Fix Liberation news website recipe

This commit is contained in:
Aurélien Chabot 2011-10-16 11:23:56 +02:00
parent 68448106a6
commit 4bc485f6a4

View File

@ -9,39 +9,72 @@ liberation.fr
from calibre.web.feeds.news import BasicNewsRecipe from calibre.web.feeds.news import BasicNewsRecipe
class Liberation(BasicNewsRecipe): class Liberation(BasicNewsRecipe):
title = u'Liberation' title = u'Liberation'
__author__ = 'Darko Miletic' __author__ = 'calibre'
description = 'News from France' description = 'Actualités'
language = 'fr' category = 'Actualités, France, Monde'
language = 'fr'
oldest_article = 7 use_embedded_content = False
max_articles_per_feed = 100 timefmt = ' [%d %b %Y]'
no_stylesheets = True max_articles_per_feed = 15
use_embedded_content = False no_stylesheets = True
remove_empty_feeds = True
filterDuplicates = True
html2lrf_options = ['--base-font-size', '10'] extra_css = '''
h1, h2, h3 {font-size:xx-large; font-family:Arial,Helvetica,sans-serif;}
p.subtitle {font-size:xx-small; font-family:Arial,Helvetica,sans-serif;}
h4, h5, h2.rubrique, {font-size:xx-small; color:#4D4D4D; font-family:Arial,Helvetica,sans-serif;}
.ref, .date, .author, .legende {font-size:xx-small; color:#4D4D4D; font-family:Arial,Helvetica,sans-serif;}
.mna-body, entry-body {font-size:medium; font-family:Arial,Helvetica,sans-serif;}
'''
keep_only_tags = [ keep_only_tags = [
dict(name='h1') dict(name='div', attrs={'class':'article'})
#,dict(name='div', attrs={'class':'object-content text text-item'}) ,dict(name='div', attrs={'class':'text-article m-bot-s1'})
,dict(name='div', attrs={'class':'article'}) ,dict(name='div', attrs={'class':'entry'})
#,dict(name='div', attrs={'class':'articleContent'}) ,dict(name='div', attrs={'class':'col_contenu'})
,dict(name='div', attrs={'class':'entry'}) ]
]
remove_tags_after = [ dict(name='div',attrs={'class':'toolbox extra_toolbox'}) ] remove_tags_after = [
dict(name='div',attrs={'class':['object-content text text-item', 'object-content', 'entry-content', 'col01', 'bloc_article_01']})
,dict(name='p',attrs={'class':['chapo']})
,dict(id='_twitter_facebook')
]
remove_tags = [ remove_tags = [
dict(name='p', attrs={'class':'clear'}) dict(name='iframe')
,dict(name='ul', attrs={'class':'floatLeft clear'}) ,dict(name='a', attrs={'class':'lnk-comments'})
,dict(name='div', attrs={'class':'clear floatRight'}) ,dict(name='div', attrs={'class':'toolbox'})
,dict(name='object') ,dict(name='ul', attrs={'class':'share-box'})
,dict(name='div', attrs={'class':'toolbox'}) ,dict(name='ul', attrs={'class':'tool-box'})
,dict(name='div', attrs={'class':'cartridge cartridge-basic-bubble cat-zoneabo'}) ,dict(name='ul', attrs={'class':'rub'})
#,dict(name='div', attrs={'class':'clear block block-call-items'}) ,dict(name='p',attrs={'class':['chapo']})
,dict(name='div', attrs={'class':'block-content'}) ,dict(name='p',attrs={'class':['tag']})
,dict(name='div',attrs={'class':['blokLies']})
,dict(name='div',attrs={'class':['alire']})
,dict(id='_twitter_facebook')
] ]
feeds = [ feeds = [
(u'La une', u'http://www.liberation.fr/rss/laune') (u'La une', u'http://rss.liberation.fr/rss/9/')
,(u'Monde' , u'http://www.liberation.fr/rss/monde') ,(u'Monde' , u'http://www.liberation.fr/rss/10/')
,(u'Sports', u'http://www.liberation.fr/rss/sports') ,(u'Économie', u'http://www.liberation.fr/rss/13/')
,(u'Politiques', u'http://www.liberation.fr/rss/11/')
,(u'Société', u'http://www.liberation.fr/rss/12/')
,(u'Cinéma', u'http://www.liberation.fr/rss/58/')
,(u'Écran', u'http://www.liberation.fr/rss/53/')
,(u'Sports', u'http://www.liberation.fr/rss/12/')
] ]
def get_masthead_url(self):
masthead = 'http://s0.libe.com/libe/img/common/logo-liberation-150.png'
br = BasicNewsRecipe.get_browser()
try:
br.open(masthead)
except:
self.log("\nCover unavailable")
masthead = None
return masthead