javaRegex解析html樣本分享_java

複製代碼 代碼如下:

package work;

import java.io.BufferedReader;
import java.io.IOException;
import java.io.InputStream;
import java.io.InputStreamReader;
import java.util.regex.Matcher;
import java.util.regex.Pattern;

import org.apache.commons.httpclient.DefaultHttpMethodRetryHandler;
import org.apache.commons.httpclient.HttpClient;
import org.apache.commons.httpclient.HttpException;
import org.apache.commons.httpclient.HttpStatus;
import org.apache.commons.httpclient.methods.GetMethod;
import org.apache.commons.httpclient.params.HttpMethodParams;

public class chuanboyi {

 public static void main(String[] args){
  // TODO Auto-generated method stub
  StringBuffer html = new StringBuffer();
  HttpClient httpclient = new HttpClient();
  //建立GET方法執行個體
  GetMethod getMethod = new GetMethod("http://www.jb51.net");
  //使用系統提供的預設恢複策略
  getMethod.getParams().setParameter(HttpMethodParams.RETRY_HANDLER, new DefaultHttpMethodRetryHandler());
  try{
   //執行GET方法
   int statusCode = httpclient.executeMethod(getMethod);
   if(statusCode != HttpStatus.SC_OK){
    System.out.println("Method is wrong " + getMethod.getStatusLine());
   }
   InputStream responseBody = getMethod.getResponseBodyAsStream();
   BufferedReader reader = new BufferedReader(new InputStreamReader(responseBody,"utf-8"));
   String line = reader.readLine();
   while(line != null){
    html.append(line).append("\n");
    line = reader.readLine();
   }
   reader.close();
   //Regex
   String regex = "<form name=\"compareForm\"[\\s\\S]+>[\\s\\S]+</form>.*<script.*>";
   String regexa ="(?<=<li>)[\\s\\S]+?(?=</li>)";
   Pattern pattern = Pattern.compile(regex);
         Matcher m = pattern.matcher(html);
         StringBuffer str = new StringBuffer();
         int i = 0;
         while(m.find()){
          str.append(m.group());
         }
         pattern = Pattern.compile(regexa);
         m = pattern.matcher(str);
         while(m.find()){
          attrs(m.group());
          i++;
         }
         System.out.println("共有"+i+"條資料！");
  }catch (HttpException e) {
   // TODO: handle exception
   System.out.println("Please check your provided http address!");
   e.printStackTrace();
  }catch (IOException e) {
   // TODO: handle exception
   System.out.println("the line is wrong!");
   e.printStackTrace();
  }finally{
   getMethod.releaseConnection();//釋放連結
  }
 }
 public static void attrs(String str){

  //擷取url的Regex
  String regexURL = "[a-z]+-[0-9]+\\.html";
  //擷取Name的Regex
  String regexName = "(?<=title=\")[[\\w-\\s][^x00-xff]]+(?=\")";
  //擷取圖片的Regex
  String regexPicture = "images.*\\.jpg";

  Pattern patternURL = Pattern.compile(regexURL);
  Pattern patternName = Pattern.compile(regexName);
  Pattern patternPicture = Pattern.compile(regexPicture);
  Matcher mURL = patternURL.matcher(str);
  Matcher mName = patternName.matcher(str);
  Matcher mPicture = patternPicture.matcher(str);
  if(mName.find()){
   System.out.println("名字:"+mName.group());
  }
  if(mURL.find()){
   System.out.println("連結:"+mURL.group());
  }
  if(mPicture.find()){
   System.out.println("圖片:"+mPicture.group());
  }
 } 
}