| public DetectEncoding ( byte buffer, int size ) : EncodingEnum | ||
| buffer | byte | |
| size | int | |
| return | EncodingEnum | |
/// <summary>
/// Returns the encoding of the input file
/// </summary>
/// <param name="srcFile"></param>
/// <returns></returns>
public static Encoding GetFileEncoding(string srcFile)
{
Encoding encoding = Encoding.Default;
if (string.IsNullOrEmpty(srcFile) || !File.Exists(srcFile))
{
return(encoding);
}
// Read in the file in binary
byte[] buffer;
try {
using (var file = new FileStream(srcFile, FileMode.Open, FileAccess.Read, FileShare.ReadWrite)) {
using (var memoryStream = new MemoryStream()) {
file.CopyTo(memoryStream);
buffer = memoryStream.ToArray();
}
}
} catch (Exception) {
return(encoding);
}
// Detect encoding
var textDetect = new TextEncodingDetect();
EncodingEnum textEnc = textDetect.DetectEncoding(buffer, buffer.Length);
switch (textEnc)
{
case EncodingEnum.Ascii:
// ASCII (chars in the 0-127 range)
encoding = Encoding.ASCII;
break;
case EncodingEnum.Ansi:
// ANSI (chars in the range 0-255 range)
encoding = Encoding.GetEncoding(1252);
break;
case EncodingEnum.Utf8Bom:
case EncodingEnum.Utf8Nobom:
// UTF-8
encoding = Encoding.UTF8;
break;
case EncodingEnum.Utf16LeBom:
case EncodingEnum.Utf16LeNobom:
// UTF-16 Little Endian
encoding = Encoding.Unicode;
break;
case EncodingEnum.Utf16BeBom:
case EncodingEnum.Utf16BeNobom:
// UTF-16 Big Endian
encoding = Encoding.BigEndianUnicode;
break;
}
return(encoding);
}
/// <summary>
/// Returns the encoding of the input file
/// </summary>
/// <param name="srcFile"></param>
/// <returns></returns>
public static Encoding GetFileEncoding(string srcFile)
{
Encoding encoding = Encoding.Default;
if (string.IsNullOrEmpty(srcFile) || !File.Exists(srcFile)) {
return encoding;
}
// Read in the file in binary
byte[] buffer;
try {
using (var file = new FileStream(srcFile, FileMode.Open, FileAccess.Read, FileShare.ReadWrite)) {
using (var memoryStream = new MemoryStream()) {
file.CopyTo(memoryStream);
buffer = memoryStream.ToArray();
}
}
} catch (Exception) {
return encoding;
}
// Detect encoding
var textDetect = new TextEncodingDetect();
EncodingEnum textEnc = textDetect.DetectEncoding(buffer, buffer.Length);
switch (textEnc) {
case EncodingEnum.Ascii:
// ASCII (chars in the 0-127 range)
encoding = Encoding.ASCII;
break;
case EncodingEnum.Ansi:
// ANSI (chars in the range 0-255 range)
encoding = Encoding.GetEncoding(1252);
break;
case EncodingEnum.Utf8Bom:
case EncodingEnum.Utf8Nobom:
// UTF-8
encoding = Encoding.UTF8;
break;
case EncodingEnum.Utf16LeBom:
case EncodingEnum.Utf16LeNobom:
// UTF-16 Little Endian
encoding = Encoding.Unicode;
break;
case EncodingEnum.Utf16BeBom:
case EncodingEnum.Utf16BeNobom:
// UTF-16 Big Endian
encoding = Encoding.BigEndianUnicode;
break;
}
return encoding;
}