Jsoup解析html實現(xiàn)招聘信息查詢功能

 <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>職位名稱</span></th>
  <th class="gsmc">公司名稱</th> 
  <th class="zwyx">職位月薪</th>
  <th class="gzdd">工作地點</th>
  <th class="gxsj">發(fā)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 >
   <a  par="ssidkey=y&amp;ss=201&amp;ff=03"  rel="external nofollow" target="_blank">Android 開發(fā)工程師</a>
  </div> </td> 

  <td class="gsmc"><a  rel="external nofollow" target="_blank">南京天洑軟件有限公司</a></td> 
  <td class="zwyx">面議</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  class="newlist_tr_detail"> 
  <td width="833px"  colspan="5"> 
  <div class="newlist_detail"> 
   <div class="clearfix"> 
   <ul> 
   <li class="newlist_deatil_two"><span>地點：南京</span><span>公司性質(zhì)：民營</span><span>公司規(guī)模：20-99人</span><span>經(jīng)驗：1-3年</span><span>學(xué)歷：大專</span></li>
   <li class="newlist_deatil_last"> 崗位職責(zé)： 1、根據(jù)需求，基于Android平臺進行程序開發(fā)； 2、根據(jù)產(chǎn)品功能模塊設(shè)計，編碼實現(xiàn)各模塊功能，并確保開發(fā)質(zhì)量； 3、編寫相關(guān)的開發(fā)文檔。 任職要求： 1、大專以上學(xué)歷， 計算機或相關(guān)專業(yè)者優(yōu)先； 2、2年以上<b>Android開發(fā)</b>經(jīng)驗； 3、熟悉Java編...</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;
 }