calibre/recipes/l_espresso.recipe

78 lines
4.6 KiB
Python

#!/usr/bin/env python
__license__ = 'GPL v3'
__author__ = 'Lorenzo Vigentini, Gabriele Marini'
__copyright__ = '2009, Lorenzo Vigentini <l.vigentini at gmail.com>'
__version__ = 'v1.02'
__date__ = '10, January 2010'
__description__ = 'Italian weekly magazine'
'''espresso.repubblica.it'''
from calibre.web.feeds.news import BasicNewsRecipe
class Espresso(BasicNewsRecipe):
__author__ = 'Lorenzo Vigentini, Gabriele Marini'
description = 'Italian weekly magazine'
cover_url = 'http://espresso.repubblica.it/images/logo_espresso.gif'
title = 'L\'Espresso '
publisher = 'Gruppo editoriale L\'Espresso'
category = 'News, politics, culture, economy, general interest'
language = 'it'
# encoding = 'cp1252'
timefmt = '[%a, %d %b, %Y]'
oldest_article = 16
max_articles_per_feed = 100
use_embedded_content = False
recursion = 10
remove_javascript = True
no_stylesheets = True
feeds = [
(u'Homepage', u'http://kpm.data.kataweb.it/kpm3eolx/rss/home'),
(u'Local', u'http://kpm.data.kataweb.it/kpm3eolx/rss/local'),
(u'Style & Design', u'http://kpm.data.kataweb.it/kpm3eolx/rss/style_design'),
(u'Opinioni', u'http://kpm.data.kataweb.it/kpm3eolx/rss/opinioni'),
(u'Rubriche', u'http://kpm.data.kataweb.it/kpm3eolx/rss/rubriche'),
(u'Limes', u'http://temi.repubblica.it/limes/feed/'),
(u'Chiesa: HomePage', u'http://data.kataweb.it/rss/chiesa/homepage/it'),
(u'Chiesa: Speciali e Focus', u'http://data.kataweb.it/rss/chiesa/speciali_e_focus/it')
]
def print_version(self,url):
print url[7:25]
if url[7:25] == 'temi.repubblica.it':
return url + '/?printpage=undefined'
elif url[7:25] == 'www.chiesa.espress':
return url
return url + '/&print=true'
keep_only_tags = [
dict(name='div', attrs={'class':['testo','copertina','occhiello','firma','didascalia','content-second-right','detail-articles','titolo-local','generic-articles']}),
dict(name='div', attrs={'class':['generic-articles','summary','detail-articles']}),
dict(name='div', attrs={'id':['content-second-right','content2']})
]
remove_tags = [
dict(name='div',attrs={'class':['servizi','aggiungi','label-web','bottom-mobile','box-abbonamenti','box-cerca','big','little','stampaweb']}),
dict(name='div',attrs={'id':['topheader','header','navigation-new','navigation','content-second-left','menutext']}),
dict(name='ul',attrs={'id':'user-utility'}),
dict(name=['script','noscript','iframe'])
]
# extra_css = '''
# h1 {font-family:Times New Roman,"Trebuchet MS",Arial,Helvetica,sans-serif; font-size:24px; font-size-adjust:none; font-stretch:normal; font-style:normal; font-variant:normal; font-weight:bold; line-height:18px;}
# h2 {font-family:Times New Roman, "Trebuchet MS",Arial,Helvetica,sans-serif; font-size:18px; font-size-adjust:none; font-stretch:normal; font-style:normal; font-variant:normal; font-weight:bold; line-height:16px; }
# h3 {color:#333333;font-family:Times New Roman, "Trebuchet MS",Arial,Helvetica,sans-serif; font-size:16px; font-size-adjust:none; font-stretch:normal; font-style:normal; font-variant:normal; font-weight:bold; line-height:14px;}
# h4 {color:#333333; font-family:Times New Roman, "Trebuchet MS",Arial,Helvetica,sans-serif;font-size:16px; font-size-adjust:none; font-stretch:normal; font-style:normal; font-variant:normal; font-weight:bold; line-height:14px; }
# h5 {color:#333333; font-family:Times New Roman, "Trebuchet MS",Arial,Helvetica,sans-serif; font-size:12px; font-size-adjust:none; font-stretch:normal; font-style:normal; font-variant:normal; font-weight:bold; line-height:14px; text-transform:uppercase;}
# .firma {color:#333333;font-family:Times New Roman, "Trebuchet MS",Arial,Helvetica,sans-serif;font-size:12px; font-size-adjust:none; font-stretch:normal; font-style:italic; font-variant:normal; font-weight:bold; line-height:15px; text-decoration:none;}
# .testo {font-family:Times New Roman, "Trebuchet MS",Arial,Helvetica,sans-serif; font-size:10px;}
# '''