private static void ConvertToUtf8WithBOM(string file)
{
var bytes = File.ReadAllBytes(file);
var asciiOnly = bytes.All(c => c <= 127);
if (asciiOnly) return;
var preamble = Encoding.UTF8.GetPreamble();
var isUtf8WithBom = bytes.Take(preamble.Length).SequenceEqual(preamble);
if (isUtf8WithBom) return;
var dd = new UTF8Prober();
var utf8DetectionResult = dd.HandleData(bytes, 0, bytes.Length);
var encoding = Encoding.UTF8;
if (utf8DetectionResult == ProbingState.NotMe) encoding = Encoding.GetEncoding(1251);
else return;
Console.WriteLine("Converting {0}. {1}", file, encoding.EncodingName);
var content = File.ReadAllText(file, encoding);
var firstNonAscii = content.Zip(Enumerable.Range(0, int.MaxValue), Tuple.Create)
.FirstOrDefault(t => t.Item1 > 127);
if (firstNonAscii != null)
{
var index = Math.Max(0, firstNonAscii.Item2 - 5);
var len = Math.Min(content.Length - index, 35);
Console.WriteLine(" non ascii text {0}", content.Substring(index, len).Replace('\r', ' ').Replace('\n', ' '));
}
File.WriteAllText(file, content, Encoding.UTF8);
}
public MBCSGroupProber()
{
probers[0] = new UTF8Prober();
probers[1] = new SJISProber();
probers[2] = new EUCJPProber();
probers[3] = new GB18030Prober();
probers[4] = new EUCKRProber();
probers[5] = new Big5Prober();
probers[6] = new EUCTWProber();
Reset();
}
public MBCSGroupProber()
{
probers[0] = new UTF8Prober();
probers[1] = new SJISProber();
probers[2] = new EUCJPProber();
probers[3] = new GB18030Prober();
probers[4] = new EUCKRProber();
probers[5] = new Big5Prober();
probers[6] = new EUCTWProber();
Reset();
}
