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

53 lines
1.5 KiB
Python

#!/usr/bin/env python2
# -*- coding: utf-8 -*-
__license__ = 'GPL v3'
__copyright__ = u'2011, Silviu Cotoar\u0103'
'''
mediafax.ro
'''
from calibre.web.feeds.news import BasicNewsRecipe
class Mediafax(BasicNewsRecipe):
title = 'Mediafax'
__author__ = u'Silviu Cotoar\u0103'
description = u'\u0218tiri din Rom\u00e2nia'
publisher = 'Mediafax'
oldest_article = 5
language = 'ro'
max_articles_per_feed = 100
no_stylesheets = True
use_embedded_content = False
category = 'Ziare,Stiri,Romania'
encoding = 'utf-8'
cover_url = 'http://storage0.dms.mpinteractiv.ro/media/1/1/1706/4134575/2/logo-mediafax-mass-media-news.jpg?width=400'
conversion_options = {
'comments' : description
,'tags' : category
,'language' : language
,'publisher' : publisher
}
keep_only_tags = [
dict(name='div', attrs={'class':'news tabs-container'})
]
remove_tags = [
dict(name='ul', attrs={'class':['CategoryNews']})
,dict(name='div', attrs={'class':['read']})
]
remove_tags_after = [ dict(name='div', attrs={'class':'cmsItemViews'}) ]
feeds = [
(u'Actualitate', u'http://www.mediafax.ro/rss/')
]
def preprocess_html(self, soup):
return self.adeify_images(soup)