using System;
using System.IO;
using System.Text;
using System.Text.RegularExpressions;
using System.Collections;

namespace DF
{
	class FileEncodingAutoDetection
	{
		private static readonly Regex coding_pep_re = new Regex(@"^(?:#|//).*coding[:=]\s*([-\.\w]+)", RegexOptions.Compiled | RegexOptions.IgnoreCase);
		private static readonly Regex coding_pep_cp_re = new Regex(@"^(?:cp-?)?(\d+)$", RegexOptions.Compiled | RegexOptions.IgnoreCase);
		private static readonly char[] line_delimeters = new char[] {'\r', '\n'}; // save runtime from allocating it over and over again

		public static Encoding DetectEncoding(FileStream fs, IList lines, int maxlines, Encoding defaultEncoding)
		{
			if(fs == null)
				throw new ArgumentNullException("fs");
			if(lines == null)
				maxlines = 0;
			if(maxlines < 0)
				throw new ArgumentOutOfRangeException("maxlines", "maxlines must be greater than zero");
			if(defaultEncoding == null)
				defaultEncoding = Encoding.Default;
			Encoding ascii = Encoding.ASCII;
			long saved_position = fs.Position;
			int len = 0;
			int utf8len = 0;
			int foundlines = 0;
			char ch1, ch2, ch = '\0';
			bool have_valid_utf8 = true;
			byte[] bytes = new byte[4096];
			StringBuilder sb = null;
			if(maxlines > 0)
				sb = new StringBuilder();
			while((len = fs.Read(bytes, 0, bytes.Length)) != 0)
			{
				// scan for first needed lines
				if(foundlines < maxlines)
				{
					string chars = new string(ascii.GetChars(bytes, 0, len));
					int linestart = 0;
					int lineend;
					while(foundlines < maxlines && linestart < chars.Length && (lineend = chars.IndexOfAny(line_delimeters)) != -1)
					{
						if(ch != '\0' && lineend == 0)
						{
							ch2 = chars[lineend];
							if((ch == '\r' && ch2 == '\n') || (ch == '\n' && ch2 == '\r'))
							{
								ch = '\0';
								linestart = 1;
								continue;
							}
						}
						if(linestart < lineend)
							sb.Append(chars, linestart, lineend - linestart);
						lines.Add(sb.ToString());
						sb.Length = 0;
						++foundlines;
						if(lineend == (chars.Length-1))
						{
							ch = chars[lineend];
							linestart = lineend + 1;
							break;
						}
						ch = '\0';
						ch1 = chars[lineend];
						ch2 = chars[lineend+1];
						if((ch1 == '\r' && ch2 == '\n') || (ch1 == '\n' && ch2 == '\r'))
							linestart = lineend + 2;
						else
							linestart = lineend + 1;
					}
					if(foundlines < maxlines && linestart < chars.Length)
						sb.Append(chars, linestart, chars.Length - linestart);
				}
				// determine if utf-8 is correct
				for(int i = 0; have_valid_utf8 && i < len; ++i)
				{
					byte b = bytes[i];
					if(b >= 0x80)
					{
						if(b < 0xc0)
						{
							if(utf8len > 0)
								utf8len--;
							else
								have_valid_utf8 = false;
						}
						else if(b >= 0xc2 && b < 0xf5)
						{
							if(utf8len > 0)
								have_valid_utf8 = false;
							else if(b < 0xe0)
								utf8len = 1;
							else if(b < 0xf0)
								utf8len = 2;
							else
								utf8len = 3;
						}
						else
						{
							have_valid_utf8 = false;
						}
					}
					else if(utf8len > 0)
						have_valid_utf8 = false;
				}
				if(!have_valid_utf8)
				{
					if(foundlines < maxlines)
						continue;
					break;
				}
			}
			fs.Position = saved_position;
			if(foundlines < maxlines)
				lines.Add(sb.ToString());
			if(have_valid_utf8)
				return Encoding.UTF8;
			return defaultEncoding;
		}

		public static StreamReader OpenText(string fname, Encoding defaultEncoding)
		{
			if(fname == null)
				throw new ArgumentNullException("fname");
			FileStream fs = File.OpenRead(fname);
			try
			{
				Encoding enc = Encoding.ASCII;
				StreamReader reader = new StreamReader(fs, enc);
				reader.Peek(); // force BOM detection
				if(reader.CurrentEncoding != enc)
					return reader; // BOM detected
				reader.DiscardBufferedData();
				reader = null;

				fs.Position = 0;
				ArrayList lines = new ArrayList();
				enc = DetectEncoding(fs, lines, 2, defaultEncoding);
				foreach(string line in lines)
				{
					Match m1 = coding_pep_re.Match(line);
					if(!m1.Success)
						continue;
					// support for ascii and mbcs encodings
					if(m1.Groups[1].Value == "ascii")
					{
						enc = Encoding.ASCII;
						break;
					}
					else if(m1.Groups[1].Value == "mbcs")
					{
						enc = Encoding.Default;
						break;
					}
					try
					{
						enc = Encoding.GetEncoding(m1.Groups[1].Value);
						break;
					}
					catch (Exception) { }
					Match m2 = coding_pep_cp_re.Match(m1.Groups[1].Value);
					if(m2.Success)
					{
						try
						{
							enc = Encoding.GetEncoding(int.Parse(m1.Groups[1].Value));
							break;
						}
						catch (Exception) { }
					}
					// throw new EncodingUnsupportedException(m1.Groups[1].Value);
				}
				reader = new StreamReader(fs, enc, false);
				return reader;
			}
			catch (Exception)
			{
				fs.Close();
				throw;
			}
		}

		public static StreamReader OpenText(string fname)
		{
			return OpenText(fname, null);
		}
	}

	class TestCase
	{
		public static void Main()
		{
			Console.WriteLine("Привет вам из си-шарпа!");
			using(StreamReader reader = FileEncodingAutoDetection.OpenText("FileEncodingAutoDetection2.cs"))
			{
				string line;
				while((line = reader.ReadLine()) != null)
					Console.WriteLine(line);
			}
		}
	}
}