private void Page_Loaded_1(object sender, RoutedEventArgs e)
{
tex.Text = Guess;
Dictionary <string, List <nodePosting> > invertedList = new Dictionary <string, List <nodePosting> >();
nodePosting p = new nodePosting();
using (BinaryReader reader = new BinaryReader(File.Open("inverted.dat", FileMode.Open)))
using (BinaryReader reader2 = new BinaryReader(File.Open("index.dat", FileMode.Open)))
{
while (reader2.BaseStream.Position < reader2.BaseStream.Length)
{
string s = reader2.ReadString();
int a = reader2.ReadInt32();
for (int i = 0; i < a; i++)
{
p.noDoc = reader.ReadInt32();
p.posting = reader.ReadInt32();
p.noPage = reader.ReadInt32();
p.nameDoc = reader.ReadString();
if (!invertedList.ContainsKey(s))
{
invertedList.Add(s, new List <nodePosting> {
p
});
}
else
{
invertedList[s].Add(p);
}
p = null;
p = new nodePosting();
}
}
if (invertedList.ContainsKey(Guess))
{
list_searching.ItemsSource = invertedList[Guess];
}
else
{
tex.Text += " not found";
}
}
}
private void Button_Click(object sender, RoutedEventArgs e)
{
try
{
OpenFileDialog openFileDialog = new OpenFileDialog();
openFileDialog.Multiselect = true; // choose multi File
openFileDialog.Filter = "Text Files|*.txt";
if (openFileDialog.ShowDialog() == true)
{
if (File.Exists(@"inverted.dat"))
{
File.Delete(@"inverted.dat");
}
if (File.Exists(@"index.dat"))
{
File.Delete(@"index.dat");
}
int noPage = 0;
int noDocument = 0;
int noWordPosition = 0;
string endPage = "*****";
Dictionary <string, List <nodePosting> > invertedList = new Dictionary <string, List <nodePosting> >();
if (!File.Exists(@"stopwords.txt"))
{
throw new Exception("1");
}
foreach (string filename in openFileDialog.FileNames)
{
if (System.IO.Path.GetExtension(filename) != ".txt")
{
throw new Exception();
}
noDocument++;//number file
noPage = 0;
noWordPosition = 0;
using (StreamReader sr = new StreamReader(filename))
{
noPage++;
string line;
string[] words;
while ((line = sr.ReadLine()) != null)
{
if (String.IsNullOrEmpty(line) || String.IsNullOrWhiteSpace(line))
{
continue;
}
line = Regex.Replace(line, @"(?:\s+|[^A-Za-z*])", " ");
words = line.Split();
foreach (string word1 in words)
{
noWordPosition++;
string word = word1.Trim().ToLower();
if (word == endPage)
{
noPage++;
continue;
}
else if (String.IsNullOrEmpty(word))
{
continue;
}
else if (stopwords.ContainsKey(word))
{
continue;
}
else
{
nodePosting temp = new nodePosting();
temp.noDoc = noDocument;
temp.posting = noWordPosition;
temp.noPage = noPage;
temp.nameDoc = filename;
if (invertedList.ContainsKey(word))
{
invertedList[word].Add(temp);
}
else
{
List <nodePosting> parts = new List <nodePosting>();
parts.Add(temp);
invertedList.Add(word, parts);
}
}
} // end words
} //end read lines
} // end using
} //end for files
//Create index file
using (BinaryWriter writer = new BinaryWriter(File.Open("inverted.dat", FileMode.Create))) // save posting file
using (BinaryWriter writer2 = new BinaryWriter(File.Open("index.dat", FileMode.Create))) // save index for read in posting file
{
foreach (KeyValuePair <string, List <nodePosting> > kvp in invertedList)
{
foreach (nodePosting nop in kvp.Value)
{
writer.Write(nop.noDoc);
writer.Write(nop.posting);
writer.Write(nop.noPage);
writer.Write(nop.nameDoc);
}
writer2.Write(kvp.Key);
writer2.Write(kvp.Value.Count);
}
}
//
butSerch.IsEnabled = true;// don
}
}//end try
catch (Exception e1)
{
butSerch.IsEnabled = false;// don
if (e1.Message == "1")
{
MessageBox.Show("stopwords.txt does not exist");
}
else
{
MessageBox.Show("Please choose a valid file");
}
}
}