diff --git a/recipes/courrierinternational.recipe b/recipes/courrierinternational.recipe index 7f8bb1315f..63c6d800cf 100644 --- a/recipes/courrierinternational.recipe +++ b/recipes/courrierinternational.recipe @@ -35,21 +35,41 @@ class CourrierInternational(BasicNewsRecipe): self.oldest_article = float(d) max_articles_per_feed = 50 + ignore_duplicate_articles = {'title'} no_stylesheets = True html2lrf_options = ['--base-font-size', '10'] - keep_only_tags = [ - dict(name='div', attrs={'class': 'article-content'}), - ] + keep_only_tags = dict(name='article') remove_tags = [ - dict(name='aside', attrs={'class': re.compile( - 'article-sitesocial|article-ad|article-tertiary|view-ci-service-article')}), - dict(name='button', attrs={'id': re.compile( - 'action-zen-off-(top|bottom)')}), + dict(attrs={'class': ['asset-read-more', 'article-secondary', 'article-aside', 'item', 'source-logo', 'source-lang', 'info-time']}) ] + extra_css = ''' + .article-lede { + margin-bottom: 1em; + } + + .caption, .credit { + font-size: 0.9em; + font-style: italic; + } + + .info-date, .info-reserved { + font-style: italic; + font-size: 0.9em; + display: block; + } + + .strapline { + font-size: 1.17em; + font-weight: bold; + margin: 0.67em 0; + line-height: 1.2; + } + ''' + needs_subscription = "optional" login_url = 'http://www.courrierinternational.com/login' @@ -72,30 +92,13 @@ class CourrierInternational(BasicNewsRecipe): feeds = [ # Some articles requiring subscription fails on download. - ('Tous', 'http://www.courrierinternational.com/feed/all/rss.xml'), - ('France', 'https://www.courrierinternational.com/feed/category/6260/rss.xml'), - ('Europe', 'https://www.courrierinternational.com/feed/category/6261/rss.xml'), - ('Ameriques', 'https://www.courrierinternational.com/feed/category/6262/rss.xml'), - ('Asie', 'https://www.courrierinternational.com/feed/category/6263/rss.xml'), - ('Afrique', 'https://www.courrierinternational.com/feed/category/6264/rss.xml'), - ('Moyen-Orient', 'https://www.courrierinternational.com/feed/category/6265/rss.xml'), - ('Expat', 'https://www.courrierinternational.com/feed/category/6994/rss.xml'), - ('Business', 'https://www.courrierinternational.com/feed/category/6991/rss.xml'), - ('Culture', 'https://www.courrierinternational.com/feed/category/6270/rss.xml'), - ('Economie', 'https://www.courrierinternational.com/feed/category/6266/rss.xml'), - ('Écrans', 'https://www.courrierinternational.com/feed/category/7342/rss.xml'), - ('Éducation', 'https://www.courrierinternational.com/feed/category/6997/rss.xml'), - ('Environnement', 'https://www.courrierinternational.com/feed/category/6267/rss.xml'), - ('Infographie', 'https://www.courrierinternational.com/feed/category/6281/rss.xml'), - ('Lifestyle', 'https://www.courrierinternational.com/feed/category/7426/rss.xml'), - ('Planète buzz', 'https://www.courrierinternational.com/feed/category/6680/rss.xml'), - ('Politique', 'https://www.courrierinternational.com/feed/category/6681/rss.xml'), - ('Santé', 'https://www.courrierinternational.com/feed/category/6682/rss.xml'), - ('Sport', 'https://www.courrierinternational.com/feed/category/6273/rss.xml'), - ('Tendances', 'https://www.courrierinternational.com/feed/category/6282/rss.xml'), - ('Réveil', 'https://www.courrierinternational.com/feed/category/8224/rss.xml'), - ('Enquêtes', 'https://www.courrierinternational.com/feed/category/6678/rss.xml'), - ('Histoire', 'https://www.courrierinternational.com/feed/category/6679/rss.xml'), - ('Sciences & Techno', 'https://www.courrierinternational.com/feed/category/6268/rss.xml'), - ('Voyage', 'https://www.courrierinternational.com/feed/category/6271/rss.xml'), + ('France', 'https://www.courrierinternational.com/feed/rubrique/france/rss.xml'), + ('Geopolitique', 'https://www.courrierinternational.com/feed/rubrique/geopolitique/rss.xml'), + ('Economie', 'https://www.courrierinternational.com/feed/rubrique/economie/rss.xml'), + ('Société', 'https://www.courrierinternational.com/feed/rubrique/societe/rss.xml'), + ('Politique', 'https://www.courrierinternational.com/feed/rubrique/politique/rss.xml'), + ('Sciences & Environnement', 'https://www.courrierinternational.com/feed/rubrique/science-environnement/rss.xml'), + ('Culture', 'https://www.courrierinternational.com/feed/rubrique/culture/rss.xml'), + ('Expat', 'https://www.courrierinternational.com/feed/rubrique/expat/rss.xml'), + ('Autres', 'http://www.courrierinternational.com/feed/all/rss.xml'), ]