mirror of
https://github.com/kovidgoyal/calibre.git
synced 2025-09-29 15:31:08 -04:00
96 lines
5.2 KiB
Plaintext
96 lines
5.2 KiB
Plaintext
__license__ = 'GPL v3'
|
|
__copyright__ = '2010, Darko Miletic <darko.miletic at gmail.com>'
|
|
'''
|
|
www.elpais.com
|
|
'''
|
|
|
|
from calibre.web.feeds.news import BasicNewsRecipe
|
|
|
|
class ElPais_RSS(BasicNewsRecipe):
|
|
title = 'El Pais'
|
|
__author__ = 'Darko Miletic'
|
|
description = 'el periodico global en Castellano'
|
|
publisher = 'EDICIONES EL PAIS, S.L.'
|
|
category = 'news, politics, finances, world, spain'
|
|
oldest_article = 2
|
|
max_articles_per_feed = 200
|
|
no_stylesheets = True
|
|
encoding = 'cp1252'
|
|
use_embedded_content = False
|
|
language = 'es'
|
|
remove_empty_feeds = True
|
|
publication_type = 'newspaper'
|
|
masthead_url = 'http://www.elpais.com/im/tit_logo.gif'
|
|
extra_css = """
|
|
body{font-family: Georgia,"Times New Roman",Times,serif }
|
|
h3{font-family: Arial,Helvetica,sans-serif}
|
|
img{margin-bottom: 0.4em; display:block}
|
|
"""
|
|
|
|
conversion_options = {
|
|
'comment' : description
|
|
, 'tags' : category
|
|
, 'publisher' : publisher
|
|
, 'language' : language
|
|
}
|
|
|
|
keep_only_tags = [dict(attrs={'class':['cabecera_noticia estirar','cabecera_noticia','','contenido_noticia']})]
|
|
remove_tags = [
|
|
dict(name=['meta','link','base','iframe','embed','object'])
|
|
,dict(attrs={'class':['info_complementa','estructura_2col_der','votos estirar','votos']})
|
|
,dict(attrs={'id':'utilidades'})
|
|
]
|
|
remove_tags_after = dict(attrs={'id':'utilidades'})
|
|
remove_attributes = ['lang','border','width','height']
|
|
|
|
feeds = [
|
|
(u'Lo ultimo' , u'http://www.elpais.com/rss/feed.html?feedId=17046')
|
|
,(u'America Latina' , u'http://www.elpais.com/rss/feed.html?feedId=17041')
|
|
,(u'Mexico' , u'http://www.elpais.com/rss/feed.html?feedId=17042')
|
|
,(u'Europa' , u'http://www.elpais.com/rss/feed.html?feedId=17043')
|
|
,(u'Estados Unidos' , u'http://www.elpais.com/rss/feed.html?feedId=17044')
|
|
,(u'Oriente proximo' , u'http://www.elpais.com/rss/feed.html?feedId=17045')
|
|
,(u'Espana' , u'http://www.elpais.com/rss/feed.html?feedId=1002' )
|
|
,(u'Andalucia' , u'http://www.elpais.com/rss/feed.html?feedId=17057')
|
|
,(u'Catalunia' , u'http://www.elpais.com/rss/feed.html?feedId=17059')
|
|
,(u'Comunidad Valenciana' , u'http://www.elpais.com/rss/feed.html?feedId=17061')
|
|
,(u'Madrid' , u'http://www.elpais.com/rss/feed.html?feedId=1016' )
|
|
,(u'Pais Vasco' , u'http://www.elpais.com/rss/feed.html?feedId=17062')
|
|
,(u'Galicia' , u'http://www.elpais.com/rss/feed.html?feedId=17063')
|
|
,(u'Opinion' , u'http://www.elpais.com/rss/feed.html?feedId=1003' )
|
|
,(u'Sociedad' , u'http://www.elpais.com/rss/feed.html?feedId=1004' )
|
|
,(u'Deportes' , u'http://www.elpais.com/rss/feed.html?feedId=1007' )
|
|
,(u'Cultura' , u'http://www.elpais.com/rss/feed.html?feedId=1008' )
|
|
,(u'Cine' , u'http://www.elpais.com/rss/feed.html?feedId=17052')
|
|
,(u'Literatura' , u'http://www.elpais.com/rss/feed.html?feedId=17053')
|
|
,(u'Musica' , u'http://www.elpais.com/rss/feed.html?feedId=17051')
|
|
,(u'Arte' , u'http://www.elpais.com/rss/feed.html?feedId=17060')
|
|
,(u'Tecnologia' , u'http://www.elpais.com/rss/feed.html?feedId=1005' )
|
|
,(u'Economia' , u'http://www.elpais.com/rss/feed.html?feedId=1006' )
|
|
,(u'Ciencia' , u'http://www.elpais.com/rss/feed.html?feedId=17068')
|
|
,(u'Salud' , u'http://www.elpais.com/rss/feed.html?feedId=17074')
|
|
,(u'Ocio' , u'http://www.elpais.com/rss/feed.html?feedId=17075')
|
|
,(u'Justicia y Leyes' , u'http://www.elpais.com/rss/feed.html?feedId=17069')
|
|
,(u'Guerras y conflictos' , u'http://www.elpais.com/rss/feed.html?feedId=17070')
|
|
,(u'Politica' , u'http://www.elpais.com/rss/feed.html?feedId=17073')
|
|
]
|
|
|
|
def print_version(self, url):
|
|
return url + '?print=1'
|
|
|
|
def preprocess_html(self, soup):
|
|
for item in soup.findAll(style=True):
|
|
del item['style']
|
|
for item in soup.findAll('a'):
|
|
if item.string is not None:
|
|
tstr = item.string
|
|
item.replaceWith(tstr)
|
|
else:
|
|
item.name='span'
|
|
for atrs in ['href','target','alt','title']:
|
|
if item.has_key(atrs):
|
|
del item[atrs]
|
|
for item in soup.findAll('img',alt=False):
|
|
item['alt'] = 'image'
|
|
return soup
|