__license__ = 'GPL v3' __copyright__ = '2010 - 2014, Darko Miletic ' ''' www.aif.ru ''' from calibre.web.feeds.news import BasicNewsRecipe class AIF_ru(BasicNewsRecipe): title = 'Arguments & Facts - Russian' __author__ = 'Darko Miletic' description = 'News from Russia' publisher = 'AIF' category = 'news, politics, Russia' oldest_article = 2 max_articles_per_feed = 100 no_stylesheets = True use_embedded_content = False encoding = 'utf8' language = 'ru' publication_type = 'magazine' masthead_url = 'http://static3.aif.ru/glossy/index/i/logo.png' extra_css = """ body{font-family: Verdana,Arial,Helvetica,sans1,sans-serif} img{display: block} """ keep_only_tags = [ dict(name='h1', attrs={'class': 'title'}), dict(name='div', attrs={'class': 'prew_tags'}), dict( name='article', attrs={'class': lambda x: x and 'articl_body' in x.split()}) ] remove_tags = [ dict(name=['iframe', 'object', 'link', 'base', 'input', 'meta']), dict(name='div', attrs={'class': 'in-topic'}), dict(name='div', attrs={ 'class': lambda x: x and 'related_article' in x.split()}), dict(name='div', attrs={'class': lambda x: x and 'articl_tag' in x.split()}) ] feeds = [(u'News', u'http://www.aif.ru/rss/all.php')]