pyppeteer用法总结

pypeteer基于asyncio异步模块

基本使用

导入模块

from pyppeteer import launch

实例化一个浏览器

browser = await launch() # 实例化一个浏览器对象
# browser = await launch({'headless':False,'dumpio':True,'ignoreDefaultArgs':['--enable-automation']})

实例化一个页面

page = await self.browser.newPage()

页面打开网站

await page.goto('https://www.baidu.com/',{'timeout':100*1000})

定位页面元素

ele = (await page.xpath('//div[text()="免密码登录"]'))[0]
await ele.click() # 点击页面元素

输入框中输入内容

await page.type('input[name=username]',uname,{'delay':100})

获取元素属性值

check = await page.Jeval('img[alt=图形验证码]','node => node.getAttribute("src")')

获取cookies

cookies = await page.evaluate('document.cookie',force_expr=True)

示例

import asyncio
from pyppeteer import launch
import traceback

class Spider:
    def __init__(self):
        self.browser = None
        self.regist_url = 'https://www.zhihu.com/signin?next=%2F'
        


    def input_time_random(self):
        return randint(100, 151)

    async def registered(self,unames:tuple):
        if not self.browser:
            self.browser = await launch({'headless':False,'dumpio':True,'ignoreDefaultArgs':['--enable-automation']})
        
        for uname in unames:
            try:
                page = await self.browser.newPage()
                

                await page.goto(self.regist_url,{'timeout':100*1000})
                # 选择免密码登录
                await (await page.xpath('//div[text()="免密码登录"]'))[0].click()
                await asyncio.sleep(1)

                # 输入用户名
                await page.type('input[name=username]',uname,{'delay': self.input_time_random() - 50})
                check = await page.Jeval('img[alt=图形验证码]','node => node.getAttribute("src")')
                print('check',check)
                # # check = await page.evaluate('img[alt=图形验证码].getAttribute("src")', force_expr=True)
                if 'null' not in check:
                     pass # 打码平台
                    
                # 点击获取验证码
                await (await page.xpath('//button[text()="获取短信验证码"]'))[0].click()
                await asyncio.sleep(60)
                cookies = await page.evaluate('document.cookie',force_expr=True)
                print('cookies',cookies)
            except:
                traceback.print_exc()


    async def run(self):
        self.browser = await launch({'headless':False,'dumpio':True,'ignoreDefaultArgs':['--enable-automation']})
        
        tasks = []
        for i in range(1):
            tasks.append(asyncio.create_task(self.registered(('XXXXX',))))


        await asyncio.wait(tasks)

if __name__=="__main__":
    spider = Spider()

    asyncio.run(spider.run())