def get_links(url, html): # ?????????
d = pq(html).make_links_absolute(base_url=url)
key_node = d('body a')
urls = set(d(node).attr('href') for node in key_node)
non_none = filter(partial(is_not, None), urls)
links = [url for url in non_none if url.startswith(base_url) and url.endswith('.shtml')]
return links
评论列表
文章目录