Sifting files generated by photorec using md5deep and tiny pieces of python.

haystack.py

#!/usr/bin/env python3

import csv
import os.path
import os
import shutil

'''Haystack.py - sifting through photorec recovered data

Concept of using photorec, md5deep, NSRL "known" file dataset and personal 
"known" file datasets (generated with md5deep) to isolate or lessen the heap
generated by poor data management practices and recovered by photorec.

This code is terrible, do not use. It was written with one hand perhaps using 
other lazy methods.
'''


def read_nsrl_results(known_files, found_files, ignore_files):
    with open(known_files, errors='ignore') as f:
        linereader = csv.reader(f, delimiter='\n')

        for line in linereader:
            # import pdb; pdb.set_trace()
            if line[0].upper() in found_files:
                with open(ignore_files, 'a') as wf:
                    wf.write("{} {}".format(line[0], found_files[line[0]]))


def md5read(infile):
    with open(infile, errors='ignore') as f:
        linereader = csv.reader(f, delimiter=' ')

        try:
            return {line[0].upper(): line[1] for line in linereader}
        except IndexError as e:
            return {line[0].upper(): None for line in linereader}


def moves_files(move_files, directory_name):
    source_dir = '/Volumes/UNTITLED/photorec'
    source_files = md5read('thing_o_hash1.txt')

    move_files = md5read(move_files)
    destination_dir = os.path.join(source_dir, directory_name)

    common_file_set = set(source_files.keys()) & set(move_files.keys())

    # import pdb; pdb.set_trace()
    if common_file_set:
        try:
            os.mkdir(destination_dir)
        except FileExistsError as e:
            print(e)

        for f in common_file_set:
            source_file = os.path.join(source_dir, source_files[f])
            try:
                shutil.move(source_file, destination_dir)
            except shutil.Error as e:
                print(e)
            except FileNotFoundError as e:
                print(e)


if __name__ == "__main__":
    found_files = md5read('thing_o_hash.txt')

    read_nsrl_results('../known/NSRLFileMD5Only.txt', found_files, 'nsrl.txt')

    # find common files
    moves_files('dropbox_files.txt', 'dropbox')
    moves_files('applications.txt', 'Applications')
    moves_files('Library.txt', 'Library')
    moves_files('System.txt', 'System')
    moves_files('nsrl.txt', 'NSRL')
    moves_files('external.txt', 'external')