Improve daily telegraph

This commit is contained in:
Kovid Goyal 2010-03-19 07:09:15 +05:30
parent d82ad7af94
commit 79396b00f4

View File

@ -6,20 +6,50 @@ __docformat__ = 'restructuredtext en'
''' '''
http://www.news.com.au/dailytelegraph/ http://www.news.com.au/dailytelegraph/
''' '''
import re
from calibre.web.feeds.news import BasicNewsRecipe from calibre.web.feeds.news import BasicNewsRecipe
class DailyTelegraph(BasicNewsRecipe): class DailyTelegraph(BasicNewsRecipe):
title = u'Daily Telegraph' title = u'Daily Telegraph'
__author__ = u'AprilHare' __author__ = u'Adrian G.'
language = 'en_AU' language = 'en_AU'
description = u'News from down under' description = u'Daily Telegraph News'
oldest_article = 2 oldest_article = 5
max_articles_per_feed = 10 max_articles_per_feed = 100
remove_tags_before = dict(name='div', attrs={'class':'article-title'}) no_stylesheets = True
remove_tags = [dict(attrs={'class':['article-source', 'article-tools']})] use_embedded_content = False
remove_tags_after = dict(attrs={'class':re.compile('share-article')}) no_javascript = True
timefmt = ' [%A, %d %B, %Y]'
encoding = 'utf-8'
keep_only_tags = [dict(name='div', attrs ={'id':'story'})]
extra_css = '''
h1{font-family:Georgia,"Times New Roman",Times,serif; font-weight:bold; font-size:large;}
.cT-storyDetails{font-family:Arial,Helvetica,sans-serif; color:#666666;font-size:x-small;}
.articleBody{font-family:Arial,Helvetica,sans-serif; color:black;font-size:small;}
.cT-imageLandscape{font-family:Arial,Helvetica,sans-serif; color:#333333 ;font-size:x-small;}
.source{font-family:Arial,Helvetica,sans-serif; color:#333333 ;font-size:xx-small;}
#content{font-family:Arial,Helvetica,sans-serif;font-size:x-small;}
.pageprint{font-family:Arial,Helvetica,sans-serif;font-size:small;}
#bylineDetails{font-family:Arial,Helvetica,sans-serif; color:#666666;font-size:x-small;}
.featurePic-wide{font-family:Arial,Helvetica,sans-serif;font-size:x-small;}
#idfeaturepic{font-family:Arial,Helvetica,sans-serif;font-size:x-small;}
h3{font-family:Georgia,"Times New Roman",Times,serif; font-size:small;}
h2{font-family:Georgia,"Times New Roman",Times,serif; font-size:small;}
h4{font-family:Georgia,"Times New Roman",Times,serif; font-size:small;}
h5{font-family:Georgia,"Times New Roman",Times,serif; font-size:small;}
body{font-family:Arial,Helvetica,sans-serif; font-size:x-small;}
'''
remove_tags = [
dict(name='div', attrs ={'id':['comments','story-related-coverage']}),
dict(name='div', attrs ={'class':['story-header-tools','story-footer','story-extras','story-related']}),
dict(name='div', attrs ={'class':['promo-image','story-extras story-extras-2']}),
dict(name='div', attrs ={'class':['assistive sidebar-jump']})
]
feeds = [ feeds = [
(u'Top Stories', u'http://feeds.news.com.au/public/rss/2.0/dtele_top_stories_253.xml'), (u'Top Stories', u'http://feeds.news.com.au/public/rss/2.0/dtele_top_stories_253.xml'),
@ -36,13 +66,6 @@ class DailyTelegraph(BasicNewsRecipe):
(u'Confidential News', u'http://feeds.news.com.au/public/rss/2.0/dtele_entertainment_confidential_252.xml'), (u'Confidential News', u'http://feeds.news.com.au/public/rss/2.0/dtele_entertainment_confidential_252.xml'),
(u'Confidential Biographies', u'http://feeds.news.com.au/public/rss/2.0/dtele_confidential_biographies_491.xml'), (u'Confidential Biographies', u'http://feeds.news.com.au/public/rss/2.0/dtele_confidential_biographies_491.xml'),
(u'Confidential Galleries', u'http://feeds.news.com.au/public/rss/2.0/dtele_confidential_galleries_483.xml'), (u'Confidential Galleries', u'http://feeds.news.com.au/public/rss/2.0/dtele_confidential_galleries_483.xml'),
(u'Confidential In-depth', u'http://feeds.news.com.au/public/rss/2.0/dtele_confidential_indepth_490.xml'),
(u'Confidential ShowBuzz', u'http://feeds.news.com.au/public/rss/2.0/dtele_confidential_showbuzz_485.xml'),
(u'Sport', u'http://feeds.news.com.au/public/rss/2.0/dtele_sport_203.xml'),
(u'AFL', u'http://feeds.news.com.au/public/rss/2.0/dtele_sports_afl_341.xml'),
(u'Cricket', u'http://feeds.news.com.au/public/rss/2.0/dtele_sports_cricket_343.xml'),
(u'Horse Racing', u'http://feeds.news.com.au/public/rss/2.0/dtele_sports_horseracing_686.xml'),
(u'NRL', u'http://feeds.news.com.au/public/rss/2.0/dtele_sports_nrl_345.xml'),
(u'Rugby Union', u'http://feeds.news.com.au/public/rss/2.0/dtele_sports_rugby_union_342.xml'),
(u'Soccer', u'http://feeds.news.com.au/public/rss/2.0/dtele_sports_soccer_344.xml')
] ]