internal static void GetUtf16SurrogatePairFromAstralScalarValue(uint scalar, out char highSurrogate, out char lowSurrogate)
{
Debug.Assert(0x10000 <= scalar && scalar <= UNICODE_LAST_CODEPOINT);
UnicodeDebug.AssertIsValidSupplementaryPlaneScalar(scalar);
// This calculation comes from the Unicode specification, Table 3-5.
highSurrogate = (char)((scalar + ((0xD800u - 0x40u) << 10)) >> 10);
lowSurrogate = (char)((scalar & 0x3FFu) + 0xDC00u);
}
/// <summary>
/// A copy of the logic in Rune.DecodeFromUtf8.
/// </summary>
public static OperationStatus DecodeScalarValueFromUtf8(ReadOnlySpan <byte> source, out uint result, out int bytesConsumed)
{
const char ReplacementChar = '\uFFFD';
// This method follows the Unicode Standard's recommendation for detecting
// the maximal subpart of an ill-formed subsequence. See The Unicode Standard,
// Ch. 3.9 for more details. In summary, when reporting an invalid subsequence,
// it tries to consume as many code units as possible as long as those code
// units constitute the beginning of a longer well-formed subsequence per Table 3-7.
int index = 0;
// Try reading input[0].
if ((uint)index >= (uint)source.Length)
{
goto NeedsMoreData;
}
uint tempValue = source[index];
if (!UnicodeUtility.IsAsciiCodePoint(tempValue))
{
goto NotAscii;
}
Finish:
bytesConsumed = index + 1;
Debug.Assert(1 <= bytesConsumed && bytesConsumed <= 4); // Valid subsequences are always length [1..4]
result = tempValue;
return(OperationStatus.Done);
NotAscii:
// Per Table 3-7, the beginning of a multibyte sequence must be a code unit in
// the range [C2..F4]. If it's outside of that range, it's either a standalone
// continuation byte, or it's an overlong two-byte sequence, or it's an out-of-range
// four-byte sequence.
if (!UnicodeUtility.IsInRangeInclusive(tempValue, 0xC2, 0xF4))
{
goto FirstByteInvalid;
}
tempValue = (tempValue - 0xC2) << 6;
// Try reading input[1].
index++;
if ((uint)index >= (uint)source.Length)
{
goto NeedsMoreData;
}
// Continuation bytes are of the form [10xxxxxx], which means that their two's
// complement representation is in the range [-65..-128]. This allows us to
// perform a single comparison to see if a byte is a continuation byte.
int thisByteSignExtended = (sbyte)source[index];
if (thisByteSignExtended >= -64)
{
goto Invalid;
}
tempValue += (uint)thisByteSignExtended;
tempValue += 0x80; // remove the continuation byte marker
tempValue += (0xC2 - 0xC0) << 6; // remove the leading byte marker
if (tempValue < 0x0800)
{
Debug.Assert(UnicodeUtility.IsInRangeInclusive(tempValue, 0x0080, 0x07FF));
goto Finish; // this is a valid 2-byte sequence
}
// This appears to be a 3- or 4-byte sequence. Since per Table 3-7 we now have
// enough information (from just two code units) to detect overlong or surrogate
// sequences, we need to perform these checks now.
if (!UnicodeUtility.IsInRangeInclusive(tempValue, ((0xE0 - 0xC0) << 6) + (0xA0 - 0x80), ((0xF4 - 0xC0) << 6) + (0x8F - 0x80)))
{
// The first two bytes were not in the range [[E0 A0]..[F4 8F]].
// This is an overlong 3-byte sequence or an out-of-range 4-byte sequence.
goto Invalid;
}
if (UnicodeUtility.IsInRangeInclusive(tempValue, ((0xED - 0xC0) << 6) + (0xA0 - 0x80), ((0xED - 0xC0) << 6) + (0xBF - 0x80)))
{
// This is a UTF-16 surrogate code point, which is invalid in UTF-8.
goto Invalid;
}
if (UnicodeUtility.IsInRangeInclusive(tempValue, ((0xF0 - 0xC0) << 6) + (0x80 - 0x80), ((0xF0 - 0xC0) << 6) + (0x8F - 0x80)))
{
// This is an overlong 4-byte sequence.
goto Invalid;
}
// The first two bytes were just fine. We don't need to perform any other checks
// on the remaining bytes other than to see that they're valid continuation bytes.
// Try reading input[2].
index++;
if ((uint)index >= (uint)source.Length)
{
goto NeedsMoreData;
}
thisByteSignExtended = (sbyte)source[index];
if (thisByteSignExtended >= -64)
{
goto Invalid; // this byte is not a UTF-8 continuation byte
}
tempValue <<= 6;
tempValue += (uint)thisByteSignExtended;
tempValue += 0x80; // remove the continuation byte marker
tempValue -= (0xE0 - 0xC0) << 12; // remove the leading byte marker
if (tempValue <= 0xFFFF)
{
Debug.Assert(UnicodeUtility.IsInRangeInclusive(tempValue, 0x0800, 0xFFFF));
goto Finish; // this is a valid 3-byte sequence
}
// Try reading input[3].
index++;
if ((uint)index >= (uint)source.Length)
{
goto NeedsMoreData;
}
thisByteSignExtended = (sbyte)source[index];
if (thisByteSignExtended >= -64)
{
goto Invalid; // this byte is not a UTF-8 continuation byte
}
tempValue <<= 6;
tempValue += (uint)thisByteSignExtended;
tempValue += 0x80; // remove the continuation byte marker
tempValue -= (0xF0 - 0xE0) << 18; // remove the leading byte marker
UnicodeDebug.AssertIsValidSupplementaryPlaneScalar(tempValue);
goto Finish; // this is a valid 4-byte sequence
FirstByteInvalid:
index = 1; // Invalid subsequences are always at least length 1.
Invalid:
Debug.Assert(1 <= index && index <= 3); // Invalid subsequences are always length 1..3
bytesConsumed = index;
result = ReplacementChar;
return(OperationStatus.InvalidData);
NeedsMoreData:
Debug.Assert(0 <= index && index <= 3); // Incomplete subsequences are always length 0..3
bytesConsumed = index;
result = ReplacementChar;
return(OperationStatus.NeedMoreData);
}