Update Liberation

This commit is contained in:
Aimylios 2024-06-23 12:51:13 +02:00
parent df4b8f9744
commit 4ca22f2b19

View File

@ -9,6 +9,8 @@ __copyright__ = '2008, Darko Miletic <darko.miletic at gmail.com>'
liberation.fr
'''
import re
from calibre.web.feeds.news import BasicNewsRecipe
@ -19,8 +21,8 @@ class Liberation(BasicNewsRecipe):
publication_type = 'newspaper'
language = 'fr'
oldest_article = 2
max_articles_per_feed = 30
oldest_article = 3
max_articles_per_feed = 10
no_stylesheets = True
remove_empty_feeds = True
ignore_duplicate_articles = {'title', 'url'}
@ -29,29 +31,41 @@ class Liberation(BasicNewsRecipe):
masthead_url = 'https://www.liberation.fr/pf/resources/images/liberation.png?d=47'
feeds = [
('Libération', 'https://www.liberation.fr/arc/outboundfeeds/rss/?outputType=xml'),
('Accueil', 'https://www.liberation.fr/arc/outboundfeeds/collection/accueil-une/?outputType=xml')
#('Libération', 'https://www.liberation.fr/arc/outboundfeeds/rss/?outputType=xml'),
('A la une', 'https://www.liberation.fr/arc/outboundfeeds/rss-all/collection/accueil-une/?outputType=xml'),
('Politique', 'https://www.liberation.fr/arc/outboundfeeds/rss-all/category/politique/?outputType=xml'),
('International', 'https://www.liberation.fr/arc/outboundfeeds/rss-all/category/international/?outputType=xml'),
('CheckNews', 'https://www.liberation.fr/arc/outboundfeeds/rss-all/category/checknews/?outputType=xml'),
('Culture', 'https://www.liberation.fr/arc/outboundfeeds/rss-all/category/culture/?outputType=xml'),
('Idées et Débats', 'https://www.liberation.fr/arc/outboundfeeds/rss-all/category/idees-et-debats/?outputType=xml'),
('Société', 'https://www.liberation.fr/arc/outboundfeeds/rss-all/category/societe/?outputType=xml'),
('Environnement', 'https://www.liberation.fr/arc/outboundfeeds/rss-all/category/environnement/?outputType=xml'),
('Economie', 'https://www.liberation.fr/arc/outboundfeeds/rss-all/category/economie/?outputType=xml'),
('Lifestyle', 'https://www.liberation.fr/arc/outboundfeeds/rss-all/category/lifestyle/?outputType=xml'),
('Portraits', 'https://www.liberation.fr/arc/outboundfeeds/rss-all/category/portraits/?outputType=xml'),
('Sports', 'https://www.liberation.fr/arc/outboundfeeds/rss-all/category/sports/?outputType=xml'),
('Sciences', 'https://www.liberation.fr/arc/outboundfeeds/rss-all/category/sciences/?outputType=xml'),
('Forums & événements', 'https://www.liberation.fr/arc/outboundfeeds/rss-all/category/forums/?outputType=xml')
]
keep_only_tags = [
dict(name='div', attrs={'class': 'left-article-section'})
dict(name='div', attrs={'class': re.compile('default__Main')})
]
remove_tags_after = [
dict(name='article', attrs={'class': re.compile('article-body-wrapper')})
]
remove_tags = [
dict(name=['button', 'source']),
dict(attrs={'class': ['ts-share-bar']}),
dict(name='div', attrs={'class': [
'article-dossier', 'color_background_green', 'display_block', 'tag-container'
]})
]
remove_attributes = [
'height', 'width'
]
extra_css = '''
h1 { font-size: 1.6em; margin-top: 0em; }
h2, h3, h4, h5, h6 { font-size: 1em; }
.image-metadata { font-size: 0.6em; margin-top: 0em; margin-bottom: 1.6em; }
'''
def get_browser(self):