telegram_crawler/clean.py

import os
root_path = os.path.abspath(os.path.dirname(__file__)).split('telegram-crawler')[0] + "telegram-crawler"
import time
from loguru import logger
import shutil

"""
定时脚本删除目标目录下文件
暂定每日执行一次 删除一天前的文件
0 0 * * * cd /opt/crawl/telegram/telegram-crawler && /root/anaconda3/envs/python3.8/bin/python clean.py > /dev/null 2>&1
cd /opt/crawl/telegram/telegram-crawler && /root/anaconda3/envs/python3.8/bin/python clean.py > /dev/null 2>&1
"""


def release_resources(days=3):
    """
    定时删除 资源文件
    :param days:
    :return:
    """
    resources_dir = 'tg_module/resources'
    profile_dir = 'tg_module/profile'

    files = []
    # 获取目录下所有文件
    resources_files = [os.path.join(resources_dir, f) for f in os.listdir(resources_dir)]
    profile_files = [os.path.join(profile_dir, f) for f in os.listdir(profile_dir)]
    files.extend(resources_files)
    files.extend(profile_files)
    now = time.time()
    delay = days * 24 * 60 * 60
    delay = days * 60 * 60

    logger.info(f"获取数量共：{len(files)}；resources: {len(resources_files)}, profile: {len(profile_files)}")
    limited_time = now - delay

    # 按照创建时间排序
    files = sorted(files, key=os.path.getctime)
    count = 0
    for i in files:
        create_time = os.path.getctime(i)
        if create_time < limited_time:  # 需要删除
            if os.path.isfile(i) or os.path.islink(i):  # 如果是文件则直接删除
                os.remove(i)
                count += 1
            else:  # 删除文件夹及其所有内容
                shutil.rmtree(i)
                logger.debug(f"删除了一个文件夹: {i}")
        else:
            break

    logger.info(f"删除文件数量：{count}")


if __name__ == '__main__':
    release_resources()