from functools import partial
from multiprocessing import Pool as ThreadPool
pageurls=[]
if maxpage:
for page in range(2, maxpage + 1):
pageurl = url + '&page=%s#mSearch' % page
pageurls.append(pageurl)
with ThreadPool() as pool:
results = pool.map(partial(get_one_page,level1=level1,level2=level2,level3=level3,name=name),pageurls)
def get_one_page(pageurl,level1,level2,level3,name):
print("正在爬取的url", pageurl)
p_list = parser_md_index(pageurl)
for p in p_list:
product_url, img, p_br, p_name, p_price, old_price = p['product_url'], p['img'], p['p_br'], p[
'p_name'], p['p_price'], p['old_price']
product_with(level1, level2, level3, name, p_br, p_name, p_price, old_price, product_url, img)
name = re.sub('\"', '\'', name)
p_name = re.sub('\"', '\'', p_name)
sql = 'INSERT INTO modern_product (level1,level2,level3,level4,p_br,p_name,p_price,old_price,product_url,img) VALUES ("{}","{}","{}","{}","{}","{}","{}","{}","{}","{}")'.format(
level1, level2, level3, name, p_br, p_name, p_price, old_price, product_url, img)
cur.execute(sql)
conn.commit()from functools import partial
from multipr