3;
expandBlob(pVal);
if( pVal->flags&MEM_Str ){
sqlite3VdbeChangeEncoding(pVal, enc & ~SQLITE_UTF16_ALIGNED);
if( (enc & SQLITE_UTF16_ALIGNED)!=0 && 1==(1&SQLITE_PTR_TO_INT(pVal->z)) ){
assert( (pVal->flags & (MEM_Ephem|MEM_Static))!=0 );
if( sqlite3VdbeMemMakeWriteable(pVal)!=SQLITE_OK ){
return 0;
}
}
sqlite3VdbeMemNulTerminate(pVal);
}else{
assert( (pVal->flags&MEM_Blob)==0 );
sqlite3VdbeMemStringify(pVal, enc);
assert( 0==(1&(int)pVal->z) );
}
assert(pVal->enc==(enc & ~SQLITE_UTF16_ALIGNED) || pVal->db==0
|| pVal->db->mallocFailed );
if( pVal->enc==(enc & ~SQLITE_UTF16_ALIGNED) ){
return pVal->z;
}else{
return 0;
}
}
/*
** Make sure the given Mem is \u0000 terminated.
*/
int sqlite3VdbeMemNulTerminate(Mem *pMem){
assert( pMem->db==0 || sqlite3_mutex_held(pMem->db->mutex) );
if( (pMem->flags & MEM_Term)!=0 || (pMem->flags & MEM_Str)==0 ){
return SQLITE_OK; /* Nothing to do */
}
if( pMem->flags & (MEM_Static|MEM_Ephem) ){
return sqlite3VdbeMemMakeWriteable(pMem);
}else{
char *z;
sqlite3VdbeMemExpandBlob(pMem);
z = sqlite3DbMallocRaw(pMem->db, pMem->n+2);
if( !z ){
return SQLITE_NOMEM;
}
memcpy(z, pMem->z, pMem->n);
z[pMem->n] = 0;
z[pMem->n+1] = 0;
if( pMem->xDel ){
pMem->xDel(pMem->z);
}else{
sqlite3_free(pMem->z);
}
pMem->xDel = 0;
pMem->z = z;
pMem->flags |= MEM_Term;
}
return SQLITE_OK;
}
/*
** Make sure the given Mem is \u0000 terminated.
*/
int sqlite3VdbeMemNulTerminate(Mem *pMem) {
/* In SQLite, a string without a nul terminator occurs when a string
** is loaded from disk (in this case the memory management is ephemeral),
** or when it is supplied by the user as a bound variable or function
** return value. Therefore, the memory management of the string must be
** either ephemeral, static or controlled by a user-supplied destructor.
*/
assert(
!(pMem->flags&MEM_Str) || /* it's not a string, or */
(pMem->flags&MEM_Term) || /* it's nul term. already, or */
(pMem->flags&(MEM_Ephem|MEM_Static)) || /* it's static or ephem, or */
(pMem->flags&MEM_Dyn && pMem->xDel) /* external management */
);
if( (pMem->flags & MEM_Term)!=0 || (pMem->flags & MEM_Str)==0 ) {
return SQLITE_OK; /* Nothing to do */
}
if( pMem->flags & (MEM_Static|MEM_Ephem) ) {
return sqlite3VdbeMemMakeWriteable(pMem);
} else {
char *z = sqliteMalloc(pMem->n+2);
if( !z ) return SQLITE_NOMEM;
memcpy(z, pMem->z, pMem->n);
z[pMem->n] = 0;
z[pMem->n+1] = 0;
pMem->xDel(pMem->z);
pMem->xDel = 0;
pMem->z = z;
}
return SQLITE_OK;
}
/*
** Make a full copy of pFrom into pTo. Prior contents of pTo are
** freed before the copy is made.
*/
int sqlite3VdbeMemCopy(Mem *pTo, const Mem *pFrom){
int rc;
if( pTo->flags & MEM_Dyn ){
sqlite3VdbeMemRelease(pTo);
}
sqlite3VdbeMemShallowCopy(pTo, pFrom, MEM_Ephem);
if( pTo->flags & MEM_Ephem ){
rc = sqlite3VdbeMemMakeWriteable(pTo);
}else{
rc = SQLITE_OK;
}
return rc;
}
/*
** Make a full copy of pFrom into pTo. Prior contents of pTo are
** freed before the copy is made.
*/
int sqlite3VdbeMemCopy(Mem *pTo, const Mem *pFrom){
int rc = SQLITE_OK;
sqlite3VdbeMemReleaseExternal(pTo);
memcpy(pTo, pFrom, MEMCELLSIZE);
pTo->flags &= ~MEM_Dyn;
if( pTo->flags&(MEM_Str|MEM_Blob) ){
if( 0==(pFrom->flags&MEM_Static) ){
pTo->flags |= MEM_Ephem;
rc = sqlite3VdbeMemMakeWriteable(pTo);
}
}
return rc;
}
/*
** Transfer the contents of pFrom to pTo. Any existing value in pTo is
** freed. If pFrom contains ephemeral data, a copy is made.
**
** pFrom contains an SQL NULL when this routine returns. SQLITE_NOMEM
** might be returned if pFrom held ephemeral data and we were unable
** to allocate enough space to make a copy.
*/
int sqlite3VdbeMemMove(Mem *pTo, Mem *pFrom) {
int rc;
if( pTo->flags & MEM_Dyn ) {
sqlite3VdbeMemRelease(pTo);
}
memcpy(pTo, pFrom, sizeof(Mem));
if( pFrom->flags & MEM_Short ) {
pTo->z = pTo->zShort;
}
pFrom->flags = MEM_Null;
pFrom->xDel = 0;
if( pTo->flags & MEM_Ephem ) {
rc = sqlite3VdbeMemMakeWriteable(pTo);
} else {
rc = SQLITE_OK;
}
return rc;
}
/* Make a copy of an sqlite3_value object
*/
sqlite3_value *sqlite3_value_dup(const sqlite3_value *pOrig){
sqlite3_value *pNew;
if( pOrig==0 ) return 0;
pNew = sqlite3_malloc( sizeof(*pNew) );
if( pNew==0 ) return 0;
memset(pNew, 0, sizeof(*pNew));
memcpy(pNew, pOrig, MEMCELLSIZE);
pNew->flags &= ~MEM_Dyn;
pNew->db = 0;
if( pNew->flags&(MEM_Str|MEM_Blob) ){
pNew->flags &= ~(MEM_Static|MEM_Dyn);
pNew->flags |= MEM_Ephem;
if( sqlite3VdbeMemMakeWriteable(pNew)!=SQLITE_OK ){
sqlite3ValueFree(pNew);
pNew = 0;
}
}
return pNew;
}
/*
** Make sure the given Mem is \u0000 terminated.
*/
int sqlite3VdbeMemNulTerminate(Mem *pMem){
if( (pMem->flags & MEM_Term)!=0 || (pMem->flags & MEM_Str)==0 ){
return SQLITE_OK; /* Nothing to do */
}
if( pMem->flags & (MEM_Static|MEM_Ephem) ){
return sqlite3VdbeMemMakeWriteable(pMem);
}else{
char *z = sqliteMalloc(pMem->n+2);
if( !z ) return SQLITE_NOMEM;
memcpy(z, pMem->z, pMem->n);
z[pMem->n] = 0;
z[pMem->n+1] = 0;
if( pMem->xDel ){
pMem->xDel(pMem->z);
}else{
sqliteFree(pMem->z);
}
pMem->xDel = 0;
pMem->z = z;
pMem->flags |= MEM_Term;
}
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 ){
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;
}
/*
** Change the value of a Mem to be a string or a BLOB.
*/
int sqlite3VdbeMemSetStr(
Mem *pMem, /* Memory cell to set to string value */
const char *z, /* String pointer */
int n, /* Bytes in string, or negative */
u8 enc, /* Encoding of z. 0 for BLOBs */
void (*xDel)(void*) /* Destructor function */
){
assert( pMem->db==0 || sqlite3_mutex_held(pMem->db->mutex) );
sqlite3VdbeMemRelease(pMem);
if( !z ){
pMem->flags = MEM_Null;
pMem->type = SQLITE_NULL;
return SQLITE_OK;
}
pMem->z = (char *)z;
if( xDel==SQLITE_STATIC ){
pMem->flags = MEM_Static;
}else if( xDel==SQLITE_TRANSIENT ){
pMem->flags = MEM_Ephem;
}else{
pMem->flags = MEM_Dyn;
pMem->xDel = xDel;
}
pMem->enc = enc;
pMem->type = enc==0 ? SQLITE_BLOB : SQLITE_TEXT;
pMem->n = n;
assert( enc==0 || enc==SQLITE_UTF8 || enc==SQLITE_UTF16LE
|| enc==SQLITE_UTF16BE );
switch( enc ){
case 0:
pMem->flags |= MEM_Blob;
pMem->enc = SQLITE_UTF8;
break;
case SQLITE_UTF8:
pMem->flags |= MEM_Str;
if( n<0 ){
pMem->n = strlen(z);
pMem->flags |= MEM_Term;
}
break;
#ifndef SQLITE_OMIT_UTF16
case SQLITE_UTF16LE:
case SQLITE_UTF16BE:
pMem->flags |= MEM_Str;
if( pMem->n<0 ){
pMem->n = sqlite3Utf16ByteLen(pMem->z,-1);
pMem->flags |= MEM_Term;
}
if( sqlite3VdbeMemHandleBom(pMem) ){
return SQLITE_NOMEM;
}
#endif /* SQLITE_OMIT_UTF16 */
}
if( pMem->flags&MEM_Ephem ){
return sqlite3VdbeMemMakeWriteable(pMem);
}
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;
}