MusicDataCollection.

package musicdatacollection;

/*
 * To change this template, choose Tools | Templates
 * and open the template in the editor.
 */

import de.umass.lastfm.*;
import java.util.Vector;
import java.io.FileWriter; 
import java.io.BufferedWriter; 
import java.io.IOException; 
import java.util.Collection; 

/**
 *
 * @author Rawan
 */
public class MusicDataCollection {

    /**
     * @param args the command line arguments
     */
    static int   numofpages = 800; 
    public static void main(String[] args) {
        Caller.getInstance().setUserAgent("RawanMG");
    Caller.getInstance().setDebugMode(true);
   String key = "74e2bf6d43f85e4a5d3a89e6eee3fd7b"; //this is the key used in the Last.fm API examples
    String user = "RawanMG";
    
     Vector<String> uniquetracks = new Vector<String>(); //to store the track name
     Vector<String> artists = new Vector<String>(); //to store the artist name
    Vector<Integer> uniqueplaycount = new Vector<Integer>(); //to store the playcount
    Vector<Integer> ulovedtracks = new Vector<Integer>(); //to store whether the track is loved or not
    Vector<String> uniqueartist = new Vector<String>(); //to store the artist vs id(index)
    
    //Testing data 
    /* Vector<String> uniquetrackstest = new Vector<String>(); //to store the track name
     Vector<String> artiststest = new Vector<String>(); //to store the artist name
    Vector<Integer> uniqueplaycounttest = new Vector<Integer>(); //to store the playcount
    Vector<Integer> ulovedtrackstest = new Vector<Integer>(); //to store whether the track is loved or not
    Vector<String> uniqueartisttest = new Vector<String>(); //to store the artist vs id(index) 
    */
   
    //Collect Loved tracks  
   Vector<String> lovedtracks = new Vector<String>(); //to store all the loved tracks
     PaginatedResult<Track> loved; 
      int c = 0;
     for(int x=1; x<=numofpages; x++){
   loved = User.getLovedTracks(user, x, key); 
   
   
     for(Track track: loved){
         
       lovedtracks.add(track.getName());
        c++; 
    }
     }
    /* loved = User.getLovedTracks(user, 2, key); 
       for(Track track: loved){
        lovedtracks.add(track.getName());
       c++; 
    }*/
     System.out.println(c); 
     PaginatedResult<Track> tracks; 
    for(int j=1; j<=numofpages; j++)
    { //First 200 tracks (page 1) 
     tracks = User.getRecentTracks(user, j, 100, key);
          
    for(Track track: tracks){
        String t= track.getName(); 
        //if the trakc is added to the vector increment the playcount
        if(uniquetracks.contains(t))
        {
            int loc = uniquetracks.indexOf(t);//location of the track in the Vector
            uniqueplaycount.set(loc, uniqueplaycount.elementAt(loc)+1); //increments the playcount
        }
        else //if the track is not in the vector add it with playcount = 1
        { uniquetracks.addElement(t);
          String artist = track.getArtist(); 
          artists.add(artist);
          uniqueplaycount.addElement(1);
          if(!uniqueartist.contains(artist))
             uniqueartist.addElement(artist); 
          if (lovedtracks.contains(t))
              ulovedtracks.add(1);
          else
              ulovedtracks.add(0); 
        }
    }
    }
   
    
    //Second 200 tracks (page 2) 
   /* tracks = User.getRecentTracks(user, 2, 200, key);
    
       for(Track track: tracks){
        String t= track.getName(); 
        //if the trakc is added to the vector increment the playcount
        if(uniquetracks.contains(t))
        {
            int loc = uniquetracks.indexOf(t);//location of the track in the Vector
            uniqueplaycount.set(loc, uniqueplaycount.elementAt(loc)+1); //increments the playcount
        }
     else //if the track is not in the vector add it with playcount = 1
        { uniquetracks.addElement(t);
          String artist = track.getArtist(); 
          artists.add(artist);
          uniqueplaycount.addElement(1);
          if(!uniqueartist.contains(artist))
             uniqueartist.addElement(artist); 
          if (lovedtracks.contains(t))
              ulovedtracks.add(1);
          else
              ulovedtracks.add(0); 
        }
    }
   //Third 200 tracks (page 3)
           tracks = User.getRecentTracks(user, 3, 200, key);
    
       for(Track track: tracks){
        String t= track.getName(); 
        //if the trakc is added to the vector increment the playcount
        if(uniquetracks.contains(t))
        {
            int loc = uniquetracks.indexOf(t);//location of the track in the Vector
            uniqueplaycount.set(loc, uniqueplaycount.elementAt(loc)+1); //increments the playcount
        }
      else //if the track is not in the vector add it with playcount = 1
        { uniquetracks.addElement(t);
          String artist = track.getArtist(); 
          artists.add(artist);
          uniqueplaycount.addElement(1);
          if(!uniqueartist.contains(artist))
             uniqueartist.addElement(artist); 
          if (lovedtracks.contains(t))
              ulovedtracks.add(1);
          else
              ulovedtracks.add(0); 
        }
    } */
  try{
    FileWriter fw = new FileWriter("musicdata.txt"); //holds track data in the form ( track_title artist_index playcount loved_or_not) 
    BufferedWriter bw = new BufferedWriter(fw); 
       System.out.println("Music Data: " +uniquetracks.size()); 
    for(int i =0; i<uniquetracks.size(); i++){
        bw.write(uniquetracks.elementAt(i) + "  " +uniqueartist.indexOf(  artists.elementAt(i))+ " " + uniqueplaycount.elementAt(i) + "  " + ulovedtracks.elementAt(i)+ "\n" ); 
    }
    bw.close();
    //50% of the data for Training
     fw = new FileWriter("trainingmusicdata.txt");
     bw = new BufferedWriter(fw); 
       System.out.println("Training Data: " + uniquetracks.size()/2); 
    for(int i =0; i<uniquetracks.size()/2; i++){
        bw.write(uniqueartist.indexOf(  artists.elementAt(i))+ "," + uniqueplaycount.elementAt(i) + "," + ulovedtracks.elementAt(i) ); 
        bw.write("\n"); 
    }
    bw.close();
    //20% of the data for Validation
        
     fw = new FileWriter("validationmusicdata.txt");
     bw = new BufferedWriter(fw); 
        
     int startv = uniquetracks.size()/2 ; //starting index for testing data
     System.out.println("startv index " + startv); 
     int sizev = (int) (uniquetracks.size()*(20.0/100.0)); //size of validation data 
     System.out.println("Validation Data: " + sizev); //the size of the data is 
     
     
    for(int i =startv; i<startv+ sizev; i++){
        bw.write(uniqueartist.indexOf(  artists.elementAt(i))+ "," + uniqueplaycount.elementAt(i) + "," + ulovedtracks.elementAt(i) ); 
        bw.write("\n"); 
    }
    bw.close();
    //30% of the data for Testing
     fw = new FileWriter("testingmusicdata.txt");
     bw = new BufferedWriter(fw); 
        
     int start = startv + sizev; //starting index for testing data
     int sizet = (int) (uniquetracks.size()*(30.0/100.0)); //size of testing data
     System.out.println("Testing Data: " + sizet); //the size of the data is 
     
     
    for(int i =start; i<uniquetracks.size(); i++){
        bw.write(uniqueartist.indexOf(  artists.elementAt(i))+ "," + uniqueplaycount.elementAt(i) + "," + ulovedtracks.elementAt(i) ); 
        bw.write("\n"); 
    }
    bw.close();
    //a new file to map artist id to artist name
    FileWriter fw2 = new FileWriter("artistdata.txt");
    bw = new BufferedWriter(fw2); 
    for(int i = 0; i<uniqueartist.size(); i++){
        bw.write(i + " " + uniqueartist.elementAt(i)); 
        bw.write("\n"); }
    bw.close();
    
    bw = new BufferedWriter(new FileWriter("filesize.txt")); 
    bw.write("Music Data: " +uniquetracks.size()); 
     bw.write("\n");
     bw.write("Training Data: " + uniquetracks.size()/2);
       bw.write("\n");
     bw.write("Testing Data: " + (uniquetracks.size()-start)); 
     bw.close(); 
    }catch(IOException e){
      System.out.println(e.getMessage()); 
    }catch(Exception e){
      System.out.println(e.getMessage()); 
      }
    }

}/*
 * To change this template, choose Tools | Templates
 * and open the template in the editor.
 */