Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- import datetime
- from calibre.ebooks.BeautifulSoup import Tag
- class AdvancedUserRecipe1286242553(BasicNewsRecipe):
- title = u'CACM'
- oldest_article = 7
- max_articles_per_feed = 100
- needs_subscription = True
- feeds = [(u'CACM', u'http://cacm.acm.org/magazine.rss')]
- no_stylesheets = True
- remove_tags = [
- dict(name='div', attrs={'class':['FeatureBox', 'ArticleComments', 'SideColumn', \
- 'LeftColumn', 'RightColumn', 'SiteSearch', 'MainNavBar','more', 'SubMenu', 'inner']})
- ]
- cover_url_pattern = 'http://cacm.acm.org/magazines/%d/%d'
- def get_browser(self):
- br = BasicNewsRecipe.get_browser()
- if self.username is not None and self.password is not None:
- br.open('https://cacm.acm.org/login')
- br.select_form(nr=1)
- br['current_member[user]'] = self.username
- br['current_member[passwd]'] = self.password
- br.submit()
- return br
- def get_cover_url(self):
- now = datetime.datetime.now()
- cover_url = None
- soup = self.index_to_soup(self.cover_url_pattern % (now.year, now.month))
- print self.cover_url_pattern % (now.year, now.month)
- cover_item = soup.find('img',attrs={'alt':'magazine cover image'})
- if cover_item:
- cover_url = cover_item['src']
- return cover_url
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement