From c9bcd9a35bfa1d15fd7f5110a20008e2edd1bdfe Mon Sep 17 00:00:00 2001 From: Kovid Goyal Date: Mon, 3 May 2010 08:31:59 -0600 Subject: [PATCH] Il Messaggero and ADN Kronos by Gabriele Marini --- resources/recipes/adnkronos.recipe | 59 ++++++++++++++++++++++++++ resources/recipes/il_messaggero.recipe | 56 ++++++++++++++++++++++++ 2 files changed, 115 insertions(+) create mode 100644 resources/recipes/adnkronos.recipe create mode 100644 resources/recipes/il_messaggero.recipe diff --git a/resources/recipes/adnkronos.recipe b/resources/recipes/adnkronos.recipe new file mode 100644 index 0000000000..13a28b86eb --- /dev/null +++ b/resources/recipes/adnkronos.recipe @@ -0,0 +1,59 @@ +#!/usr/bin/env python +__license__ = 'GPL v3' +__author__ = 'Gabriele Marini, based on Darko Miletic' +__copyright__ = '2009-2010, Darko Miletic ' +description = 'Italian daily newspaper - 02-05-2010' + +''' +http://www.adnkronos.com/ +''' + +from calibre.web.feeds.news import BasicNewsRecipe + +class Adnkronos(BasicNewsRecipe): + __author__ = 'Gabriele Marini' + description = 'News agency' + cover_url = 'http://www.adnkronos.com/IGN6/img/popup_ign.jpg' + title = u'Adnkronos' + publisher = 'Adnkronos Group - ews agency' + category = 'News, politics, culture, economy, general interest' + + language = 'it' + timefmt = '[%a, %d %b, %Y]' + + oldest_article = 7 + max_articles_per_feed = 80 + use_embedded_content = False + recursion = 10 + + remove_javascript = True + def get_article_url(self, article): + link = article.get('id', article.get('guid', None)) + return link + + extra_css = ' .newsAbstract{font-style: italic} ' + keep_only_tags = [dict(name='div', attrs={'class':['breadCrumbs','newsTop','newsText']}) + ] + + + remove_tags = [ + dict(name='div', attrs={'class':['leogoo','leogoo2']}) + ] + + + feeds = [ + (u'Prima Pagina', u'http://rss.adnkronos.com/RSS_PrimaPagina.xml'), + (u'Ultima Ora', u'http://rss.adnkronos.com/RSS_Ultimora.xml'), + (u'Politica', u'http://rss.adnkronos.com/RSS_Politica.xml'), + (u'Esteri', u'http://rss.adnkronos.com/RSS_Esteri.xml'), + (u'Cronoca', u'http://rss.adnkronos.com/RSS_Cronaca.xml'), + (u'Economia', u'http://rss.adnkronos.com/RSS_Economia.xml'), + (u'Finanza', u'http://rss.adnkronos.com/RSS_Finanza.xml'), + (u'CyberNews', u'http://rss.adnkronos.com/RSS_CyberNews.xml'), + (u'Spettacolo', u'http://rss.adnkronos.com/RSS_Spettacolo.xml'), + (u'Cultura', u'http://rss.adnkronos.com/RSS_Cultura.xml'), + (u'Sport', u'http://rss.adnkronos.com/RSS_Sport.xml'), + (u'Sostenibilita', u'http://rss.adnkronos.com/RSS_Sostenibilita.xml'), + (u'Salute', u'http://rss.adnkronos.com/RSS_Salute.xml') + ] + diff --git a/resources/recipes/il_messaggero.recipe b/resources/recipes/il_messaggero.recipe new file mode 100644 index 0000000000..e2c90075d5 --- /dev/null +++ b/resources/recipes/il_messaggero.recipe @@ -0,0 +1,56 @@ +#!/usr/bin/env python +__license__ = 'GPL v3' +__author__ = 'Gabriele Marini, based on Darko Miletic' +__copyright__ = '2009, Darko Miletic ' +description = 'Italian daily newspaper - v1.01 (04, January 2010)' + +''' +http://www.messaggero.it/ +''' + +from calibre.web.feeds.news import BasicNewsRecipe + +class IlMessaggero(BasicNewsRecipe): + __author__ = 'Gabriele Marini' + description = 'Italian News' + + cover_url = 'http://www.ilmessaggero.it/img_tst/logomsgr.gif' + title = u'Il Messaggero' + publisher = 'Caltagirone Editore' + category = 'News, politics, culture, economy, general interest' + + language = 'it' + timefmt = '[%a, %d %b, %Y]' + + oldest_article = 5 + max_articles_per_feed = 100 + use_embedded_content = False + recursion = 10 + + remove_javascript = True + + + keep_only_tags = [dict(name='h1', attrs={'class':'titoloLettura2'}), + dict(name='h2', attrs={'class':'sottotitLettura'}), + dict(name='span', attrs={'class':'testoArticoloG'}) + ] + + + + feeds = [ + (u'HomePage', u'http://www.ilmessaggero.it/rss/home.xml'), + (u'Primo Piano', u'http://www.ilmessaggero.it/rss/initalia_primopiano.xml'), + (u'Cronaca Bianca', u'http://www.ilmessaggero.it/rss/initalia_cronacabianca.xml'), + (u'Cronaca Nera', u'http://www.ilmessaggero.it/rss/initalia_cronacanera.xml'), + (u'Economia e Finanza', u'http://www.ilmessaggero.it/rss/economia.xml'), + (u'Politica', u'http://www.ilmessaggero.it/rss/initalia_politica.xml'), + (u'Scienza e Tecnologia', u'http://www.ilmessaggero.it/rss/scienza.xml'), + (u'Cinema', u'http://www.ilmessaggero.it/rss.php?refresh_ce#'), + (u'Viaggi', u'http://www.ilmessaggero.it/rss.php?refresh_ce#'), + (u'Roma', u'http://www.ilmessaggero.it/rss/roma.xml'), + (u'Cultura e Tendenze', u'http://www.ilmessaggero.it/rss/roma_culturaspet.xml'), + (u'Sport', u'http://www.ilmessaggero.it/rss/sport.xml'), + (u'Calcio', u'http://www.ilmessaggero.it/rss/sport_calcio.xml'), + (u'Motori', u'http://www.ilmessaggero.it/rss/sport_motori.xml') + ] +