calibre/recipes/toi.recipe

46 lines
1.4 KiB
Plaintext

from calibre.web.feeds.news import BasicNewsRecipe
class TimesOfIndia(BasicNewsRecipe):
title = u'Times of India'
language = 'en_IN'
__author__ = 'Kovid Goyal'
oldest_article = 1 #days
max_articles_per_feed = 25
no_stylesheets = True
keep_only_tags = [dict(attrs={'class':'maintable12'})]
remove_tags = [
dict(style=lambda x: x and 'float' in x),
dict(attrs={'class':'prvnxtbg'}),
]
feeds = [
('Top Stories',
'http://timesofindia.indiatimes.com/rssfeedstopstories.cms'),
('India',
'http://timesofindia.indiatimes.com/rssfeeds/-2128936835.cms'),
('World',
'http://timesofindia.indiatimes.com/rssfeeds/296589292.cms'),
('Mumbai',
'http://timesofindia.indiatimes.com/rssfeeds/-2128838597.cms'),
('Entertainment',
'http://timesofindia.indiatimes.com/rssfeeds/1081479906.cms'),
('Cricket',
'http://timesofindia.indiatimes.com/rssfeeds/4719161.cms'),
('Sunday TOI',
'http://timesofindia.indiatimes.com/rssfeeds/1945062111.cms'),
('Life and Style',
'http://timesofindia.indiatimes.com/rssfeeds/2886704.cms'),
('Business',
'http://timesofindia.indiatimes.com/rssfeeds/1898055.cms'),
('Mad Mad World',
'http://timesofindia.indiatimes.com/rssfeeds/2178430.cms'),
('Most Read',
'http://timesofindia.indiatimes.com/rssfeedmostread.cms')
]
def print_version(self, url):
return url + '?prtpage=1'
def preprocess_html(self, soup):
return soup