private void loadDictionaryButton_Click()
{
string dictPath = @"C:\Users\ירדן\Desktop\לימודים\שנה ד\סמס1\אחזור מידע\data\posting";
bool stemm = false;
d = new Dictionary <string, TermInfo>();
docs = new Dictionary <string, DocumentInfo>();
string p = "";
if (stemm)
{
p = dictPath + @"\DictionaryStemm.txt";
}
else
{
p = dictPath + @"\Dictionary.txt";
}
if (!File.Exists(p))
{
System.Windows.Forms.MessageBox.Show("Dictionary not found !");
}
else
{
Thread tDocs = new Thread(() => loadDocs(dictPath, stemm));
tDocs.Start();
string[] tempDic = File.ReadAllLines(p);
for (int i = 0; i < tempDic.Length; i += 2)
{
string[] t = tempDic[i].Split(',');
string term = t[0];
TermInfo ti = new TermInfo(t[0], int.Parse(t[1]), int.Parse(t[2]), t[3][0]);
ti.setPointer(int.Parse(t[4]));
t = tempDic[i + 1].Split(',');
foreach (string next in t)
{
ti.nextString.Add(next);
}
d[term] = ti;
}
tDocs.Join();
System.Windows.Forms.MessageBox.Show("load data succeeded");
}
}
/// <summary>
/// build mini postings for dictionary
/// </summary>
/// <param name="terms"></param>
public void buildInvertedIndex(Dictionary <string, TermInfo> terms)
{
List <string> sortTerms = terms.Keys.ToList();
sortTerms.Sort();
string fullPath = path + @"\" + idx + ".txt";
using (FileStream fs = File.Create(fullPath))
{
using (StreamWriter sw = new StreamWriter(fs))
{
string newL = "";
foreach (string t in sortTerms)
{
// newL = String.Join(",", terms[t].locations);
newL = terms[t].getStringOfDocs();
sw.WriteLine(t);
sw.Write(newL);
sw.WriteLine();
//Dictionary :
int listSize = terms[t].locations.Count;
int frecCurrPost = terms[t].corpusF;
if (fullD.ContainsKey(t))
{
fullD[t].docF += listSize;
fullD[t].corpusF += frecCurrPost;
fullD[t].addNextString(terms[t].nextStringFull);
}
else
{
TermInfo newT = new TermInfo(t, listSize, frecCurrPost, terms[t].type);
newT.nextStringFull = terms[t].nextStringFull;
fullD[t] = newT;
}
}
terms.Clear();
}
}
idx++;
}
/// <summary>
/// read amount of files, for each file split to docs, parse the document, and build mini dictionary
/// </summary>
/// <param name="amount"></param>
/// <returns></returns>
public Dictionary <string, TermInfo> readBatch(int amount)
{
if (files.Length > 0)
{
Dictionary <string, TermInfo> termsInFiles = new Dictionary <string, TermInfo>();
for (int i = 0; i < amount; i++)
{
string text = System.IO.File.ReadAllText(files[idxFile]);
List <string> docs = getDocs(text);
foreach (string doc in docs)
{
parser = new Parse(stopWords, stemm);
Dictionary <string, int> termsInCurrDoc = parser.parseDoc(doc); // term and his tf from the doc
DocumentInfo d = parser.getDoc(); // doc details
d.numTermsInDoc = parser.countTerms;
#region doc FBIS-3366
/* string s = d.docID.Trim(' ');
* if (s.Equals("FBIS3-3366"))
* {
*
* using (StreamWriter sw = new StreamWriter(p+"\\Y.txt"))
* {
* foreach (string ter in termsInCurrDoc.Keys)
* {
* sw.WriteLine(String.Join(",",new string[]{ter , termsInCurrDoc[ter].ToString() }));
* }
*
* }
* }*/
#endregion
// add Terms From Doc To dictionary
foreach (KeyValuePair <string, int> curr in termsInCurrDoc)
{
if (!termsInFiles.ContainsKey(curr.Key))
{
TermInfo t = new TermInfo();
t.term = curr.Key;
t.locations[d.docID] = curr.Value;
t.type = parser.termType[t.term];
t.corpusF = curr.Value;
t.docF = 1;
if (parser.nextTerm.ContainsKey(curr.Key))
{
t.addNextString(parser.nextTerm[curr.Key]);
}
t.inFirstThird[d.docID] = parser.termsInFirstThirdOfText[curr.Key];
t.inLast10[d.docID] = parser.termsInLast10OfText[curr.Key];
termsInFiles[curr.Key] = t;
}
else
{
termsInFiles[curr.Key].locations[d.docID] = curr.Value;
termsInFiles[curr.Key].corpusF += curr.Value;
termsInFiles[curr.Key].docF++;
if (parser.nextTerm.ContainsKey(curr.Key))
{
termsInFiles[curr.Key].addNextString(parser.nextTerm[curr.Key]);
}
termsInFiles[curr.Key].inFirstThird[d.docID] = parser.termsInFirstThirdOfText[curr.Key];
termsInFiles[curr.Key].inLast10[d.docID] = parser.termsInLast10OfText[curr.Key];
}
}
docsInDataBase.Add(d);
countDocL += d.numTermsInDoc;
}
idxFile++;
}
return(termsInFiles);
}
return(null);
}
/// <summary>
/// load dictionary to memory
/// </summary>
/// <param name="sender"></param>
/// <param name="e"></param>
private void loadDictionaryButton_Click(object sender, RoutedEventArgs e)
{
string dictPath = PathPostingDictionary.Text;
bool stemm = false;
if (stemmingCheck.HasContent && stemmingCheck.IsChecked == true)
{
stemm = true;
}
d = new Dictionary <string, TermInfo>();
docs = new Dictionary <string, DocumentInfo>();
string p = "";
if (stemm)
{
p = dictPath + @"\DictionaryStemm.txt";
}
else
{
p = dictPath + @"\Dictionary.txt";
}
if (!File.Exists(p))
{
System.Windows.Forms.MessageBox.Show("Dictionary not found !");
}
else
{
Thread tDocs = new Thread(() => loadDocs(dictPath, stemm));
tDocs.Start();
string[] tempDic = File.ReadAllLines(p);
for (int i = 0; i < tempDic.Length; i += 2)
{
string[] t = tempDic[i].Split(',');
string term = t[0];
TermInfo ti = new TermInfo(t[0], int.Parse(t[1]), int.Parse(t[2]), t[3][0]);
ti.setPointer(int.Parse(t[4]));
t = tempDic[i + 1].Split(',');
foreach (string next in t)
{
ti.nextString.Add(next);
}
/* t = tempDic[i + 2].Split(',');
* foreach (string next in t)
* {
* ti.synForTerm.Add(next);
* }*/
d[term] = ti;
}
/*
* foreach (string term in tempDic)
* {
* string[] t = term.Split(',');
* TermInfo ti = new TermInfo(t[0], int.Parse(t[1]), int.Parse(t[2]), t[3][0]);
* ti.setPointer(int.Parse(t[4]));
* d[t[0]] = ti;
* }*/
tDocs.Join();
System.Windows.Forms.MessageBox.Show("load data succeeded");
}
}
