public static InvertedIndex CreateIndex(string[] files)
{
var result = new InvertedIndex();
for (var i = 0; i < files.Length; i++)
{
var extracted = _cut.Extract(files[i]);
var tokens = DefaultAnalyzer.Analyzer(extracted.Text.Split());
foreach (var token in tokens)
{
result.append(token, files[i]);
}
}
foreach (KeyValuePair <string, Dictionary <string, int> > kvp in result.data)
{
Console.Write("Term = {0}", kvp.Key);
foreach (KeyValuePair <string, int> res in kvp.Value)
{
Console.Write(" Posting-List = {0}, Frequency = {1}", res.Key, res.Value);
}
Console.WriteLine();
}
return(result);
}
private void Form1_Load(object sender, EventArgs e)
{
var files = Directory.EnumerateFiles(@"D:\Documents\DataSet\", "*.*", SearchOption.AllDirectories)
.Where(s => s.EndsWith(".pdf") || s.EndsWith(".doc") ||
s.EndsWith(".docx") || s.EndsWith(".ppt") || s.EndsWith(".ppts") ||
s.EndsWith(".xls") || s.EndsWith(".xlsx") || s.EndsWith(".txt") ||
s.EndsWith(".html") || s.EndsWith(".xml"));
string[] filearr = files.ToArray();
Console.WriteLine("Clicked!!" + filearr.Length);
res = StringIndexer.CreateIndex(filearr);
Console.WriteLine("\n\nWorked");
AutoCompleteStringCollection sourceName = new AutoCompleteStringCollection();
foreach (KeyValuePair <string, Dictionary <string, int> > kvp in res)
{
sourceName.Add(kvp.Key);
}
textBox1.AutoCompleteCustomSource = sourceName;
}
/// <summary>
///
/// </summary>
/// <param name="inder"></param>
/// <param name="queryRep"></param>
/// <returns></returns>
public static Dictionary <string, Dictionary <string, int> > matchingFunction(InvertedIndex inder, string queryRep)
{
var res = from result in inder.data
where result.Key.Equals(queryRep)
select result;
return(res.ToDictionary(p => p.Key, p => p.Value));
}