Bangkok Biz News and Matichon by Anat Ruangrassamee

This commit is contained in:
Kovid Goyal 2010-11-26 07:50:25 -07:00
parent 64ede2a0ea
commit 69f89114c2
2 changed files with 47 additions and 0 deletions

View File

@ -0,0 +1,25 @@
from calibre.web.feeds.news import BasicNewsRecipe
class AdvancedUserRecipe1290689337(BasicNewsRecipe):
__author__ = 'Anat R.'
language = 'th'
title = u'Bangkok Biz News'
oldest_article = 7
max_articles_per_feed = 100
no_stylesheets = True
remove_javascript = True
use_embedded_content = False
feeds = [(u'Headlines',
u'http://www.bangkokbiznews.com/home/services/rss/home.xml'),
(u'Politics', u'http://www.bangkokbiznews.com/home/services/rss/politics.xml'),
(u'Business', u'http://www.bangkokbiznews.com/home/services/rss/business.xml'),
(u'Finance', u' http://www.bangkokbiznews.com/home/services/rss/finance.xml'),
(u'Technology', u' http://www.bangkokbiznews.com/home/services/rss/it.xml')]
remove_tags_before = dict(name='div', attrs={'class':'box-Detailcontent'})
remove_tags_after = dict(name='p', attrs={'class':'allTags'})
remove_tags = []
remove_tags.append(dict(name = 'div', attrs = {'id': 'content-tools'}))
remove_tags.append(dict(name = 'p', attrs = {'class':'allTags'}))
remove_tags.append(dict(name = 'div', attrs = {'id':'morePic'}))
remove_tags.append(dict(name = 'ul', attrs = {'class':'tabs-nav'}))

View File

@ -0,0 +1,22 @@
from calibre.web.feeds.news import BasicNewsRecipe
class AdvancedUserRecipe1290412756(BasicNewsRecipe):
__author__ = 'Anat R.'
title = u'Matichon'
oldest_article = 7
language = 'th'
max_articles_per_feed = 100
no_stylesheets = True
remove_javascript = True
use_embedded_content = False
feeds = [(u'News', u'http://www.matichon.co.th/rss/news_article.xml'),
(u'Columns', u'http://www.matichon.co.th/rss/news_columns.xml'),
(u'Politics', u'http://www.matichon.co.th/rss/news_politic.xml'),
(u'Business', u'http://www.matichon.co.th/rss/news_business.xml'),
(u'World', u'http://www.matichon.co.th/rss/news_world.xml'),
(u'Sports', u'http://www.matichon.co.th/rss/news_sport.xml'),
(u'Entertainment', u'http://www.matichon.co.th/rss/news_entertainment.xml')]
keep_only_tags = []
keep_only_tags.append(dict(name = 'h3', attrs = {'class' : 'read-h'}))
keep_only_tags.append(dict(name = 'p', attrs = {'class' : 'read-time'}))
keep_only_tags.append(dict(name = 'div', attrs = {'class' : 'news-content'}))