from calibre.web.feeds.recipes import BasicNewsRecipe class iHeuteRecipe(BasicNewsRecipe): __author__ = 'FunThomas' title = u'iDnes.cz' publisher = u'MAFRA a.s.' description = 'iDNES.cz Zprávy, Technet, Komiksy a další' oldest_article = 3 max_articles_per_feed = 2 feeds = [ (u'Zprávy', u'http://servis.idnes.cz/rss.asp?c=zpravodaj'), (u'Sport', u'http://servis.idnes.cz/rss.asp?c=sport'), (u'Technet', u'http://servis.idnes.cz/rss.asp?c=technet'), (u'Mobil', u'http://servis.idnes.cz/rss.asp?c=mobil'), (u'Ekonomika', u'http://servis.idnes.cz/rss.asp?c=ekonomikah'), (u'Cestování', u'http://servis.idnes.cz/rss.asp?c=iglobe'), (u'Komixy', u'http://servis.idnes.cz/rss.asp?c=komiksy') ] encoding = 'cp1250' language = 'cs' cover_url = 'http://g.idnes.cz/u/loga-n4/idnes.gif' remove_javascript = True no_stylesheets = True remove_attributes = ['width', 'height'] remove_tags = [dict(name='div', attrs={'id': ['zooming']}), dict(name='div', attrs={ 'class': ['related', 'mapa-wrapper']}), dict(name='table', attrs={'id': ['opener-img', 'portal']}), dict(name='table', attrs={'class': ['video-16ku9']})] remove_tags_after = [ dict(name='div', attrs={'id': ['related', 'related2']})] keep_only_tags = [dict(name='div', attrs={'class': [ 'art-full adwords-text', 'dil-day', 'art-full']}), dict(name='table', attrs={'class': ['kemel-box']})] def print_version(self, url): print_url = url split_url = url.split("?") if (split_url[0].rfind('dilbert.asp') != -1): # dilbert komix print_url = print_url.replace('.htm', '.gif&tisk=1') print_url = print_url.replace('.asp', '.aspx') elif (split_url[0].rfind('kemel.asp') == -1): # not Kemel komix print_url = 'http://zpravy.idnes.cz/tiskni.asp?' + split_url[1] # kemel kemel print page doesn't work return print_url extra_css = ''' h1 {font-size:125%; font-weight:bold} h3 {font-size:110%; font-weight:bold} '''