from calibre.web.feeds.news import BasicNewsRecipe class StrategyBusinessRecipe(BasicNewsRecipe): __license__ = 'GPL v3' __author__ = 'kwetal' language = 'en' version = 1 title = u'Strategy+Business' publisher = u' Booz & Company' category = u'Business' description = (u'Business magazine for senior business executives and the people who influence them.' 'Go to http://www.strategy-business.com/registration to sign up for a free account') oldest_article = 13 * 7 # 3 months max_articles_per_feed = 100 use_embedded_content = False remove_empty_feeds = True needs_subscription = True no_stylesheets = True remove_javascript = True def get_browser(self): br = BasicNewsRecipe.get_browser(self) br.open('http://www.strategy-business.com/registration') for i, f in enumerate(br.forms()): if 'gatekeeper_edit' in f.name: br.select_form(name=f.name) for c in f.controls: if c.name.endswith('_email'): br[c.name] = self.username elif c.name.endswith('_password'): br[c.name] = self.password br.submit().read() break return br extra_css = ''' body{font-family:verdana,arial,helvetica,geneva,sans-serif ;} a {text-decoration: none; color: blue;} h1 {margin: 0em; padding: 0em;} h2 {font-size: medium; font-weight: bold;} #sb-date {font-size: xx-small; color: #696969} #category {font-style: italic; font-size: small; color: black; margin: 0em; padding: 0em;} #byline {font-size: small; color: #666666} div.profiles {font-size: small; font-style: italic; color: #696969} div.profiles h2 {font-size: medium; font-style: normal; font-weight: bold; color: black} ''' feeds = [] feeds.append( (u'Finance', u'http://feeds.feedburner.com/StrategyBusiness-Finance?format=xml')) feeds.append((u'Global Perspective', u'http://feeds.feedburner.com/StrategyBusiness-GlobalPerspective?format=xml')) feeds.append( (u'Innovation', u'http://feeds.feedburner.com/StrategyBusiness-Innovation?format=xml')) feeds.append((u'Marketing And Sales', u'http://feeds.feedburner.com/StrategyBusiness-MarketingAndSales?format=xml')) feeds.append((u'Operations And Manufacturing', u'http://feeds.feedburner.com/StrategyBusiness-OperationsAndManufacturing?format=xml')) feeds.append((u'Organizations And People', u'http://feeds.feedburner.com/StrategyBusiness-OrganizationsAndPeople?format=xml')) feeds.append((u'Strategy And Leadership', u'http://feeds.feedburner.com/StrategyBusiness-StrategyAndLeadership?format=xml')) feeds.append((u'Sustainability', u'http://feeds.feedburner.com/StrategyBusiness-Sustainability?format=xml')) feeds.append((u'Auto, Airlines And Transport', u'http://feeds.feedburner.com/StrategyBusiness-AutoAirlinesAndTransport?format=xml')) feeds.append((u'Consumer Products', u'http://feeds.feedburner.com/StrategyBusiness-ConsumerProducts?format=xml')) feeds.append( (u'Energy', u'http://feeds.feedburner.com/StrategyBusiness-Energy?format=xml')) feeds.append( (u'Health Care', u'http://feeds.feedburner.com/StrategyBusiness-HealthCare?format=xml')) feeds.append( (u'Technology', u'http://feeds.feedburner.com/StrategyBusiness-Technology?format=xml')) feeds.append((u'Thought Leaders', u'http://feeds.feedburner.com/StrategyBusiness-ThoughtLeaders?format=xml')) feeds.append((u'Business Literature', u'http://feeds.feedburner.com/StrategyBusiness-BusinessLiterature?format=xml')) feeds.append((u'Recent Research', u'http://feeds.feedburner.com/StrategyBusiness-RecentResearch?format=xml')) keep_only_tags = [ dict(name='h1'), dict(attrs={'class': ['introAndByline', 'content', 'resources']}), ] remove_tags = [] remove_tags.append(dict(name='img', attrs={'class': 'content1'})) remove_tags.append( dict(name='img', attrs={'src': '/media/image/end_of_story.gif'})) remove_tags.append(dict(name='div', attrs={'class': [ 'sb-adarea468', 'GigyaShare', 'moreBlogLinks', 'clearboth', 'GigyaCommentsContainer']})) remove_tags.append(dict(name='div', attrs={'id': 'sb-paging'})) remove_tags.append(dict(name='div', attrs={'id': 'textsize'})) remove_tags.append( dict(name='div', id=lambda x: x and x.startswith('div-gpt-ad-'))) def get_article_url(self, article): url = BasicNewsRecipe.get_article_url(self, article) return url.partition('?')[0] + '?pg=all'