//reset the load and gui of part 2
private void Reset_Part2_Click(object sender, RoutedEventArgs e)
{
//reset GUI
r = null;
p = null;
ind = null;
showcatch.IsEnabled = false;
showDic.IsEnabled = false;
runQuery.IsEnabled = false;
QueryTextBox.Clear();
isWikiExp.IsChecked = false;
isSumCheck.IsChecked = false;
stemmcheckBox.IsChecked = false;
//Delete Last File
try
{
if (File.Exists(Results.LastFile))
{
File.Delete(Results.LastFile);
}
}
catch (IOException)
{
Debug.WriteLine("CANT DELETE!!!!!");
}
//Garbage Collector
GC.Collect();
GC.WaitForPendingFinalizers();
//notification
System.Windows.Forms.MessageBox.Show("The Memory is clean!", "Done!", MessageBoxButtons.OK, MessageBoxIcon.Warning);
}
public Ranker(string path, Parser p, ReadFile r, Indexer ind, bool isStem)
{
this.path = path;
avgDL = 0;
k = 1.3;
this.isStem = isStem;
this.r = r;
this.p = p;
this.ind = ind;
DocWeight = new Dictionary <string, double>();
Docsmaxtf = new Dictionary <string, int>();
docslength = new Dictionary <string, double>();
string file;
//check if we can do fast loading
if (isStem)
{
file = path + @"\rankerStem";
}
else
{
file = path + @"\ranker";
}
if (File.Exists(file))
{
//if File exiest loading dictionaries in order to run faster
using (FileStream fs = new FileStream(file, FileMode.Open))
{
IFormatter bf = new BinaryFormatter();
DocWeight = (Dictionary <string, double>)bf.Deserialize(fs); //read object
Docsmaxtf = (Dictionary <string, int>)bf.Deserialize(fs); //read object
docslength = (Dictionary <string, double>)bf.Deserialize(fs); //read object
avgDL = (double)bf.Deserialize(fs);
}
}
else
{
//calculate the dictionaries in order to run- long process
calculateWeight();
//after calculate save it for next time
using (FileStream fs = new FileStream(file, FileMode.Create))
{
IFormatter bf = new BinaryFormatter();
bf.Serialize(fs, DocWeight); //write object
bf.Serialize(fs, Docsmaxtf); //write object
bf.Serialize(fs, docslength); //write object
bf.Serialize(fs, avgDL); //write object
}
}
}
private void StartEngine()
{
//Delete Exists Directory
if (Directory.Exists(pathclose + @"\AfterPost"))
{
Directory.Delete(pathclose + @"\AfterPost", true);
}
if (Directory.Exists(pathclose + @"\tempPost"))
{
Directory.Delete(pathclose + @"\tempPost", true);
}
/* if (Directory.Exists(pathclose + @"\CacheDic"))
* Directory.Delete(pathclose + @"\CacheDic", true);*/
//initlizing reader,parser,indexer
r = new ReadFile(pathopen);
p = new Parser(pathopen + @"\stop_words.txt", isStem);
ind = new Indexer(pathclose, isStem);
//For program timing
DateTime start = DateTime.Now;
for (int i = 0; i < r.files.Count; i++)
{
Debug.WriteLine(i);
//return dictionary<DocNO, TEXT>
d = r.ProccessDocs(r.files[i]);
foreach (string s in d.Keys)
{
//Parse DOC
Dictionary <string, termInfo> docdic = p.Parse(d[s]);
//make temp Post file
ind.tempPosting(docdic, s.Trim(' '), p.maxterm, p.maxtf);
}
}
StreamWriter sw = new StreamWriter(pathopen + @"\docsMap.txt");
sw.WriteLine(r.sb.ToString());
sw.Close();
ind.tempPost(); //Write the last dictionary
p.clearSteam(); // cleans the stemmers dictionary
ind.mergefile(); // merge and split
ind.writeTextDic(); // WriteDic For show
TimeSpan ts3 = DateTime.Now - start;
//The requested popup of this run
System.Windows.Forms.MessageBox.Show("Number Of Docs Indexed : " + ind.TotalDoc + "\nTime Of Running : " + ts3.TotalSeconds + "\nIndex Size[bytes] : " + ind.PostSize + "\nCache Size[bytes] : " + ind.cacheSize, "Done!", MessageBoxButtons.OK, MessageBoxIcon.Information);
}
private void Reset_Click(object sender, RoutedEventArgs e)
{
//restart gui
Corpustextbox.IsEnabled = true;
CorpusFolderBrowse.IsEnabled = true;
pathopen = "";
Corpustextbox.Clear();
if (ind != null)
{
ind.swrite.Close();
}
if (Directory.Exists(pathclose + "\\AfterPost"))
{
Directory.Delete(pathclose + "\\AfterPost", true);
}
ind = null;
r = null;
p = null;
//DELETE DICTIONERY
string s;
if (isStem)
{
s = "\\CacheDic\\dicStem.txt";
}
else
{
s = "\\CacheDic\\dic.txt";
}
if (File.Exists(pathclose + s))
{
File.Delete(pathclose + s);
}
showDic.IsEnabled = false;
showcatch.IsEnabled = false;
//Garbage Collector
GC.Collect();
GC.WaitForPendingFinalizers();
//notification
System.Windows.Forms.MessageBox.Show("The Memory is clean!", "Done!", MessageBoxButtons.OK, MessageBoxIcon.Warning);
}
private void startButton_Click(object sender, RoutedEventArgs e)
{
corpusFolder = InputPathTxt.Text;
postingsFolder = postingsTextBox.Text;
if (corpusFolder == "" || postingsFolder == "")
{
System.Windows.MessageBox.Show("Path Required!", "Error");
}
else
{
labelClock.Visibility = Visibility.Visible;
imageClock.Visibility = Visibility.Visible;
Stopwatch stopWatch = new Stopwatch();
stopWatch.Start();
System.Windows.MessageBox.Show("The process started");
rf = new ReadFile(corpusFolder, postingsFolder, stemmer);
rf.MainProcess();
foreach (string lang in rf.languages)
{
comboBoxLang.Items.Add(lang);
}
string time = (stopWatch.ElapsedMilliseconds / 1000).ToString();
labelClock.Visibility = Visibility.Hidden;
imageClock.Visibility = Visibility.Hidden;
Dictionary <string, TermData> theDictionary = rf.mainDictionary;
List <string> termList = rf.termList;
dictionary = new Dictionary <string, int>();
System.Windows.MessageBox.Show("Finish indexing in " + time + " seconds\n"
+ "number of Documents: " + rf.docsCounter + "\n"
+ "number of terms: " + theDictionary.Count
, "Finish Inexing");
// add the term to dictionary for show
foreach (string term in termList)
{
dictionary.Add(term, theDictionary[term].NumInCorpus);
}
stopWatch.Stop();
}
}
//delete all the folders and reset the class Read File
private void resetButton_Click(object sender, RoutedEventArgs e)
{
//delete all files from the posting's directory
if (postingsFolder == "")
{
System.Windows.MessageBox.Show("posting directory path required", "Error");
}
else
{
System.IO.DirectoryInfo di = new DirectoryInfo(postingsFolder);
foreach (DirectoryInfo dir in di.GetDirectories())
{
dir.Delete(true);
}
rf = null;
System.Windows.MessageBox.Show("Reset Completed", "Error");
}
}
//Load all files of part 2-corpus, stop words, cache, dictionary ,posting and rank
private void Load2_click(object sender, RoutedEventArgs e)
{
//Folder Chooser
var dlg = new FolderBrowserDialog();
System.Windows.Forms.DialogResult result = dlg.ShowDialog(this.GetIWin32Window());
//change the source path
if (dlg.SelectedPath != "")
{
pathopen = dlg.SelectedPath;
pathclose = dlg.SelectedPath;
//init all the first part objects
ind = new Indexer(pathclose, isStem);
p = new Parser(pathopen + @"\stop_words.txt", isStem);
r = new ReadFile(pathopen + @"\corpus\");
string dic;
string cache;
if (isStem)//check if stem
{
dic = pathclose + @"\CacheDic\dicStem.dicx";
cache = pathclose + @"\CacheDic\cacheStem.chex";
}
else
{
dic = pathclose + @"\CacheDic\dic.dicx";
cache = pathclose + @"\CacheDic\cache.chex";
}
try
{
//load dic
using (FileStream fs = new FileStream(dic, FileMode.Open))
{
IFormatter bf = new BinaryFormatter();
ind.dic = (Dictionary <string, DicRecord>)bf.Deserialize(fs);//read object
}
//load cache
using (FileStream fs = new FileStream(cache, FileMode.Open))
{
IFormatter bf = new BinaryFormatter();
ind.cache = (Dictionary <string, List <PostingInfo> >)bf.Deserialize(fs);//read object
}
}
catch (IOException)
{
//cant find load and cache files in currect folder
System.Windows.Forms.MessageBox.Show("Files Missing, can't Load", "ERROR!", MessageBoxButtons.OK, MessageBoxIcon.Error);
return;
}
//for vieiwing
ind.writeTextChache();
showcatch.IsEnabled = true;
ind.writeTextDic();
showDic.IsEnabled = true;
//new ranker and load the dictionaries of the class if the file exists in the selected folder
rank = new Ranker(pathclose, p, r, ind, isStem);
searcher = new Searcher(p, ind, rank, pathopen);
//open the run btn
runQuery.IsEnabled = true;
//notify when finished
System.Windows.Forms.MessageBox.Show("Ready To search!!", "Done!", MessageBoxButtons.OK, MessageBoxIcon.Information);
}
}
