统计单词

package stastic;

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

/**
 * @Author:wangwei
 * @Create on :2008-1-19
 */
public class StatisticsWord {
 public static void main(String args[]) throws IOException {
  BufferedReader buf = new BufferedReader(new FileReader(
    "f://article.txt")); // 假设文章英语article.txt存于F盘下
  StringBuffer sbuf = new StringBuffer();
  String line = null;
  while ((line = buf.readLine()) != null) {
   sbuf.append(line);// 追加到缓冲字符串中
  }
  buf.close();
  Pattern expression = Pattern.compile("[a-zA-Z]+"); // 使用正则表达式匹配单词
  String string1 = sbuf.toString().toLowerCase();
  Matcher matcher = expression.matcher(string1);
  TreeMap myTreeMap = new TreeMap();// 创建树映射 存放键/值对
  int n = 0; // 单词总数
  Object word = null; // 文章中的单词
  Object num = null; // 出现的次数
  while (matcher.find()) {
   word = matcher.group();
   if (word.equals("a") || word.equals("an") || word.equals("the"))
    continue;
   n++;
   if (myTreeMap.containsKey(word)) {// 过滤重复单词
    num = myTreeMap.get(word);// 得到单词出现的次数
    Integer count = (Integer) num;
    myTreeMap.put(word, new Integer(count.intValue() + 1));
   } else {
    myTreeMap.put(word, new Integer(1));// 第一次出现的单词添加到映射中
   }
  }
  Iterator iter = myTreeMap.keySet().iterator();// 得到树映射键集合的迭代器
  Object key = null;
  Integer countWord[] = new Integer[n];
  String wordArray[] = new String[n];
  int index = 0;
  while (iter.hasNext()) {// 使用迭代器遍历树映射的键
   key = iter.next();
   countWord[index++] = (Integer) myTreeMap.get(key);
   wordArray[index - 1] = (String) key;
  }
  // 按出现频率高低排序
  for (int i = 0; i < (index - 1); i++) // 不同单词的总数为index-1;
  {
   for (int j = i + 1; j < index; j++) {
    if ((countWord[j].compareTo(countWord[i])) > 0) {
     Integer temp = countWord[i];
     countWord[i] = countWord[j];
     countWord[j] = temp;
    }
   }
  }
  BufferedWriter bufw = new BufferedWriter(new FileWriter(
    "f://result.txt"));// 保存结果到F盘下result.txt中
  bufw.write("该文章频率最高的三个单词为：");
  bufw.newLine();
  System.out.println("该文章频率最高的三个单词为：");
  for (int cnt = 0; cnt < 3; cnt++) {
   System.out.println(wordArray[index - cnt - 1] + " 单词个数为："
     + countWord[cnt]);
   bufw.write(wordArray[index - cnt - 1] + " 单词个数为：" + countWord[cnt]);
   bufw.newLine();
  }
  bufw.close();
 }
}