/// <summary>
/// Detect the character encoding form this byte array.
/// </summary>
/// <param name="bytes"></param>
/// <returns></returns>
public static DetectionResult DetectFromBytes(byte[] bytes)
{
var detector = new CharsetDetector();
detector.Feed(bytes, 0, bytes.Length);
return(detector.DataEnd());
}
private static void ReadStream(Stream stream, long?maxBytes, CharsetDetector detector)
{
const int bufferSize = 1024;
byte[] buff = new byte[bufferSize];
int read;
long readTotal = 0;
var toRead = CalcToRead(maxBytes, readTotal, bufferSize);
while ((read = stream.Read(buff, 0, toRead)) > 0)
{
detector.Feed(buff, 0, read);
if (maxBytes != null)
{
readTotal += read;
if (readTotal >= maxBytes)
{
return;
}
toRead = CalcToRead(maxBytes, readTotal, bufferSize);
}
if (detector._done)
{
return;
}
}
}
/// <summary>
/// Detect the character encoding form this byte array.
/// </summary>
/// <param name="bytes"></param>
/// <returns></returns>
public static DetectionResult DetectFromBytes(byte[] bytes)
{
if (bytes == null)
{
throw new ArgumentNullException(nameof(bytes));
}
var detector = new CharsetDetector();
detector.Feed(bytes, 0, bytes.Length);
return(detector.DataEnd());
}
/// <summary>
/// Detect the character encoding by reading the stream.
///
/// Note: stream position is not reset before and after.
/// </summary>
/// <param name="stream">The steam. </param>
public static DetectionResult DetectFromStream(Stream stream)
{
var detector = new CharsetDetector();
byte[] buff = new byte[1024];
int read;
while ((read = stream.Read(buff, 0, buff.Length)) > 0 && !detector._done)
{
detector.Feed(buff, 0, read);
}
return(detector.DataEnd());
}
/// <summary>
/// Detect the character encoding by reading the stream.
///
/// Note: stream position is not reset before and after.
/// </summary>
/// <param name="stream">The steam. </param>
/// <param name="maxBytesToRead">max bytes to read from <paramref name="stream"/>. If <c>null</c>, then no max</param>
/// <exception cref="ArgumentOutOfRangeException"><paramref name="maxBytesToRead"/> 0 or lower.</exception>
public static DetectionResult DetectFromStream(Stream stream, long?maxBytesToRead)
{
if (stream == null)
{
throw new ArgumentNullException(nameof(stream));
}
if (maxBytesToRead <= 0)
{
throw new ArgumentOutOfRangeException(nameof(maxBytesToRead));
}
var detector = new CharsetDetector();
ReadStream(stream, maxBytesToRead, detector);
return(detector.DataEnd());
}
/// <summary>
/// Detect the character encoding form this byte array.
/// It searchs for BOM from bytes[offset].
/// </summary>
/// <param name="bytes">The byte array containing the text</param>
/// <param name="offset">The zero-based byte offset in buffer at which to begin reading the data from</param>
/// <param name="len">The maximum number of bytes to be read</param>
/// <returns></returns>
public static DetectionResult DetectFromBytes(byte[] bytes, int offset, int len)
{
if (bytes == null)
{
throw new ArgumentNullException(nameof(bytes));
}
if (offset < 0)
{
throw new ArgumentOutOfRangeException(nameof(offset));
}
if (len < 0)
{
throw new ArgumentOutOfRangeException(nameof(len));
}
if (bytes.Length < offset + len)
{
throw new ArgumentException($"{nameof(len)} is greater than the number of bytes from {nameof(offset)} to the end of the array.");
}
var detector = new CharsetDetector();
detector.Feed(bytes, offset, len);
return(detector.DataEnd());
}
