The source code for Apache Xerces: ConvertUTF16toUCS4 is:
ConversionResult ConvertUTF16toUCS4(
UTF16 **sourceStart, UTF16 *sourceEnd,
UCS4 **targetStart, const UCS4 *targetEnd)
{
ConversionResult result = ok;
register UTF16 *source = *sourceStart;
register UCS4 *target = *targetStart;
while (source < sourceEnd)
{
register UCS4 ch;
ch = *source++;
if (ch >= kSurrogateHighStart && ch <= kSurrogateHighEnd && source < sourceEnd)
{
register UCS4 ch2 = *source;
if (ch2 >= kSurrogateLowStart && ch2 <= kSurrogateLowEnd)
{
ch = ((ch - kSurrogateHighStart) << halfShift) + (ch2 - kSurrogateLowStart) + halfBase;
++source;
};
};
if (target >= targetEnd)
{
result = targetExhausted;
break;
};
*target++ = ch;
};
*sourceStart = source;
*targetStart = target;
return result;
};
I am trying to convert a UTF16 encoded surrogate pairs into UCS4 encoded data. I am using WindowsOS and a little endian machine.
If you look closely, you can see that, after the conversion, they are assigning
target to *targetStart
. So wouldn't it be pointing to the last element of the target instead of the first element of the target?
When I remove the statement *targetStart = target;
from my code, it is working as expected. Is this a bug in the API or am I missing something?