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

   1 author: //p[contains(@class, "byline")]/a[contains(@class, "author")]
   2
   3 date: //span[contains(@class, "publish-date")]/time[@pubdate]/@datetime
   4
   5 body: //div[contains(@class, 'entry-content')]
   6 # for vergecasts, e.g. http://www.theverge.com/2013/8/22/4648566/the-vergecast-090-august-22th-2013-video
   7 body: //article
   8 prune: no
   9 #tidy: no
  10
  11 strip: //article/header
  12 strip: //*[@id='sticky-menu']
  13 strip: //aside
  14 strip: //nav
  15 strip: //img[contains(@class, 'vox-lazy-load')]
  16 # deal with bad parsing
  17 strip: //div[contains(@class, 'story-image')]//div[contains(., 'function(')]
  18 strip: //div[contains(@class, 'm-linkset')]
  19 strip: //div[contains(@class, 'm-entry__sidebar')]
  20 strip: //ul[contains(@class, 'm-article__sources')]
  21 strip: //div[contains(@class, 'chorus-emc__content')]
  22
  23
  24 strip_id_or_class: gallery
  25 strip_id_or_class: article-meta
  26 strip_id_or_class: story-navigation
  27 strip_id_or_class: slegend
  28 strip_id_or_class: related-product-meta
  29 strip_id_or_class: comments
  30 strip_id_or_class: ui-jump-list
  31 strip_id_or_class: pullquote
  32 strip_id_or_class: m-ad
  33 strip_id_or_class: social-sharing
  34 strip_id_or_class: m-video-entry__excerpt
  35 strip_id_or_class: hidden
  36
  37 replace_string(<noscript>): <div>
  38 replace_string(</noscript>): </div>
  39
  40 find_string: <script
  41 replace_string: <div style="display:none"
  42 find_string: </script>
  43 replace_string: </div>
  44
  45 strip: //q
  46
  47 strip: //a[contains(@class, 'entry-section-title')]
  48
  49 test_url: http://www.theverge.com/2012/2/29/2821763/lytro-review
  50 test_url: http://www.theverge.com/2011/11/3/2534861/nokia-lumia-800-review
  51 test_url: http://www.theverge.com/2013/2/24/4026114/barnes-noble-shifting-focus-away-from-nook-hardware
  52 test_url: http://www.theverge.com/2014/6/19/5824072/top-shelf-living-the-dream
  53 test_url: http://www.theverge.com/rss/frontpage