public bool buildIndexDic()
{
indexDic = new SortedDictionary <string, PosInfo>();
var fm = new FileManager();
var tp = new TextParser(Algorithm.InvertedIndexSearch);
foreach (var doc in FileManager.docsIds)
{
var rawText = fm.ReadFileText(doc.Value);
if (rawText != string.Empty)
{
var textTokens = tp.parseText(rawText);
var tokenPosition = 0;
foreach (var token in textTokens)
{
PosInfo pi;
if (indexDic.TryGetValue(token, out pi)) //already existed
{
if (indexDic[token].PostingDic.Keys.ToList().Contains(doc.Key)) //doc exist
{ //append new position
indexDic[token].PostingDic[doc.Key].Add(tokenPosition);
}
else //doc not exist
{ //append new doc & position then increase Freq.
indexDic[token].PostingDic.Add(doc.Key, new List <int>()
{
tokenPosition
});
indexDic[token].Frequency++;
}
}
else // New token
{
pi = new PosInfo();
pi.Frequency = 1;
pi.PostingDic = new SortedDictionary <int, List <int> >();
pi.PostingDic.Add(doc.Key, new List <int>()
{
tokenPosition
});
indexDic.Add(token, pi);
}
tokenPosition++;
}
}
}
return(true);
}
public void replaceQueryTokens(ref List <string> tokens, int docsCount, dynamic indexDictionary)
{
for (int i = 0; i < tokens.Count; i++)
{
switch (tokens[i])
{
case "(":
case ")":
break;
case "NOT":
tokens[i] = "~";
break;
case "AND":
tokens[i] = "&";
break;
case "OR":
tokens[i] = "|";
break;
default:
if (_Algorithm == Algorithm.InvertedIndexSearch)
{
var wi = new WordInfo();
//search dictionary for the word
if (indexDictionary.TryGetValue(tokens[i], out wi))
{
//get base2 equivelant word string
tokens[i] = PosToBase2(wi.Posting, docsCount);
}
else //word not found
{
tokens[i] = "0";
}
}
else if (_Algorithm == Algorithm.BooleanSearch)
{
bool[] boolVector;
//search dictionary for the word
if (indexDictionary.TryGetValue(tokens[i], out boolVector))
{
//get base2 equivelant word string
tokens[i] = BoolToBase2(boolVector.ToList());
}
else //word not found
{
tokens[i] = "0";
}
}
else if (_Algorithm == Algorithm.PositionalIndexSearch)
{
var pi = new PosInfo();
//search dictionary for the word
if (indexDictionary.TryGetValue(tokens[i], out pi))
{
//get base2 equivelant word string
tokens[i] = PosToBase2(pi.PostingDic.Keys.ToList(), docsCount);
}
else //word not found
{
tokens[i] = "0";
}
}
break;
}
}
}
