Пример #1
0
 public void getTitle(Response response) {
   JXDocument doc = response.document();
   try {
     logger.info(
         "url:{} {}",
         response.getUrl(),
         doc.sel("//h1[@class='postTitle']/a/text()|//a[@id='cb_post_title_url']/text()"));
     // do something
   } catch (Exception e) {
     e.printStackTrace();
   }
 }
Пример #2
0
 @Override
 public void start(Response response) {
   JXDocument doc = response.document();
   try {
     List<Object> urls = doc.sel("//a[@class='titlelnk']/@href");
     logger.info("{}", urls.size());
     for (Object s : urls) {
       push(new Request(s.toString(), "getTitle"));
     }
   } catch (Exception e) {
     e.printStackTrace();
   }
 }