inc/3rdparty/site_config/standard/online.wsj.com.txt

   1 title: //meta[@property="og:title"]/@content
   2 body: //div[@id='article_story_body']
   3
   4 author: //h3[@class='byline']/a
   5 # for slid show content
   6 body: //ul[@id='imageSlide']//li[@class='firstSlide']//img | (//div[@class='txt_body']//p)[1]
   7 date: //li[@class='dateStamp']/small
   8
   9 strip_id_or_class: insetFullBracket
  10 strip_id_or_class: insettipBox
  11 #strip_id_or_class: legacyInset
  12 strip_id_or_class: recipeACShopAndBuyText
  13
  14 strip: //div[contains(@class, 'insetContent')]//cite
  15 strip: //*[contains(@style, 'visibility: hidden;')]
  16 strip: //div[contains(@class, 'insetContent') and not(contains(@class, 'image'))]
  17
  18 prune: no
  19 tidy: no
  20
  21 test_url: http://online.wsj.com/article/SB10001424052970203363504577185322849515102.html
  22 # slide show
  23 test_url: http://online.wsj.com/article/SB10001424052970204791104577110550376458164.html