Fix #4768 (Jerusalem Post Failed Feed)

This commit is contained in:
Kovid Goyal 2010-02-08 21:17:35 -07:00
parent 3b28deb26e
commit 17c0f31808

View File

@ -10,22 +10,19 @@ class JerusalemPost(BasicNewsRecipe):
__author__ = 'Kovid Goyal'
max_articles_per_feed = 10
no_stylesheets = True
remove_tags_before = {'class':'byline'}
remove_tags = [
{'class':['artAdBlock clearboth', 'tbartop', 'divdot_vrttbox',
'slideshow']},
dict(id=['artFontButtons', 'artRelatedBlock']),
]
remove_tags_after = {'id':'artTxtBlock'}
remove_tags_before = {'class':'jp-grid-content'}
remove_tags_after = {'id':'body_val'}
feeds = [ ('Front Page', 'http://www.jpost.com/servlet/Satellite?pagename=JPost/Page/RSS&cid=1123495333346'),
('Israel News', 'http://www.jpost.com/servlet/Satellite?pagename=JPost/Page/RSS&cid=1178443463156'),
('Middle East News', 'http://www.jpost.com/servlet/Satellite?pagename=JPost/Page/RSS&cid=1123495333498'),
('International News', 'http://www.jpost.com/servlet/Satellite?pagename=JPost/Page/RSS&cid=1178443463144'),
('Editorials', 'http://www.jpost.com/servlet/Satellite?pagename=JPost/Page/RSS&cid=1123495333211'),
]
def postprocess_html(self, soup, first):
for tag in soup.findAll(name=['table', 'tr', 'td']):
tag.name = 'div'
return soup
def preprocess_html(self, soup):
for x in soup.findAll(name=['form', 'input']):
x.name = 'div'
for x in soup.findAll('body', style=True):
del x['style']
return soup