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 # stesie@
  12 # - removed copyright box
  13 # - removed "print more" box
  14
  15 # set body
  16 tidy: no
  17 # body in single page view
  18 body: //div[@id="spArticleContent"]
  19 # body in default view
  20 body: //div[@id="spArticleSection"]
  21 body: //div[contains(@class, 'article-section')] | //div[@id='js-article-top-wide-asset'] | //p[contains(@class, 'article-intro')] | //div[contains(@class, 'js-module-box-image')]
  22 # body in "Fotostrecke"
  23 body: //div[@id="spBigaContent"]
  24
  25 # set date in single page view
  26 date: //div[@id="spArticleContent"]/h3
  27 # strip date
  28 strip: //div[@id="spArticleContent"]/h3
  29 # set date in "Fotostrecke"
  30 date: //div[@id="spBigaDatum"]
  31
  32 # title in default view
  33 title: //h2[contains(@class, 'article-title')]
  34 #set title in single page view
  35 title: //div[@id='spArticleContent']/h2
  36 # strip title
  37 strip: //div[@id='spArticleContent']/h1
  38 strip: //div[@id='spArticleContent']/h2
  39 #set title in "Fotostrecke"
  40 title: //div[@class='spBigaHeadline']
  41
  42 # set author
  43 author: //p[@class="spAuthor"]/a
  44 author: substring-after(//p[@class="spAuthor"], 'Von ')
  45 # strip author
  46 strip: //p[@class='spAuthor']
  47
  48 # remove captions
  49 strip: //*/span[@class='spPicLayerText']
  50 strip: //*/div[@class='spPanoPlayerPaneControl']
  51 strip: //*/div[@class='spCredit']
  52 strip: //*/div[@class='spCredit']/following-sibling::p
  53
  54 # remove ads
  55 strip: //div[@class='spMInline']
  56
  57 # remove photogalleries and extras
  58 strip: //div[contains(@class, 'spPhotoGallery')]
  59 strip: //div[@class='spPhotoGallery']/following-sibling::br
  60 strip: //div[@class='spAssetAlignleft']
  61 strip: //div[contains(@class,'spAsset')]
  62 strip: //br[@clear='all']
  63
  64 # remove community functions
  65 strip: //div[@id='spSocialBookmark']
  66 strip: //div[contains(@class, 'spCommunityBox')]
  67 strip: //div[contains(@class, 'spArticleNewsfeedBox')]
  68 strip: //div[@class='spArticleCredit']
  69
  70 # remove clutter in "Fotostrecke"
  71 strip: //div[@id='spBreadcrumb']
  72 strip: //div[@id='spBigaLatestEntries']
  73 strip: //div[contains(@class, 'spBigaNavi')]
  74 strip: //div[@class='spDottedLine']
  75
  76 strip: //div[@class='asset-box article-print-more']
  77 strip: //div[@class='article-copyright']
  78 strip: //span[@class='image-buttons']
  79
  80 # Use link to print article for single page view
  81 single_page_link: //a[contains(@href, '-druck')]
  82 if_page_contains: //div[contains(@class, 'multi-pager-control')]
  83
  84 # Clean up title in print view
  85 find_string: <title>Druckversion -
  86 replace_string: <title>
  87
  88 # use next link in "Fotostrecke"
  89 next_page_link: //a[@class='spBigaControlForw']
  90 test_url: http://www.spiegel.de/politik/deutschland/0,1518,787602,00.html
  91
  92 # regular article
  93 test_url: http://www.spiegel.de/wirtschaft/soziales/griechenland-was-den-griechischen-buergern-nun-droht-a-1042682.html
  94
  95 # multipage article
  96 test_url: http://www.spiegel.de/spiegel/a-710880.html