crawler_kanyikan/main/run.py

# -*- coding: utf-8 -*-
# @Author: wangkun
# @Time: 2022/4/18
import datetime
import os
import random
import sys
import time
sys.path.append(os.getcwd())
from main.download_sendtime import DownloadSendtime
from main.download_up import DownloadUp
from main.common import Common
from main.download_play import DownloadPlay
from main.get_feeds import get_feeds


class Main:

    @classmethod
    def download_play_job(cls):
        """
        播放量脚本任务
        """
        while True:
            play_now = datetime.datetime.now()
            # 指定时间结束抓取视频
            if play_now.hour == 0:
                Common.crawler_log().info("结束抓取播放量视频\n")
                time.sleep(3)
                break
            else:
                # 获取视频信息，并下载
                get_feeds()
                DownloadPlay.download_play_video("prod")
                # 请求随机间隔时间
                time.sleep(random.randint(31, 40))

        # 删除 charles 缓存文件
        Common.del_charles_files()
        # 删除多余日志
        Common.del_logs()
        # 统计累计下载数量
        Common.kanyikan_download_count()

    @classmethod
    def download_up_job(cls):
        """
        上升榜脚本任务
        """
        while True:
            up_now = datetime.datetime.now()
            # 指定时间结束抓取视频
            if up_now.hour > 19:
                Common.crawler_log().info("结束抓取上升榜视频\n")
                time.sleep(3)
                break
            else:
                # 获取视频信息，并下载
                get_feeds()
                DownloadUp.download_up_video("prod")
                # 请求随机间隔时间
                time.sleep(random.randint(31, 40))

        # 删除 charles 缓存文件
        Common.del_charles_files()
        # 删除多余日志
        Common.del_logs()
        # 统计累计下载数量
        Common.kanyikan_download_count()

    @classmethod
    def download_sendtime_job(cls):
        """
        发布时间榜脚本任务
        """
        while True:
            sendtime_now = datetime.datetime.now()
            # 指定时间结束抓取视频
            if sendtime_now.hour > 9:
                Common.crawler_log().info("结束抓取3天榜单视频\n")
                time.sleep(3)
                break
            else:
                # 获取视频信息，并下载
                get_feeds()
                DownloadSendtime.download_sendtime_video("prod")
                # 请求随机间隔时间
                time.sleep(random.randint(31, 40))

        # 删除 charles 缓存文件
        Common.del_charles_files()
        # 删除多余日志
        Common.del_logs()
        # 统计累计下载数量
        Common.kanyikan_download_count()

    @classmethod
    def main(cls):
        """
        - 1.凌晨0点-10点      3日内播放大于2万 爬取
        - 2.早上10点-20点     内容上升榜 爬取
        - 3.晚上20点-24点     15万播放爬取
        """
        while True:
            while True:
                main_now = datetime.datetime.now()
                if 9 >= main_now.hour >= 0:
                    Common.crawler_log().info("开始抓取3天榜单视频\n")
                    time.sleep(1)
                    cls.download_sendtime_job()
                elif 19 >= main_now.hour >= 10:
                    Common.crawler_log().info("开始抓取上升榜视频\n")
                    time.sleep(1)
                    cls.download_up_job()
                elif 24 >= main_now.hour >= 20:
                    Common.crawler_log().info("开始抓取播放量视频\n")
                    time.sleep(1)
                    cls.download_play_job()
                else:
                    Common.crawler_log().info("结束今天抓取及上传任务\n")
                    time.sleep(60)
                    break


if __name__ == "__main__":
    main = Main()
    main.main()