wordcount.java

Hash函数实例

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

class Counter 
{
  private int i = 1;
  int read() { return i; }
  void increment() { i++; }
}

public class WordCount
{
   FileReader file;//文件
   private TreeMap[] counts=new TreeMap[197];//存各个单词
   int mod=0,mostword=0,totalword=0,distinctword=0;//分别是最多，总数，不同单词
    WordCount(String filename)throws FileNotFoundException 
    {
      try 
      {   
          for(int i=0;i<197;i++)//初始化队列
          {
          	counts[i]= new TreeMap();
          }
          file = new FileReader(filename); //打开一个文件
      
      } catch(FileNotFoundException e) 
      {
        System.err.println("Could not open " + filename);
        throw e;
      }
    }
  void cleanup() //关闭文件
  {
    try 
    {
        file.close();
     } catch(IOException e) 
     {
         System.err.println("file.close() unsuccessful");
     }
  }
  void countWords() //记单词的总数
  {
    try 
    {
    
    	    BufferedReader st1 =new BufferedReader(file);//把文件读入缓冲
    		String line;
    		 
       		while((line = st1.readLine())!=null)//文件是否结束
    		{
    			
    		    StringTokenizer st=new  StringTokenizer(line," ,.;?!/\"\f\t,'(',')'");
    			while(st.hasMoreTokens())
    			  {
    			  	    String s = st.nextToken();//取下一个单词
    			  	    totalword ++;//单词总数加一
    			  	   
    			  	    if(s.hashCode()>0)//取Hash码，如为负取为正
    			  	     mod=(s.hashCode()%197);
    			  	     else
    			  	      mod=((-s.hashCode())%197);
    			  	     
    			  		if(counts[mod].containsKey(s))//对单词记数
    			  		((Counter)counts[mod].get(s)).increment();
    			  		else
    			  		counts[mod].put(s,new Counter());//放入队列
                   }
    
    		
      	  }
      
       
    } catch(IOException e) 
    {
      System.err.println("st.nextToken() unsuccessful");
    }
  }
  Collection values() //返回值
  {
    return this.values();
  }
  Set keySet()//返回一个Set
  { 
    return this.keySet(); 
  }
  Counter getCounter(String s,int i) //返回一个Counter对象
  {
    return (Counter)counts[i].get(s);//
  }
/*  public static void main(String[] args) 
       throws FileNotFoundException 
   {
      WordCount wc = new WordCount("file.txt");
      wc.countWords();
      int listlong=0;
      for(int i=0;i<197;i++)//找出最冲突最多的次数
      {
      	if(wc.counts[i].size()>0) //是否有单词
      	{
      	  if(listlong<wc.counts[i].size())//记录最多的冲突
      	  {
      	     listlong=wc.counts[i].size();
       	  }
      	  wc.distinctword += wc.counts[i].size();//不同单词数
      	  Iterator keys = wc.counts[i].keySet().iterator();
      	  System.out.print("counts["+i+"]:");
          while(keys.hasNext())//某队列是否完成
          {
        	String key = (String)keys.next();//取出关键字
        	System.out.print(key+" "+wc.getCounter(key,i).read()+"; ");//输出单词及次数
        	if(wc.getCounter(key,i).read()>wc.mostword)//找出出现最多的单词次数
        	  wc.mostword = wc.getCounter(key,i).read();//记录出现最多的单词次数
          }
          System.out.println();      
        }
         
      }
      
      System.out.println("不同的单词共有"+wc.distinctword);
      
      System.out.println("总单词数"+wc.totalword);
      System.out.println("冲突最多的地址有单词"+listlong+"个冲突最多的地址分别是：");	
      
      for(int i=0;i<197;i++)//输出冲突最多的
      {	 
        if(wc.counts[i].size()==listlong) //找出冲突最多的
        {  
            Iterator keys1 = wc.counts[i].keySet().iterator();
            System.out.print("counts["+i+"]:");//输出
        	while(keys1.hasNext())
           {
              String key1 = (String)keys1.next();//取出关键字
              System.out.print(key1+" :"+wc.getCounter(key1,i).read()+";");//输出单词及次数
           }
            System.out.println();
         }
      }
     System.out.println("出现最多的单词是:");
      for(int i=0;i<197;i++)//统计出现最多的单词
      {	 
        if(wc.counts[i].size()>0) //是否有单词
        {  
            Iterator keys1 = wc.counts[i].keySet().iterator();//把SET转换成迭代器
            while(keys1.hasNext())//某队列是否完成
           {
              String key1 = (String)keys1.next();//取出关键字
              if(wc.getCounter(key1,i).read() == wc.mostword)//找出出现最多的单词
                System.out.print(key1+" :"+wc.getCounter(key1,i).read()+";");//输出单词及次数
           }
         }
      }
      System.out.println();
      wc.cleanup();//关闭文件
  }*/
}