import re from calibre.web.feeds.news import BasicNewsRecipe class Overclock_pl(BasicNewsRecipe): title = u'Overclock.pl' oldest_article = 7 max_articles_per_feed = 100 __author__ = 'fenuks' description = u'Vortal poświęcony tematyce hardware, kładący największy nacisk na podkręcanie / overclocking (włącznie z extreme) i chłodzenie / cooling (air cooling, water cooling, freon cooling, dry ice, liquid nitrogen).' # noqa category = 'IT' language = 'pl' masthead_url = 'http://www.overclock.pl/gfx/logo_m.png' cover_url = 'http://www.overclock.pl/gfx/logo_m.png' no_stylesheets = True remove_empty_feeds = True preprocess_regexps = [(re.compile(ur'Komentarze do aktualności:.*?', re.DOTALL), lambda match: ''), (re.compile(ur'

Nawigacja

', re.DOTALL), lambda match: '')] keep_only_tags = [ dict(name='div', attrs={'class': 'news'}), dict(id='articleContent')] remove_tags = [dict(name='span', attrs={'class': 'info'}), dict( attrs={'class': 'shareit'})] feeds = [(u'Aktualno\u015bci', u'http://www.overclock.pl/rss.news.xml'), (u'Testy i recenzje', u'http://www.overclock.pl/rss.articles.xml')] def print_version(self, url): if 'articles/show' in url: return url.replace('show', 'showall') else: return url