home *** CD-ROM | disk | FTP | other *** search
/ Chip 2011 November / CHIP_2011_11.iso / Programy / Narzedzia / Calibre / calibre-0.8.18.msi / file_280 / emg_en.recipe < prev    next >
Encoding:
Text File  |  2011-09-09  |  1.6 KB  |  46 lines
  1. __license__   = 'GPL v3'
  2. __copyright__ = '2010, Darko Miletic <darko.miletic at gmail.com>'
  3. '''
  4. emg.rs/en/news
  5. '''
  6.  
  7. from calibre.web.feeds.news import BasicNewsRecipe
  8.  
  9. class emportal_en(BasicNewsRecipe):
  10.     title                 = 'Ekonom:east News'
  11.     __author__            = 'Darko Miletic'
  12.     description           = 'Daily business news from Serbia.'
  13.     publisher             = 'Ekonom:east Media Group'
  14.     category              = 'Business, SEE, Serbia, Belgrade, news, Ekonomist, EMportal'
  15.     oldest_article        = 2
  16.     max_articles_per_feed = 200
  17.     no_stylesheets        = True
  18.     encoding              = 'utf8'
  19.     use_embedded_content  = False
  20.     language              = 'en'
  21.     remove_empty_feeds    = True
  22.     masthead_url          = 'http://www.emg.rs/img/emportal-rss.png'
  23.     extra_css             = ' body{font-family: Arial,Helvetica,sans-serif } '
  24.  
  25.     conversion_options = {
  26.                           'comment'   : description
  27.                         , 'tags'      : category
  28.                         , 'publisher' : publisher
  29.                         , 'language'  : language
  30.                         }
  31.  
  32.     remove_tags        = [dict(attrs={'class':['text-share']})]
  33.     keep_only_tags     = [dict(attrs={'class':'text'})]
  34.     remove_tags_after  = dict(attrs={'class':'text-share'})
  35.     remove_attributes  = ['width','height']
  36.  
  37.     feeds = [(u'Serbia' , u'http://www.emg.rs/en/news/serbia/rss.xml')]
  38.  
  39.     def print_version(self, url):
  40.         return url.replace('.html','.print.html')
  41.  
  42.     def preprocess_html(self, soup):
  43.         for item in soup.findAll(style=True):
  44.             del item['style']
  45.         return soup
  46.