#!/usr/bin/env python2 # -*- coding: utf-8 -*- __license__ = 'GPL v3' __copyright__ = u'2011, Silviu Cotoar\u0103' ''' historia.ro ''' from calibre.web.feeds.news import BasicNewsRecipe class HistoriaRo(BasicNewsRecipe): title = u'Historia' __author__ = u'Silviu Cotoar\u0103' description = '' publisher = 'Historia' oldest_article = 5 language = 'ro' max_articles_per_feed = 100 no_stylesheets = True use_embedded_content = False category = 'Ziare,Reviste,Istorie' encoding = 'utf-8' cover_url = 'http://www.historia.ro/sites/all/themes/historia/images/historia.png' conversion_options = { 'comments': description, 'tags': category, 'language': language, 'publisher': publisher } keep_only_tags = [ dict(name='div', attrs={'class': 'c_antet_title'}), dict(name='a', attrs={ 'class': 'overlaybox'}), dict(name='div', attrs={'class': 'art_content'}) ] remove_tags = [ dict(name='div', attrs={'class': ['fl_left']}), dict(name='div', attrs={ 'id': ['article_toolbar']}), dict(name='div', attrs={'class': ['zoom_cont']}) ] feeds = [ (u'Feeds', u'http://www.historia.ro/rss.xml') ] def preprocess_html(self, soup): return self.adeify_images(soup)