inc/3rdparty/site_config/standard/abcnews.go.com.txt

   1 title: //h1[@class='headline']
   2 body: //div[@id='storyText']
   3 # for video entries
   4 body: //img[@id='ff-img'] | //div[@id='meta']//div[contains(@class, 'overview')]
   5 author: //div[@class='byline']
   6 date: //div[@class='date']
   7 strip: //*[@id='date_partner']
   8
   9 strip: //div[@class='breadcrumb']
  10 strip: //div[contains(@class,'show_tools')]
  11 strip: //div[@id='sponsoredByAd']
  12 strip: //div[contains(@class,'rel_container')]
  13 strip: //p[a[starts-with(@href, 'http://www.twitter.com')]]
  14 strip: //p[a[starts-with(@href, 'http://www.facebook.com')]]
  15 strip: //p[contains(., 'Click here to return to')]
  16 #strip_id_or_class: media
  17 strip_id_or_class: mediaplayer
  18
  19 replace_string(<link rel="image_src" href="http): <img id="ff-img" src="http
  20
  21 prune: no
  22
  23 single_page_link: concat(//li[@class='pager']//a/@href, '&singlePage=true')
  24
  25 test_url: http://abcnews.go.com/Politics/newt-gingrich-rocky-rollout-presidential-campaign-recover/story?id=13632744
  26 # multi-page
  27 test_url: http://abcnews.go.com/Blotter/family-freed-american-hostage-somalia-seals-obama/story?id=15439544