Fix ZDNet

This commit is contained in:
Kovid Goyal 2012-08-07 15:29:27 +05:30
parent ca750b3394
commit b25cc34477

View File

@ -12,7 +12,7 @@ class cdnet(BasicNewsRecipe):
title = 'zdnet' title = 'zdnet'
description = 'zdnet security' description = 'zdnet security'
__author__ = 'Oliver Niesner' __author__ = 'Oliver Niesner, Krittika Goyal'
language = 'en' language = 'en'
use_embedded_content = False use_embedded_content = False
@ -20,41 +20,42 @@ class cdnet(BasicNewsRecipe):
max_articles_per_feed = 40 max_articles_per_feed = 40
no_stylesheets = True no_stylesheets = True
encoding = 'latin1' encoding = 'latin1'
auto_cleanup = True
remove_tags = [dict(id='eyebrows'), #remove_tags = [dict(id='eyebrows'),
dict(id='header'), #dict(id='header'),
dict(id='search'), #dict(id='search'),
dict(id='nav'), #dict(id='nav'),
dict(id='blog-author-info'), #dict(id='blog-author-info'),
dict(id='post-tags'), #dict(id='post-tags'),
dict(id='bio-naraine'), #dict(id='bio-naraine'),
dict(id='bio-kennedy'), #dict(id='bio-kennedy'),
dict(id='author-short-disclosure-kennedy'), #dict(id='author-short-disclosure-kennedy'),
dict(id=''), #dict(id=''),
dict(name='div', attrs={'class':'banner'}), #dict(name='div', attrs={'class':'banner'}),
dict(name='div', attrs={'class':'int'}), #dict(name='div', attrs={'class':'int'}),
dict(name='div', attrs={'class':'talkback clear space-2'}), #dict(name='div', attrs={'class':'talkback clear space-2'}),
dict(name='div', attrs={'class':'content-1 clear'}), #dict(name='div', attrs={'class':'content-1 clear'}),
dict(name='div', attrs={'class':'space-2'}), #dict(name='div', attrs={'class':'space-2'}),
dict(name='div', attrs={'class':'space-3'}), #dict(name='div', attrs={'class':'space-3'}),
dict(name='div', attrs={'class':'thumb-2 left'}), #dict(name='div', attrs={'class':'thumb-2 left'}),
dict(name='div', attrs={'class':'hotspot'}), #dict(name='div', attrs={'class':'hotspot'}),
dict(name='div', attrs={'class':'hed hed-1 space-1'}), #dict(name='div', attrs={'class':'hed hed-1 space-1'}),
dict(name='div', attrs={'class':'view-1 clear content-3 space-2'}), #dict(name='div', attrs={'class':'view-1 clear content-3 space-2'}),
dict(name='div', attrs={'class':'hed hed-1 space-1'}), #dict(name='div', attrs={'class':'hed hed-1 space-1'}),
dict(name='div', attrs={'class':'hed hed-1'}), #dict(name='div', attrs={'class':'hed hed-1'}),
dict(name='div', attrs={'class':'post-header'}), #dict(name='div', attrs={'class':'post-header'}),
dict(name='div', attrs={'class':'lvl-nav clear'}), #dict(name='div', attrs={'class':'lvl-nav clear'}),
dict(name='div', attrs={'class':'t-share-overlay overlay-pop contain-overlay-4'}), #dict(name='div', attrs={'class':'t-share-overlay overlay-pop contain-overlay-4'}),
dict(name='p', attrs={'class':'tags'}), #dict(name='p', attrs={'class':'tags'}),
dict(name='span', attrs={'class':'follow'}), #dict(name='span', attrs={'class':'follow'}),
dict(name='span', attrs={'class':'int'}), #dict(name='span', attrs={'class':'int'}),
dict(name='h4', attrs={'class':'h s-4'}), #dict(name='h4', attrs={'class':'h s-4'}),
dict(name='a', attrs={'href':'http://www.twitter.com/ryanaraine'}), #dict(name='a', attrs={'href':'http://www.twitter.com/ryanaraine'}),
dict(name='div', attrs={'class':'special1'})] #dict(name='div', attrs={'class':'special1'})]
remove_tags_after = [dict(name='div', attrs={'class':'clear'})] #remove_tags_after = [dict(name='div', attrs={'class':'clear'})]
feeds = [ ('zdnet', 'http://feeds.feedburner.com/zdnet/security') ] feeds = [ ('zdnet', 'http://feeds.feedburner.com/zdnet/security') ]
@ -63,6 +64,3 @@ class cdnet(BasicNewsRecipe):
for item in soup.findAll(style=True): for item in soup.findAll(style=True):
del item['style'] del item['style']
return soup return soup