home *** CD-ROM | disk | FTP | other *** search
/ Chip 2011 November / CHIP_2011_11.iso / Programy / Narzedzia / Calibre / calibre-0.8.18.msi / file_280 / curierulnational.recipe < prev    next >
Encoding:
Text File  |  2011-09-09  |  1.7 KB  |  55 lines
  1. # -*- coding: utf-8 -*-
  2. #!/usr/bin/env  python
  3.  
  4. __license__   = 'GPL v3'
  5. __copyright__ = u'2011, Silviu Cotoar\u0103'
  6. '''
  7. curierulnational.ro
  8. '''
  9.  
  10. from calibre.web.feeds.news import BasicNewsRecipe
  11.  
  12. class CurierulNal(BasicNewsRecipe):
  13.     title                 = u'Curierul Na\u0163ional'
  14.     __author__            = u'Silviu Cotoar\u0103'
  15.     description           = ''
  16.     publisher             = 'Curierul Na\u0163ional'
  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.curierulnational.ro/logo.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':'col1'})
  35.             , dict(name='img', attrs={'id':'placeholder'})
  36.                      ]
  37.  
  38.     remove_tags = [
  39.              dict(name='p', attrs={'id':['alteArticole']})
  40.            , dict(name='div', attrs={'id':['textSize']})
  41.            , dict(name='ul', attrs={'class':['unit-rating']})
  42.            , dict(name='div', attrs={'id':['comments']})
  43.                   ]
  44.  
  45.     remove_tags_after = [
  46.               dict(name='ul', attrs={'class':'unit-rating'})
  47.             ]
  48.  
  49.     feeds          = [
  50.             (u'Feeds', u'http://www.curierulnational.ro/feed.xml')
  51.                  ]
  52.  
  53.     def preprocess_html(self, soup):
  54.         return self.adeify_images(soup)
  55.