calibre/recipes/tagesan.recipe
2016-01-08 20:24:03 +05:30

52 lines
2.0 KiB
Plaintext

from calibre.web.feeds.news import BasicNewsRecipe
class AdvancedUserRecipe1450864843(BasicNewsRecipe):
title = u'Tagesanzeiger'
__author__ = 'noxxx'
language = 'de'
oldest_article = 1
max_articles_per_feed = 8
no_stylesheets = True
auto_cleanup = True
scale_news_images_to_device = True
compress_news_images = True
remove_javascript = True
encoding = 'utf-8'
remove_empty_feeds = True
ignore_duplicate_articles = {'title', 'url'}
feeds = [
(u'Front', u'http://www.tagesanzeiger.ch/rss.html'),
(u'Zurich', u'http://www.tagesanzeiger.ch/zuerich/rss.html'),
(u'Schweiz', u'http://www.tagesanzeiger.ch/schweiz/rss.html'),
(u'International', u'http://www.tagesanzeiger.ch/ausland/rss.html'),
(u'Wirtschaft', u'http://www.tagesanzeiger.ch/wirtschaft/rss.html'),
(u'Sport', u'http://www.tagesanzeiger.ch/sport/rss.html'),
(u'Kultur', u'http://www.tagesanzeiger.ch/kultur/rss.html'),
(u'Leben', u'http://www.tagesanzeiger.ch/leben/rss.html'),
(u'Panorama', u'http://www.tagesanzeiger.ch/panorama/rss.html'),
(u'Digital', u'http://www.tagesanzeiger.ch/digital/rss.html')
]
keep_only_tags = [dict(name='h1'),
dict(name='h3'),
dict(name='div', attrs={'id': 'topElement'}),
dict(name='div', attrs={'id': 'mainContent'})
]
auto_cleanup_keep = {'//div[@id="article-image"]',
'//*[@class="standard"]'
}
conversion_options = {'base_font_size': 12}
extra_css = 'h1 { font-size: 150%; font-weight: bold; } \
h2 { font-size: 120%; font-weight: bold; } \
h3 { font-size: 100%; font-style: italic; } \
h4, h5, h6 { font-size: 100%; font-style: italic; }'
def get_cover_url(self):
self.cover_url = 'http://newsmail.cleartext.ch/referenz_tagesanzeiger.jpg'
return getattr(self, 'cover_url', None)