diff --git a/recipes/cicero.recipe b/recipes/cicero.recipe index de6676ae44..beb323cf31 100644 --- a/recipes/cicero.recipe +++ b/recipes/cicero.recipe @@ -1,38 +1,47 @@ from calibre.web.feeds.news import BasicNewsRecipe class BasicUserRecipe1316245412(BasicNewsRecipe): - title = u'Cicero Online' description = u'Magazin f\xfcr politische Kultur' publisher = 'Ringier Publishing GmbH' category = 'news, politics, Germany' language = 'de' encoding = 'UTF-8' - __author__ = 'Armin Geller' # 2011-09-17 + __author__ = 'Armin Geller' # Upd. 2011-09-19 - oldest_article = 7 + oldest_article = 7 max_articles_per_feed = 100 no_stylesheets = True auto_cleanup = False +# remove_javascript = True + remove_tags = [ - dict(name='div', attrs={'id':["header", "navigation", "skip-link", "header-print", "header-print-url", "meta-toolbar", "footer"]}), - dict(name='div', attrs={'class':["region region-sidebar-first column sidebar", "breadcrumb", "breadcrumb-title", "meta", "comment-wrapper", - "field field-name-field-show-teaser-right field-type-list-boolean field-label-above"]}), - dict(name='div', attrs={'title':["Dossier Auswahl"]}), - dict(name='h2', attrs={'class':["title comment-form"]}), - dict(name='form', attrs={'class':["comment-form user-info-from-cookie"]}), - ] + dict(name='div', attrs={'id':["header", "navigation", "skip-link", "header-print", "header-print-url", "meta-toolbar", "footer"]}), + dict(name='div', attrs={'class':["region region-sidebar-first column sidebar", "breadcrumb", "breadcrumb-title", "meta", "comment-wrapper", + "field field-name-field-show-teaser-right field-type-list-boolean field-label-above"]}), + dict(name='div', attrs={'title':["Dossier Auswahl"]}), + dict(name='h2', attrs={'class':["title comment-form"]}), + dict(name='form', attrs={'class':["comment-form user-info-from-cookie"]}), + # 2011-09-19 clean-up on first feed historical caricature- and video preview pictures and social icons + dict(name='table', attrs={'class':["mcx-social-horizontal", "page-header"]}), # 2011-09-19 + dict(name='div', attrs={'class':["page-header", "view view-alle-karikaturen view-id-alle_karikaturen view-display-id-default view-dom-id-1", + "pagination", + "view view-letzte-videos view-id-letzte_videos view-display-id-default view-dom-id-1"]}), # 2011-09-19 + ] feeds = [ - (u'Das gesamte Portfolio', u'http://www.cicero.de/rss.xml'), - (u'Berliner Republik', u'http://www.cicero.de/berliner-republik.xml'), - (u'Weltb\xfchne', u'http://www.cicero.de/weltbuehne.xml'), - (u'Kapital', u'http://www.cicero.de/kapital.xml'), - (u'Salon', u'http://www.cicero.de/salon.xml'), - (u'Blogs', u'http://www.cicero.de/blogs.xml'), #seems not to be in use at the moment - ] + (u'Das gesamte Portfolio', u'http://www.cicero.de/rss.xml'), + (u'Berliner Republik', u'http://www.cicero.de/berliner-republik.xml'), + (u'Weltb\xfchne', u'http://www.cicero.de/weltbuehne.xml'), + (u'Kapital', u'http://www.cicero.de/kapital.xml'), + (u'Salon', u'http://www.cicero.de/salon.xml'), + (u'Blogs', u'http://www.cicero.de/blogs.xml'), #seems not to be in use at the moment + ] def print_version(self, url): - return url + '?print' + return url + '?print' + +# def get_cover_url(self): +# return 'http://www.cicero.de/sites/all/themes/cicero/logo.png' # need to find a good logo on their home page!