from calibre.web.feeds.news import BasicNewsRecipe class IDGNow(BasicNewsRecipe): title = 'IDG Now!' __author__ = 'Diniz Bortolotto' description = 'Posts do IDG Now!' oldest_article = 7 max_articles_per_feed = 20 encoding = 'utf8' publisher = 'Now!Digital Business Ltda.' category = 'technology, telecom, IT, Brazil' language = 'pt_BR' publication_type = 'technology portal' use_embedded_content = False extra_css = '.headline {font-size: x-large;} \n .fact { padding-top: 10pt }' def get_article_url(self, article): link = article.get('link', None) if link is None: return article if link.split('/')[-1] == "story01.htm": link = link.split('/')[-2] a = ['0B', '0C', '0D', '0E', '0F', '0G', '0I', '0N', '0L0S', '0A', '0J3A'] b = ['.', '/', '?', '-', '=', '&', '_', '.com', 'www.', '0', ':'] for i in range(0, len(a)): link = link.replace(a[i], b[i]) link = link.split('&')[-3] link = link.split('=')[1] link = link + "/IDGNoticiaPrint_view" return link feeds = [ (u'Ultimas noticias', u'http://rss.idgnow.com.br/c/32184/f/499640/index.rss'), (u'Computa\xe7\xe3o Corporativa', u'http://rss.idgnow.com.br/c/32184/f/499643/index.rss'), (u'Carreira', u'http://rss.idgnow.com.br/c/32184/f/499644/index.rss'), (u'Computa\xe7\xe3o Pessoal', u'http://rss.idgnow.com.br/c/32184/f/499645/index.rss'), (u'Internet', u'http://rss.idgnow.com.br/c/32184/f/499646/index.rss'), (u'Mercado', u'http://rss.idgnow.com.br/c/32184/f/419982/index.rss'), (u'Seguran\xe7a', u'http://rss.idgnow.com.br/c/32184/f/499647/index.rss'), (u'Telecom e Redes', u'http://rss.idgnow.com.br/c/32184/f/499648/index.rss') ] reverse_article_order = True