沪深上证所有股票代码抓取（系列一）

这个比较简单直接上代码

from selenium import webdriver

import pymysql

import re


def main():
    urls = ['https://www.banban.cn/gupiao/list_sh.html',
            'https://www.banban.cn/gupiao/list_sz.html',
            'https://www.banban.cn/gupiao/list_cyb.html']
    
    classs = ['上证','深证','创业']
    
    driver = webdriver.Chrome()
    
    conn = pymysql.connect(
            host='127.0.0.1',
            user='root'
            ,password='123456',
            port=3306,
            db='stocks',
            charset='utf8'
            )
    
    cur = conn.cursor()
    
    
    for index in range(len(urls)):
        class_ = classs[index]
        driver.get(urls[index])
        contents = driver.find_elements_by_xpath('//*[@id="ctrlfscont"]/ul/li')
        for content in contents:
            content = content.text
            cur.execute('insert into stock values("%s","%s","%s",%s)'%(class_,content.split('(')[0],re.compile(r'\d+').findall(content)[0],0))
    cur.connection.commit()
    cur.close()
    driver.close()
    
       
    
if __name__ == '__main__':
    main()

抓取结果

下一篇 所有股票历史数据抓取