mirror of
https://github.com/kovidgoyal/calibre.git
synced 2025-09-29 15:31:08 -04:00
61 lines
2.7 KiB
Python
61 lines
2.7 KiB
Python
__license__ = 'GPL v3'
|
|
__copyright__ = '2008-2014, Karthik <hashkendistro@gmail.com>, Darko Miletic <darko.miletic at gmail.com>'
|
|
'''
|
|
economictimes.indiatimes.com
|
|
'''
|
|
|
|
|
|
from calibre.web.feeds.news import BasicNewsRecipe, classes
|
|
|
|
|
|
class TheEconomicTimes(BasicNewsRecipe):
|
|
title = 'The Economic Times India'
|
|
__author__ = 'Karthik, Darko Miletic'
|
|
description = 'Financial news from India'
|
|
publisher = 'economictimes.indiatimes.com'
|
|
category = 'news, finances, politics, India'
|
|
oldest_article = 1
|
|
max_articles_per_feed = 50
|
|
no_stylesheets = True
|
|
use_embedded_content = False
|
|
simultaneous_downloads = 1
|
|
encoding = 'utf-8'
|
|
language = 'en_IN'
|
|
remove_attributes = ['style', 'height', 'width']
|
|
publication_type = 'newspaper'
|
|
masthead_url = 'http://economictimes.indiatimes.com/photo/2676871.cms'
|
|
ignore_duplicate_articles = {'title', 'url'}
|
|
extra_css = '.summary {font-weight:normal; font-size:normal; }'
|
|
|
|
keep_only_tags = [
|
|
dict(name='h1'),
|
|
classes('artByline artSyn artImg artText publisher publish_on slideshowPackage'),
|
|
]
|
|
remove_tags = [
|
|
classes(
|
|
'story_title storyCollection shareBar sr_widget_free jsSrWidgetFree srwidgetfree_3'
|
|
' sr_paid jsSrWidgetPaid ar_wrp arwd_ld_chk adBox custom_ad mgid orn_free_r bold'
|
|
),
|
|
]
|
|
|
|
feeds = [(u'Top Stories', u'http://economictimes.indiatimes.com/rssfeedstopstories.cms'),
|
|
(u'News', u'http://economictimes.indiatimes.com/News/rssfeeds/1715249553.cms'),
|
|
(u'Market', u'http://economictimes.indiatimes.com/Markets/markets/rssfeeds/1977021501.cms'),
|
|
(u'Personal Finance',
|
|
u'http://economictimes.indiatimes.com/rssfeeds/837555174.cms'),
|
|
(u'Infotech', u'http://economictimes.indiatimes.com/Infotech/rssfeeds/13357270.cms'),
|
|
(u'Job', u'http://economictimes.indiatimes.com/Infotech/rssfeeds/107115.cms'),
|
|
(u'Opinion', u'http://economictimes.indiatimes.com/opinion/opinionshome/rssfeeds/897228639.cms'),
|
|
(u'Features', u'http://economictimes.indiatimes.com/Features/etfeatures/rssfeeds/1466318837.cms'),
|
|
(u'Environment',
|
|
u'http://economictimes.indiatimes.com/rssfeeds/2647163.cms'),
|
|
(u'NRI', u'http://economictimes.indiatimes.com/rssfeeds/7771250.cms')
|
|
]
|
|
|
|
def preprocess_html(self, soup):
|
|
for image in soup.findAll('img', attrs={'src': True}):
|
|
image['src'] = image['src'].replace("width-300", "width-640")
|
|
for img in soup.findAll('img', attrs={'data-original': True}):
|
|
img['src'] = img['data-original'].replace('photo', 'thumb').replace('quality-100', 'quality-100,width-600,resizemode-4')
|
|
return soup
|