import re

import requests

from selenium import webdriver

from selenium.webdriver.common.keys import Keys

from selenium.webdriver.common.by import By

import pyperclip

import time

from bs4 import BeautifulSoup

import requests

from requests import Session, request, post, get

import os

import prefect

from prefect import task, Flow

import mysql.connector

from requests.cookies import cookiejar_from_dict

from urllib.parse import urlencode

# log_in to naver

@task(log_stdout=True)

def login_naver(driver_login, naver_id, naver_pw):

    driver_login.get('https://nid.naver.com/nidlogin.login')

    time.sleep(0.5)

    tag_id = driver.find_element(By.NAME, 'id')

    tag_id.clear()

    tag_pw = driver.find_element(By.NAME, 'pw')

    tag_pw.clear()

    time.sleep(1)

    tag_id.click()

    pyperclip.copy(naver_id)

    tag_id.send_keys(naver_id)

    time.sleep(0.5)

    tag_pw.click()

    pyperclip.copy(naver_pw)

    tag_pw.send_keys(naver_pw)

    time.sleep(0.5)

    login_btn = driver.find_element(By.ID, 'log.login')

    login_btn.click()

    time.sleep(2)

# scraping from naver mail

@task(log_stdout=True)

def scrap_naver_mail(url, url_specific):

    logger = prefect.context.get("logger")

    session = requests.Session()

    logger.info('going in to naver_mail')

    res = session.get('https://mail.naver.com/' target="_blank">https://mail.naver.com/')

    # naver_cookie = {_.split('=')[0]: _.split('=')[1]

    #                 for _ in re.split('[; ]{2}', 'NNB=NR5HUBJVOLHGE; nid_inf=107699324; NID_AUT=gv7DKVpnQvFYKG49zB2fFP1uHwZxvWiHWHyDQhoWFxj5mHOvzCemyQdnpcmnAn2I; NID_JKL=xF4oI/diivVbteBpSa7v6jOMakQvddayVnkpweF1HuE=; NMUSER=uXYlKqEwaqbsFq2rFxE/1Hk5prG9W4bwKAU/aqurFxbraxbmFqvsFxuZKAnmaqns3xndFogdFqusHou9a9vsxonOaxRpa9vs6xRpaqRVaqns1rejL9Us6xRVaqnD16lvpB2RFVl5WLl5MBp0bSloWrdnaAvmKARqp6FTW43CbNvR16lvpB2RFVl5WLl5MBp0bSloWrdnaAvmKAn=; NID_SES=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')}

    # mail_text_response = get(url='https://mail.naver.com', cookies=naver_cookie)

    # logger.info(f'got mail info{mail_text_response.text}')

    # mail_cookies = {_.split('=')[0]: _.split('=')[1] for _ in re.split('[;]{2}',

    #                                   'NID_SES=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; path=/; domain=.naver.com')}

    # for k, v in naver_cookie.items():

    #     session.cookies.set(k, v)

    # mail_params = dict()

    # for _ in url_specific.split('&'):

    #     mail_params[_.split('=')[0]] = _.split('=')[1]

    # resp = post(url, data=mail_params, cookies=mail_cookies)

    # resp.json()

    # logger.info(resp.text)

    # for _ in session.cookies.get_dict():

    #     logger.info(f'key of naver_cookies: {_}')

    # logger.info(naver_cookie)

    # for _ in resp.json()['mailData']:

    #     print('{}|{}|{}'.format(_['subject'], _['receivedTime']))

    # return resp

@task(log_stdout=True)

def store_naver_mail(records):

    logger = prefect.context.get('logger')

    logger.info('connecting to the database')

    connection = mysql.connector.connect(

        host=os.getenv('DATABASE_HOST', 'database'),

        user=os.getenv('DATABASE_USER', 'root'),

        password=os.getenv('DATABASE_PASSWORD', 'mariadb'),

        database=os.getenv('DATABASE_SCHEMA', 'practice'),

    )

    cursor = connection.cursor()

    sql = 'insert into scraping_mail(mail_title, contents, received_date, classify) values(%s, %s, %d, %s)'

    logger.info('inserting mail scrap data')

    cursor.executemany(sql, records)

    connection.commit()

with Flow("Naver mail scraping") as flow:

    driver = webdriver.Chrome('\chromedriver.exe')

    naver_id = os.getenv('NAVER_ID')

    naver_pw = os.getenv('NAVER_PW')

    login_naver(naver_id, naver_pw)

    url_common = \

    url_common_specific = 

    url_spam = 'https://mail.naver.com/' target="_blank">https://mail.naver.com/json/list/?'

    url_spam_specific = 

    mail_data = scrap_naver_mail(url_common, url_common_specific) + scrap_naver_mail(url_spam, url_spam_specific)

    store_naver_mail(mail_data)

flow.run()

(URL은 정보때문에 뺐습니다)

scrap_naver_mail 이거때문에 안돌아가는거 같은데 혹시 교정해주실 굇수분 있나여