public void AddFiles(string[] input, VectorTree tree)
{
ClearFiles(input, tree);
foreach (var file in Directory.GetFiles(input[1], input[2]))
{
var n = AddFile(file, tree);
Console.Write("{0} {1} ", file, n);
}
Console.WriteLine(tree.Visualize(0, 0));
var size = tree.Size(0, 0);
Console.WriteLine();
Console.WriteLine("depth {0} width {1} count: {2}, merges: {3}", size.depth, size.width, tree.Count, tree.MergeCount);
using (var treeStream = File.Create("tree.bin"))
using (var wordStream = File.Create("word.bin"))
using (var posStream = File.Create("pos.bin"))
{
tree.GetNode(0, 0).Serialize(treeStream, wordStream, posStream);
}
//using (var treeStream = File.OpenRead("tree.bin"))
//using (var wordStream = File.OpenRead("word.bin"))
//{
// var deserialized = VectorTree.Load(treeStream, wordStream);
// Console.WriteLine(deserialized.Visualize());
// var deserializedSize = deserialized.Size();
// Console.WriteLine("depth {0} width {1}", deserializedSize.depth, deserializedSize.width);
//}
}
private int AddDocument(string documentId, string[] input, VectorTree tree)
{
var docId = Hash(documentId);
var dic = new SortedList <double, string>();
var count = tree.Count;
foreach (var word in input)
{
if (string.IsNullOrWhiteSpace(word))
{
continue;
}
tree.GetNode(0, 0).Add(word, docId);
var wordvec = new VectorNode(word).TermVector;
if (wordvec.CosAngle(tree.Find(0, 0, word).TermVector) < VectorNode.IdenticalAngle)
{
throw new Exception("error");
}
//Console.Write("{0} [", word);
//foreach (var c in word.Components())
//{
// Console.Write("{0}:{1}, ", c.Key, c.Value);
//}
//Console.WriteLine("]");
}
return(tree.Count - count);
}