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

   1 title: //meta[@property="og:title"]/@content
   2 title: //h1
   3 title: //*[@class='posttitle']
   4 author: //*[@class='entryAuthor']/a[1]
   5 author://*[@class='member-title']
   6 author://li[@class='author']/a[contains(@href, '/author/')]
   7 date: substring-after(//div[@class='entryAuthor'], '·')
   8 date: substring-before(//*[@class='entryDate'], '|')
   9 body: //div[@class='entry']
  10 strip: //span[contains(@class, 'nextprev')]
  11 #strip_id_or_class: ngg-galleryoverview
  12 # ngg-galleryoverview is the whole content sometimes, e.g. http://www.wired.com/underwire/2011/12/best-mixtapes-of-2011/?pid=5736&viewall=true
  13
  14 strip: //p[span[contains(@class, 'contentjump')]]
  15 strip: //text()[contains(., 'nextpage')]
  16
  17 prune: no
  18
  19 single_page_link: //a[contains(@href, '/all/1') and contains(@class, 'contentjumpall')]
  20
  21 test_url: http://www.wired.com/cloudline/2011/10/meet-arms-cortex-a15-the-future-of-the-ipad-and-possibly-the-macbook-air/
  22 test_url: http://www.wired.com/threatlevel/2012/05/ff_counterfeiter/all/1