calibre/recipes/romanialibera.recipe
Kovid Goyal 29cd8d64ea
Change shebangs to python from python2
Also remove a few other miscellaneous references to python2
2020-08-22 18:47:51 +05:30

49 lines
1.7 KiB
Python

#!/usr/bin/env python
# -*- coding: utf-8 -*-
__license__ = 'GPL v3'
__copyright__ = u'2011, Silviu Cotoar\u0103'
'''
romanialibera.ro
'''
from calibre.web.feeds.news import BasicNewsRecipe
class RomaniaLibera(BasicNewsRecipe):
title = u'Rom\u00e2nia Liber\u0103'
__author__ = u'Silviu Cotoar\u0103'
description = u'Rom\u00e2nia Liber\u0103'
publisher = u'Rom\u00e2nia Liber\u0103'
oldest_article = 5
language = 'ro'
max_articles_per_feed = 100
no_stylesheets = True
use_embedded_content = False
category = 'Ziare,Stiri'
encoding = 'utf-8'
cover_url = 'http://www.romanialibera.ro/templates/lilac/images/sigla_1.gif'
conversion_options = {
'comments': description, 'tags': category, 'language': language, 'publisher': publisher
}
keep_only_tags = [
dict(name='div', attrs={'id': 'articol'})
]
remove_tags = [
dict(name='div', attrs={'id': ['art_actions']}), dict(name='div', attrs={'class': ['stats']}), dict(name='div', attrs={'class': ['data']}), dict(name='div', attrs={'class': ['autori']}), dict(name='div', attrs={'class': ['banda_explicatii_text']}), dict(name='td', attrs={'class': ['connect_widget_vertical_center connect_widget_button_cell']}), dict(name='div', attrs={'class': ['aceeasi_tema']}), dict(name='div', attrs={'class': ['art_after_text']}), dict(name='div', attrs={'class': ['navigare']}), dict(name='div', attrs={'id': ['art_text_left']}) # noqa
]
remove_tags_after = [
dict(name='div', attrs={'class': 'art_after_text'})
]
feeds = [
(u'Feeds', u'http://www.romanialibera.ro/rss.xml')
]
def preprocess_html(self, soup):
return self.adeify_images(soup)