home *** CD-ROM | disk | FTP | other *** search
/ Maximum CD 2010 November / maximum-cd-2010-11.iso / DiscContents / calibre-0.7.13.msi / file_4053 < prev    next >
Encoding:
Text File  |  2010-07-02  |  1.5 KB  |  36 lines
  1. from calibre.web.feeds.news import BasicNewsRecipe
  2.  
  3. class AdvancedUserRecipe1278049615(BasicNewsRecipe):
  4.     title          = u'Statesman'
  5.     pubisher  = 'http://www.statesman.com/'
  6.     description           = 'Austin Texas Daily Newspaper'
  7.     category              = 'News, Austin, Texas'
  8.     __author__            = 'rty'
  9.     oldest_article = 3
  10.  
  11.     max_articles_per_feed = 100
  12.  
  13.     feeds          = [(u'News', u'http://www.statesman.com/section-rss.do?source=news&includeSubSections=true'),
  14.     (u'Business', u'http://www.statesman.com/section-rss.do?source=business&includeSubSections=true'),
  15.     (u'Life', u'http://www.statesman.com/section-rss.do?source=life&includesubsection=true'),
  16.     (u'Editorial', u'http://www.statesman.com/section-rss.do?source=opinion&includesubsections=true'),
  17.     (u'Sports', u'http://www.statesman.com/section-rss.do?source=sports&includeSubSections=true')
  18.     ]
  19.     masthead_url = "http://www.statesman.com/images/cmg-logo.gif"
  20.     #temp_files = []
  21.     #articles_are_obfuscated = True
  22.  
  23.     remove_javascript = True
  24.     use_embedded_content   = False
  25.     no_stylesheets = True
  26.     language = 'en'
  27.     encoding               = 'utf-8'
  28.     conversion_options = {'linearize_tables':True}
  29.     remove_tags = [
  30.                     dict(name='div', attrs={'id':'cxArticleOptions'}),
  31.                         ]
  32.     keep_only_tags = [
  33.      dict(name='div', attrs={'class':'cxArticleHeader'}),
  34.                      dict(name='div', attrs={'id':'cxArticleBodyText'}),
  35.                                ]
  36.