dkremer

#!/usr/bin/env python3

import hashlib
import os
import sys

if __name__='__main__':
    print("this script will remove some files on your computer")
    print("are you SURE you want to continue ? y for yes and n for no")
    response = input()
    if response != 'y' :
        print("Ok, now exiting")
        sys.exit(1)
    else :
        print("Ok, now searching for duplicata")
        main()
        sys.exit(0)

def main():

    file_tree = os.walk(os.curdir)  # path iterator
    checksum_list = list()  # list files with including checksum

    for a_folder_tuple in file_tree : #  doing the checksum list

        list_files_folder = a_folder_tuple[2]
        current_dir = a_folder_tuple[0]
        print("check_sim is now checking the folder",current_dir)

        for a_pic in list_files_folder :

            string_filename = current_dir+os.sep+a_pic
            fd_pic = open(string_filename,'rb')  #read only, binary mode
            input_buffer = fd_pic.read()  # checksum on the whole file
            fd_pic.close()  # throw this file descriptor and close the file
            my_checksum = hashlib.md5(input_buffer)  # compute checksum
            #keep filename and checksum in this list
            checksum_list.append( ( my_checksum.hexdigest(), string_filename ) )

    checksum_list.sort( key = lambda x : x[0] )  # alphabetically sorted by filename

    same_files = [ ]  # the same files stored

    for i in range(len(checksum_list)-1) :  # checking for duplicate files
        same_files.append( checksum_list[i] )  # and deleting them

        # Match different files ? Remove the identic files previous
        if checksum_list[i+1][0] != checksum_list[i][0] :
            if len(same_files) > 1 :
                files_to_delete = [ k[1] for k in same_files ]
                print("This files are same",files_to_delete)
                for k in range(len(files_to_delete)-1):
                    print("deleting",files_to_delete[k])
                    os.remove(files_to_delete[k])
                print("keeping",files_to_delete[k+1])
            same_files = [ ]  # erase duplicata list