getHash

#!/usr/bin/env python
#
# Hash some files, has been setup to ignore certain files that I don't care about
# new_hash, pre_hash take a path to log files if you're so inclined
# Will deposit the .md5 files in the same directory as the files you're hashing
#
# By default this will not rehash any files that have an associated .md5 file. This can be changed
# by removing the "elif os.path.exists(tmp + ".md5"): block of code. That block of code could also
# be a good location to put new code in that compares the old hash value with the new one to track
# data integrity changes (say after restoring a RAID volume or moving large amounts of data to a new
# drive, etc...)

# Sean Grimes
# 01.15.16

import os
import formatter
import hashlib
import sys
from os import walk

def md5(fname):
	hash = hashlib.md5()
	with open(fname, "rb") as f:
		for chunk in iter(lambda: f.read(4096), b""):
			hash.update(chunk)
	return hash.hexdigest()


file_directory = os.getcwd()
file_directory = os.path.abspath(file_directory)
paths_to_hash = []
path = []
new_hash = "/Users/PATH/TO/STORE/new_hashes.txt"
pre_hash = "/Users/PATH/TO/STOREold_hashes.txt"
new_file = open(pre_hash, 'w')
new_file.write("\n")
new_file.close()

for root, dirs, filenames in os.walk(file_directory, topdown=False):
	for name in filenames:
		tmp = os.path.join(root, name)
		f = tmp+".md5"
		print tmp
		
		# Check if we already have a hash for this file
		if tmp.endswith('.md5'):
			continue
		elif ".DS_Store" in tmp:
			continue
		elif ".Spotlight-V100" in tmp:
			continue
		elif ".Trashes" in tmp:
			continue
		elif ".fseventsd" in tmp:
			continue
		elif os.path.exists(tmp + ".md5"):
			pre = open(pre_hash, 'a')
			str = tmp + " already had a hash."
			pre.write(tmp),
			pre.write(" already had a hash.\n")
			continue
		else:
			new2 = open(new_hash, 'a')
			new2.write(tmp),
			new2.write(" has now been hashed.\n")
			file = open(f, 'w')
			file.write(md5(tmp))