View Javadoc
1   package us.codecraft.webmagic.samples;
2   
3   import us.codecraft.webmagic.Page;
4   import us.codecraft.webmagic.Site;
5   import us.codecraft.webmagic.Spider;
6   import us.codecraft.webmagic.processor.PageProcessor;
7   
8   import java.util.List;
9   
10  /**
11   * @author code4crafter@gmail.com <br>
12   * Date: 13-4-21
13   * Time: 下午8:08
14   */
15  public class NjuBBSProcessor implements PageProcessor {
16      @Override
17      public void process(Page page) {
18          List<String> requests = page.getHtml().regex("<a[^<>]*href=(bbstcon\\?board=Pictures&file=[^>]*)").all();
19          page.addTargetRequests(requests);
20          page.putField("title",page.getHtml().xpath("//div[@id='content']//h2/a"));
21          page.putField("content",page.getHtml().smartContent());
22      }
23  
24      @Override
25      public Site getSite() {
26          return Site.me().setDomain("bbs.nju.edu.cn");
27      }
28  
29      public static void main(String[] args) {
30          Spider.create(new NjuBBSProcessor()).addUrl("http://bbs.nju.edu.cn/board?board=Pictures").run();
31      }
32  }