home *** CD-ROM | disk | FTP | other *** search
/ Maximum CD 2010 November / maximum-cd-2010-11.iso / DiscContents / calibre-0.7.13.msi / file_3763 < prev    next >
Encoding:
Text File  |  2010-08-04  |  850 b   |  27 lines

  1. from calibre.web.feeds.news import BasicNewsRecipe
  2. import re
  3.  
  4. class NatureNews(BasicNewsRecipe):
  5.     title          = u'Nature News'
  6.     language       = 'en'
  7.     __author__     = 'Krittika Goyal, Starson17'
  8.     oldest_article = 31 #days
  9.     remove_empty_feeds    = True
  10.     max_articles_per_feed = 50
  11.  
  12.     no_stylesheets = True
  13.     remove_tags_before = dict(name='h1', attrs={'class':'heading entry-title'})
  14.     remove_tags_after  = dict(name='h2', attrs={'id':'comments'})
  15.     remove_tags = [
  16.        dict(name='h2', attrs={'id':'comments'}),
  17.        dict(attrs={'alt':'Advertisement'}),
  18.        dict(name='div', attrs={'class':'ad'}),
  19.     ] 
  20.  
  21.     preprocess_regexps = [
  22.         (re.compile(r'<p>ADVERTISEMENT</p>', re.DOTALL|re.IGNORECASE), lambda match: ''),
  23.         ]
  24.  
  25.     feeds = [('Nature News', 'http://feeds.nature.com/news/rss/most_recent')]
  26.  
  27.