datahandler.java

数据挖掘中的Aprior算法java实现

package datamining;

import java.io.*;

/**
 * A class for handling a database for datamining purposes.
 * Is used for getting data from the database.
 */
public class DataHandler {

  String filename;		// the filename of the database
  BufferedReader reader;	// the buffer to read the input to 
  int samplesize;
  int partsize;			// the size of a part when using Partition
  int counter;
  boolean locked;
  
  /**
   * The default constructor for creating a DataHandler object.
   */
  public DataHandler() { 
  
  }

  /**
   * Constructor for creating a DataHandler object with parameters.
   *
   * @param	filename	the name of the database file
   */
  public DataHandler(String filename) {
    this.filename = filename;
    this.samplesize = 1;
    this.partsize = -1;
    this.counter = 0; 
    this.locked = false;
  }

  /**
   * Method for opening the connection to the database. Checks
   * whether it is allowed to (re)open the connection. Checks also
   * if it should reopen a certain part of the database (partition algorithm).
   */
  public void open() {
    if (!locked) {
      try {
        this.reader = new BufferedReader(new FileReader(filename));
	reader.mark(1000*partsize);
      } catch (Exception e) {}
    } else if (partsize > 0) {
      try {
        reader.reset();
      } catch (Exception e) {
        System.out.println("could not reset to mark");
      }
      counter = 0;
      locked = false;
    } else {
      locked = false;
    }
  }
  
  /**
   * Method for reading a transaction from the database.
   * It returns the transaction as an array of integers.
   *
   * @return		the transaction as an array of integers.
   * 			an array of length 0 is returned if there was no
   * 			transaction or the part was read to the end.
   */
  public int[] read() {
    int[] items = new int[0];
    if (!locked) {
      try {
        for(int i = this.samplesize; i > 1; i--) {
          System.out.println("skipping sample");
          reader.readLine();
        }
        items = toIntArray(reader.readLine().trim().split(" "));
	counter++;
      } catch (Exception e) {
	// eof
        items = new int[0];
	if (counter < partsize)
          locked = true;
      }
      if (counter == partsize) {
        locked = true;
      }
    }
    return items;
  }

  /**
   * Sets now manieth transaction is given from the database.
   */ 
  public void setSample(int size) {
    this.samplesize = size;
  }

  /**
   * Sets the size of a part.
   */
  public void setPartsize(int size) {
    this.partsize = size;
  }

  /**
   * Advances the mark in the database to the current point, which is
   * the new beginning of the new part.
   */
  public void nextPart() {
    try {
      reader.mark(1000*partsize);
      counter = 0;
    } catch (Exception e) {
      System.out.println("mark could not be set");
    }
  }

  /**
   * Converts a String array to integer array.
   *
   * @param	str	the array of Strings to be converted
   * @return		the array of integers
   */
  private int[] toIntArray(String[] str) {
    int[] intArray = new int[str.length];
    for(int i=0; i < str.length; i++) {
      intArray[i] = new Integer(str[i].trim()).intValue();	
    }
    return intArray;
  }

}