Exemple #1
0
 def execute(self, html):
     data, page = html
     soup = BeautifulSoup(data,convertEntities=BeautifulSoup.XML_ENTITIES, fromEncoding='utf-8')
     if Retriever.is_acquisition_feed(soup):
         self.tasks = [ Entry(unicode(entry), self.link) for entry in Retriever.get_entries(soup) ]
     else: #navigation feed
         entries = Retriever.get_entries(soup)
         links = filter( lambda l: l!=None, [Retriever.get_catalog_link(entry) for entry in entries] )
         links = filter( Retriever.is_permitted_link, links)
         links = map( lambda link: make_correct_link(self.link, link), links)
         self.tasks = [ Page(link) for link in links]
     return True