inc/3rdparty/site_config/standard/npr.org.txt

   1 title: //div[contains(@class, 'storytitle')]//h1
   2 author: //p[@class="byline"]/span
   3 body: //div[@id='storyspan02']//*[@class='duration' or @class='download' or contains(@class, 'photo')] | //div[@id='storytext'] | //div[@class='transcript']
   4 date: //meta[@name="date"]/@content
   5
   6 strip: //div[@class='enlarge_measure']
   7 strip: //div[@class='enlarge_html']
   8 strip: //a[@class='enlargeicon']
   9 strip: //div[contains(@class, 'bookedition')]
  10 strip: //div[@class='textsize']
  11 strip: //ul[@class='genres']
  12 strip: //span[@class='bull']
  13 strip_id_or_class: secondary
  14 strip_id_or_class: con1col
  15 strip: //h3[@class='conheader']
  16
  17 replace_string(<a name="more">&nbsp;</a>): <!-- no more -->
  18 replace_string(<div class="transcript">): <div class="transcript"><h2>Transcript</h2>
  19
  20 prune: no
  21 strip://div[@class="ecommercepop"]
  22 strip://span[@class="bull"]
  23 strip://span[@class="purchaseLink"]
  24 strip://div[@class="enlarge_html"]
  25 strip://div[@class="enlarge_measure"]
  26 strip://div[@class="container con1col small"]
  27 strip://a[contains(@class, "enlargebtn")]
  28 strip://div[contains(@class, "bucketwrap internallink")]
  29
  30 test_url: http://www.npr.org/blogs/thetwo-way/2011/07/12/137799301/sports-loses-its-escapist-gleam-in-a-summer-of-court-dates
  31 test_url: http://www.npr.org/2012/07/04/156190948/feeling-under-siege-catholic-leadership-shifts-right
  32 test_url: http://www.npr.org/2012/12/13/166480907/the-years-best-sci-fi-crosses-galaxies-and-genres