public bool MoveNext()
{
// We don't need to worry about tearing since this enumerator is a ref struct.
if (_currentCharPair > char.MaxValue)
{
// There was a surrogate pair smuggled in here from a previous operation.
// Shift out the high surrogate value and return immediately.
_currentCharPair >>= 16;
return(true);
}
if (_remainingUtf8Bytes.IsEmpty)
{
return(false);
}
// TODO_UTF8STRING: Since we assume Utf8String instances are well-formed, we may instead
// call an optimized version of the "decode" routine below which skips well-formedness checks.
OperationStatus status = Rune.DecodeFromUtf8(_remainingUtf8Bytes, out Rune currentRune, out int bytesConsumed);
Debug.Assert(status == OperationStatus.Done, "Somebody fed us invalid data?");
if (currentRune.IsBmp)
{
// Common case - BMP scalar value.
_currentCharPair = (uint)currentRune.Value;
}
else
{
// Uncommon case - supplementary plane (astral) scalar value.
// We'll smuggle the two UTF-16 code units into a single 32-bit value,
// with the leading surrogate packed into the low 16 bits of the value,
// and the trailing surrogate packed into the high 16 bits of the value.
UnicodeUtility.GetUtf16SurrogatesFromSupplementaryPlaneScalar((uint)currentRune.Value, out char leadingCodeUnit, out char trailingCodeUnit);
_currentCharPair = (uint)leadingCodeUnit + ((uint)trailingCodeUnit << 16);
}
// TODO_UTF8STRING: We can consider unsafe slicing below if we wish since we know we're
// not going to overrun the end of the span.
_remainingUtf8Bytes = _remainingUtf8Bytes.Slice(bytesConsumed);
return(true);
}
public bool MoveNext()
{
// We don't need to worry about tearing since this enumerator is a ref struct.
if (_remainingUtf8Bytes.IsEmpty)
{
return(false);
}
// TODO_UTF8STRING: Since we assume Utf8Span instances are well-formed, we may instead
// call an optimized version of the "decode" routine below which skips well-formedness checks.
OperationStatus status = Rune.DecodeFromUtf8(_remainingUtf8Bytes, out _currentRune, out int bytesConsumed);
Debug.Assert(status == OperationStatus.Done, "Somebody fed us invalid data?");
// TODO_UTF8STRING: We can consider unsafe slicing below if we wish since we know we're
// not going to overrun the end of the span.
_remainingUtf8Bytes = _remainingUtf8Bytes.Slice(bytesConsumed);
return(true);
}
internal sealed override OperationStatus DecodeFirstRune(ReadOnlySpan <byte> bytes, out Rune value, out int bytesConsumed)
{
return(Rune.DecodeFromUtf8(bytes, out value, out bytesConsumed));
}