import hashlib
import sys
import os
import glob
from os.path import isfile, join

def sha256(fname):
	hash = hashlib.sha256()
	with open(fname, "rb") as f:
		for chunk in iter(lambda: f.read(4096), b""):
			hash.update(chunk)
	return hash.hexdigest()

def get_files(path):
	return [file for file in glob.glob(path + '/**/*.*', recursive=True)]

def main(args):
	hashes = {}
	files = get_files(args[1])
	should_delete = []

	# .sort() sorts by filename, whatsapp stores files in yyyy-mm-dd format
	# so sorting alphabetically is fine.
	files.sort()

	# Sort by modified date (when pic was taken I think)
	# files.sort(key=lambda x: os.path.getmtime(x), reverse=False)

	for file in files:
		checksum = sha256(file)

		if checksum in hashes:
			should_delete.append(file)
		else:
			hashes[checksum] = file

	print("Duplicate files:\n")

	for file in should_delete:
		print(file)

	yesno = input("Delete these {} files? y/n\n".format(len(should_delete)))

	if yesno.lower().startswith('y'):
		for file in should_delete:
			print("Deleting '{}'".format(file))
			os.remove(file)

		print('Done.')

	print('Exiting.')
	exit(0)

if __name__ == "__main__":
	if len(sys.argv) != 2:
		print('Invalid argument')
		print('Usage: python3 {} directory'.format(__file__))
		exit(1)

	main(sys.argv)