from calibre.web.feeds.news import BasicNewsRecipe, classes class DeutscheWelle_bs(BasicNewsRecipe): title = 'Deutsche Welle' __author__ = 'unkn0wn' description = 'Vijesti iz Njemacke i svijeta' publisher = 'Deutsche Welle' category = 'news, politics, Germany' oldest_article = 2 max_articles_per_feed = 100 use_embedded_content = False no_stylesheets = True language = 'bs' publication_type = 'newsportal' remove_empty_feeds = True remove_javascript = True masthead_url = 'http://www.dw-world.de/skins/std/channel1/pics/dw_logo1024.gif' ignore_duplicate_articles = {'title', 'url'} remove_attributes = ['height', 'width', 'style'] keep_only_tags = [ dict(name='article') ] recipe_specific_options = { 'days': { 'short': 'Oldest article to download from this news source. In days ', 'long': 'For example, 0.5, gives you articles from the past 12 hours', 'default': str(oldest_article) } } def __init__(self, *args, **kwargs): BasicNewsRecipe.__init__(self, *args, **kwargs) d = self.recipe_specific_options.get('days') if d and isinstance(d, str): self.oldest_article = float(d) remove_tags = [ dict(name=['footer', 'source']), dict(attrs={'data-tracking-name':'sharing-icons-inline'}), classes('kicker advertisement vjs-wrapper') ] feeds = [ (u'Politika', u'http://rss.dw-world.de/rdf/rss-bos-pol'), (u'Evropa', u'http://rss.dw-world.de/rdf/rss-bos-eu'), (u'Kiosk', u'http://rss.dw-world.de/rdf/rss-bos-eu'), (u'Ekonomija i Nuka', u'http://rss.dw-world.de/rdf/rss-bos-eco'), (u'Kultura', u'http://rss.dw-world.de/rdf/rss-bos-cul'), (u'Sport', u'http://rss.dw-world.de/rdf/rss-bos-sp') ] def preprocess_html(self, soup): for img in soup.findAll('img', srcset=True): img['src'] = img['srcset'].split()[6] return soup