calibre/recipes/lesoir_be.recipe
2015-01-23 19:08:21 +05:30

50 lines
1.7 KiB
Python

#!/usr/bin/env python2
__license__ = 'GPL v3'
__copyright__ = '2008-2011, Lionel Bergeret <lbergeret at gmail.com>'
'''
lesoir.be
'''
from calibre import strftime
from calibre.web.feeds.news import BasicNewsRecipe
class LeSoirBe(BasicNewsRecipe):
title = u'Le Soir'
__author__ = u'Lionel Bergeret'
description = u'News from Belgium in French'
publisher = u'lesoir.be'
category = 'news, Belgium'
oldest_article = 3
language = 'fr'
masthead_url = 'http://pdf.lesoir.be/pdf/images/SOIR//logo.gif'
max_articles_per_feed = 20
no_stylesheets = True
use_embedded_content = False
timefmt = ' [%d %b %Y]'
keep_only_tags = [
dict(name = 'div', attrs = {'id': 'story_head'})
,dict(name = 'div', attrs = {'id': 'story_body'})
]
remove_tags = [
dict(name='form', attrs={'id':'story_actions'})
,dict(name='div', attrs={'id':'sb-share'})
,dict(name='div', attrs={'id':'sb-subscribe'})
]
feeds = [
(u'Belgique' , u'http://www.lesoir.be/actualite/belgique/rss.xml' )
,(u'France' , u'http://www.lesoir.be/actualite/france/rss.xml' )
,(u'Monde' , u'http://www.lesoir.be/actualite/monde/rss.xml' )
,(u'Regions' , u'http://www.lesoir.be/regions/rss.xml' )
,(u'Vie du Net' , u'http://www.lesoir.be/actualite/vie_du_net/rss.xml' )
,(u'Petite Gazette' , u'http://www.lesoir.be/actualite/sciences/rss.xml' )
]
def get_cover_url(self):
cover_url = strftime( 'http://pdf.lesoir.be/pdf/%Y-%m-%d_BRUX_UNE_1.PDF')
return cover_url