home *** CD-ROM | disk | FTP | other *** search
/ Chip 2011 November / CHIP_2011_11.iso / Programy / Narzedzia / Calibre / calibre-0.8.18.msi / file_280 / faznet.recipe < prev    next >
Text File  |  2011-09-09  |  3KB  |  64 lines

  1. __license__   = 'GPL v3'
  2. __copyright__ = '2008-2011, Kovid Goyal <kovid at kovidgoyal.net>, Darko Miletic <darko at gmail.com>'
  3. '''
  4. Profile to download FAZ.NET
  5. '''
  6.  
  7. from calibre.web.feeds.news import BasicNewsRecipe
  8.  
  9. class FazNet(BasicNewsRecipe):
  10.     title                 = 'FAZ.NET'
  11.     __author__            = 'Kovid Goyal, Darko Miletic'
  12.     description           = 'Frankfurter Allgemeine Zeitung'
  13.     publisher             = 'Frankfurter Allgemeine Zeitung GmbH'
  14.     category              = 'news, politics, Germany'
  15.     use_embedded_content  = False
  16.     language = 'de'
  17.  
  18.     max_articles_per_feed = 30
  19.     no_stylesheets        = True
  20.     encoding              = 'utf-8'
  21.     remove_javascript     = True
  22.  
  23.     html2lrf_options = [
  24.                           '--comment', description
  25.                         , '--category', category
  26.                         , '--publisher', publisher
  27.                         ]
  28.  
  29.     html2epub_options = 'publisher="' + publisher + '"\ncomments="' + description + '"\ntags="' + category + '"'
  30.  
  31.     keep_only_tags = [dict(name='div', attrs={'class':'Article'})]
  32.  
  33.     remove_tags = [
  34.                      dict(name=['object','link','embed','base'])
  35.                     ,dict(name='div',
  36.                         attrs={'class':['LinkBoxModulSmall','ModulVerlagsInfo',
  37.                             'ArtikelServices', 'ModulLesermeinungenFooter',
  38.                             'ModulArtikelServices', 'BoxTool Aufklappen_Grau',
  39.                             'SocialMediaUnten', ]}),
  40.                     dict(id=['KurzLinkMenu', 'ArtikelServicesMenu']),
  41.                   ]
  42.  
  43.     feeds = [
  44.               ('FAZ.NET Aktuell', 'http://www.faz.net/s/RubF3CE08B362D244869BE7984590CB6AC1/Tpl~Epartner~SRss_.xml'),
  45.               ('Politik', 'http://www.faz.net/s/RubA24ECD630CAE40E483841DB7D16F4211/Tpl~Epartner~SRss_.xml'),
  46.               ('Wirtschaft', 'http://www.faz.net/s/RubC9401175958F4DE28E143E68888825F6/Tpl~Epartner~SRss_.xml'),
  47.               ('Feuilleton', 'http://www.faz.net/s/RubCC21B04EE95145B3AC877C874FB1B611/Tpl~Epartner~SRss_.xml'),
  48.               ('Sport', 'http://www.faz.net/s/Rub9F27A221597D4C39A82856B0FE79F051/Tpl~Epartner~SRss_.xml'),
  49.               ('Gesellschaft', 'http://www.faz.net/s/Rub02DBAA63F9EB43CEB421272A670A685C/Tpl~Epartner~SRss_.xml'),
  50.               ('Finanzen', 'http://www.faz.net/s/Rub4B891837ECD14082816D9E088A2D7CB4/Tpl~Epartner~SRss_.xml'),
  51.               ('Wissen', 'http://www.faz.net/s/Rub7F4BEE0E0C39429A8565089709B70C44/Tpl~Epartner~SRss_.xml'),
  52.               ('Reise', 'http://www.faz.net/s/RubE2FB5CA667054BDEA70FB3BC45F8D91C/Tpl~Epartner~SRss_.xml'),
  53.               ('Technik & Motor', 'http://www.faz.net/s/Rub01E4D53776494844A85FDF23F5707AD8/Tpl~Epartner~SRss_.xml'),
  54.               ('Beruf & Chance', 'http://www.faz.net/s/RubB1E10A8367E8446897468EDAA6EA0504/Tpl~Epartner~SRss_.xml')
  55.             ]
  56.  
  57.     def preprocess_html(self, soup):
  58.         mtag = '<meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>'
  59.         soup.head.insert(0,mtag)
  60.         del soup.body['onload']
  61.         for item in soup.findAll(style=True):
  62.             del item['style']
  63.         return soup
  64.