crawler_gzh/main/run_gzh.py

# -*- coding: utf-8 -*-
# @Author: wangkun
# @Time: 2022/8/17
import datetime
import os
import sys
import time

# import time
sys.path.append(os.getcwd())
from main.common import Common
# from main.gzh import GZH
from main.gzh_new import GZH


class Main:
    # @classmethod
    # def main(cls, env):
    #     while True:
    #         if 20 >= datetime.datetime.now().hour >= 8:
    #             # 获取所有用户的公众号文章信息
    #             Common.logger('gzh').info("开始抓取公众号任务\n")
    #             GZH.search_user_by_word('gzh')
    #             GZH.get_all_gzh('gzh', env)
    #             # 清除日志
    #             Common.del_logs('gzh')
    #             # 翻页初始化
    #             GZH.gzh_count = []
    #             Common.logger('gzh').info("公众号抓取任务结束，休眠{}小时\n", 24-datetime.datetime.now().hour)
    #             time.sleep(3600 * (24-datetime.datetime.now().hour))
    #         else:
    #             pass

    @classmethod
    def main_new(cls, env):
        while True:
            if 19 >= datetime.datetime.now().hour >= 8:
                # 获取所有用户的公众号文章信息
                Common.logger('gzh').info("开始抓取公众号任务\n")
                GZH.search_user_by_word('gzh', env)
                # 清除日志
                Common.del_logs('gzh')
                # 翻页初始化
                GZH.gzh_count = []
                Common.logger('gzh').info("公众号抓取任务结束，休眠 1 小时\n")
                time.sleep(600)
                # Common.logger('gzh').info("公众号抓取任务结束，休眠{}小时\n", 24-datetime.datetime.now().hour)
                # time.sleep(3600 * (24-datetime.datetime.now().hour))
            else:
                pass


if __name__ == "__main__":
    Main.main_new('prod')