home *** CD-ROM | disk | FTP | other *** search
/ Maximum CD 2010 November / maximum-cd-2010-11.iso / DiscContents / calibre-0.7.13.msi / file_3708 < prev    next >
Encoding:
Text File  |  2009-10-21  |  1.7 KB  |  50 lines
  1. #!/usr/bin/env  python
  2.  
  3. __license__   = 'GPL v3'
  4. __copyright__ = '2009, Darko Miletic <darko.miletic at gmail.com>'
  5. '''
  6. dn.se
  7. '''
  8. from calibre.web.feeds.news import BasicNewsRecipe
  9.  
  10. class DN_se(BasicNewsRecipe):
  11.     title                 = 'Dagens Nyheter'
  12.     __author__            = 'Darko Miletic'
  13.     description           = 'News from Sweden'
  14.     publisher             = 'Dagens Nyheter'
  15.     category              = 'news, politics, Sveden'
  16.     oldest_article        = 2
  17.     delay                 = 1
  18.     max_articles_per_feed = 100
  19.     no_stylesheets        = True
  20.     use_embedded_content  = False
  21.     encoding              = 'utf-8'
  22.     language              = 'sv'
  23.  
  24.     conversion_options = {
  25.                           'comment'   : description
  26.                         , 'tags'      : category
  27.                         , 'publisher' : publisher
  28.                         , 'language'  : language
  29.                         }
  30.  
  31.     feeds = [
  32.                (u'Nyheter', u'http://www.dn.se/m/rss/toppnyheter')
  33.               ,(u'Ekonomi', u'http://www.dn.se/ekonomi-rss'      )
  34.               ,(u'Sport'  , u'http://www.dn.se/sport-rss'        )
  35.               ,(u'Debatt' , u'http://www.dn.se/debatt-rss'       )
  36.               ,(u'Ledare' , u'http://www.dn.se/ledare-rss'       )
  37.               ,(u'Kultur' , u'http://www.dn.se/kultur-rss'       )
  38.             ]
  39.  
  40.     keep_only_tags = [dict(name='div', attrs={'id':'article'})]
  41.     remove_tags_before = dict(name='h1')
  42.     remove_tags_after = dict(name='div',attrs={'id':'byline'})
  43.     remove_tags = [
  44.                      dict(name=['object','link','base'])
  45.                     ,dict(name='div',attrs={'id':'hook'})
  46.                   ]
  47.  
  48.     
  49.     
  50.