home *** CD-ROM | disk | FTP | other *** search
/ Maximum CD 2010 November / maximum-cd-2010-11.iso / DiscContents / calibre-0.7.13.msi / file_4150 < prev    next >
Encoding:
Text File  |  2009-10-14  |  1.1 KB  |  46 lines

  1. __license__   = 'GPL v3'
  2. __copyright__ = '2008, Kovid Goyal <kovid at kovidgoyal.net>'
  3.  
  4. '''
  5. Fetch zdnet.
  6. '''
  7.  
  8. from calibre.web.feeds.news import BasicNewsRecipe
  9.  
  10.  
  11. class cdnet(BasicNewsRecipe):
  12.  
  13.     title = 'zdnet'
  14.     description = 'zdnet security'
  15.     __author__ = 'Oliver Niesner'
  16.     language = 'en'
  17.  
  18.     use_embedded_content   = False
  19.     timefmt = ' [%d %b %Y]'
  20.     max_articles_per_feed = 40
  21.     no_stylesheets = True
  22.     encoding = 'latin1'
  23.  
  24.  
  25.  
  26.     remove_tags = [dict(id='eyebrows'),
  27.            dict(id='header'),
  28.            dict(id='search'),
  29.            dict(id='nav'),
  30.            dict(id=''),
  31.            dict(name='div', attrs={'class':'banner'}),
  32.            dict(name='p', attrs={'class':'tags'}),
  33.            dict(name='a', attrs={'href':'http://www.twitter.com/ryanaraine'}),
  34.            dict(name='div', attrs={'class':'special1'})]
  35.     remove_tags_after = [dict(name='div', attrs={'class':'bloggerDesc clear'})]
  36.  
  37.     feeds =  [ ('zdnet', 'http://feeds.feedburner.com/zdnet/security') ]
  38.  
  39.  
  40.     def preprocess_html(self, soup):
  41.         for item in soup.findAll(style=True):
  42.             del item['style']
  43.         return soup
  44.  
  45.  
  46.