inc/3rdparty/site_config/standard/spiegel.de.txt

   1 # A. Niepel, narya.de@...
   2 # - added single_page_link
   3 # - added author for default and single page view
   4 # - added date for single page view
   5 # fforst@...
   6 # - Fixed it
   7 # bode2104@...
   8 # - Fixed single_page_link
   9 # - Included intro text in single page view
  10 # - Added body in default view
  11
  12 # set body
  13 tidy: no
  14 # body in single page view
  15 body: //div[@id="spArticleContent"]
  16 # body in default view
  17 body: //div[@id="spArticleSection"]
  18 # body in "Fotostrecke"
  19 body: //div[@id="spBigaContent"]
  20
  21 # set date in single page view
  22 date: //div[@id="spArticleContent"]/h3
  23 # strip date
  24 strip: //div[@id="spArticleContent"]/h3
  25 # set date in "Fotostrecke"
  26 date: //div[@id="spBigaDatum"]
  27
  28 #set title in single page view
  29 title: //div[@id='spArticleContent']/h2
  30 # strip title
  31 strip: //div[@id='spArticleContent']/h1
  32 strip: //div[@id='spArticleContent']/h2
  33 #set title in "Fotostrecke"
  34 title: //div[@class='spBigaHeadline']
  35
  36 # set author
  37 author: //p[@class="spAuthor"]/a
  38 author: substring-after(//p[@class="spAuthor"], 'Von ')
  39 # strip author
  40 strip: //p[@class='spAuthor']
  41
  42 # remove captions
  43 strip: //*/span[@class='spPicLayerText']
  44 strip: //*/div[@class='spPanoPlayerPaneControl']
  45 strip: //*/div[@class='spCredit']
  46 strip: //*/div[@class='spCredit']/following-sibling::p
  47
  48 # remove ads
  49 strip: //div[@class='spMInline']
  50
  51 # remove photogalleries and extras
  52 strip: //div[@class='spPhotoGallery']
  53 strip: //div[@class='spPhotoGallery']/following-sibling::br
  54 strip: //div[@class='spAssetAlignleft']
  55 strip: //div[contains(@class,'spAsset')]
  56 strip: //br[@clear='all']
  57
  58 # remove community functions
  59 strip: //div[@id='spSocialBookmark']
  60 strip: //div[contains(@class, 'spCommunityBox')]
  61 strip: //div[contains(@class, 'spArticleNewsfeedBox')]
  62 strip: //div[@class='spArticleCredit']
  63
  64 # remove clutter in "Fotostrecke"
  65 strip: //div[@id='spBreadcrumb']
  66 strip: //div[@id='spBigaLatestEntries']
  67 strip: //div[contains(@class, 'spBigaNavi')]
  68 strip: //div[@class='spDottedLine']
  69
  70 # Use link to print article for single page view
  71 single_page_link: //a[contains(@href, '-druck')]
  72
  73 # use next link in "Fotostrecke"
  74 next_page_link: //a[@class='spBigaControlForw']
  75 test_url: http://www.spiegel.de/politik/deutschland/0,1518,787602,00.html