private void Add(Document document)
{
var words = _tokenizer.Tokenize(document.Text).ToArray();
for (int i = 0; i < words.Length; i++)
{
Add(new Tuple<string, int>(words[i], i), document);
}
}
private void Add(Tuple<string, int> termWithPosition, Document document)
{
var term = termWithPosition.Item1;
if (!_dictionary.ContainsKey(term))
{
_dictionary[term] = new Dictionary<Document, HashSet<int>>();
}
if (!_dictionary[term].ContainsKey(document))
{
_dictionary[term][document] = new HashSet<int>();
}
_dictionary[term][document].Add(termWithPosition.Item2);
}
public int TrackDocumentId(Document document)
{
File.AppendAllLines(_documentIdFile, new[] {$"{document.FilePath}{SplitSymbol}{_index}"});
return _index++;
}