深度学习，递归地移动某目录下所有文件到别处，用于划分测试集

dl_move_files_to_test.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
#
# 递归移动某个文件夹下的所有文件到另一个文件，作为深度学习划分测试集
# 2022-06-18
#
# 参数：
# python3 main.py -s <SRC-DIR> -d <DEST-DIR> -p <MOVE-PERCENT>
# 例子：
# python3 main.py E:\image-src E:\image-dest -p 90

import shutil
import os
import argparse
import sys
import random

g_dirty_run = False

# src_dir源目录，如E:\image-src
# dest_dir目标目录，如E:\image-dest
# percent需要移动文件百分比，对文件进行移动，对目录递归创建目录
def move_files(src_dir, dest_dir, percent):
	if not os.path.exists(dest_dir):
		os.makedirs(dest_dir)
	
	listdir_results = os.listdir(src_dir)
	src_filenames = [] # 本目录下的文件
	for listdir in listdir_results:
		src_filename = os.path.join(src_dir, listdir)
		if os.path.isfile(src_filename):
			src_filenames.append(src_filename)
		elif os.path.isdir(src_filename):
			dest_filename = os.path.join(dest_dir, listdir)
			move_files(src_filename, dest_filename, percent)
	
	src_file_count = len(src_filenames)
	if 0 == src_file_count:
		return

	move_count = int(src_file_count * percent)
	print('src dir={0}, {1}/{2} files to be moved'.format(src_dir, move_count, src_file_count))
	if g_dirty_run is False and move_count > 0:
		random.shuffle(src_filenames)
		#print(src_filenames)
		
		src_filenames = src_filenames[:move_count]
		for src_filename in src_filenames:
			shutil.move(src_filename, dest_dir)

if __name__ == "__main__":
	script_file = os.path.basename(sys.argv[0])
	parser = argparse.ArgumentParser(prog=script_file,
	                                 formatter_class=lambda prog: argparse.HelpFormatter(prog,max_help_position=80),
	                                 description='A simple file mover to split dataset to test')
	parser.add_argument('-s','--src-dir', help='source dir', required=True)
	parser.add_argument('-d','--dest-dir', help='dest dir', required=True)
	parser.add_argument('-p','--move-percent', help='move percent, range is (1,100]', required=True)
	parser.add_argument("--dirty-run", help="run for brief preview, would not move any files", action="store_true")
	
	args = vars(parser.parse_args())
	
	src_dir, dest_dir, move_percent, g_dirty_run = args['src_dir'], args['dest_dir'], float(args['move_percent']), args['dirty_run']

	if g_dirty_run:
		print('Dirty run mode, no files will be moved, but new folders will be created')

	if move_percent > 0 and move_percent <= 100:
		move_files(src_dir, dest_dir, move_percent / 100)
	else:
		print('Error, move percent must in range (1,100]')
		sys.exit(-1)