mirror of
https://github.com/kovidgoyal/calibre.git
synced 2025-09-29 15:31:08 -04:00
190 lines
7.2 KiB
Python
190 lines
7.2 KiB
Python
#!/usr/bin/env python
|
|
# vim:fileencoding=utf-8
|
|
|
|
__license__ = 'GPL v3'
|
|
__copyright__ = '2010-2013, Darko Miletic <darko.miletic at gmail.com>'
|
|
'''
|
|
www.kommersant.ru
|
|
'''
|
|
|
|
from calibre.web.feeds.news import BasicNewsRecipe, classes
|
|
|
|
|
|
class Kommersant_ru(BasicNewsRecipe):
|
|
title = 'Kommersant'
|
|
__author__ = 'Darko Miletic (with fixes by bugmen00t)'
|
|
description = 'News from Russia'
|
|
publisher = 'Kommersant'
|
|
category = 'news, politics, Russia'
|
|
oldest_article = 7
|
|
max_articles_per_feed = 50
|
|
no_stylesheets = True
|
|
use_embedded_content = False
|
|
language = 'ru'
|
|
publication_type = 'newspaper'
|
|
cover_url = 'https://iv.kommersant.ru/ContentFlex/images/logo.png'
|
|
|
|
remove_tags_before = dict(name='header', attrs={'class': 'doc_header'})
|
|
|
|
remove_tags_after = dict(
|
|
name='div', attrs={'class': 'doc__text document_authors'}
|
|
)
|
|
|
|
remove_tags = [
|
|
dict(name='ul', attrs={'class': 'crumbs'}),
|
|
dict(name='div', attrs={'class': 'hide_desktop'}),
|
|
dict(name='div', attrs={'class': 'incut incut--right'}),
|
|
dict(name='div', attrs={'class': 'incut incut--left'}),
|
|
dict(name='div', attrs={'class': 'incut incut--center'}),
|
|
dict(name='div', attrs={'class': 'ba'}),
|
|
dict(name='div', attrs={'id': 'lenta'}),
|
|
dict(name='div', attrs={'class': 'layout basement_news__body'}),
|
|
dict(name='footer', attrs={'class': 'footer'}),
|
|
dict(name='section', attrs={'class': 'potd'}),
|
|
dict(name='footer', attrs={'class': 'doc_footer'}),
|
|
classes('ui-modal__hide'),
|
|
dict(name='div', attrs={'class': 'adv_interscroll hide_desktop'}),
|
|
dict(name=['button', 'source']),
|
|
]
|
|
|
|
def preprocess_html(self, soup):
|
|
for tag in soup.findAll(attrs={'data-lazyimage-src': True}):
|
|
tag['src'] = tag['data-lazyimage-src']
|
|
return soup
|
|
|
|
feeds = [
|
|
(
|
|
'\u0413\u043B\u0430\u0432\u043D\u043E\u0435',
|
|
'https://www.kommersant.ru/rss/main.xml'
|
|
),
|
|
(
|
|
'\u0413\u0430\u0437\u0435\u0442\u0430 "\u041A\u043E\u043C\u043C\u0435\u0440\u0441\u0430\u043D\u0442"',
|
|
'https://www.kommersant.ru/rss/daily.xml'
|
|
),
|
|
(
|
|
'\u041B\u0435\u043D\u0442\u0430 \u043D\u043E\u0432\u043E\u0441\u0442\u0435\u0439',
|
|
'https://www.kommersant.ru/RSS/news.xml'
|
|
),
|
|
(
|
|
'\u041C\u0430\u0442\u0435\u0440\u0438\u0430\u043B\u044B \u0441 \u0441\u0430\u0439\u0442\u0430',
|
|
'https://www.kommersant.ru/RSS/corp.xml'
|
|
),
|
|
(
|
|
'\u0420\u0430\u0434\u0438\u043E \u041A\u043E\u043C\u043C\u0435\u0440\u0441\u0430\u043D\u0442\u044A-FM',
|
|
'https://www.kommersant.ru/RSS/radio.xml'
|
|
),
|
|
(
|
|
'\u0422\u0435\u043C\u0430\u0442\u0438\u0447\u0435\u0441\u043A\u0438\u0435 \u043F\u0440\u0438\u043B\u043E\u0436\u0435\u043D\u0438\u044F',
|
|
'https://www.kommersant.ru/RSS/tema.xml'
|
|
),
|
|
(
|
|
'\u0416\u0443\u0440\u043D\u0430\u043B \u00AB\u041E\u0413\u041E\u041D\u0401\u041A\u00BB',
|
|
'https://www.kommersant.ru/RSS/ogoniok.xml'
|
|
),
|
|
(
|
|
'\u0416\u0443\u0440\u043D\u0430\u043B \u00AB\u041A\u043E\u043C\u043C\u0435\u0440\u0441\u0430\u043D\u0442\u044A WEEKEND\u00BB',
|
|
'https://www.kommersant.ru/RSS/weekend.xml'
|
|
),
|
|
(
|
|
'Журнал «Коммерсантъ АВТОПИЛОТ»',
|
|
'https://www.kommersant.ru/RSS/auto.xml'
|
|
),
|
|
(
|
|
'\u041F\u043E\u043B\u0438\u0442\u0438\u043A\u0430',
|
|
'https://www.kommersant.ru/rss/section-politics.xml'
|
|
),
|
|
(
|
|
'\u042D\u043A\u043E\u043D\u043E\u043C\u0438\u043A\u0430',
|
|
'https://www.kommersant.ru/RSS/section-economics.xml'
|
|
),
|
|
(
|
|
'\u0411\u0438\u0437\u043D\u0435\u0441',
|
|
'https://www.kommersant.ru/rss/section-business.xml'
|
|
),
|
|
(
|
|
'\u0412 \u043C\u0438\u0440\u0435',
|
|
'https://www.kommersant.ru/rss/section-world.xml'
|
|
),
|
|
(
|
|
'\u041F\u0440\u043E\u0438\u0441\u0448\u0435\u0441\u0442\u0432\u0438\u044F',
|
|
'https://www.kommersant.ru/rss/section-accidents.xml'
|
|
),
|
|
(
|
|
'\u041E\u0431\u0449\u0435\u0441\u0442\u0432\u043E',
|
|
'https://www.kommersant.ru/rss/section-society.xml'
|
|
),
|
|
(
|
|
'\u041A\u0443\u043B\u044C\u0442\u0443\u0440\u0430',
|
|
'https://www.kommersant.ru/rss/section-culture.xml'
|
|
),
|
|
(
|
|
'\u0421\u043F\u043E\u0440\u0442',
|
|
'https://www.kommersant.ru/rss/section-sport.xml'
|
|
), ('Hi-Tech', 'https://www.kommersant.ru/RSS/section-hitech.xml'),
|
|
(
|
|
'\u0410\u0432\u0442\u043E',
|
|
'https://www.kommersant.ru/RSS/Autopilot_on.xml'
|
|
),
|
|
(
|
|
'\u0421\u0442\u0438\u043B\u044C',
|
|
'https://www.kommersant.ru/RSS/section-style.xml'
|
|
),
|
|
(
|
|
'\u0421\u0430\u043D\u043A\u0442-\u041F\u0435\u0442\u0435\u0440\u0431\u0443\u0440\u0433',
|
|
'https://www.kommersant.ru/rss/regions/piter_all.xml'
|
|
),
|
|
(
|
|
'\u0412\u043E\u0440\u043E\u043D\u0435\u0436',
|
|
'https://www.kommersant.ru/rss/regions/vrn_all.xml'
|
|
),
|
|
(
|
|
'\u0415\u043A\u0430\u0442\u0435\u0440\u0438\u043D\u0431\u0443\u0440\u0433',
|
|
'https://www.kommersant.ru/rss/regions/ekaterinburg_all.xml'
|
|
),
|
|
(
|
|
'\u0418\u0436\u0435\u0432\u0441\u043A',
|
|
'https://www.kommersant.ru/rss/regions/izhevsk_all.xml'
|
|
),
|
|
(
|
|
'\u041A\u0430\u0437\u0430\u043D\u044C',
|
|
'https://www.kommersant.ru/rss/regions/kazan_all.xml'
|
|
),
|
|
(
|
|
'\u041A\u0440\u0430\u0441\u043D\u043E\u0434\u0430\u0440',
|
|
'https://www.kommersant.ru/rss/regions/krasnodar_all.xml'
|
|
),
|
|
(
|
|
'\u041A\u0440\u0430\u0441\u043D\u043E\u044F\u0440\u0441\u043A',
|
|
'https://www.kommersant.ru/rss/regions/krasnoyarsk_all.xml'
|
|
),
|
|
(
|
|
'\u041D\u0438\u0436\u043D\u0438\u0439 \u041D\u043E\u0432\u0433\u043E\u0440\u043E\u0434',
|
|
'https://www.kommersant.ru/rss/regions/nnov_all.xml'
|
|
),
|
|
(
|
|
'\u041D\u043E\u0432\u043E\u0441\u0438\u0431\u0438\u0440\u0441\u043A',
|
|
'https://www.kommersant.ru/rss/regions/novosibirsk_all.xml'
|
|
),
|
|
(
|
|
'\u041F\u0435\u0440\u043C\u044C',
|
|
'https://www.kommersant.ru/rss/regions/perm_all.xml'
|
|
),
|
|
(
|
|
'\u0420\u043E\u0441\u0442\u043E\u0432-\u043D\u0430-\u0414\u043E\u043D\u0443',
|
|
'https://www.kommersant.ru/rss/regions/rostov_all.xml'
|
|
),
|
|
(
|
|
'\u0421\u0430\u043C\u0430\u0440\u0430',
|
|
'https://www.kommersant.ru/rss/regions/samara_all.xml'
|
|
),
|
|
(
|
|
'\u0421\u0430\u0440\u0430\u0442\u043E\u0432',
|
|
'https://www.kommersant.ru/rss/regions/saratov_all.xml'
|
|
),
|
|
('\u0423\u0444\u0430', 'https://www.kommersant.ru/rss/regions/ufa_all.xml'),
|
|
(
|
|
'\u0427\u0435\u043B\u044F\u0431\u0438\u043D\u0441\u043A',
|
|
'https://www.kommersant.ru/rss/regions/chelyabinsk_all.xml'
|
|
)
|
|
]
|