private bool IsBigEndian(EncodingDetectionContext context, int numBytesPerUnit)
{
return(context.EolIndexes.Any(eolIndex => {
bool isPrecededByNullBytes = IsPrecededByNullBytes(context.Data, eolIndex, numNullBytes: numBytesPerUnit - 1);
bool isLastByteInUnit = eolIndex % numBytesPerUnit == numBytesPerUnit - 1;
return isLastByteInUnit && isPrecededByNullBytes;
}));
}
private bool IsLittleEndian(EncodingDetectionContext context, int numBytesPerUnit)
{
return(context.EolIndexes.Any(eolIndex => {
bool isFollowedByNullBytes = IsFollowedByNullBytes(context.Data, context.DataLength, eolIndex, numNullBytes: numBytesPerUnit - 1);
bool isFirstByteInUnit = eolIndex % numBytesPerUnit == 0;
return isFirstByteInUnit && isFollowedByNullBytes;
}));
}
private bool IsUtf8(EncodingDetectionContext context)
{
// If there is any 0 within the data then can't be utf-8
if (context.Data.Take(context.DataLength).Any(b => b == ByteCode.Null))
{
return(false);
}
return(context.EolIndexes.Any(eolIndex => {
// Check eol is not sorrounded by null bytes, checking also corner cases (eol is first or last bytes)
bool notPrecededByNull = (eolIndex == 0 || context.Data[eolIndex - 1] != ByteCode.Null);
bool notFollowedByNull = (eolIndex == context.DataLength - 1 || context.Data[eolIndex + 1] != ByteCode.Null);
return notPrecededByNull && notFollowedByNull;
}));
}
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));
}
/// <summary>
/// Returns the encoding of the data if can be detected through any of the end of line bytes.
/// </summary>
/// <param name="data">Data where to detect encoding.</param>
/// <param name="dataLength">Length of the data.</param>
/// <returns>The encoding if detected.</returns>
internal EncodingType GetEncodingFromEolBytes(byte[] data, int dataLength)
{
var eolIndexes = data.FindEolByteIndexes();
if (!eolIndexes.Any())
{
// If there are not eol bytes then the encoding is not needed because the file will not need eol replacement.
return(EncodingType.None);
}
var context = new EncodingDetectionContext(data, dataLength, eolIndexes);
foreach (var encodingToCheck in encodingsToCheckOrdered)
{
var encoding = encodingToCheck.Key;
var checkEncodingFunction = encodingToCheck.Value;
if (checkEncodingFunction(context))
{
return(encoding);
}
}
return(EncodingType.None);
}
private bool IsBigEndianEncoding(EncodingDetectionContext context, EncodingType encoding)
{
return(IsFromEncoding(context, encoding, IsBigEndian));
}
private bool IsUtf16BE(EncodingDetectionContext context)
{
return(IsBigEndianEncoding(context, EncodingType.Utf16BE));
}
private bool IsUtf16LE(EncodingDetectionContext context)
{
return(IsLittleEndianEncoding(context, EncodingType.Utf16LE));
}
