home *** CD-ROM | disk | FTP | other *** search
/ Maximum CD 2010 November / maximum-cd-2010-11.iso / DiscContents / calibre-0.7.13.msi / file_3698 < prev    next >
Encoding:
Text File  |  2010-04-05  |  3.6 KB  |  94 lines
  1. #!/usr/bin/env  python
  2.  
  3. __license__   = 'GPL v3'
  4. __copyright__ = '2009, Darko Miletic <darko.miletic at gmail.com>'
  5. '''
  6. diariocordoba.com
  7. '''
  8.  
  9. from calibre.web.feeds.news import BasicNewsRecipe
  10.  
  11.  
  12. class Diariosur(BasicNewsRecipe):
  13.     title                 = u'Diario Cordoba'
  14.     __author__            = u'Francisco'
  15.     description           = u'News Cordoba'
  16.     oldest_article        = 5
  17.     max_articles_per_feed = 100
  18.     delay                 = 0
  19.     timeout               = 120
  20.     no_stylesheets        = True
  21.     use_embedded_content  = False
  22.     encoding              = 'UTF-8'
  23.     remove_javascript     = True
  24.     language              = 'es'
  25.     extra_css              = 'body{font-family: Arial,Helvetica,sans-serif}'
  26.  
  27.  
  28.     remove_attributes = ['height','width']
  29.  
  30.     keep_only_tags = [
  31.                        dict(name='div', attrs={'id':'contenidos'})
  32.                      ]
  33.  
  34.     remove_tags = [
  35.                    dict(name='div', attrs={'class':'Recorte'}),
  36.                   dict(name='div', attrs={'id':'comentarios'}),
  37.                   dict(name='div', attrs={'id':'CajaAccesoCuentaUsuario'}),
  38.                   dict(name='div', attrs={'id':'cajacomparte'}),
  39.                   dict(name='div', attrs={'class':'FormularioDeAcceso'}),
  40.                   dict(name='div', attrs={'class':'TextoFormularioDeAcceso'}),
  41.                   dict(name='div', attrs={'class':'Recorte'}),
  42.                   dict(name='div', attrs={'id':'IframeCajaComparte'}),
  43.                   dict(name='div', attrs={'id':'CintilloComentario'}),
  44.                   dict(name='div', attrs={'id':'EscribeComentario'}),
  45.                   dict(name='div', attrs={'class':'Nota'}),
  46.                   dict(name='div', attrs={'id':'FormularioComentario'}),
  47.                   dict(name='div', attrs={'id':'Comparte'}),
  48.                   dict(name='iframe', attrs={'id':'IframeCajaComparte'}),
  49.                   dict(name='ul', attrs={'class':'herramientasDeNoticia'}),
  50.                   dict(name='div', attrs={'id':'NoticiaEnPapel'}),
  51.                   dict(name='div', attrs={'class':'navegaNoticias'}),
  52.                   dict(name='p', attrs={'class':'RecorteEnNoticias'}),
  53.                   dict(name='ul', attrs={'class':'herramientasDeNoticia'}),
  54.                   dict(name='div', attrs={'class':'navegaNoticias'}),
  55.                   dict(name='div', attrs={'id':'CajaComparte'}),
  56.                   dict(name='div', attrs={'id':'Comparte'})
  57.  
  58.  
  59.  
  60.                   ]
  61.  
  62.  
  63.  
  64.  
  65.     html2lrf_options = [
  66.                           '--comment', description
  67.                         , '--base-font-size', '6'
  68.                         , '--category', 'news, Spain'
  69.                           , '--ignore-tables'
  70.                         ]
  71.  
  72.  
  73.  
  74.  
  75.     feeds = [
  76.      (u'Ultima Hora' ,'http://www.diariocordoba.com/rss/ultimahora.xml'),
  77.      (u'Tema del Dia' ,'http://www.diariocordoba.com/rss/106.xml'),
  78.      (u'Local' ,'http://www.diariocordoba.com/rss/101.xml'),
  79.      (u'Provincia' ,'http://www.diariocordoba.com/rss/102.xml'),
  80.      (u'Andalucia' ,'http://www.diariocordoba.com/rss/1.xml'),
  81.      (u'Opinion' ,'http://www.diariocordoba.com/rss/100.xml'),
  82.      (u'Deportes' ,'http://www.diariocordoba.com/rss/4.xml'),
  83.      (u'Espa\xc3\xb1a' ,'http://www.diariocordoba.com/rss/7.xml'),
  84.      (u'Internacional' ,'http://www.diariocordoba.com/rss/6.xml'),
  85.      (u'Economia' ,'http://www.diariocordoba.com/rss/5.xml'),
  86.      (u'Cultura' ,'http://www.diariocordoba.com/rss/3.xml'),
  87.      (u'Sociedad' ,'http://www.diariocordoba.com/rss/103.xml'),
  88.      (u'Gente' ,'http://www.diariocordoba.com/rss/204.xml'),
  89.      (u'Noticias Curiosas' ,'http://www.diariocordoba.com/rss/205.xml'),
  90.      (u'Tecnologia' ,'http://www.diariocordoba.com/rss/206.xml')
  91.  
  92.  
  93.                  ]
  94.