/// <summary>
/// Creates a <see cref="Rune"/> from the provided UTF-16 code unit.
/// </summary>
/// <exception cref="ArgumentOutOfRangeException">
/// If <paramref name="ch"/> represents a UTF-16 surrogate code point
/// U+D800..U+DFFF, inclusive.
/// </exception>
public Rune(char ch)
{
uint expanded = ch;
if (UnicodeUtility.IsSurrogateCodePoint(expanded))
{
ThrowHelper.ThrowArgumentOutOfRangeException(ExceptionArgument.ch);
}
_value = expanded;
}
/// <summary>
/// Attempts to create a <see cref="Rune"/> from the provided input value.
/// </summary>
public static bool TryCreate(char ch, out Rune result)
{
uint extendedValue = ch;
if (!UnicodeUtility.IsSurrogateCodePoint(extendedValue))
{
result = UnsafeCreate(extendedValue);
return(true);
}
else
{
result = default;
return(false);
}
}
// returns a negative number on failure
private static int ReadRuneFromString(string input, int index)
{
if (input is null)
{
ThrowHelper.ThrowArgumentNullException(ExceptionArgument.input);
}
if ((uint)index >= (uint)input.Length)
{
ThrowHelper.ThrowArgumentOutOfRange_IndexException();
}
// Optimistically assume input is within BMP.
uint returnValue = input[index];
if (UnicodeUtility.IsSurrogateCodePoint(returnValue))
{
if (!UnicodeUtility.IsHighSurrogateCodePoint(returnValue))
{
return(-1);
}
// Treat 'returnValue' as the high surrogate.
//
// If this becomes a hot code path, we can skip the below bounds check by reading
// off the end of the string using unsafe code. Since strings are null-terminated,
// we're guaranteed not to read a valid low surrogate, so we'll fail correctly if
// the string terminates unexpectedly.
index++;
if ((uint)index >= (uint)input.Length)
{
return(-1); // not an argument exception - just a "bad data" failure
}
uint potentialLowSurrogate = input[index];
if (!UnicodeUtility.IsLowSurrogateCodePoint(potentialLowSurrogate))
{
return(-1);
}
returnValue = UnicodeUtility.GetScalarFromUtf16SurrogatePair(returnValue, potentialLowSurrogate);
}
return((int)returnValue);
}
// returns a negative number on failure
internal static int ReadFirstRuneFromUtf16Buffer(ReadOnlySpan <char> input)
{
if (input.IsEmpty)
{
return(-1);
}
// Optimistically assume input is within BMP.
uint returnValue = input[0];
if (UnicodeUtility.IsSurrogateCodePoint(returnValue))
{
if (!UnicodeUtility.IsHighSurrogateCodePoint(returnValue))
{
return(-1);
}
// Treat 'returnValue' as the high surrogate.
if (1 >= (uint)input.Length)
{
return(-1); // not an argument exception - just a "bad data" failure
}
uint potentialLowSurrogate = input[1];
if (!UnicodeUtility.IsLowSurrogateCodePoint(potentialLowSurrogate))
{
return(-1);
}
returnValue = UnicodeUtility.GetScalarFromUtf16SurrogatePair(returnValue, potentialLowSurrogate);
}
return((int)returnValue);
}