예제 #1
0
    def get_link(url):
        if url not in Fetcher.l_visit:
            Fetcher.visit.put(url)
            Fetcher.l_visit.add(url)
            Fetcher.l_unvisit.remove(url)

            try:
                response = urlopen(url)
                if 'text/html' in response.getheader('Content-Type'):
                    data = response.read().decode('latin-1')
                parser = Parser(url)
                parser.feed(data)
                data = parser.page_links()
                update(data)
            except:
                print("oi")