calibre/recipes/miradasalsur.recipe
Kovid Goyal 952b95d3ad pep8
2013-06-27 17:34:01 +05:30

68 lines
3.1 KiB
Plaintext

__copyright__ = '2009-2013, Darko Miletic <darko.miletic at gmail.com>'
'''
sur.infonews.com
'''
import datetime
from calibre.web.feeds.news import BasicNewsRecipe
class MiradasAlSur(BasicNewsRecipe):
title = 'Miradas al Sur'
__author__ = 'Darko Miletic'
description = 'Semanario Argentino'
publisher = 'ElArgentino.com'
category = 'news, politics, Argentina'
oldest_article = 7
max_articles_per_feed = 100
no_stylesheets = True
use_embedded_content = False
encoding = 'utf-8'
language = 'es_AR'
remove_empty_feeds = True
masthead_url = 'http://sur.infonews.com/sites/default/files/www_miradas_al_sur_com_logo.gif'
extra_css = """
body{font-family: Arial,Helvetica,sans-serif}
h1{font-family: Georgia,Times,serif}
.field-field-story-author{color: gray; font-size: small}
"""
conversion_options = {
'comment' : description
, 'tags' : category
, 'publisher' : publisher
, 'language' : language
, 'series' : title
}
keep_only_tags = [dict(name='div', attrs={'id':['content-header', 'content-area']})]
remove_tags = [
dict(name=['link','meta','iframe','embed','object']),
dict(name='form', attrs={'class':'fivestar-widget'}),
dict(attrs={'class':lambda x: x and 'terms-inline' in x.split()})
]
feeds = [
(u'Politica' , u'http://sur.infonews.com/taxonomy/term/1/0/feed'),
(u'Internacional' , u'http://sur.infonews.com/taxonomy/term/2/0/feed'),
(u'Informe Especial' , u'http://sur.infonews.com/taxonomy/term/14/0/feed'),
(u'Delitos y pesquisas', u'http://sur.infonews.com/taxonomy/term/6/0/feed'),
(u'Lesa Humanidad' , u'http://sur.infonews.com/taxonomy/term/7/0/feed'),
(u'Cultura' , u'http://sur.infonews.com/taxonomy/term/8/0/feed'),
(u'Deportes' , u'http://sur.infonews.com/taxonomy/term/9/0/feed'),
(u'Contratapa' , u'http://sur.infonews.com/taxonomy/term/10/0/feed'),
]
def get_cover_url(self):
# determine the series number, unfortunately not gonna happen now
#self.conversion_options.update({'series_index':seriesnr})
cover_url = None
cdate = datetime.date.today()
todayweekday = cdate.isoweekday()
if (todayweekday != 7):
cdate -= datetime.timedelta(days=todayweekday)
cover_page_url = cdate.strftime('http://sur.infonews.com/ediciones/%Y-%m-%d/tapa')
soup = self.index_to_soup(cover_page_url)
cover_item = soup.find('img', attrs={'class':lambda x: x and 'imagecache-tapa_edicion_full' in x.split()})
if cover_item:
cover_url = cover_item['src']
return cover_url