import os
import hashlib

def search_file(dir_name = ''):
	' 扫描目录、删除重复文件 '
	file_path = []
	for path in os.walk(dir_name):
		for file in path[2]:
			file_path.append(path[0] + '\\' + file)
	return file_path

def generate_md5(content):
	' 生成MD5 '
	md5 = hashlib.md5(content)
	return md5.hexdigest()

def check_del(file_lst = []):
	md5_lst = []
	for file in file_lst:
		print('Check', file)
		f = open(file, 'rb')
		md5 = generate_md5(f.read(1024))
		f.close()
		if md5 in md5_lst:
			os.remove(file)
			print('Delete!')
			continue
		md5_lst.append(md5)


if __name__ == '__main__':
	check_del(search_file(r' Target folder '))