public bool buildIndexDic()
{
indexDic = new SortedDictionary <string, WordInfo>();
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);
foreach (var token in textTokens)
{
WordInfo wi;
if (indexDic.TryGetValue(token, out wi))
{
if (!indexDic[token].Posting.Contains(doc.Key))
{
indexDic[token].Frequency++;
indexDic[token].Posting.Add(doc.Key);
}
}
else // New token
{
wi = new WordInfo();
wi.Frequency = 1;
wi.Posting = new List <int>()
{
doc.Key
};
indexDic.Add(token, wi);
}
}
}
}
//Sort all postings
foreach (var record in indexDic)
{
record.Value.Posting.Sort();
}
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;
}
}
}