/***************************************************
  Starter code for CS182
  Assignment 6: Model Merging (Computational Credit)
  **************************************************/

import java.util.*;
import java.io.*;

/** Data serves mainly as a wrapper to the getData method, which
 *  extracts data from a file with 1 sentence per line into a Vector; 
 *  each element in the returned Vector is itself a Vector of Strings.
 *
 *  Example use:
 *
 *  Vector data = Data.getData("a6-training-data.txt");
 *
 */

public class Data {

  /** Read in file with name filename and build an initial grammar. */
  public static Vector getData(String filename){
    Vector data = new Vector();
				
    try {
      // create a BufferedReader based on the file
      BufferedReader datafile = new BufferedReader(new FileReader(filename));
						
      // read in a line
      String line = datafile.readLine();
      while (line != null) {
        String sent = line.trim(); // remove any whitespace around line
        if (!sent.equals("")) {
          // uncomment below to print out sentences read in 
          //	  System.out.println("Read sentence: " + sent);
          data.add(string2Vector(sent));
        }
        line = datafile.readLine();
      }
    }
    catch(java.io.IOException exception) {
      System.out.println("IO exception for file " + filename);
    } // end catch
				
    return data;
  }
		
  /** Take a String and use StringTokenizer to split it into a Vector
      of words. */
  public static Vector string2Vector(String s) {
    Vector v = new Vector();
    StringTokenizer st = new StringTokenizer(s);
    while (st.hasMoreTokens()) {
      v.add(st.nextToken());
    }
    return v;
  }
}