__license__ = 'GPL v3' __copyright__ = '2010, Darko Miletic ' ''' information.dk ''' import re from calibre.web.feeds.news import BasicNewsRecipe class Information_dk(BasicNewsRecipe): title = 'Information - Denmark' __author__ = 'Darko Miletic' description = 'News from Denmark' publisher = 'information.dk' category = 'news, politics, Denmark' oldest_article = 2 max_articles_per_feed = 100 no_stylesheets = True remove_empty_feeds = True use_embedded_content = False encoding = 'utf8' language = 'da' conversion_options = { 'comment': description, 'tags': category, 'publisher': publisher, 'language': language } feeds = [ (u'Nyheder', u'https://www.information.dk/feed'), (u'Politik', u'https://www.information.dk/emne/politik/feed'), (u'Indland', u'https://www.information.dk/indland/feed'), (u'Udland', u'https://www.information.dk/udland/feed'), (u'Debat', u'https://www.information.dk/debat/feed'), (u'Kunst og kultur', u'https://www.information.dk/kultur/feed'), (u'Literatur', u'https://www.information.dk/litteratur/feed'), (u'Film', u'https://www.information.dk/emne/film/feed'), (u'Kunst', u'https://www.information.dk/emne/kunst/feed') ] keep_only_tags = [dict(name='article', attrs={}), dict(name='div', attrs={'class': 'field-name-body'})] remove_tags_before = dict(name='h1', attrs={'class': 'print-title'}) remove_tags_after = dict(name='div', attrs={'class': 'print-footer'}) field_name_best_from_inf = '^.*?field-name-best-from-inf.*?$' related_stuff = '^.*?related-stuff.*?$' view_iv_seneste = '^.*?view-iv-seneste.*?$' view_iv_news_queue = '^.*?view-iv-news-queue.*?$' view_footer = '^.*?view-footer.*?$' field_name_recommendations = '^.*?field-name-recommendations.*?$' field_name_comments_label = '^.*?field-name-comments-label.*?$' field_name_no_comments = '^.?field-name-no-comments.*?$' comment = '^.*?comment.*?$' remove_tags = [dict(name=['object', 'link']), dict(name='div', attrs={'id': 'node-article-full-group-date'}), dict(name='div', attrs={'id': 'node-article-full-group-right'}), dict(name='div', attrs={'id': 'node-article-non-subscriber-full-group-date'}), dict(name='div', attrs={'id': 'node-article-non-subscriber-full-group-right'}), dict(name='div', attrs={'class': re.compile(field_name_best_from_inf, re.IGNORECASE)}), dict(name='div', attrs={'class': re.compile(related_stuff, re.IGNORECASE)}), dict(name='div', attrs={'class': re.compile(view_iv_seneste, re.IGNORECASE)}), dict(name='div', attrs={'class': re.compile(view_iv_news_queue, re.IGNORECASE)}), dict(name='div', attrs={'class': re.compile(view_footer, re.IGNORECASE)}), dict(name='div', attrs={'class': re.compile(field_name_recommendations, re.IGNORECASE)}), dict(name='div', attrs={'class': re.compile(field_name_comments_label, re.IGNORECASE)}), dict(name='div', attrs={'class': re.compile(field_name_no_comments, re.IGNORECASE)}), dict(name='article', attrs={'class': re.compile(comment, re.IGNORECASE)}), dict(name='h2', attrs={'class': 'pane-title'}), dict(name='div', attrs={'id': 'comments'})]