Jsoup解析html实现查询功能 Jsoup解析html实现招聘信息查询功能

 <div class="newlist_list_content" id="newlist_list_content_table"> 
 <table width="853" class="newlist" cellpadding="0" cellspacing="0">
  <tbody>
  <tr>
  <th class="zwmc"><span>职位名称</span></th>
  <th class="gsmc">公司名称</th> 
  <th class="zwyx">职位月薪</th>
  <th class="gzdd">工作地点</th>
  <th class="gxsj">发布日期</th>
  </tr>
  </tbody>
</table> 

 <table cellpadding="0" cellspacing="0" width="853" class="newlist"> 
  <tbody>
  <tr> 
  <td class="zwmc"> <input type="checkbox" name="vacancyid" value="CC415107716J90250224000_635_1_03_201__1_" onclick="zlapply.uncheckAll('allvacancyid')" />
  <div style="width:300px;float:left">
   <a style="font-weight: bold" par="ssidkey=y&ss=201&ff=03" href="http://jobs.zhaopin.com/415107716250224.htm" rel="external nofollow" target="_blank">Android 开发工程师</a>
  </div> </td> 

  <td class="gsmc"><a href="http://special.zhaopin.com/pagepublish/41510771/index.html" rel="external nofollow" target="_blank">南京天洑软件有限公司</a></td> 
  <td class="zwyx">面议</td> 
  <td class="gzdd">南京</td> 
  <td class="gxsj"><span>10-24</span><a class="newlist_list_xlbtn" href="javascript:;" rel="external nofollow" ></a></td> 
  </tr> 
  <tr style="display: none" class="newlist_tr_detail"> 
  <td width="833px" style="line-height: 0;" colspan="5"> 
  <div class="newlist_detail"> 
   <div class="clearfix"> 
   <ul> 
   <li class="newlist_deatil_two"><span>地点：南京</span><span>公司性质：民营</span><span>公司规模：20-99人</span><span>经验：1-3年</span><span>学历：大专</span></li>
   <li class="newlist_deatil_last"> 岗位职责： 1、根据需求，基于Android平台进行程序开发； 2、根据产品功能模块设计，编码实现各模块功能，并确保开发质量； 3、编写相关的开发文档。 任职要求： 1、大专以上学历， 计算机或相关专业者优先； 2、2年以上<b>Android开发</b>经验； 3、熟悉Java编...</li> 
   </ul> 
   <dl> 
   <dt> 
   <a href="javascript:void(0)" rel="external nofollow" onclick="return zlapply.searchjob.ajaxApplyBrig1('CC415107716J90250224000_635','ssi','_1_03_201__2_')"> <img src="/assets/images/newlist_sqimg_03.jpg" /> </a> 
   </dt> 
   <dd>
   <a href="javascript:zlapply.searchjob.saveOne('CC415107716J90250224000_635')" rel="external nofollow" ><img src="/assets/images/newlist_scimg_06.jpg" /></a>
   </dd> 
   </dl> 
   </div> 
  </div> </td>
  </tr> 
  </tbody>
</table> 

 public static List<HtmlFeed> parse(String html) {
   Document doc = Jsoup.parse(html);
   Elements elements = doc.getElementsByClass("newlist").select("tr");  
   List<HtmlFeed> list=new ArrayList<HtmlFeed>();

  for (Element ele : elements) {
   if (!ele.select("td").toString().equals("")) {
    String job_url = ele.getElementsByClass("zwmc").select("a").attr("href");
    String job = ele.getElementsByClass("zwmc").text();
    String company = ele.getElementsByClass("gsmc").text();
    String addr = ele.getElementsByClass("gzdd").text();
    String date = ele.getElementsByClass("gxsj").text();

     HtmlFeed feed = new HtmlFeed();
 if (!job_url.toString().equals("")&&!job.toString().equals ("")&&!addr.toString().equals("")&&!company.toString().equals("")&&!date.toString().equals("")) {
  feed.setJob_url(job_url.toString());
  feed.setJob(job.toString());
  feed.setAddr(addr.toString());
  feed.setCompany(company.toString());
  feed.setDate(date.toString());

   list.add(feed);
     }
  }

  }
  return list;
 }