diff --git a/recipes/soldiers.recipe b/recipes/soldiers.recipe index fb96e5a2ed..a1e9e5ca23 100644 --- a/recipes/soldiers.recipe +++ b/recipes/soldiers.recipe @@ -15,6 +15,8 @@ class Soldiers(BasicNewsRecipe): max_articles_per_feed = 100 no_stylesheets = True use_embedded_content = False + auto_cleanup = True + auto_cleanup_keep = '//div[@id="mediaWrapper"]' simultaneous_downloads = 1 delay = 4 max_connections = 1 @@ -31,14 +33,14 @@ class Soldiers(BasicNewsRecipe): , 'language' : language } - keep_only_tags = [dict(name='div', attrs={'id':['storyHeader','textArea']})] + #keep_only_tags = [dict(name='div', attrs={'id':['storyHeader','textArea']})] - remove_tags = [ - dict(name='div', attrs={'id':['addThis','comment','articleFooter']}) - ,dict(name=['object','link']) - ] + #remove_tags = [ + #dict(name='div', attrs={'id':['addThis','comment','articleFooter']}) + #,dict(name=['object','link']) + #] - feeds = [(u'Frontpage', u'http://www.army.mil/rss/feeds/soldiersfrontpage.xml' )] + feeds = [(u'Frontpage', u'http://www.army.mil/rss/2/' )] def get_cover_url(self):