piaoquan_crawler/gongzhonghao/gongzhonghao_main/run_gongzhonghao5_author_scheduling.py

# -*- coding: utf-8 -*-
# @Author: wangkun
# @Time: 2023/4/23
import argparse
import os
import sys
import time
sys.path.append(os.getcwd())
from common.public import task_fun
from common.common import Common
from common.scheduling_db import MysqlHelper
from gongzhonghao.gongzhonghao_author.gongzhonghao5_author import GongzhonghaoAuthor5


def main(log_type, crawler, task, oss_endpoint, env):
    while True:
        cmd = 'ps -ef | grep "run_gongzhonghao5_author_scheduling.py" | grep -v "grep"'
        result = os.popen(cmd).read()
        if len(result) != 0:
            Common.logger(log_type, crawler).info("公众号_5抓取未完成，无需启动新进程")
            time.sleep(1)
        else:
            break
    task_dict = task_fun(task)['task_dict']
    rule_dict = task_fun(task)['rule_dict']
    task_id = task_dict['task_id']
    select_user_sql = f"""select * from crawler_user_v3 where task_id={task_id}"""
    user_list = MysqlHelper.get_values(log_type, crawler, select_user_sql, env, action="")
    Common.logger(log_type, crawler).info(f"调度任务:\n{task_dict}")
    Common.logger(log_type, crawler).info(f"抓取规则:\n{rule_dict}")
    Common.logger(log_type, crawler).info(f"用户列表:\n{user_list}")
    Common.logger(log_type, crawler).info(f'开始抓取 {task_dict["task_name"]} 定向榜\n')
    GongzhonghaoAuthor5.get_all_videos(log_type=log_type,
                                        crawler=crawler,
                                        user_list=user_list,
                                        rule_dict=rule_dict,
                                        oss_endpoint=oss_endpoint,
                                        env=env)
    Common.del_logs(log_type, crawler)
    Common.logger(log_type, crawler).info('抓取完一轮\n')


if __name__ == "__main__":
    parser = argparse.ArgumentParser()  ## 新建参数解释器对象
    parser.add_argument('--log_type', type=str)  ## 添加参数,注明参数类型
    parser.add_argument('--crawler')  ## 添加参数
    parser.add_argument('--task')  ## 添加参数
    parser.add_argument('--oss_endpoint')  ## 添加参数
    parser.add_argument('--env')  ## 添加参数
    args = parser.parse_args()  ### 参数赋值，也可以通过终端赋值
    main(log_type=args.log_type,
         crawler=args.crawler,
         task=args.task,
         oss_endpoint=args.oss_endpoint,
         env=args.env)