calibre/resources/recipes/jpost.recipe
2010-02-08 21:17:35 -07:00

29 lines
1.2 KiB
Plaintext

from calibre.web.feeds.news import BasicNewsRecipe
class JerusalemPost(BasicNewsRecipe):
title = 'Jerusalem Post'
description = 'News from Israel and the Middle East'
use_embedded_content = False
language = 'en'
__author__ = 'Kovid Goyal'
max_articles_per_feed = 10
no_stylesheets = True
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 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