#!/usr/bin/env python
# coding:UTF-8
# -*- coding:utf-8 -*-

#去除重复和图片大小小于2MB的图片

import os,re,hashlib,platform
from colorama import init,Fore,Back,Style

init(autoreset=True)

dirPath = input("输入图片所在目录(默认当前目录): ")
if dirPath == '':
    dirPath = os.getcwd()

if platform.system().lower() == 'linux':
    pathsign = '/'
elif platform.system().lower() == 'windows':
    pathsign = '\\'

picPath = dirPath + pathsign

piclist = filter(lambda x:re.match(r'\S+.(jpg|png)$',x),os.listdir(picPath))

md5code_list = []
delfile_list = []
for i in piclist:
    if os.path.isfile(picPath+i):
        #fsize = os.path.getsize(picPath+i)
        #print('图片[ '+i+' ]的大小容量为: '+str(fsize))
        #if fsize < 2097152:
        #    delfile_list.append(i)
        #    print('图片[ '+Fore.RED+i+Fore.RESET+' ]尺寸小于2MB,将被删除!')
        with open(picPath+pathsign+i,'rb') as f:
            data = f.read()
            md5code = hashlib.md5(data).hexdigest()
            print('图片[ '+Fore.BLUE+i+Fore.RESET+' ]的md5值为: '+Fore.GREEN+md5code)
            if md5code not in md5code_list:
                md5code_list.append(md5code)
            else:
                delfile_list.append(i)
                print('图片[ '+Fore.RED+i+Fore.RESET+' ]重复,将被删除!')

del_count = len(delfile_list)
for i in set(delfile_list):   #如果判断图片大小也加入删除列表,可能有重复元素,用set过滤
    os.remove(picPath+pathsign+i)
print('删除图片数量为: '+str(del_count))