优化重复文件处理

getFiles.py

@@ -11,6 +11,27 @@ def get_file_list(_path):
     return file_list
 
 
+def get_empty_folder_list(path):
+    folder_list = []
+    for f in os.walk(path):
+        if len(f[2]) == 0 and len(f[1]) == 0:
+            folder_list.append(f[0])
+    return folder_list
+
+
+def remove_empty_folder(path):
+    del_folder_list = []
+    del_folder_list.extend(get_empty_folder_list(path))
+
+    while len(del_folder_list) > 0:
+        for f in del_folder_list:
+            if os.path.exists(f):
+                print(f"删除空文件夹 {f}")
+                os.rmdir(f)
+        del_folder_list = []
+        del_folder_list.extend(get_empty_folder_list(path))
+
+
 if __name__ == "__main__":
     result = get_file_list(r"E:\Backup\Tools\Normal\浏览器")
     for r in result:

main.py

@@ -1,14 +1,20 @@
-from getFiles import get_file_list
+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():
-    disk = "Myxx_Backup_16T_3"
-    cate = "leshe"
-    path = r"X:\leshe"
 
+def main_process():
     _exist_file_list = []
 
     if disk == "" or cate == "" or path == "":
@@ -32,10 +38,10 @@ def main_process():
 
     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()}")
+        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}")
+        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}")
@@ -45,13 +51,20 @@ def main_process():
     return _exist_file_list
 
 
-if __name__ == "__main__":
+def handle_exist(_exist_file_list):
-    exist_file_list = main_process()
+    if len(_exist_file_list) > 0:
-    if len(exist_file_list) > 0:
         with open('dup.txt', 'w', encoding='utf-8') as dup_file:
-            for ef in exist_file_list:
+            for ef in _exist_file_list:
-                print(ef)
+                dup_file.write(ef + "\r\n")
-                dup_file.write(ef)
+                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)