diff --git a/recipes/jakarta_globe.recipe b/recipes/jakarta_globe.recipe index f6f1715268..d33bd81fbe 100644 --- a/recipes/jakarta_globe.recipe +++ b/recipes/jakarta_globe.recipe @@ -6,29 +6,23 @@ class JakartaGlobe(BasicNewsRecipe): max_articles_per_feed = 100 feeds = [ - (u'News', u'http://www.thejakartaglobe.com/pages/getrss/getrss-news.php'), - (u'Business', u'http://www.thejakartaglobe.com/pages/getrss/getrss-business.php'), - (u'Technology', u'http://www.thejakartaglobe.com/pages/getrss/getrss-tech.php'), - (u'My Jakarta', u'http://www.thejakartaglobe.com/pages/getrss/getrss-myjakarta.php'), - (u'International', u'http://www.thejakartaglobe.com/pages/getrss/getrss-international.php'), - (u'Life and Times', u'http://www.thejakartaglobe.com/pages/getrss/getrss-lifeandtimes.php'), - ] + (u'News', u'http://www.thejakartaglobe.com/news/feed/'), + (u'Business', u'http://www.thejakartaglobe.com/business/feed/'), + (u'Opinion', u'http://www.thejakartaglobe.com/opinion/feed/'), + (u'Count me in', u'http://www.thejakartaglobe.com/count-me-in/feed/'), + (u'International', u'http://www.thejakartaglobe.com/international/feed/'), + (u'Sports', u'http://www.thejakartaglobe.com/sports/feed/'), + ] __author__ = 'rty' pubisher = 'JakartaGlobe.com' description = 'JakartaGlobe, Indonesia, Newspaper' category = 'News, Indonesia' - remove_javascript = True use_embedded_content = False no_stylesheets = True + auto_cleanup = True language = 'en_ID' encoding = 'utf-8' conversion_options = {'linearize_tables':True} masthead_url = 'http://www.thejakartaglobe.com/pages/2010/images/jak-globe-logo.jpg' - keep_only_tags = [ - dict(name='div', attrs={'class':'story'}), - dict(name='span', attrs={'class':'headline'}), - dict(name='div', attrs={'class':'story'}), - dict(name='p', attrs={'id':'bodytext'}) - ] diff --git a/recipes/jakarta_post.recipe b/recipes/jakarta_post.recipe index d8d609469d..000c91628b 100644 --- a/recipes/jakarta_post.recipe +++ b/recipes/jakarta_post.recipe @@ -27,12 +27,11 @@ class JakartaPost(BasicNewsRecipe): use_embedded_content = False no_javascript = True remove_empty_feeds = True + auto_cleanup = True timefmt = ' [%A, %d %B, %Y]' encoding = 'utf-8' - keep_only_tags = [dict(name='div', attrs ={'id':'news-main'})] - extra_css = ''' h1{font-family:Georgia,"Times New Roman",Times,serif; font-weight:bold; font-size:large;} .cT-storyDetails{font-family:Arial,Helvetica,sans-serif; color:#666666;font-size:x-small;} @@ -51,10 +50,6 @@ class JakartaPost(BasicNewsRecipe): body{font-family:Arial,Helvetica,sans-serif; font-size:x-small;} ''' - remove_tags = [ - dict(name='div', attrs ={'class':['text-size']}), - ] - feeds = [ (u'Breaking News', u'http://www.thejakartapost.com/breaking/feed'),