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

   1 body: //div[@id='cnnContentContainer']//div[contains(@class, 'cnn_strycntntlft')]
   2 title: //div[@class="cnn_storyarea"]/h1
   3 author: //div[@class="cnnByline"]/strong
   4 date: substring-after(//div[@class="cnn_strytmstmp"], 'Sun')
   5 date: substring-after(//div[@class="cnn_strytmstmp"], 'Mon')
   6 date: substring-after(//div[@class="cnn_strytmstmp"], 'Tue')
   7 date: substring-after(//div[@class="cnn_strytmstmp"], 'Wed')
   8 date: substring-after(//div[@class="cnn_strytmstmp"], 'Thu')
   9 date: substring-after(//div[@class="cnn_strytmstmp"], 'Fri')
  10 date: substring-after(//div[@class="cnn_strytmstmp"], 'Sat')
  11 strip: //div[@class="cnn_storyarea"]/h1
  12 strip_id_or_class: cnnByline
  13 strip_id_or_class: cnn_strytmstmp
  14 strip_id_or_class: cnn_strycaptiontxt
  15 strip_id_or_class: cnn_strybtntoolsbttm
  16 strip_id_or_class: cnn_strybtntools
  17 strip_id_or_class: cnn_strybtmcntnt
  18 strip_id_or_class: sharebar
  19 #strip_id_or_class: cnn_containerwht
  20 strip_id_or_class: cnn_stryathrtmp
  21 replace_string(<a name="em0"></a>): <!-- a name -->
  22 test_url: http://www.cnn.com/2012/05/13/us/new-york-police-policy/index.html?eref=rss_topstories
  23 test_url: http://rss.cnn.com/rss/edition.rss