home *** CD-ROM | disk | FTP | other *** search
/ Maximum CD 2011 January / maximum-cd-2011-01.iso / DiscContents / calibre-0.7.26.msi / file_4129 < prev    next >
Encoding:
Text File  |  2010-09-30  |  2.1 KB  |  51 lines
  1. __license__   = 'GPL v3'
  2. __copyright__ = '2010, Darko Miletic <darko.miletic at gmail.com>'
  3. '''
  4. kurier.at
  5. '''
  6.  
  7. from calibre.web.feeds.news import BasicNewsRecipe
  8.  
  9. class Kurier(BasicNewsRecipe):
  10.     title                 = 'Kurier'
  11.     __author__            = 'Darko Miletic'
  12.     description           = 'News from Austria'
  13.     publisher             = 'KURIER'
  14.     category              = 'news, politics, Austria'
  15.     oldest_article        = 2
  16.     max_articles_per_feed = 200
  17.     no_stylesheets        = True
  18.     encoding              = 'cp1252'
  19.     use_embedded_content  = False
  20.     language              = 'de_AT'
  21.     remove_empty_feeds    = True
  22.     publication_type      = 'newspaper'
  23.     extra_css             = ' body{font-family: Verdana,Helvetica,sans-serif } img{margin-bottom: 0.4em} .bild_us{font-size: x-small} '
  24.  
  25.     conversion_options = {
  26.                           'comment'   : description
  27.                         , 'tags'      : category
  28.                         , 'publisher' : publisher
  29.                         , 'language'  : language
  30.                         }
  31.  
  32.     remove_tags = [dict(attrs={'class':['contenttabs','drucken','versenden','leserbrief','kommentieren','addthis_button']})]
  33.     keep_only_tags    = [dict(attrs={'id':'content'})]
  34.     remove_tags_after = dict(attrs={'id':'author'})
  35.     remove_attributes = ['width','height']
  36.  
  37.     feeds = [
  38.                (u'Nachrichten', u'http://kurier.at/rss/nachrichten_nachrichten_rss.xml'   )
  39.               ,(u'Techno'     , u'http://kurier.at/rss/techno_techno_rss.xml'   )
  40.               ,(u'Wirtschaft' , u'http://kurier.at/rss/wirtschaft_wirtschaft_rss.xml'   )
  41.               ,(u'Kultur'     , u'http://kurier.at/rss/kultur_kultur_rss.xml'   )
  42.               ,(u'Freizeit'   , u'http://kurier.at/rss/freizeit_freizeit_rss.xml'   )
  43.               ,(u'Wetter'     , u'http://kurier.at/rss/oewetter_rss.xml'   )
  44.               ,(u'Verkehr'    , u'http://kurier.at/rss/verkehr_rss.xml'   )
  45.             ]
  46.  
  47.     def preprocess_html(self, soup):
  48.         for item in soup.findAll(style=True):
  49.             del item['style']
  50.         return self.adeify_images(soup)
  51.