mirror of
https://github.com/kovidgoyal/calibre.git
synced 2025-09-29 15:31:08 -04:00
48 lines
2.2 KiB
Plaintext
48 lines
2.2 KiB
Plaintext
from calibre.web.feeds.news import BasicNewsRecipe
|
|
|
|
|
|
class KurierLubelski(BasicNewsRecipe):
|
|
title = u'Kurier Lubelski'
|
|
__author__ = 'fenuks'
|
|
description = u'Gazeta Regionalna Kurier Lubelski. Najnowsze Wiadomości Lublin. Czytaj Informacje Lublin!'
|
|
category = 'newspaper'
|
|
language = 'pl'
|
|
encoding = 'iso-8859-2'
|
|
masthead_url = 'http://s.polskatimes.pl/g/logo_naglowek/kurierlubelski.png?24'
|
|
oldest_article = 7
|
|
max_articles_per_feed = 100
|
|
remove_empty_feeds = True
|
|
no_stylesheets = True
|
|
use_embedded_content = False
|
|
ignore_duplicate_articles = {'title', 'url'}
|
|
remove_tags_after = dict(attrs={'src': 'http://nm.dz.com.pl/dz.png'})
|
|
remove_tags = [dict(id='mat-podobne'), dict(name='a', attrs={
|
|
'class': 'czytajDalej'}), dict(attrs={'src': 'http://nm.dz.com.pl/dz.png'})]
|
|
|
|
feeds = [
|
|
(u'Wiadomo\u015bci', u'http://kurierlubelski.feedsportal.com/c/32980/f/533785/index.rss?201302'),
|
|
(u'Region', u'http://www.kurierlubelski.pl/rss/kurierlubelski_region.xml?201302'),
|
|
(u'Sport', u'http://kurierlubelski.feedsportal.com/c/32980/f/533786/index.rss?201302'),
|
|
(u'Kultura', u'http://kurierlubelski.feedsportal.com/c/32980/f/533787/index.rss?201302'),
|
|
(u'Rozmaito\u015bci', u'http://www.kurierlubelski.pl/rss/kurierlubelski_rozmaitosci.xml?201302'),
|
|
|
|
(u'Dom', u'http://www.kurierlubelski.pl/rss/kurierlubelski_dom.xml?201302'),
|
|
(u'Serwisy', u'http://www.kurierlubelski.pl/rss/kurierlubelski_serwisy.xml?201302'),
|
|
(u'Motofakty', u'http://www.kurierlubelski.pl/rss/kurierlubelski_motofakty.xml?201302'),
|
|
(u'M\xf3j Reporter', u'http://www.kurierlubelski.pl/rss/kurierlubelski_mojreporter.xml?201302'),
|
|
(u'Praca', u'http://www.kurierlubelski.pl/rss/kurierlubelski_praca.xml?201302')]
|
|
|
|
def print_version(self, url):
|
|
return url.replace('artykul', 'drukuj')
|
|
|
|
def skip_ad_pages(self, soup):
|
|
if 'Advertisement' in soup.title:
|
|
nexturl = soup.find('a')['href']
|
|
return self.index_to_soup(nexturl, raw=True)
|
|
|
|
def get_cover_url(self):
|
|
soup = self.index_to_soup(
|
|
'http://www.prasa24.pl/gazeta/kurier-lubelski/')
|
|
self.cover_url = soup.find(id='pojemnik').img['src']
|
|
return getattr(self, 'cover_url', self.cover_url)
|