ritornelo.py

import os
import time
import hashlib
import csv

#	Cache Handler
cache_file = "/tmp/cache_md5"

def load_cache():
	cache = {}
	with open(cache_file, newline='') as csvfile:
		spamreader = csv.reader(csvfile, delimiter=';', quotechar='|')
		for row in spamreader:
			file, hashed_sum = row
			cache[file] = hashed_sum
	return cache

def save_cache(data):
	print("Saving cache at " + cache_file + "...")
	with open(cache_file, 'w', newline='') as csvfile:
		spamwriter = csv.writer(csvfile, delimiter=';',
			quotechar='|', quoting=csv.QUOTE_MINIMAL)
		for row in data:
			file, hash_sum = row
			spamwriter.writerow([file, hash_sum])
	print("Cache saved with success")

#	Render engine
terminal_size = os.get_terminal_size().columns
def render(pair, done, total):
	start, end = pair
	spacing = " " * (terminal_size - (len(start) + len(end) + 6))
	info = start + spacing + "=>    " + end
	percent = round(100 * done/total, 1)
	progress = "*** Current status: " + str(percent) + "% ***" + '\t'
	print('\r' + info, end='')
	print('\n' + progress, end='')

#	Indexing all files
path_list = os.walk(".")
targets = []
for root, dirs, files in path_list:
	for file in files:
		path = os.path.join(root, file)
		if not (".git" in path or ".config" in path or "Work"):
			targets.append(path)


total = len(targets)
print(str(total) + " files found")

#	Getting MD5 sum for each file
cache = load_cache()
hash_by_path = []
init_sums = time.time()
for file in targets:
	hash_sum = ''
	if file in cache:
		hash_sum = cache[file]
	else:
		data = open(file, 'rb').read()
		hash_sum = hashlib.md5(data).hexdigest()
	pair = (file, hash_sum)
	render(pair, len(hash_by_path), total)
	hash_by_path.append(pair)
end_sums = time.time()
print("MD5 Hash sums done with success")
save_cache(hash_by_path)

#	Reverse index
by_hash = {}
for pair in hash_by_path:
	file, hash_sum = pair
	if not hash_sum in by_hash:
		by_hash[hash_sum] = [file]
	else:
		by_hash[hash_sum].append(file)

#	Check identical files
def input_loop(question, options):
	while True:
		print(question)
		answer = input("Input: ")
		if answer in options:
			return answer
		else:
			print("Bad answer")

for hash_sum in by_hash:
	files = by_hash[hash_sum]
	length = len(files)
	if length != 1:
		while True:
			print('\t *** Time to work ***')
			#	Show repetitions
			i = 1
			for file in files:
				print(str(i) + ')\t' + file)
				i += 1
			#	Stay or jump
			instructions = ["q - quit", "n - next", "s - select"] 
			for instruction in instructions:
				print(instruction)
			stay = input_loop("Work on that one or go next", ["q", "n", "s"])
			match stay:
				case "q":
					exit()
				case "n":
					break
				case "s":
					#	Choose file
					question = "Choose a file to decide: [a - all] [1-" + str(length) + "]"
					options = [str(number) for number in range(1, length + 1)]
					options.append("a")
					file_input = input_loop(question, options)
					selected_files = files if file_input == "a" else [files[int(file_input) - 1]]
					#	Select action
					instructions = ["d - delete forever", "k - keep as it is", "t - send to trash"]
					for instruction in instructions:
						print(instruction)
					question = "Choose an action to that file"
					options = ["d", "k", "t"]
					selected_action = input_loop(question, options)
					#	Confirm
					print("File: " + str(selected_files))
					print("Action: " + selected_action) 
					confirm = input_loop("Are you sure? [y/n]", ["y", "n"])
					if confirm == "y":
						match selected_action:
							case "d":
								for file in selected_files:
									print("Deleting " + file)
									os.remove(file)
							case "k":
								print("Doing nothing...")
							case "t":
								print("Sending to trash...")