home *** CD-ROM | disk | FTP | other *** search
/ Maximum CD 2010 November / maximum-cd-2010-11.iso / DiscContents / calibre-0.7.13.msi / file_4009 < prev    next >
Encoding:
Text File  |  2009-12-05  |  1.4 KB  |  43 lines
  1.  
  2. __license__   = 'GPL v3'
  3. __copyright__ = '2009, Darko Miletic <darko.miletic at gmail.com>'
  4. '''
  5. en.rian.ru
  6. '''
  7.  
  8. from calibre.web.feeds.news import BasicNewsRecipe
  9.  
  10. class Ria_eng(BasicNewsRecipe):
  11.     title                 = 'Ria Novosti'
  12.     __author__            = 'Darko Miletic'
  13.     description           = 'News from Russia in English'
  14.     language              = 'en'
  15.     publisher             = 'en.rian.ru'
  16.     category              = 'news, politics, Russia'
  17.     oldest_article        = 3
  18.     max_articles_per_feed = 100
  19.     no_stylesheets        = True
  20.     use_embedded_content  = False
  21.     encoding              = 'utf-8'
  22.  
  23.     conversion_options = {
  24.                           'comment'          : description
  25.                         , 'tags'             : category
  26.                         , 'publisher'        : publisher
  27.                         , 'language'         : language
  28.                         }
  29.  
  30.  
  31.     keep_only_tags = [dict(name='div', attrs={'class':'article'})]
  32.  
  33.     remove_tags = [
  34.                      dict(name=['object','link','iframe','base'])
  35.                     ,dict(name='div',attrs={'class':['related','mmban','view-story']})
  36.                     ,dict(name='span',attrs={'class':'copyright'})
  37.                   ]
  38.     remove_tags_after = dict(name='div',attrs={'class':'text'})
  39.  
  40.  
  41.     feeds = [(u'Online news', u'http://en.rian.ru/export/rss2/archive/index.xml')]
  42.  
  43.