From 79396b00f42fe5172cfc03f1fd16f6d356abb0a0 Mon Sep 17 00:00:00 2001 From: Kovid Goyal Date: Fri, 19 Mar 2010 07:09:15 +0530 Subject: [PATCH] Improve daily telegraph --- resources/recipes/daily_telegraph.recipe | 87 +++++++++++++++--------- 1 file changed, 55 insertions(+), 32 deletions(-) diff --git a/resources/recipes/daily_telegraph.recipe b/resources/recipes/daily_telegraph.recipe index 61054e1db0..5e1a2f7bfb 100644 --- a/resources/recipes/daily_telegraph.recipe +++ b/resources/recipes/daily_telegraph.recipe @@ -6,43 +6,66 @@ __docformat__ = 'restructuredtext en' ''' http://www.news.com.au/dailytelegraph/ ''' -import re from calibre.web.feeds.news import BasicNewsRecipe class DailyTelegraph(BasicNewsRecipe): title = u'Daily Telegraph' - __author__ = u'AprilHare' + __author__ = u'Adrian G.' language = 'en_AU' - description = u'News from down under' - oldest_article = 2 - max_articles_per_feed = 10 - remove_tags_before = dict(name='div', attrs={'class':'article-title'}) - remove_tags = [dict(attrs={'class':['article-source', 'article-tools']})] - remove_tags_after = dict(attrs={'class':re.compile('share-article')}) - + description = u'Daily Telegraph News' + oldest_article = 5 + max_articles_per_feed = 100 + no_stylesheets = True + use_embedded_content = False + 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 = [ - (u'Top Stories', u'http://feeds.news.com.au/public/rss/2.0/dtele_top_stories_253.xml'), - (u'National News', u'http://feeds.news.com.au/public/rss/2.0/dtele_national_news_202.xml'), - (u'World News', u'http://feeds.news.com.au/public/rss/2.0/dtele_world_news_204.xml'), - (u'NSW and ACT', u'http://feeds.news.com.au/public/rss/2.0/dtele_nswact_225.xml'), - (u'Arts', u'http://feeds.news.com.au/public/rss/2.0/dtele_art_444.xml'), - (u'Business News', u'http://feeds.news.com.au/public/rss/2.0/dtele_business_226.xml'), - (u'Entertainment News', u'http://feeds.news.com.au/public/rss/2.0/dtele_entertainment_news_201.xml'), - (u'Lifestyle News', u'http://feeds.news.com.au/public/rss/2.0/dtele_lifestyle_227.xml'), - (u'Music', u'http://feeds.news.com.au/public/rss/2.0/dtele_music_441.xml'), - (u'Property Confidential', u'http://feeds.news.com.au/public/rss/2.0/dtele_property_confidential_463.xml'), - (u'Property - Your Space', u'http://feeds.news.com.au/public/rss/2.0/dtele_property_yourspace_462.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 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') + (u'Top Stories', u'http://feeds.news.com.au/public/rss/2.0/dtele_top_stories_253.xml'), + (u'National News', u'http://feeds.news.com.au/public/rss/2.0/dtele_national_news_202.xml'), + (u'World News', u'http://feeds.news.com.au/public/rss/2.0/dtele_world_news_204.xml'), + (u'NSW and ACT', u'http://feeds.news.com.au/public/rss/2.0/dtele_nswact_225.xml'), + (u'Arts', u'http://feeds.news.com.au/public/rss/2.0/dtele_art_444.xml'), + (u'Business News', u'http://feeds.news.com.au/public/rss/2.0/dtele_business_226.xml'), + (u'Entertainment News', u'http://feeds.news.com.au/public/rss/2.0/dtele_entertainment_news_201.xml'), + (u'Lifestyle News', u'http://feeds.news.com.au/public/rss/2.0/dtele_lifestyle_227.xml'), + (u'Music', u'http://feeds.news.com.au/public/rss/2.0/dtele_music_441.xml'), + (u'Property Confidential', u'http://feeds.news.com.au/public/rss/2.0/dtele_property_confidential_463.xml'), + (u'Property - Your Space', u'http://feeds.news.com.au/public/rss/2.0/dtele_property_yourspace_462.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 Galleries', u'http://feeds.news.com.au/public/rss/2.0/dtele_confidential_galleries_483.xml'), ] + +