public MatchedDocument[] Search(string query)
{
var queryNotWords = GetAllowedNormalizedWords(parser.GetAllNotWords(query)).ToHashSet <string>();
queryNotWords.Add("not");
var queryWords = parser.GetAllWords(query)
.Where(word => !queryNotWords.Contains(word));
queryWords = GetAllowedNormalizedWords(queryWords);
var queryPhrases = GetAllowedNormalizedWords(parser.GetAllPhrases(query))
.Select(parser.RemoveDelimeters)
.Select(phrase => GetAllowedNormalizedWords(parser.GetAllWords(phrase)));
if (!queryWords.Any() && !queryPhrases.Any())
{
return(new MatchedDocument[0]);
}
var matchedDocuments = documents
.Where(doc => queryWords.All(doc.Value.Contains) &&
queryNotWords.All(word => !doc.Value.Contains(word)) &&
queryPhrases.All(phrase => doc.Value.Contains(phrase)));
return(matchedDocuments
.Select(doc => new MatchedDocument(doc.Key, TFIDF(queryWords, doc.Value)))
.ToArray());
}
