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

/*
根據(jù)指定的規(guī)則，通過構(gòu)造正則表達(dá)式獲取網(wǎng)址
*/

public class Urls
{
    private String startUrl;                                         //開始采集網(wǎng)址
    String  urlContent;
    String ContentArea;
    private String strAreaBegin ,strAreaEnd ;            //采集區(qū)域開始采集字符串和結(jié)束采集字符串
    private String stringInUrl,stringNotInUrl;       
    String strContent;//獲得的采集內(nèi)容
    String[] allUrls;                                                            //采集到的所有網(wǎng)址
    private String  regex;                                                 //采集規(guī)則

    UrlAndTitle   urlAndTitle=new UrlAndTitle();    //存儲網(wǎng)址和標(biāo)題                   

   
    public static void main(String[] args)
    {
         Urls myurl=new Urls("<body","/body>");
         myurl.getStartUrl("http://www.zuzwn.com/");
         myurl.getUrlContent();
         myurl.getContentArea();
         myurl.getStartUrl("http://www.zuzwn.com/");
         myurl.getStringNotInUrl("google");
         myurl.Urls();

        //System.out.println("startUrl:"+myurl.startUrl);
        //System.out.println("urlcontent:"+myurl.urlContent);
        //System.out.println("ContentArea:"+myurl.ContentArea);

    }

   
    //初始化構(gòu)造函數(shù) strAreaBegin 和strAreaEnd

    public Urls (String strAreaBegin,String strAreaEnd)
    {
        this.strAreaBegin=strAreaBegin;
        this.strAreaEnd=strAreaEnd;
  }

  //
    public void Urls()
    {
        int i=0;
        //String regex ="<a href="?'?http://[a-zA-Z0-9]+/.[a-zA-Z0-9]+/.[a-zA-Z]+/?[/.?[/S|/s]]+[a>]$";
        String regex ="<a.*?/a>";
         //String regex ="http://.*?>";
        Pattern pt=Pattern.compile(regex);
        Matcher mt=pt.matcher(ContentArea);
        while(mt.find())
         {
                 System.out.println(mt.group());
                 i++;

                 //獲取標(biāo)題
                 Matcher title=Pattern.compile(">.*?</a>").matcher(mt.group());
                 while(title.find())
                 {
                      System.out.println("標(biāo)題:"+title.group().replaceAll(">|</a>",""));
                 }

                 //獲取網(wǎng)址
                 Matcher myurl=Pattern.compile("href=.*?>").matcher(mt.group());
                 while(myurl.find())
                 {
                      System.out.println("網(wǎng)址:"+myurl.group().replaceAll("href=|>",""));
                 }

                 System.out.println();

                
         }

        System.out.println("共有"+i+"個(gè)符合結(jié)果");

    }   

    //獲得開始采集網(wǎng)址
    public void getStartUrl(String startUrl)
    {
        this.startUrl=startUrl;
    }

    //獲得網(wǎng)址所在內(nèi)容;
    public void getUrlContent()
    {

        StringBuffer is=new StringBuffer();
        try
        {
            URL myUrl=new URL(startUrl);
            BufferedReader br= new BufferedReader(
                                                        new InputStreamReader(myUrl.openStream()));

            String s;                                               
            while((s=br.readLine())!=null)
            {
                is.append(s);
            }                                           
            urlContent=is.toString();
        }
    catch(Exception e)

    {
        System.out.println("網(wǎng)址文件未能輸出");
        e.printStackTrace();
    }

       
    }

   
    //獲得網(wǎng)址所在的匹配區(qū)域部分
    public void getContentArea()
    {
         int pos1=0,pos2=0;
         pos1= urlContent.indexOf(strAreaBegin)+strAreaBegin.length();
         pos2=urlContent.indexOf(strAreaEnd,pos1);
         ContentArea=urlContent.substring(pos1,pos2);
    }

    //以下兩個(gè)函數(shù)獲得網(wǎng)址應(yīng)該要包含的關(guān)鍵字及不能包含的關(guān)鍵字
    //這里只做初步的實(shí)驗(yàn)。后期，保護(hù)的關(guān)鍵字及不能包含的關(guān)鍵字應(yīng)該是不只一個(gè)的。
    public void getStringInUrl(String stringInUrl)
    {
         this.stringInUrl=stringInUrl;       

    }

    public void getStringNotInUrl(String stringNotInUrl)
    {
        this.stringNotInUrl=stringNotInUrl;
    }

    //獲取采集規(guī)則

    //獲取url網(wǎng)址
    public void getUrl()
    {

    }

    public String getRegex()
    {
        return regex;

    }

    class UrlAndTitle
    {
        String myURL;
        String title;
    }
}