calibre/recipes/osw.recipe

42 lines
1.4 KiB
Python

#!/usr/bin/env python
__license__ = 'GPL v3'
__author__ = 'teepel <teepel44@gmail.com>'
'''
http://www.osw.waw.pl - Osrodek studiow wschodnich
'''
from calibre.web.feeds.news import BasicNewsRecipe
class OSW_Recipe(BasicNewsRecipe):
language = 'pl'
title = u'Ośrodek Studiów Wschodnich'
__author__ = 'teepel <teepel44@gmail.com>'
INDEX='http://www.osw.waw.pl'
description = u'Ośrodek Studiów Wschodnich im. Marka Karpia. Centre for Eastern Studies.'
category = u'News'
oldest_article = 7
max_articles_per_feed = 100
cover_url=''
remove_empty_feeds= True
no_stylesheets=True
remove_javascript = True
simultaneous_downloads = 5
keep_only_tags =[]
#this line should show title of the article, but it doesnt work
keep_only_tags.append(dict(name = 'h1', attrs = {'class' : 'print-title'}))
keep_only_tags.append(dict(name = 'div', attrs = {'class' : 'print-submitted'}))
keep_only_tags.append(dict(name = 'div', attrs = {'class' : 'print-content'}))
remove_tags =[]
remove_tags.append(dict(name = 'table', attrs = {'id' : 'attachments'}))
remove_tags.append(dict(name = 'div', attrs = {'class' : 'print-submitted'}))
feeds = [(u'OSW', u'http://www.osw.waw.pl/pl/rss.xml')]
def print_version(self, url):
return url.replace('http://www.osw.waw.pl/pl/', 'http://www.osw.waw.pl/pl/print/')