public void GlobalSetup()
{
Program.NativeGlobalInit();
_binaryDir = Path.GetFullPath(AppDomain.CurrentDomain.BaseDirectory);
_sampleBaseDir = Path.GetFullPath(Path.Combine(_binaryDir, "..", "..", "..", "..", "..", "..", "..", "Samples"));
_sampleDir = Path.Combine(_sampleBaseDir, "EncDetect");
_magicFile = Path.Combine(_binaryDir, "magic.mgc");
_magic = Magic.Open(_magicFile);
_autoitDetect = new AdvTextEncDetect();
foreach (string srcFileName in SrcFileNames)
{
string srcFile = Path.Combine(_sampleDir, srcFileName);
long fileSize = new FileInfo(srcFile).Length;
byte[] bytes = new byte[fileSize];
using (FileStream fs = new FileStream(srcFile, FileMode.Open, FileAccess.Read, FileShare.Read))
{
fs.Read(bytes, 0, bytes.Length);
}
SrcFiles[srcFileName] = bytes;
}
}
public static Encoding DetectEncoding(ReadOnlySpan <byte> span, out TextEncoding textEnc)
{
textEnc = AdvDetect.DetectEncoding(span);
return(AdvTextEncDetect.TextEncodingToBclEncoding(textEnc));
}
public static bool IsText(byte[] buffer, int offset, int count)
{
if (buffer.Length < offset + count)
{
throw new ArgumentOutOfRangeException(nameof(buffer));
}
// [Stage 1] Contains unicode BOM -> text
if (3 <= offset + count &&
buffer[offset] == Utf8Bom[0] && buffer[offset + 1] == Utf8Bom[1] && buffer[offset + 2] == Utf8Bom[2])
{
return(true);
}
if (2 <= offset + count)
{
if (buffer[offset] == Utf16LeBom[0] && buffer[offset + 1] == Utf16LeBom[1])
{
return(true);
}
if (buffer[offset] == Utf16BeBom[0] && buffer[offset + 1] == Utf16BeBom[1])
{
return(true);
}
}
// [Stage 2] Check if a chunk can be decoded as system default ANSI locale.
// Many multi-byte encodings have 'unused area'. If a file contains one of these area, treat it as a binary.
// Ex) EUC-KR's layout : https://en.wikipedia.org/wiki/CP949#/media/File:Unified_Hangul_Code.svg
bool isText = true;
Encoding ansiEnc = Encoding.GetEncoding(DefaultAnsi.CodePage, new EncoderExceptionFallback(), new DecoderExceptionFallback());
try
{
// ReSharper disable once ReturnValueOfPureMethodIsNotUsed
ansiEnc.GetChars(buffer, offset, count);
}
catch (DecoderFallbackException)
{ // Failure
isText = false;
}
// [Stage 3]
// Problem: Some encodings make use of 128-255 area, so every byte is valid. (e.g. Windows-1252 / CP437)
// To counter these issue, if file is seems to be text, check again with AutoIt.Common.TextEncodingDetect
if (isText)
{
AdvTextEncDetect detect = new AdvTextEncDetect();
byte[] idxZeroBuffer;
if (offset == 0)
{
idxZeroBuffer = buffer;
}
else
{
idxZeroBuffer = new byte[count];
Array.Copy(buffer, offset, idxZeroBuffer, 0, count);
}
switch (detect.DetectEncoding(idxZeroBuffer, 0, idxZeroBuffer.Length))
{
// Binary
case TextEncoding.None:
// PEBakery mandates unicode text to have BOM.
// They must have been filtered out in stage 1.
case TextEncoding.Utf16LeBom:
case TextEncoding.Utf16BeBom:
case TextEncoding.Utf8Bom:
// Treat unicode text file without a BOM as a binary.
case TextEncoding.Utf16LeNoBom:
case TextEncoding.Utf16BeNoBom:
case TextEncoding.Utf8NoBom:
isText = false;
break;
}
}
return(isText);
}
public static Encoding DetectEncoding(byte[] buffer, int offset, int count, out TextEncoding textEnc)
{
textEnc = AdvDetect.DetectEncoding(buffer, offset, count);
return(AdvTextEncDetect.TextEncodingToBclEncoding(textEnc));
}
