home *** CD-ROM | disk | FTP | other *** search
/ Chip 2011 November / CHIP_2011_11.iso / Programy / Narzedzia / Calibre / calibre-0.8.18.msi / file_280 / romanialibera.recipe < prev    next >
Encoding:
Text File  |  2011-09-09  |  2.1 KB  |  60 lines
  1. # -*- coding: utf-8 -*-
  2. #!/usr/bin/env  python
  3.  
  4. __license__   = 'GPL v3'
  5. __copyright__ = u'2011, Silviu Cotoar\u0103'
  6. '''
  7. romanialibera.ro
  8. '''
  9.  
  10. from calibre.web.feeds.news import BasicNewsRecipe
  11.  
  12. class RomaniaLibera(BasicNewsRecipe):
  13.     title                 = u'Rom\u00e2nia Liber\u0103'
  14.     __author__            = u'Silviu Cotoar\u0103'
  15.     description           = u'Rom\u00e2nia Liber\u0103'
  16.     publisher             = u'Rom\u00e2nia Liber\u0103'
  17.     oldest_article        = 5
  18.     language              = 'ro'
  19.     max_articles_per_feed = 100
  20.     no_stylesheets        = True
  21.     use_embedded_content  = False
  22.     category              = 'Ziare,Stiri'
  23.     encoding              = 'utf-8'
  24.     cover_url             = 'http://www.romanialibera.ro/templates/lilac/images/sigla_1.gif'
  25.  
  26.     conversion_options = {
  27.                              'comments'   : description
  28.                             ,'tags'       : category
  29.                             ,'language'   : language
  30.                             ,'publisher'  : publisher
  31.                          }
  32.  
  33.     keep_only_tags = [
  34.             dict(name='div', attrs={'id':'articol'})
  35.                      ]
  36.  
  37.     remove_tags = [
  38.              dict(name='div', attrs={'id':['art_actions']})
  39.            , dict(name='div', attrs={'class':['stats']})
  40.            , dict(name='div', attrs={'class':['data']})
  41.             , dict(name='div', attrs={'class':['autori']})
  42.             , dict(name='div', attrs={'class':['banda_explicatii_text']})
  43.            , dict(name='td', attrs={'class':['connect_widget_vertical_center connect_widget_button_cell']})
  44.            , dict(name='div', attrs={'class':['aceeasi_tema']})
  45.            , dict(name='div', attrs={'class':['art_after_text']})
  46.            , dict(name='div', attrs={'class':['navigare']})
  47.             , dict(name='div', attrs={'id':['art_text_left']})
  48.                            ]
  49.  
  50.     remove_tags_after = [
  51.               dict(name='div', attrs={'class':'art_after_text'})
  52.             ]
  53.  
  54.     feeds          = [
  55.             (u'Feeds', u'http://www.romanialibera.ro/rss.xml')
  56.                  ]
  57.  
  58.     def preprocess_html(self, soup):
  59.         return self.adeify_images(soup)
  60.