mirror of
https://github.com/kovidgoyal/calibre.git
synced 2025-06-23 15:30:45 -04:00
61 lines
2.1 KiB
Plaintext
61 lines
2.1 KiB
Plaintext
from calibre.web.feeds.news import BasicNewsRecipe
|
|
|
|
|
|
class AdvancedUserRecipe1271637235(BasicNewsRecipe):
|
|
|
|
title = u'Thairath'
|
|
__author__ = 'Anat R.'
|
|
language = 'th'
|
|
|
|
oldest_article = 7
|
|
|
|
max_articles_per_feed = 100
|
|
no_stylesheets = True
|
|
|
|
remove_javascript = True
|
|
|
|
use_embedded_content = False
|
|
feeds = [
|
|
(u'News', u'http://www.thairath.co.th/rss/news.xml'),
|
|
(u'Politics', u'http://www.thairath.co.th/rss/pol.xml'),
|
|
(u'Economy', u'http://www.thairath.co.th/rss/eco.xml'),
|
|
(u'International', u'http://www.thairath.co.th/rss/oversea.xml'),
|
|
(u'Sports', u'http://www.thairath.co.th/rss/sport.xml'),
|
|
(u'Life', u'http://www.thairath.co.th/rss/life.xml'),
|
|
(u'Education', u'http://www.thairath.co.th/rss/edu.xml'),
|
|
(u'Tech', u'http://www.thairath.co..th/rss/tech.xml'),
|
|
(u'Entertainment', u'http://www.thairath.co.th/rss/ent.xml')]
|
|
keep_only_tags = []
|
|
|
|
keep_only_tags.append(dict(name='h1', attrs={'id': 'title'}))
|
|
|
|
keep_only_tags.append(dict(name='ul', attrs={'class':
|
|
'detail-info'}))
|
|
|
|
keep_only_tags.append(dict(name='img', attrs={'class':
|
|
'detail-image'}))
|
|
|
|
keep_only_tags.append(dict(name='div', attrs={'class':
|
|
'entry'}))
|
|
remove_tags = []
|
|
remove_tags.append(dict(name='div', attrs={'id':
|
|
'menu-holder'}))
|
|
|
|
remove_tags.append(dict(name='div', attrs={'class':
|
|
'addthis_toolbox addthis_default_style'}))
|
|
|
|
remove_tags.append(dict(name='div', attrs={'class': 'box top-item'}))
|
|
|
|
remove_tags.append(
|
|
dict(name='div', attrs={'class': 'column-200 column-margin-430'}))
|
|
|
|
remove_tags.append(dict(name='div', attrs={'id':
|
|
'detail-related'}))
|
|
|
|
remove_tags.append(dict(name='div', attrs={'id': 'related'}))
|
|
|
|
remove_tags.append(dict(name='id', attrs={'class': 'footer'}))
|
|
|
|
remove_tags.append(
|
|
dict(name="ul", attrs={'id': 'banner-highlights-images'}))
|