jsoup相关的工具类

package com.kinth.datacrawl.admin.web.webmagic;

import java.io.IOException;

import org.jsoup.Jsoup;
import org.jsoup.nodes.Document;

/**
 * jsoup相关的工具类
 * @author mbm
 *
 */
public class JsoupUtils {

    /**
     * 对url进行连接尝试，如果连接成功则返回true，否则返回false
     * @param url
     * @return
     */
    public static boolean attempConnect(String url){
        try {
            Jsoup.connect(url)
                 .timeout(3*1000)
                 .get();
            return true;
        } catch (Exception e) {
            return false;
        }
    }

    /**
     * 下载指定url的文档对象
     * @param url
     * @return
     * @throws IOException 
     */
    public static Document downloadDocument(String url) throws IOException{
        Document document = Jsoup.connect(url).timeout(5*1000).get();
        return document;
    }
}