From ccc5b092be6d6fb35477dd668219cb68b0c57a31 Mon Sep 17 00:00:00 2001 From: Kovid Goyal Date: Wed, 22 Jan 2014 14:59:27 +0530 Subject: [PATCH] Update Indian Express --- recipes/indian_express.recipe | 54 ++++++++++------------------------- 1 file changed, 15 insertions(+), 39 deletions(-) diff --git a/recipes/indian_express.recipe b/recipes/indian_express.recipe index 4723b334cd..fcb0dd313a 100644 --- a/recipes/indian_express.recipe +++ b/recipes/indian_express.recipe @@ -4,58 +4,34 @@ class IndianExpress(BasicNewsRecipe): title = u'Indian Express' language = 'en_IN' __author__ = 'Krittika Goyal' - oldest_article = 1 #days + oldest_article = 1 # days max_articles_per_feed = 25 - encoding = 'cp1252' + encoding = 'utf-8' no_stylesheets = True - remove_tags_before = dict(name='div', attrs={'class':'top_head'}) - #remove_tags_after = dict(name='td', attrs={'class':'newptool1'}) - remove_tags = [ - dict(name='iframe'), - dict(name='div', attrs={'class':['pagination_new', 'comments_desc', - 'commenttext', 'postcomm', 'content_right', 'bookmarks_div', - 'comment_box', 'bookmarks_div_bot', 'box']}), - dict(name='div', attrs={'id':['footer', 'tab_innerhc', 'discussion', - 'google_new', 'header_new', 'slidebox']}), - dict(name='a', attrs={'class':'nobdr'}), - dict(name='h3', attrs={'class':['readcomm']}), - #dict(name='span', text=':'), - ] - + use_embedded_content = False + auto_cleanup = True feeds = [ ('Front Page', - 'http://syndication.indianexpress.com/rss/33/front-page.xml'), -('Markets', - 'http://syndication.indianexpress.com/rss/793/markets.xml'), + 'http://indianexpress.com/print/front-page/feed/'), ('Editorials', - 'http://syndication.indianexpress.com/rss/35/editorials.xml'), + 'http://indianexpress.com/section/opinion/editorials/feed/'), ('Crime', - 'http://syndication.indianexpress.com/rss/801/crime-&-justice.xml'), + 'http://indianexpress.com/section/india/crime/feed/'), ('Cricket', - 'http://syndication.indianexpress.com/rss/777/cricket.xml'), + 'http://indianexpress.com/section/sports/cricket/feed/'), ('Health', - 'http://syndication.indianexpress.com/rss/697/health.xml'), + 'http://www.indianexpress.com/lifestyle/health/feed/'), ('Asia', - 'http://syndication.indianexpress.com/rss/790/asia.xml'), + 'http://indianexpress.com/section/world/asia/'), ('Politics', - 'http://syndication.indianexpress.com/rss/799/politics.xml'), + 'http://indianexpress.com/section/india/politics/feed/'), ('Mumbai', - 'http://syndication.indianexpress.com/rss/707/mumbai.xml'), + 'http://www.indianexpress.com/cities/mumbai/feed/'), ('Op-Ed', - 'http://syndication.indianexpress.com/rss/36/oped.xml'), -('Economy', - 'http://syndication.indianexpress.com/rss/794/economy.xml'), + 'http://indianexpress.com/section/opinion/feed/'), ('Lifestyle', - 'http://syndication.indianexpress.com/rss/713/lifestyle.xml'), -('Letters to the Editor', - 'http://syndication.indianexpress.com/rss/40/letters-to-editor.xml'), -('Movie Reviews', - 'http://syndication.indianexpress.com/rss/665/movie-reviews.xml'), + 'http://indianexpress.com/section/lifestyle/feed/'), ('Bollywood', - 'http://syndication.indianexpress.com/rss/887/bollywood.xml'), + 'http://indianexpress.com/section/entertainment/bollywood/feed/'), ] - - def print_version(self, url): - return url+'/0' -