calibre/recipes/stiintasitehnica.recipe
2015-01-23 19:08:21 +05:30

57 lines
2.1 KiB
Python

#!/usr/bin/env python2
# -*- coding: utf-8 -*-
__license__ = 'GPL v3'
__copyright__ = u'2011, Silviu Cotoar\u0103'
'''
stiintasitehnica.com
'''
from calibre.web.feeds.news import BasicNewsRecipe
class Stiintasitehnica(BasicNewsRecipe):
title = u'\u0218tiin\u021b\u0103 \u015fi Tehnic\u0103'
__author__ = u'Silviu Cotoar\u0103'
description = u'\u0218tiin\u021b\u0103 \u015fi Tehnic\u0103'
publisher = u'\u0218tiin\u021b\u0103 \u015fi Tehnic\u0103'
oldest_article = 50
language = 'ro'
max_articles_per_feed = 100
no_stylesheets = True
use_embedded_content = False
category = u'Ziare,Reviste,Stiinta,Tehnica'
encoding = 'utf-8'
cover_url = 'http://www.stiintasitehnica.com/images/logo.jpg'
conversion_options = {
'comments' : description
,'tags' : category
,'language' : language
,'publisher' : publisher
}
keep_only_tags = [
dict(name='div', attrs={'id':'mainColumn2'})
]
remove_tags = [
dict(name='span', attrs={'class':['redEar']})
, dict(name='table', attrs={'class':['connect_widget_interactive_area']})
, dict(name='div', attrs={'class':['panel-overlay']})
, dict(name='div', attrs={'id':['pointer']})
, dict(name='img', attrs={'class':['nav-next', 'nav-prev']})
, dict(name='table', attrs={'class':['connect_widget_interactive_area']})
, dict(name='hr', attrs={'class':['dotted']})
]
remove_tags_after = [
dict(name='hr', attrs={'class':['dotted']})
]
feeds = [
(u'Feeds', u'http://www.stiintasitehnica.com/rss/stiri.xml')
]
def preprocess_html(self, soup):
return self.adeify_images(soup)