#!/usr/bin/env python # -*- coding: utf-8 -*- __license__ = 'GPL v3' __copyright__ = u'2011, Silviu Cotoar\u0103' ''' sapteseri.ro ''' from calibre.web.feeds.news import BasicNewsRecipe class SapteSeri(BasicNewsRecipe): title = u'Sapte Seri' __author__ = u'Silviu Cotoar\u0103' description = u'Sapte Seri' publisher = u'Sapte Seri' oldest_article = 5 language = 'ro' max_articles_per_feed = 100 no_stylesheets = True use_embedded_content = False category = 'Ziare,Oras,Distractie,Fun' encoding = 'utf-8' remove_empty_feeds = True remove_javascript = True cover_url = 'http://www.sapteseri.ro/Images/logo.jpg' conversion_options = { 'comments': description, 'tags': category, 'language': language, 'publisher': publisher } keep_only_tags = [ dict(name='h1', attrs={'id': 'title'}), dict(name='div', attrs={'class': 'mt10 mb10'}), dict( name='div', attrs={'class': 'mb20 mt10'}), dict(name='div', attrs={'class': 'mt5 mb20'}) ] remove_tags = [ dict(name='div', attrs={'id': ['entityimgworking']}) ] feeds = [ (u'Ce se intampla azi in Bucuresti', u'http://www.sapteseri.ro/ro/feed/ce-se-intampla-azi/bucuresti/') ] def preprocess_html(self, soup): return self.adeify_images(soup)