private void Convert(byte[] data, int dataLength, EncodingType encoding, bool hasBom, byte[] outputData, out int outputLength)
{
int outputIndex = 0;
if (hasBom)
{
CopyBomTo(outputData, encoding, out outputIndex);
}
int numBytesPerUnit = encoding.GetNumBytesPerUnit();
int startDataIndex = outputIndex;
for (int dataIndex = startDataIndex; dataIndex <= dataLength - numBytesPerUnit; dataIndex += numBytesPerUnit)
{
var currentUnit = GetUnit(data, dataIndex, numBytesPerUnit);
var nextUnit = GetUnit(data, dataIndex + numBytesPerUnit, numBytesPerUnit);
ProcessUnit(outputData, currentUnit, encoding, ref outputIndex);
if (currentUnit.IsCR(encoding) && nextUnit.IsLF(encoding))
{
// If CRLF then the do not process the next LF unit because the end of line has been inserted while processing the CR
dataIndex += numBytesPerUnit;
}
}
outputLength = outputIndex;
}
private bool IsFromEncoding(
EncodingDetectionContext context,
EncodingType encoding,
Func <EncodingDetectionContext, int, bool> IsFromEndiannes)
{
int numBytesPerUnit = encoding.GetNumBytesPerUnit();
// If there is a entire null unit within the data then can't be this encoding
if (ContainsNullBytesUnit(context.Data, context.DataLength, numBytesPerUnit))
{
return(false);
}
return(IsFromEndiannes(context, numBytesPerUnit));
}
private static byte[] ToUnit(this byte byteChar, EncodingType encoding)
{
var nullBytes = Enumerable.Range(0, encoding.GetNumBytesPerUnit() - 1)
.Select(i => ByteCode.Null);
var byteCharArray = new byte[1] {
byteChar
};
if (encoding.IsBigEndian())
{
return(nullBytes.Concat(byteCharArray).ToArray());
}
return(byteCharArray.Concat(nullBytes).ToArray());
}
public void GetEncoding_WhenDataEndsByEolButDataIsNotCompletelyFilled(byte[] eolBytes, EncodingType expectedEncoding)
{
// Arrange
int bufferLength = 10 * expectedEncoding.GetNumBytesPerUnit();
byte[] data = new byte[bufferLength];
int dataLength = bufferLength / 2;
FillDataWithDummyValues(data, dataLength);
eolBytes.CopyTo(data, dataLength - eolBytes.Length);
CreateSut();
// Act
var encoding = sut.GetEncoding(data, dataLength).encoding;
// Assert
Assert.Equal(expectedEncoding, encoding);
}