mirror of
https://github.com/kovidgoyal/calibre.git
synced 2025-09-29 15:31:08 -04:00
47 lines
3.1 KiB
Plaintext
47 lines
3.1 KiB
Plaintext
from calibre.web.feeds.news import BasicNewsRecipe
|
|
import re
|
|
|
|
class Gildia(BasicNewsRecipe):
|
|
title = u'Gildia.pl'
|
|
__author__ = 'fenuks'
|
|
description = u'Fantastyczny Portal Kulturalny - newsy, recenzje, galerie, wywiady. Literatura, film, gry komputerowe i planszowe, komiks, RPG, sklep. Nie lekceważ potęgi wyobraźni!'
|
|
cover_url = 'http://www.film.gildia.pl/_n_/portal/redakcja/logo/logo-gildia.pl-500.jpg'
|
|
category = 'culture'
|
|
cover_url = 'http://gildia.pl/images/logo-main.png'
|
|
language = 'pl'
|
|
oldest_article = 8
|
|
max_articles_per_feed = 100
|
|
remove_empty_feeds=True
|
|
no_stylesheets=True
|
|
ignore_duplicate_articles = {'title', 'url'}
|
|
preprocess_regexps = [(re.compile(ur'</?sup>'), lambda match: '') ]
|
|
remove_tags=[dict(name='div', attrs={'class':'backlink'}), dict(name='div', attrs={'class':'im_img'}), dict(name='div', attrs={'class':'addthis_toolbox addthis_default_style'})]
|
|
keep_only_tags=dict(name='div', attrs={'class':'widetext'})
|
|
feeds = [(u'Gry', u'http://www.gry.gildia.pl/rss'), (u'Literatura', u'http://www.literatura.gildia.pl/rss'), (u'Film', u'http://www.film.gildia.pl/rss'), (u'Horror', u'http://www.horror.gildia.pl/rss'), (u'Konwenty', u'http://www.konwenty.gildia.pl/rss'), (u'Plansz\xf3wki', u'http://www.planszowki.gildia.pl/rss'), (u'Manga i anime', u'http://www.manga.gildia.pl/rss'), (u'Star Wars', u'http://www.starwars.gildia.pl/rss'), (u'Techno', u'http://www.techno.gildia.pl/rss'), (u'Historia', u'http://www.historia.gildia.pl/rss'), (u'Magia', u'http://www.magia.gildia.pl/rss'), (u'Bitewniaki', u'http://www.bitewniaki.gildia.pl/rss'), (u'RPG', u'http://www.rpg.gildia.pl/rss'), (u'LARP', u'http://www.larp.gildia.pl/rss'), (u'Muzyka', u'http://www.muzyka.gildia.pl/rss'), (u'Nauka', u'http://www.nauka.gildia.pl/rss')]
|
|
|
|
|
|
def skip_ad_pages(self, soup):
|
|
content = soup.find('div', attrs={'class':'news'})
|
|
if 'recenzj' in soup.title.string.lower():
|
|
for link in content.findAll(name='a'):
|
|
if 'recenzj' in link['href'] or 'muzyka/plyty' in link['href']:
|
|
return self.index_to_soup(link['href'], raw=True)
|
|
if 'fragmen' in soup.title.string.lower():
|
|
for link in content.findAll(name='a'):
|
|
if 'fragment' in link['href']:
|
|
return self.index_to_soup(link['href'], raw=True)
|
|
|
|
|
|
def preprocess_html(self, soup):
|
|
for a in soup('a'):
|
|
if a.has_key('href') and 'http://' not in a['href'] and 'https://' not in a['href']:
|
|
if '/gry/' in a['href']:
|
|
a['href']='http://www.gry.gildia.pl' + a['href']
|
|
elif u'książk' in soup.title.string.lower() or u'komiks' in soup.title.string.lower():
|
|
a['href']='http://www.literatura.gildia.pl' + a['href']
|
|
elif u'komiks' in soup.title.string.lower():
|
|
a['href']='http://www.literatura.gildia.pl' + a['href']
|
|
else:
|
|
a['href']='http://www.gildia.pl' + a['href']
|
|
return soup
|