mirror of
https://github.com/kovidgoyal/calibre.git
synced 2025-09-29 15:31:08 -04:00
69 lines
3.0 KiB
Plaintext
69 lines
3.0 KiB
Plaintext
from calibre.web.feeds.news import BasicNewsRecipe
|
|
|
|
|
|
class GazetaWspolczesna(BasicNewsRecipe):
|
|
title = u'Gazeta Wsp\xf3\u0142czesna'
|
|
__author__ = 'fenuks'
|
|
description = u'Gazeta Współczesna - portal regionalny.'
|
|
category = 'newspaper'
|
|
language = 'pl'
|
|
encoding = 'iso-8859-2'
|
|
extra_css = 'ul {list-style: none; padding:0; margin:0;}'
|
|
INDEX = 'http://www.wspolczesna.pl'
|
|
masthead_url = INDEX + '/images/top_logo.png'
|
|
oldest_article = 7
|
|
max_articles_per_feed = 100
|
|
remove_empty_feeds = True
|
|
no_stylesheets = True
|
|
use_embedded_content = False
|
|
ignore_duplicate_articles = {'title', 'url'}
|
|
|
|
feeds = [
|
|
(u'Wszystkie', u'http://www.wspolczesna.pl/rss.xml'),
|
|
(u'August\xf3w', u'http://www.wspolczesna.pl/augustow.xml'),
|
|
(u'Bia\u0142ystok', u'http://www.wspolczesna.pl/bialystok.xml'),
|
|
(u'Bielsk Podlaski', u'http://www.wspolczesna.pl/bielsk.xml'),
|
|
(u'E\u0142k', u'http://www.wspolczesna.pl/elk.xml'),
|
|
(u'Grajewo', u'http://www.wspolczesna.pl/grajewo.xml'),
|
|
(u'Go\u0142dap', u'http://www.wspolczesna.pl/goldap.xml'),
|
|
(u'Hajn\xf3wka', u'http://www.wspolczesna.pl/hajnowka.xml'),
|
|
(u'Kolno', u'http://www.wspolczesna.pl/kolno.xml'),
|
|
(u'\u0141om\u017ca', u'http://www.wspolczesna.pl/lomza.xml'),
|
|
(u'Mo\u0144ki', u'http://www.wspolczesna.pl/monki.xml'),
|
|
(u'Olecko', u'http://www.wspolczesna.pl/olecko.xml'),
|
|
(u'Ostro\u0142\u0119ka', u'http://www.wspolczesna.pl/ostroleka.xml'),
|
|
(u'Powiat Bia\u0142ostocki', u'http://www.wspolczesna.pl/powiat.xml'),
|
|
(u'Sejny', u'http://www.wspolczesna.pl/sejny.xml'),
|
|
(u'Siemiatycze', u'http://www.wspolczesna.pl/siemiatycze.xml'),
|
|
(u'Sok\xf3\u0142ka', u'http://www.wspolczesna.pl/sokolka.xml'),
|
|
(u'Suwa\u0142ki', u'http://www.wspolczesna.pl/suwalki.xml'),
|
|
(u'Wysokie Mazowieckie', u'http://www.wspolczesna.pl/wysokie.xml'),
|
|
(u'Zambr\xf3w', u'http://www.wspolczesna.pl/zambrow.xml'),
|
|
(u'Sport', u'http://www.wspolczesna.pl/sport.xml'),
|
|
(u'Praca', u'http://www.wspolczesna.pl/praca.xml'),
|
|
(u'Dom', u'http://www.wspolczesna.pl/dom.xml'),
|
|
(u'Auto', u'http://www.wspolczesna.pl/auto.xml'),
|
|
(u'Zdrowie', u'http://www.wspolczesna.pl/zdrowie.xml')]
|
|
|
|
keep_only_tags = [dict(id='article')]
|
|
|
|
def get_cover_url(self):
|
|
soup = self.index_to_soup(
|
|
self.INDEX + '/apps/pbcs.dll/section?Category=JEDYNKI')
|
|
nexturl = self.INDEX + soup.find(id='covers').find('a')['href']
|
|
soup = self.index_to_soup(nexturl)
|
|
self.cover_url = self.INDEX + \
|
|
soup.find(id='cover').find(name='img')['src']
|
|
return getattr(self, 'cover_url', self.cover_url)
|
|
|
|
def decode_feedportal_url(self, url):
|
|
link = url.rpartition('l/0L0S')[2][:-12]
|
|
replaces = (('0B', '.'), ('0C', '/'), ('0H', ','),
|
|
('0D', '?'), ('0F', '='), ('0A', '0'), ('0I', '_'))
|
|
for t in replaces:
|
|
link = link.replace(*t)
|
|
return 'http://' + link
|
|
|
|
def print_version(self, url):
|
|
return self.decode_feedportal_url(url) + '&Template=printpicart'
|