Move updated files from src directory to dst directory

sync-files.py

#!/usr/bin/env python
from __future__ import print_function

import os
import sys
from collections import defaultdict
from fnmatch import fnmatch
from pprint import pprint
from hashlib import sha1


def sha1sum(filename):
    return sha1(open(filename).read()).hexdigest()


def build_destinations(dst, ext):
    print("# Building destination lookup for '{0}'".format(dst), file=sys.stderr)
    d = defaultdict(set)
    for root, _, files in os.walk(dst):
        for f in files:
            if fnmatch(f, ext):
                d[f].add(root)
    return d


def duplicates(d):
    # filename is a duplicate if it appears in value more than once
    print("# Finding duplicates", file=sys.stderr)
    duplicate_keys = [k for k, v in d.items() if len(v) > 1]
    for k in duplicate_keys:
        print("{0} {1}: {2}".format(dup_count, k, d[k]), file=sys.stderr)
    return len(duplicate_keys)


def process_files(src, d, ext):
    print("# Calculating moves and deletions", file=sys.stderr)
    for filename in os.listdir(src):
        if fnmatch(filename, ext) and filename in d:
            # print("Considering {0}".format(filename), file=sys.stdout)
            src_fullpath = os.path.join(src, filename)
            dest = d[filename]
            assert len(dest) == 1
            dst_dir = list(dest)[0]
            dst_fullpath = os.path.join(dst_dir, filename)
            if sha1sum(src_fullpath) == sha1sum(dst_fullpath):
                print("rm '{0}'".format(src_fullpath))
            else:
                print("mv '{0}' '{1}/.'".format(src_fullpath, dst_dir), file=sys.stderr)


def sync_files(src, dst, ext='*.pdf'):
    src = os.path.expanduser(src)
    dst = os.path.expanduser(dst)
    # Build dictionary of destinations
    d = build_destinations(dst, ext)
    # Find out if there are any collisions on name
    assert not duplicates(d)
    # Figure out where files shgould be moved to
    process_files(src, d, ext)


if __name__ == '__main__':
    sync_files('~/Downloads', '~/Dropbox/Documents/programming')