inc/3rdparty/site_config/standard/sportsillustrated.cnn.com.txt

   1 # main sportsillustrated.com articles
   2 #
   3 body: //div[@id="cnnStoryContent"]
   4 title: //div[@id="cnnStoryHeadline"]//h1
   5 author: //div[@id="cnnSubBanner"]//strong
   6 date: substring-after(//div[@id="cnnTimeStamp"], "Updated: ")
   7 date: substring-after(//div[@id="cnnTimeStamp"], "Posted: ")
   8
   9 # kill ugly font buttons
  10 strip: //div[@id="cnnSCFontButtons"]
  11
  12 # kill misc filler videos & etc
  13 strip: //div[@class="cnnDivideContent"]
  14 strip: //*[@class="cnnTMbox"]
  15
  16 # si vault articles
  17 # -------------
  18 body: //div[@class="siv_artPara"]
  19 title: //div[@class="siv_artHeader"]//h1
  20 author: //div[@class="byline"]
  21 date: //div[@class="date"]
  22
  23 next_page_link: //div[@id='cnnStoryContinue']/a
  24 strip_id_or_class: cnnstorypagination
  25
  26 test_url: http://sportsillustrated.cnn.com/2012/writers/peter_king/02/27/combine/index.html