private static uint ConsumeDataWithoutExistingPartialSequence(ReadOnlySpan <byte> utf8Bytes)
{
var indexOfFirstInvalidSequence = Utf8Utility.GetIndexOfFirstInvalidUtf8Sequence(utf8Bytes);
if (indexOfFirstInvalidSequence < 0)
{
// Successfully consumed entire buffer without error.
return(DefaultSequence);
}
else
{
// Couldn't consume entire buffer; is this due to a partial buffer or truly invalid data?
utf8Bytes = utf8Bytes.Slice(indexOfFirstInvalidSequence);
var validity = Utf8Utility.PeekFirstSequence(utf8Bytes, out int numBytesConsumed, out _);
if (validity == SequenceValidity.Incomplete)
{
// Saw a partial (not invalid) sequence, remember it for next time.
Debug.Assert(1 <= numBytesConsumed && numBytesConsumed <= 3);
// Put all partial data into the high 3 bytes, making room for us to
// write the count of partial bytes in the buffer as the low byte.
Span <byte> partialSequenceAsBytes = stackalloc byte[sizeof(uint)];
BinaryPrimitives.WriteMachineEndian(partialSequenceAsBytes, ref numBytesConsumed);
if (BitConverter.IsLittleEndian)
{
utf8Bytes.Slice(0, numBytesConsumed).CopyTo(partialSequenceAsBytes.Slice(1));
}
else
{
utf8Bytes.Slice(0, numBytesConsumed).CopyTo(partialSequenceAsBytes);
}
return(BinaryPrimitives.ReadMachineEndian <uint>(partialSequenceAsBytes));
}
else
{
// Truly invalid data.
// (Shouldn't have gotten 'Empty' or 'WellFormed'.)
Debug.Assert(validity == SequenceValidity.Invalid);
return(InvalidSequence);
}
}
}
private static uint ConsumeDataWithExistingPartialSequence(uint partialSequence, ReadOnlySpan <byte> utf8Bytes)
{
Debug.Assert(partialSequence != 0 && !IsInvalidPartialSequence(partialSequence));
int partialSequenceOriginalByteCount = (byte)partialSequence;
if (BitConverter.IsLittleEndian)
{
// When we turn this into a Span<byte>, want MSB to be the first byte of the partial sequence
partialSequence >>= 8;
}
// Copy as much data as we can from the input buffer to our partial sequence.
Span <byte> partialSequenceAsBytes = stackalloc byte[sizeof(uint)];
BinaryPrimitives.WriteMachineEndian(partialSequenceAsBytes, ref partialSequence);
int numBytesToCopyFromInputToPartialSequence = Math.Min(4 - partialSequenceOriginalByteCount, utf8Bytes.Length);
utf8Bytes.Slice(0, numBytesToCopyFromInputToPartialSequence).CopyTo(partialSequenceAsBytes.Slice(partialSequenceOriginalByteCount));
int partialSequenceNewByteCount = partialSequenceOriginalByteCount + numBytesToCopyFromInputToPartialSequence;
// And check for validity of the new (hopefully complete) partial sequence.
var validity = Utf8Utility.PeekFirstSequence(partialSequenceAsBytes.Slice(0, partialSequenceNewByteCount), out int numBytesConsumed, out _);
Debug.Assert(1 <= numBytesConsumed && numBytesConsumed <= 4);
if (validity == SequenceValidity.WellFormed)
{
// This is the happy path; we've consumed some set of bytes from the input
// buffer and it has caused the partial sequence to validate. Let's calculate
// how many bytes from the input buffer were required to complete the sequence,
// then strip them off the incoming data.
// n.b. This might not be the same as numBytesToCopyFromInputToPartialSequence.
int numBytesRequiredFromInputBufferToFinishPartialSequence = numBytesConsumed - partialSequenceOriginalByteCount;
return(ConsumeDataWithoutExistingPartialSequence(utf8Bytes.Slice(numBytesRequiredFromInputBufferToFinishPartialSequence)));
}
else if (validity == SequenceValidity.Incomplete)
{
// We've consumed all data available to us and we still have an incomplete sequence.
// It's still valid (until we see invalid bytes), so squirrel away what we've seen
// and report success to our caller.
Debug.Assert(numBytesConsumed < 4);
Debug.Assert(numBytesConsumed == partialSequenceNewByteCount);
// Put all partial data into the high 3 bytes, making room for us to
// write the count of partial bytes in the buffer as the low byte.
partialSequence = BinaryPrimitives.ReadMachineEndian <uint>(partialSequenceAsBytes);
if (BitConverter.IsLittleEndian)
{
return((partialSequence << 8) | (uint)numBytesConsumed);
}
else
{
return((partialSequence & unchecked ((uint)~0xFF)) | (uint)numBytesConsumed);
}
}
else
{
// Truly invalid data.
// (Shouldn't have gotten 'Empty' or 'WellFormed'.)
Debug.Assert(validity == SequenceValidity.Invalid);
return(InvalidSequence);
}
}