private static HuffmanCoder CreateFromText(string text)
{
var leafNodes = text.ToCharArray()
.GroupBy(c => c)
.Select(g => new TreeNode(g.Key, ((double)g.Count()) / text.Length));
return(HuffmanCoder.Create(leafNodes));
}
public static byte[] Encode(string text)
{
var encoder = HuffmanCoder.CreateFromText(text);
var signature = encoder.ToSignature();
var contentLength = text.Length;
var content = text.SelectMany(c => encoder.EncodingTable[c]);
var headerBytes = Encoding.UTF8.GetBytes($"{signature}\n{contentLength}\n");
var contentBytes = Utilities.BoolsToBytes(content);
return(headerBytes.Concat(contentBytes).ToArray());
}
private static HuffmanCoder CreateFromSignature(string signature)
{
var leafNodes = signature.Split("|^").Select(s =>
{
var ss = s.Split(",^");
var c = Utilities.StrToChar(ss[0]);
var probValue = double.Parse(ss[1]);
return(new TreeNode(c, probValue));
});
return(HuffmanCoder.Create(leafNodes));
}
public static string Decode(byte[] encodedBytes)
{
using (var ms = new MemoryStream(encodedBytes))
using (var sr = new StreamReader(ms))
{
var signature = sr.ReadLine();
var contentLength = sr.ReadLine();
var byteCount = Encoding.UTF8.GetByteCount($"{signature}\n{contentLength}\n");
ms.Position = byteCount;
var encodedContent = new byte[ms.Length - ms.Position];
ms.Read(encodedContent, 0, encodedContent.Length);
var huffman = HuffmanCoder.CreateFromSignature(signature);
return(huffman.Decode(encodedContent, int.Parse(contentLength)));
}
}
static void Main(string[] args)
{
var path = Path.GetDirectoryName(Assembly.GetExecutingAssembly().Location);
var inputFile = Path.Combine(path, "ulysses.txt");
var encodedOutputFile = Path.Combine(path, "ulysses-encoded.txt");
var decodedOutputFile = Path.Combine(path, "ulysses-decoded.txt");
var ulysses = File.ReadAllText(inputFile);
// Encode the input file
var encoded = HuffmanCoder.Encode(ulysses);
File.WriteAllBytes(encodedOutputFile, encoded);
// Read in the output file and ensure it can be decoded
var decoded = HuffmanCoder.Decode(File.ReadAllBytes(encodedOutputFile));
File.WriteAllText(decodedOutputFile, decoded);
// Returns true yay
var same = string.Equals(ulysses, decoded);
}