calibre/recipes/informacje_usa.recipe
2012-02-20 11:11:32 +05:30

19 lines
1.1 KiB
Plaintext

from calibre.web.feeds.news import BasicNewsRecipe
import re
class Informacje_USA(BasicNewsRecipe):
title = u'Informacje USA'
oldest_article = 7
max_articles_per_feed = 100
__author__ = 'fenuks'
description = u'portal wiadomości amerykańskich'
category = 'news'
language = 'pl'
masthead_url= 'http://www.informacjeusa.com/wp-content/add_images/top_logo_5_2010.jpg'
cover_url='http://www.informacjeusa.com/wp-content/add_images/top_logo_5_2010.jpg'
no_stylesheets = True
preprocess_regexps = [(re.compile(ur'<p>Zobacz:.*?</p>', re.DOTALL), lambda match: ''), (re.compile(ur'<p><a href=".*?Zobacz także:.*?</a></p>', re.DOTALL), lambda match: ''), (re.compile(ur'<p><p>Zobacz też:.*?</a></p>', re.DOTALL), lambda match: '')]
keep_only_tags=[dict(name='div', attrs={'class':'box box-single'})]
remove_tags_after= dict(attrs={'class':'tags'})
remove_tags= [dict(attrs={'class':['postmetadata', 'tags', 'banner']}), dict(name='a', attrs={'title':['Drukuj', u'Wyślij']})]
feeds = [(u'Informacje', u'http://www.informacjeusa.com/feed/')]