Revision of recursive_checksum.py

1 file changed, 3 insertions, 1 deletion

recursive_checksum.py

			@@ -45,4 +45,6 @@ def main():
45	45
46	46		# Pretty prints json to output file
47	47		with open(os.path.join(rootdir, CHECKSUMS), "w") as f:
48		-	json.dump(output, f, indent=4)
	48	+	json.dump(output, f, indent=4)
	49	+
	50	+	main()

1 file changed, 48 insertions

recursive_checksum.py(file created)

		@@ -0,0 +1,48 @@
1	+	#!/usr/bin/env python
2	+
3	+	import os
4	+	import json
5	+	from hashlib import md5
6	+
7	+	rootdir = os.getcwd()
8	+
9	+	CHECKSUMS = "checksums.json"
10	+
11	+	try:
12	+	with open(CHECKSUMS) as f:
13	+	output = json.load(f)
14	+	except:
15	+	output = {}
16	+
17	+	def md5sum(filename):
18	+	""" Opens a file and progressively generates an MD5 hash
19	+	from its contents, avoiding loading the complete
20	+	contents into ram at once
21	+	http://stackoverflow.com/a/24847608 """
22	+	hash = md5()
23	+	with open(filename, "rb") as f:
24	+	for chunk in iter(lambda: f.read(128 * hash.block_size), b""):
25	+	hash.update(chunk)
26	+	return hash.hexdigest()
27	+
28	+	def main():
29	+	""" Iterate over files in subdirectories of current dir, use
30	+	md5sum() to generate checksums for each, append to a dict
31	+	for inclusion in checksums.json (name configurable) or
32	+	checks an existing checksums.json for conflicts """
33	+	for folder, subs, files in os.walk(rootdir):
34	+	for filename in files:
35	+	if filename == CHECKSUMS:
36	+	continue
37	+	fn = os.path.join(folder, filename)[len(rootdir)+1:]
38	+	if fn in output:
39	+	if output[fn] != md5sum(os.path.join(folder, filename)):
40	+	# replace with code to redownload given file?
41	+	print("Mismatch on {}".format(fn))
42	+	else:
43	+	output[os.path.join(folder, filename)[len(rootdir)+1:]] = \
44	+	md5sum(os.path.join(folder, filename))
45	+
46	+	# Pretty prints json to output file
47	+	with open(os.path.join(rootdir, CHECKSUMS), "w") as f:
48	+	json.dump(output, f, indent=4)

Newer Older