diff --git a/recipes/reuters.recipe b/recipes/reuters.recipe index 313e6b8607..ceaec26c0c 100644 --- a/recipes/reuters.recipe +++ b/recipes/reuters.recipe @@ -1,5 +1,3 @@ -import re - from calibre.web.feeds.news import BasicNewsRecipe @@ -14,6 +12,8 @@ class Reuters(BasicNewsRecipe): max_articles_per_feed = 10 no_stylesheets = True remove_javascript = True + auto_cleanup = True + auto_cleanup_keep = '//div[@class="relatedPhoto landscape"]' extra_css = ''' body{font-family:arial,helvetica,sans;} @@ -23,29 +23,7 @@ class Reuters(BasicNewsRecipe): .timestamp{font-size:x-small; } ''' - keep_only_tags = [dict(name='div', attrs={'class':'column2 gridPanel grid8'})] - - remove_tags = [dict(name='div', attrs={'id':['recommendedArticles','relatedNews','relatedVideo','relatedFactboxes']}), - dict(name='p', attrs={'class':['relatedTopics']}), - dict(name='a', attrs={'id':['fullSizeLink']}), - dict(name='div', attrs={'class':['photoNav','relatedTopicButtons','articleComments','gridPanel grid8','footerHalf gridPanel grid1','gridPanel grid2','gridPanel grid3']}), - # Remove the Tweet, Share this, Email and Print links below article title too! - dict(name='div', attrs={'class':['columnRight']}), - ] - - preprocess_regexps = [ (re.compile(i[0], re.IGNORECASE | re.DOTALL), i[1]) for i in - [ - ##(r'.*?' , lambda match : ''), - (r'
.*?', lambda match : ''), - (r'.*?', lambda match : ''), - (r'.*?', lambda match : ''), - (r'.*?
', lambda match : ''), - (r'

Share:

.*?', lambda match : ''), - (r'
.*?
', lambda match : '
'), - ]] - - - feeds = [ ('Top Stories', 'http://feeds.reuters.com/reuters/topNews?format=xml'), + feeds = [('Top Stories', 'http://feeds.reuters.com/reuters/topNews?format=xml'), ('US News', 'http://feeds.reuters.com/reuters/domesticNews?format=xml'), ('World News', 'http://feeds.reuters.com/reuters/worldNews?format=xml'), ('Politics News', 'http://feeds.reuters.com/reuters/politicsNews?format=xml'),