Check if file contains ASCII printable characters.

check_ascii_file.py

#!/usr/bin/env python3

import os
import sys

PRINTABLE_ASCII = [
	9, 10, 13, (32, 126)
]


def to_complete_list(mixed_list):
	complete_list = list()

	for item in mixed_list:
		if isinstance(item, tuple):
			if len(item) != 2:
				print("WARN: Invalid format list")
				continue

			minor = item[0]
			major = item[1]

			if not isinstance(minor, int) or not isinstance(major, int):
				print("WARN: Invalid format list")
				continue

			for i in range(minor, major + 1):
				complete_list.append(i)

		elif isinstance(item, int):
			complete_list.append(item)

		else:
			print("WARN: Invalid format list")
	
	return complete_list


def main(args):
	
	if len(args) <= 1:
		print("ERROR: Invalid parameters")
		exit(1)

	filename = args[1]

	if not os.path.isfile(filename):
		print("ERROR: Filename {} not found".format(filename))
		exit(2)

	printable_ascii_list = to_complete_list(PRINTABLE_ASCII)

	# Binary read
	file_desc = open(filename, "rb")

	read_bytes = 0
	count_line = 1

	flag_continue = True
	while flag_continue:
		byte = file_desc.read(1)
		read_bytes += 1

		d_byte = int.from_bytes(byte, "big")

		if d_byte == 0:
			# End of file
			print("INFO: End of file '{}', read bytes '{}', number of lines '{}'".format(filename, read_bytes, count_line))
			flag_continue = False

		elif d_byte == 10:
			# New line
			count_line += 1

		elif not d_byte in printable_ascii_list:
			print("WARN: File '{}', read bytes '{}', non printable char '{}' at line '{}'".format(filename, read_bytes, d_byte, count_line))
			flag_continue = False
			
	file_desc.close()


if __name__ == "__main__":
	main(sys.argv)