// start the run in a separate thread
private void Start_Click(object sender, RoutedEventArgs e)
{
reset.IsEnabled = true;
readfile = new ReadFile();
Start.IsEnabled = false;
stem = checkBox.IsChecked.Value;
Start.Content = "Running...";
Thread readFromFolderThread = new Thread(delegate()
{
DoEverything();
});
readFromFolderThread.Start();
stem = checkBox.IsChecked.Value;
}
private void reset_button(object sender, RoutedEventArgs e)
{
resetPressed = true;
if (readfile != null)
{
readfile.resetReadFile();
}
if (indexer != null)
{
indexer.resetIndexer();
}
Start.Content = "START";
StatsTextBlock.Items.Clear();
readfile = new ReadFile();
indexer = null;
if (File.Exists(PostingsPath + @"\DocPostings.txt"))
{
File.Delete(PostingsPath + @"\DocPostings.txt");
}
if (File.Exists(PostingsPath + @"\TermPostings.txt"))
{
File.Delete(PostingsPath + @"\TermPostings.txt");
}
if (File.Exists(PostingsPath + @"\DocPostingsStemming.txt"))
{
File.Delete(PostingsPath + @"\DocPostingsStemming.txt");
}
if (File.Exists(PostingsPath + @"\TermPostingsStemming.txt"))
{
File.Delete(PostingsPath + @"\TermPostingsStemming.txt");
}
stop();
if (postingsPathOn && dataPathOn)
{
selectQueryFile.IsEnabled = false;
QueryEntry.IsEnabled = false;
Start.IsEnabled = true;
save_result_button.IsEnabled = false;
}
}
// the function that runs readfile, parser, indexer and postings
private void DoEverything()
{
this.Dispatcher.Invoke((Action)(() =>
{
Start.Content = "Parsing...";
}));
Stopwatch mainwatch = new Stopwatch();
mainwatch.Start();
readfile.readFilesFromFolder(DataSetPath, stem);// read the files and parse
if (resetPressed)
{
stop();
return;
}
indexer = new Indexer();
TimeSpan readfileTime = readfile.stopwatch.Elapsed;
int docCount = readfile.documents.Count;
this.Dispatcher.Invoke((Action)(() =>
{
Start.Content = "Indexing Terms...";
StatsTextBlock.Items.Clear();
StatsTextBlock.Items.Add("Read file + Parse Time : " + readfileTime);
StatsTextBlock.Items.Add("Number of documents parsed: " + docCount);
StatsTextBlock.Items.Add("Total Time so far: " + readfileTime);
}));
indexer.createTermAndDictionaries(readfile.documents);//creates the dictionaries (inverted index)
TimeSpan indexerTime = indexer.stopwatch1.Elapsed;
readfile = null;
int termsCount = indexer.sortedTermDictionary.Count;
if (resetPressed)
{
stop();
return;
}
this.Dispatcher.Invoke((Action)(() =>
{
Start.Content = "Creating postings...";
StatsTextBlock.Items.Clear();
StatsTextBlock.Items.Add("Read file + Parse Time : " + readfileTime);
StatsTextBlock.Items.Add("Number of documents parsed: " + docCount);
StatsTextBlock.Items.Add("inverted Index Creation Time : " + indexerTime);
StatsTextBlock.Items.Add("Number of terms indexed : " + termsCount);
StatsTextBlock.Items.Add("Total Time so far: " + (readfileTime + indexerTime));
}));
Stopwatch lastmainwatch = new Stopwatch();
lastmainwatch.Start();
// writes the documents postings file
string Stemming = "";
if (stem)
{
Stemming = "Stemming";
}
using (StreamWriter sw = File.AppendText(@PostingsPath + @"\DocPostings" + Stemming + ".txt"))
{
string json = JsonConvert.SerializeObject(indexer.docDictionary, Formatting.Indented);
sw.Write(json);
}
this.Dispatcher.Invoke((Action)(() =>
{
Start.Content = "Creating postings...";
StatsTextBlock.Items.Clear();
StatsTextBlock.Items.Add("Read file + Parse Time : " + readfileTime);
StatsTextBlock.Items.Add("Number of documents parsed: " + docCount);
StatsTextBlock.Items.Add("inverted Index Creation Time : " + indexerTime);
StatsTextBlock.Items.Add("Number of terms indexed : " + termsCount);
StatsTextBlock.Items.Add("Documents postings created.");
StatsTextBlock.Items.Add("Total Time so far : " + (readfileTime + indexerTime + lastmainwatch.Elapsed));
}));
indexer.docDictionary.Clear();
if (resetPressed)
{
stop();
return;
}
// writes the terms postings file
indexer.CreateTermsPostings(@PostingsPath, stem);
indexer.sortedTermDictionary.Clear();
// the text file from which the inverted index is loaded and showed
indexer = null;
readfile = null;
this.Dispatcher.Invoke((Action)(() =>
{
StatsTextBlock.Items.Clear();
StatsTextBlock.Items.Add("Read file + Parse Time : " + readfileTime);
StatsTextBlock.Items.Add("Number of documents parsed: " + docCount);
StatsTextBlock.Items.Add("inverted Index Creation Time : " + indexerTime);
StatsTextBlock.Items.Add("Number of terms indexed : " + termsCount);
StatsTextBlock.Items.Add("Documents postings created.");
StatsTextBlock.Items.Add("Terms postings created.");
StatsTextBlock.Items.Add("Posting Files creation Time: " + lastmainwatch.Elapsed.ToString());
StatsTextBlock.Items.Add("Total Time : " + (mainwatch.Elapsed));
if (File.Exists(@PostingsPath + @"\TermPostings.txt"))
{
ReadTermsPostingsButton.IsEnabled = true;
}
ShowInvertedIndex.IsEnabled = true;
}));
if (resetPressed)
{
stop();
return;
}
this.Dispatcher.Invoke((Action)(() =>
{
Start.Content = "START (press RESET)";
}));
lastmainwatch.Stop();
}
