Fix #862381 (SMH Recipe downloads extraneous "Video feedback" form (fix attached))

This commit is contained in:
Kovid Goyal 2011-09-29 08:36:30 -06:00
parent 893d7d1f03
commit a862073a00

View File

@ -22,10 +22,10 @@ class Smh_au(BasicNewsRecipe):
remove_empty_feeds = True
masthead_url = 'http://images.smh.com.au/2010/02/02/1087188/smh-620.jpg'
publication_type = 'newspaper'
extra_css = """
h1{font-family: Georgia,"Times New Roman",Times,serif }
body{font-family: Arial,Helvetica,sans-serif}
.cT-imageLandscape,.cT-imagePortrait{font-size: x-small}
extra_css = """
h1{font-family: Georgia,"Times New Roman",Times,serif }
body{font-family: Arial,Helvetica,sans-serif}
.cT-imageLandscape,.cT-imagePortrait{font-size: x-small}
"""
conversion_options = {
@ -35,16 +35,16 @@ class Smh_au(BasicNewsRecipe):
, 'language' : language
}
remove_tags = [
dict(name='div', attrs={'id':['googleAds','moreGoogleAds','comments']})
,dict(name='div', attrs={'class':'cT-imageMultimedia'})
,dict(name=['object','embed','iframe'])
]
remove_tags_after = [dict(name='div',attrs={'class':'articleBody'})]
keep_only_tags = [dict(name='div',attrs={'id':'content'})]
remove_tags = [
dict(attrs={'class':'hidden'}),
dict(name=['link','meta','base','embed','object','iframe'])
remove_tags = [
dict(name='div',
attrs={'id':['googleAds','moreGoogleAds','comments',
'video-player-content']}),
dict(name='div', attrs={'class':'cT-imageMultimedia'}),
dict(name=['object','embed','iframe']),
dict(attrs={'class':'hidden'}),
dict(name=['link','meta','base','embed','object','iframe'])
]
remove_attributes = ['width','height','lang']
@ -84,4 +84,4 @@ class Smh_au(BasicNewsRecipe):
if not item.has_key('alt'):
item['alt'] = 'image'
return soup