files_md5/main.py

import os
import win32con
import win32api
from getFiles import get_file_list, remove_empty_folder
from getMd5 import get_file_md5
from db import *
from time import time as t

disk = ("Data_sjry"
        ""
        ""
        "")
cate = "sjry"
path = r"F:\Temp\sjry\2023-10"


def main_process():
    _exist_file_list = []

    if disk == "" or cate == "" or path == "":
        print("检查参数")
        return

    print(f"开始获取文件列表：{t()}")
    file_list = get_file_list(path)
    print(f"结束获取文件列表：{t()}")

    unchecked_list = []
    cate_file_list = get_path_by_cate(cate)

    total_len = len(file_list)
    for (index, file) in enumerate(file_list):
        if file[2:] in cate_file_list:
            print(f"{index}/{total_len}文件已经存在: {file}")
        else:
            print(f"{index}/{total_len}未处理文件: {file}")
            unchecked_list.append(file)

    file_list_len = len(unchecked_list)
    for (i, f) in enumerate(unchecked_list):
        print(f"({i + 1}/{file_list_len}) - 开始处理: {f} - {t()}")
        t_start = t()
        md5 = get_file_md5(f)
        print(f"完成MD5计算耗时：{t() - t_start}")
        if exist := is_exist(md5):
            _exist_file_list.append(f)
            print(f"重复文件！！！md5为({md5})的文件已经存在，文件路径:{exist['path']} -> {f}")
            continue
        insert_data((cate, disk, f, md5))
        print(f"文件的md5({md5})成功插入: {f}")
    return _exist_file_list


def handle_exist(_exist_file_list):
    if len(_exist_file_list) > 0:
        with open('dup.txt', 'w', encoding='utf-8') as dup_file:
            for ef in _exist_file_list:
                dup_file.write(ef + "\r\n")
                win32api.SetFileAttributes(ef, win32con.FILE_ATTRIBUTE_NORMAL)
                os.remove(ef)
                print(f"删除： {ef}")
            dup_file.close()
    else:
        print("没有找到重复文件")


if __name__ == "__main__":
    exist_file_list = main_process()
    handle_exist(exist_file_list)
    remove_empty_folder(path)