Fix Il Sole 24 Ore

This commit is contained in:
Kovid Goyal 2011-04-20 15:35:40 -06:00
parent 7b2ea64d43
commit 6502429c6d

View File

@ -1,71 +1,65 @@
#!/usr/bin/env python
__license__ = 'GPL v3'
__author__ = 'Lorenzo Vigentini & Edwin van Maastrigt'
__copyright__ = '2009, Lorenzo Vigentini <l.vigentini at gmail.com> and Edwin van Maastrigt <evanmaastrigt at gmail.com>'
__description__ = 'Financial news daily paper - v1.02 (30, January 2010)'
__author__ = 'Marco Saraceno'
__copyright__ = '2010, Marco Saraceno <marcosaraceno at gmail.com>'
description = 'Italian daily newspaper - v 1.1 (Mar14,2011)'
'''
http://www.ilsole24ore.com/
http://www.ilsole24ore.com
'''
from calibre.web.feeds.news import BasicNewsRecipe
class IlSole24Ore(BasicNewsRecipe):
__author__ = 'Marco Saraceno'
description = 'Italian financial daily newspaper'
class ilsole24Ore(BasicNewsRecipe):
author = 'Lorenzo Vigentini & Edwin van Maastrigt'
description = 'Financial news daily paper'
cover_url = 'http://www.ilsole24ore.com/img2007/print_header.gif'
title = u'il Sole 24 Ore New'
publisher = 'italiaNews'
category = 'News, finance, economy, politics'
cover_url = 'http://www.shopping24.ilsole24ore.com/ProductRelated/rds/img/logo_sole.gif'
title = u'Il Sole 24 Ore'
publisher = 'Gruppo editoriale GRUPPO 24ORE'
category = 'News, politics, culture, economy, financial, Italian'
language = 'it'
timefmt = '[%a, %d %b, %Y]'
oldest_article = 2
max_articles_per_feed = 50
max_articles_per_feed = 100
use_embedded_content = False
extra_css = '.headline {font-size: x-large;} \n .fact { padding-top: 10pt }'
remove_tags = [
dict(name='div', attrs={'class':['header','titolo']}),
dict(name='table', attrs={'class':['footer1024','footerdown']}),
]
remove_javascript = True
no_stylesheets = True
def get_article_url(self, article):
return article.get('id', article.get('guid', None))
link = article.get('link', None)
if link is None:
return article
if link.split('/')[-1]=="story01.htm":
link=link.split('/')[-2]
a=['0B','0C','0D','0E','0F','0G','0N' ,'0L0S','0A']
b=['.' ,'/' ,'?' ,'-' ,'=' ,'&' ,'.com','www.','0']
for i in range(0,len(a)):
link=link.replace(a[i],b[i])
link="http://"+link
return link
feeds = [
(u'Notizie Italia', u'http://www.ilsole24ore.com/rss/notizie/italia.xml'),
(u'Notizie Europa', u'http://www.ilsole24ore.com/rss/notizie/europa.xml'),
(u'Notizie USA', u'http://www.ilsole24ore.com/rss/notizie/usa.xml'),
(u'Notizie Americhe', u'http://www.ilsole24ore.com/rss/notizie/americhe.xml'),
(u'Notizie Medio Oriente e Africa', u'http://www.ilsole24ore.com/rss/notizie/medio-oriente-e-africa.xml'),
(u'Notizie Asia e Oceania', u'http://www.ilsole24ore.com/rss/notizie/asia-e-oceania.xml'),
(u'Commenti', u'http://www.ilsole24ore.com/rss/commenti-e-idee.xml'),
(u'Norme e tributi', u'http://www.ilsole24ore.com/rss/norme-e-tributi.xml'),
(u'Finanza', u'http://www.ilsole24ore.com/rss/finanza-e-mercati.xml'),
(u'Economia', u'http://www.ilsole24ore.com/rss/economia.xml'),
(u'Tecnologia', u'http://www.ilsole24ore.com/rss/tecnologie.xml'),
(u'Cultura', u'http://www.ilsole24ore.com/rss/cultura.xml'),
]
def print_version(self, url):
link, sep, params = url.rpartition('?')
if link is None:
return link.replace('_1.php', '_php')
return link.replace('.shtml', '_PRN.shtml')
keep_only_tags = [
dict(name='div', attrs={'class':'txt'})
]
# remove_tags = [dict(name='br')]
feeds = [
(u'Prima pagina', u'http://www.ilsole24ore.com/rss/primapagina.xml'),
(u'Norme e tributi', u'http://www.ilsole24ore.com/rss/norme-tributi.xml'),
(u'Finanza e mercati', u'http://www.ilsole24ore.com/rss/finanza-mercati.xml'),
(u'Economia e lavoro', u'http://www.ilsole24ore.com/rss/economia-lavoro.xml'),
(u'Italia', u'http://www.ilsole24ore.com/rss/italia.xml'),
(u'Mondo', u'http://www.ilsole24ore.com/rss/mondo.xml'),
(u'Tecnologia e business', u'http://www.ilsole24ore.com/rss/tecnologia-business.xml'),
(u'Cultura e tempo libero', u'http://www.ilsole24ore.com/rss/tempolibero-cultura.xml'),
(u'Sport', u'http://www.ilsole24ore.com/rss/sport.xml'),
(u'Professionisti 24', u'http://www.ilsole24ore.com/rss/prof_home.xml'),
(u'Ambiente e Sicurezza',u'http://www.ilsole24ore.com/rss/prof_as.xml')
]
extra_css = '''
html, body, table, tr, td, h1, h2, h3, h4, h5, h6, p, a, span, br, img {margin:0;padding:0;border:0;font-size:12px;font-family:"Georgia","Times New Roman";}
.linkHighlight {color:#0292c6;}
.txt {border-bottom:1px solid #7c7c7c;padding-bottom:20px};text-align:justify;font-family:"serif"}
.txt p {line-height:18px;}
.txt span {line-height:22px;}
.title h3 {color:#7b7b7b;}
.title h4 {color:#08526e;font-size:26px;font-family:"Times New Roman";font-weight:normal;}
'''
return url.replace('.shtml', '_PRN.shtml')