Fix #786299 (Updated recipe for El Mundo)

This commit is contained in:
Kovid Goyal 2011-05-21 14:20:45 -06:00
parent 746b9b1d63
commit 13387225e0
2 changed files with 33 additions and 15 deletions

View File

@ -1,6 +1,6 @@
__license__ = 'GPL v3'
__copyright__ = '2009, Darko Miletic <darko.miletic at gmail.com>'
__copyright__ = '2009-2011, Darko Miletic <darko.miletic at gmail.com>'
'''
elmundo.es
'''
@ -10,15 +10,24 @@ from calibre.web.feeds.news import BasicNewsRecipe
class ElMundo(BasicNewsRecipe):
title = 'El Mundo'
__author__ = 'Darko Miletic'
description = 'News from Spain'
publisher = 'El Mundo'
description = 'Lider de informacion en espaniol'
publisher = 'Unidad Editorial Informacion General S.L.U.'
category = 'news, politics, Spain'
oldest_article = 2
max_articles_per_feed = 100
no_stylesheets = True
use_embedded_content = False
encoding = 'iso8859_15'
language = 'es'
language = 'es_ES'
masthead_url = 'http://estaticos03.elmundo.es/elmundo/iconos/v4.x/v4.01/bg_h1.png'
publication_type = 'newspaper'
extra_css = """
body{font-family: Arial,Helvetica,sans-serif}
.metadata_noticia{font-size: small}
h1,h2,h3,h4,h5,h6,.subtitulo {color: #3F5974}
.hora{color: red}
.update{color: gray}
"""
conversion_options = {
'comments' : description
@ -30,22 +39,31 @@ class ElMundo(BasicNewsRecipe):
keep_only_tags = [dict(name='div', attrs={'class':'noticia'})]
remove_tags_before = dict(attrs={'class':['titular','antetitulo'] })
remove_tags_after = dict(name='div' , attrs={'id':['desarrollo_noticia','tamano']})
remove_attributes = ['lang','border']
remove_tags = [
dict(name='div', attrs={'class':['herramientas','publicidad_google']})
,dict(name='div', attrs={'id':'modulo_multimedia' })
,dict(name='ul', attrs={'class':'herramientas' })
,dict(name=['object','link'])
,dict(name=['object','link','embed','iframe','base','meta'])
]
feeds = [
(u'Portada' , u'http://rss.elmundo.es/rss/descarga.htm?data2=4' )
,(u'Deportes' , u'http://rss.elmundo.es/rss/descarga.htm?data2=14')
,(u'Economia' , u'http://rss.elmundo.es/rss/descarga.htm?data2=7' )
,(u'Espana' , u'http://rss.elmundo.es/rss/descarga.htm?data2=8' )
,(u'Internacional' , u'http://rss.elmundo.es/rss/descarga.htm?data2=9' )
,(u'Cultura' , u'http://rss.elmundo.es/rss/descarga.htm?data2=6' )
,(u'Ciencia/Ecologia', u'http://rss.elmundo.es/rss/descarga.htm?data2=5' )
,(u'Comunicacion' , u'http://rss.elmundo.es/rss/descarga.htm?data2=26')
,(u'Television' , u'http://rss.elmundo.es/rss/descarga.htm?data2=76')
(u'Portada' , u'http://estaticos.elmundo.es/elmundo/rss/portada.xml' )
,(u'Deportes' , u'http://estaticos.elmundo.es/elmundodeporte/rss/portada.xml')
,(u'Economia' , u'http://estaticos.elmundo.es/elmundo/rss/economia.xml' )
,(u'Espana' , u'http://estaticos.elmundo.es/elmundo/rss/espana.xml' )
,(u'Internacional' , u'http://estaticos.elmundo.es/elmundo/rss/internacional.xml' )
,(u'Cultura' , u'http://estaticos.elmundo.es/elmundo/rss/cultura.xml' )
,(u'Ciencia/Ecologia', u'http://estaticos.elmundo.es/elmundo/rss/ciencia.xml' )
,(u'Comunicacion' , u'http://estaticos.elmundo.es/elmundo/rss/comunicacion.xml' )
,(u'Television' , u'http://estaticos.elmundo.es/elmundo/rss/television.xml' )
]
def preprocess_html(self, soup):
for item in soup.findAll(style=True):
del item['style']
return soup
def get_article_url(self, article):
return article.get('guid', None)

Binary file not shown.

Before

Width:  |  Height:  |  Size: 550 B

After

Width:  |  Height:  |  Size: 1.1 KiB