calibre/recipes/ksiazka_pl.recipe
2011-09-21 10:50:11 -06:00

29 lines
1.3 KiB
Plaintext

from calibre.web.feeds.news import BasicNewsRecipe
import re
class Ksiazka_net_pl(BasicNewsRecipe):
title = u'ksiazka.net.pl'
__author__ = 'fenuks'
description = u'Ksiazka.net.pl - book vortal'
cover_url = 'http://www.ksiazka.net.pl/fileadmin/templates/ksiazka.net.pl/images/1PortalKsiegarski-logo.jpg'
category = 'books'
language = 'pl'
oldest_article = 8
max_articles_per_feed = 100
no_stylesheets= True
#extra_css = 'img {float: right;}'
preprocess_regexps = [(re.compile(ur'Podoba mi się, kupuję:'), lambda match: '<br />')]
remove_tags_before= dict(name='div', attrs={'class':'m-body'})
remove_tags_after= dict(name='div', attrs={'class':'m-body-link'})
remove_tags=[dict(attrs={'class':['mk_library-icon', 'm-body-link', 'tagi']})]
feeds = [(u'Wiadomości', u'http://www.ksiazka.net.pl/?id=wiadomosci&type=100'),
(u'Książki', u'http://www.ksiazka.net.pl/?id=ksiazki&type=100'),
(u'Rynek', u'http://www.ksiazka.net.pl/?id=rynek&type=100')]
def image_url_processor(self, baseurl, url):
if (('file://' in url) and ('www.ksiazka.net.pl/' not in url)):
return 'http://www.ksiazka.net.pl/' + url[8:]
elif 'http://' not in url:
return 'http://www.ksiazka.net.pl/' + url
else:
return url