Update Live Mint

This commit is contained in:
Kovid Goyal 2022-02-17 22:41:31 +05:30
parent d2baa161c7
commit bbfa51852c
No known key found for this signature in database
GPG Key ID: 06BC317B515ACE7C

View File

@ -1,4 +1,6 @@
from calibre.web.feeds.news import BasicNewsRecipe #!/usr/bin/env python
from calibre.web.feeds.news import BasicNewsRecipe, classes
class LiveMint(BasicNewsRecipe): class LiveMint(BasicNewsRecipe):
@ -7,14 +9,28 @@ class LiveMint(BasicNewsRecipe):
__author__ = 'Krittika Goyal' __author__ = 'Krittika Goyal'
oldest_article = 1 # days oldest_article = 1 # days
max_articles_per_feed = 50 max_articles_per_feed = 50
encoding = 'utf-8'
use_embedded_content = False use_embedded_content = False
no_stylesheets = True no_stylesheets = True
auto_cleanup = True remove_attributes = ['style', 'height', 'width']
keep_only_tags = [
dict(name='h1'),
dict(name='picture'),
dict(name='figcaption'),
classes('articleInfo FirstEle summary highlights paywall'),
]
remove_tags = [
classes(
'trendingSimilarHeight moreNews mobAppDownload label msgError msgOk'
)
]
feeds = [ feeds = [
('Companies', 'https://www.livemint.com/rss/companies'), ('Companies', 'https://www.livemint.com/rss/companies'),
('Opinion', 'https://www.livemint.com/rss/opinion'), ('Opinion', 'https://www.livemint.com/rss/opinion'),
('Money', 'https://www.livemint.com/rss/money'), ('Money', 'https://www.livemint.com/rss/money'),
('Economy', 'https://www.livemint.com/rss/economy/'),
('Politics', 'https://www.livemint.com/rss/politics'), ('Politics', 'https://www.livemint.com/rss/politics'),
('Science', 'https://www.livemint.com/rss/science'), ('Science', 'https://www.livemint.com/rss/science'),
('Industry', 'https://www.livemint.com/rss/industry'), ('Industry', 'https://www.livemint.com/rss/industry'),
@ -30,3 +46,8 @@ class LiveMint(BasicNewsRecipe):
('Budget', 'https://www.livemint.com/rss/budget'), ('Budget', 'https://www.livemint.com/rss/budget'),
('Elections', 'https://www.livemint.com/rss/elections'), ('Elections', 'https://www.livemint.com/rss/elections'),
] ]
def preprocess_html(self, soup):
for img in soup.findAll('img', attrs={'data-src': True}):
img['src'] = img['data-src']
return soup