# coding=utf-8 ''' Created on 2016年3月11日 @author: ChenHao ''' from bs4 import BeautifulSoup fin = open("../spider_download/initPage/1.html", "r") html_cont = fin.read() soup = BeautifulSoup(html_cont, 'html.parser', from_encoding='utf-8') print(soup) new_product_urls = set() table = soup.find("table", id="ctl00_ContentMain_SearchResultsGrid_grid")