calibre/resources/recipes/clarin.recipe

58 lines
2.5 KiB
Plaintext

__license__ = 'GPL v3'
__copyright__ = '2008-2010, Darko Miletic <darko.miletic at gmail.com>'
'''
clarin.com
'''
from calibre import strftime
from calibre.web.feeds.news import BasicNewsRecipe
class Clarin(BasicNewsRecipe):
title = 'Clarin'
__author__ = 'Darko Miletic'
description = 'Noticias de Argentina y mundo'
publisher = 'Grupo Clarin'
category = 'news, politics, Argentina'
oldest_article = 2
max_articles_per_feed = 100
use_embedded_content = False
no_stylesheets = True
cover_url = strftime('http://www.clarin.com/diario/%Y/%m/%d/portada.jpg')
encoding = 'cp1252'
language = 'es'
masthead_url = 'http://www.clarin.com/shared/v10/img/Hd/lg_Clarin.gif'
extra_css = ' body{font-family: Arial,Helvetica,sans-serif} h2{font-family: Georgia,"Times New Roman",Times,serif; font-size: xx-large} .Volan,.Pie,.Autor{ font-size: x-small} .Copete,.Hora{font-size: large} '
conversion_options = {
'comment' : description
, 'tags' : category
, 'publisher': publisher
, 'language' : language
}
remove_tags = [
dict(name='a' , attrs={'class':'Imp' })
,dict(name='div' , attrs={'class':'Perma' })
,dict(name='h1' , text='Imprimir' )
]
feeds = [
(u'Ultimo Momento', u'http://www.clarin.com/diario/hoy/um/sumariorss.xml')
,(u'El Pais' , u'http://www.clarin.com/diario/hoy/elpais.xml' )
,(u'Opinion' , u'http://www.clarin.com/diario/hoy/opinion.xml' )
,(u'El Mundo' , u'http://www.clarin.com/diario/hoy/elmundo.xml' )
,(u'Sociedad' , u'http://www.clarin.com/diario/hoy/sociedad.xml' )
,(u'La Ciudad' , u'http://www.clarin.com/diario/hoy/laciudad.xml' )
,(u'Policiales' , u'http://www.clarin.com/diario/hoy/policiales.xml' )
,(u'Deportes' , u'http://www.clarin.com/diario/hoy/deportes.xml' )
]
def print_version(self, url):
rest = url.partition('-0')[-1]
lmain = rest.partition('.')[0]
lurl = u'http://www.servicios.clarin.com/notas/jsp/clarin/v9/notas/imprimir.jsp?pagid=' + lmain
return lurl