fileanalyser.java

1．文件和目录管理

/**
 * Class FileAnylser used to stastic words,figures and interpunctions in an
 * english text file
 */

package file;
import java.awt.*;
import java.awt.event.*;
import javax.swing.*;
import java.io.*;
import java.io.File;
import java.util.*;

public class FileAnalyser {
	
    //private FileReader reader;
    private StreamTokenizer stok;
	private File f;
	private TreeMap map;
	private TreeSet set;
	
	private int words;
	private int numbers;
	private int interpunctions;
	private String[] postfix = {"asm","bat","bak","c","cfm","cpp","doc",
								"diz","h","htm","html","ihtml","ini",
								"jav","java","jcp","jsp","jcw","me",
								"pas","shtml","syn","tld","txt","xml","xsl"};
	/**
	 * construct a FileAnalyser
	 */
	public FileAnalyser() {
		
		map = new TreeMap();
		set = new TreeSet();
		for(int i = 0;i < postfix.length;i++)
			set.add(postfix[i]);
	}
		
	/**
	 * method to stastic words,figures and interpunctions in an
 	 * english text file
 	 * @param fileName the file name to stastic 
	 */
	public void stastic(String fileName) {	
		
		int tok;
		String s;
		words = 0;
		numbers = 0;
		interpunctions = 0;
		
		try {
			openFile(fileName);	
			FileInputStream reader = new FileInputStream(f);
			stok = new StreamTokenizer(reader);	
			stok.ordinaryChar('.');
      		stok.ordinaryChar('-');		
		
			while((tok = stok.nextToken()) != stok.TT_EOF) {
				if(tok == StreamTokenizer.TT_WORD) {
					s = (String)stok.sval;
					count(map,s);
					words++;
				}
				else if((tok ==StreamTokenizer.TT_NUMBER)) {
				
					s = Double.toString(stok.nval);
					numbers++;
					count(map,s);
					}
				else {
					s = String.valueOf((char)stok.ttype);
					count(map,s);
					interpunctions++;				
				}
				System.out.println(stok.lineno());
			}
		} catch(IOException ioe){
		} catch(Exception e) {
		}
		
	}
	
	/**
	 * private method to check a word whether is ever counted
	 * if is a new word put it into the map and set its value 1
	 * else increment the value of the word
	 * @param map the collection of a file to stastic,stores the words
	 * @param word the file encounts 
	 */
	
	private void count(Map map,String word) {
		
		Integer counter;
		counter = (Integer)map.get(word);
		if(counter == null) {
			
			counter = new Integer(1);
		}
		else {
			
			counter = new Integer(counter.intValue() + 1);
		}
		map.put(word,counter);
	}
	
	/**
	 * method to search how many times a word appears in the file
	 * @param word the word to search
	 * @return the times the word appears 
	 */
	public int frequency(String word) {
		
		Integer counter  = (Integer)map.get(word);
		if(counter == null) {
			
			return 0;
		}
		else return counter.intValue();
	}
	
	/**
	 * method to open file throws FileNameEmptyException,NotTextFileException
	 * NoSuchFileException,IsDirectoryException
	 * @param fileName the file name of a file to open
	 */
	public void openFile(String fileName) throws FileNameEmptyException
		,NotTextFileException,NoSuchFileException,IsDirectoryException {
		
	
		fileName = fileName.trim();
		if(fileName.length() == 0){
		
			throw new FileNameEmptyException();
		}
		int dot = fileName.indexOf(".") + 1;
		String s = fileName.substring(dot,fileName.length());
		if(!set.contains(s))
			throw new NotTextFileException(fileName);
		f = new File(fileName);
		if(!f.exists()) {
				
			throw new NoSuchFileException(fileName);
		}
		else if(f.isDirectory())
			throw new IsDirectoryException(fileName);	
	}
	
	/**
	 * method to get the token in the file
	 */
	public Object[] getContent() {
		
		return map.keySet().toArray();
	}
	/**
	 * method to get the total words in the file
	 */
	public int getTotalWords() {
		
		return words;
	}
	/**
	 * method to get the total figures in the file
	 */
	public int getTotalNumbers() {
		
		return numbers;
		
	}
	
	/**
	 * method to get the total interpunctions in the file
	 */
	public int getTotalInterpunctions() {
		
		return interpunctions;
	}
}