home *** CD-ROM | disk | FTP | other *** search
/ Maximum CD 2010 November / maximum-cd-2010-11.iso / DiscContents / calibre-0.7.13.msi / file_4057 < prev    next >
Encoding:
Text File  |  2010-06-08  |  4.6 KB  |  79 lines

  1. __license__   = 'GPL v3'
  2. __copyright__ = '2008, Kovid Goyal <kovid at kovidgoyal.net>'
  3.  
  4. '''
  5. Fetch sueddeutsche.
  6. '''
  7. from calibre.web.feeds.news import BasicNewsRecipe
  8.  
  9.  
  10. class Sueddeutsche(BasicNewsRecipe):
  11.  
  12.     title = u'S\xfcddeutsche'
  13.     description = 'News from Germany'
  14.     __author__ = 'Oliver Niesner and Sujata Raman'
  15.     use_embedded_content   = False
  16.     timefmt = ' [%d %b %Y]'
  17.     oldest_article = 7
  18.     max_articles_per_feed = 50
  19.     no_stylesheets = True
  20.     language = 'de'
  21.  
  22.     encoding = 'utf-8'
  23.     remove_javascript = True
  24.  
  25.  
  26.     remove_tags = [ dict(name='link'), dict(name='iframe'),
  27.                     dict(name='div', attrs={'id':["bookmarking","themenbox","artikelfoot","CAD_AD",
  28.                           "SKY_AD","NT1_AD","navbar1","sdesiteheader"]}),
  29.  
  30.                     dict(name='div', attrs={'class':["similar-article-box","artikelliste","nteaser301bg",
  31.                                  "pages closed","basebox right narrow"]}),
  32.  
  33.                     dict(name='div', attrs={'class':["articleDistractor","listHeader","listHeader2","hr2",
  34.                              "item","videoBigButton","articlefooter full-column",
  35.                                                      "bildbanderolle full-column","footerCopy padleft5"]}),
  36.  
  37.                     dict(name='p', attrs={'class':["ressortartikeln","artikelFliestext","entry-summary"]}),
  38.                     dict(name='div', attrs={'style':["position:relative;"]}),
  39.                     dict(name='span', attrs={'class':["nlinkheaderteaserschwarz","artikelLink","r10000000"]}),
  40.                     dict(name='table', attrs={'class':["stoerBS","kommentare","footer","pageBoxBot","pageAktiv","bgcontent"]}),
  41.                     dict(name='ul', attrs={'class':["breadcrumb","articles","activities","sitenav"]}),
  42.                     dict(name='td', attrs={'class':["artikelDruckenRight"]}),
  43.                     dict(name='p', text = "ANZEIGE")
  44.                      ]
  45.  
  46.     extra_css = '''
  47.                     h2{font-family:Arial,Helvetica,sans-serif; font-size: x-small; color: #003399;}
  48.                     a{font-family:Arial,Helvetica,sans-serif; font-size: x-small; font-style:italic;}
  49.                     .dachzeile p{font-family:Arial,Helvetica,sans-serif; font-size: x-small; }
  50.                     h1{ font-family:Arial,Helvetica,sans-serif;  font-size:x-large; font-weight:bold;}
  51.                     .artikelTeaser{font-family:Arial,Helvetica,sans-serif; font-size: x-small; font-weight:bold; }
  52.                     body{font-family:Arial,Helvetica,sans-serif; }
  53.                     .photo {font-family:Arial,Helvetica,sans-serif; font-size: x-small; color: #666666;}                 '''
  54.  
  55.     #feeds = [(u'Topthemen', u'http://suche.sueddeutsche.de/query/politik/-docdatetime/drilldown/%C2%A7documenttype%3AArtikel?output=rss')]
  56.  
  57.     feeds = [(u'Wissen', u'http://suche.sueddeutsche.de/query/wissen/nav/%C2%A7ressort%3AWissen/sort/-docdatetime?output=rss'),
  58.              (u'Politik', u'http://suche.sueddeutsche.de/query/politik/nav/%C2%A7ressort%3APolitik/sort/-docdatetime?output=rss'),
  59.              (u'Wirtschaft', u'http://suche.sueddeutsche.de/query/wirtschaft/nav/%C2%A7ressort%3AWirtschaft/sort/-docdatetime?output=rss'),
  60.              (u'Finanzen', u'http://suche.sueddeutsche.de/query/finanzen/nav/%C2%A7ressort%3AGeld/sort/-docdatetime?output=rss'),
  61.              (u'Kultur', u'http://suche.sueddeutsche.de/query/kultur/nav/%C2%A7ressort%3AKultur/sort/-docdatetime?output=rss'),
  62.              (u'Sport', u'http://suche.sueddeutsche.de/query/sport/nav/%C2%A7ressort%3ASport/sort/-docdatetime?output=rss'),
  63.              (u'Bayern', u'http://suche.sueddeutsche.de/query/bayern/nav/%C2%A7ressort%3ABayern/sort/-docdatetime?output=rss'),
  64.              (u'Panorama', u'http://suche.sueddeutsche.de/query/panorama/sort/-docdatetime?output=rss'),
  65.              (u'Leben&Stil', u'http://suche.sueddeutsche.de/query/stil/nav/%C2%A7ressort%3A%22Leben%20%26%20Stil%22/sort/-docdatetime?output=rss'),
  66.              (u'Gesundheit', u'http://suche.sueddeutsche.de/query/gesundheit/nav/%C2%A7ressort%3AGesundheit/sort/-docdatetime?output=rss'),
  67.              (u'Auto&Reise', u'http://suche.sueddeutsche.de/query/automobil/nav/%C2%A7ressort%3A%22Auto%20%26%20Mobil%22/sort/-docdatetime?output=rss'),
  68.              (u'Computer', u'http://suche.sueddeutsche.de/query/computer/nav/%C2%A7ressort%3AComputer/sort/-docdatetime?output=rss'),
  69.              (u'Job&Karriere', u'http://suche.sueddeutsche.de/query/job/nav/%C2%A7ressort%3A%22Job%20%26%20Karriere%22/sort/-docdatetime?output=rss'),
  70.              (u'Reise', u'http://suche.sueddeutsche.de/query/reise/nav/%C2%A7ressort%3AReise/sort/-docdatetime?output=rss')
  71.              ]
  72.  
  73.     def print_version(self, url):
  74.         return url.replace('/text/', '/text/print.html')
  75.  
  76.  
  77.  
  78.  
  79.