home *** CD-ROM | disk | FTP | other *** search
/ Chip 2011 November / CHIP_2011_11.iso / Programy / Narzedzia / Calibre / calibre-0.8.18.msi / file_280 / cnetnews.recipe < prev    next >
Encoding:
Text File  |  2011-09-09  |  1.7 KB  |  44 lines
  1.  
  2. __license__   = 'GPL v3'
  3. __copyright__ = '2009, Darko Miletic <darko.miletic at gmail.com>'
  4. '''
  5. news.cnet.com
  6. '''
  7.  
  8. from calibre.web.feeds.news import BasicNewsRecipe
  9.  
  10. class CnetNews(BasicNewsRecipe):
  11.     title                 = 'CNET News'
  12.     __author__            = 'Darko Miletic'
  13.     description           = 'Tech news and business reports by CNET News. Focused on information technology, core topics include computers, hardware, software, networking, and Internet media.'
  14.     publisher             = 'CNET'
  15.     category              = 'news, IT, USA'
  16.     oldest_article        = 2
  17.     max_articles_per_feed = 100
  18.     no_stylesheets        = True
  19.     encoding              = 'cp1252'
  20.     use_embedded_content  = False
  21.     language              = 'en'
  22.  
  23.     conversion_options = {
  24.                           'comment'   : description
  25.                         , 'tags'      : category
  26.                         , 'publisher' : publisher
  27.                         , 'language'  : language
  28.                         }
  29.  
  30.     remove_tags = [
  31.                     dict(name='div', attrs={'id':'tweetmemeAndFacebook'})
  32.                     ,dict(name='ul', attrs={'class':'contentTools'})
  33.                     ,dict(name='aside', attrs={'id':'filed'})
  34.                     ,dict(name='div', attrs={'class':'postLinks'})
  35.                     ,dict(name='span', attrs={'class':'shareButton'})
  36.                     ,dict(name='span', attrs={'class':'printButton'})
  37.                     ,dict(name='span', attrs={'class':'emailButton'})
  38.                     ,dict(name='div', attrs={'class':'editorBio'})
  39.                   ]
  40.     keep_only_tags = dict(name='div', attrs={'class':'post'})
  41.  
  42.     feeds = [(u'News', u'http://news.cnet.com/2547-1_3-0-20.xml')]
  43.  
  44.