humanize.py

import struct
import typing


def humanize(val: bytes, words: typing.List[str]) -> typing.List[str]:
    return [words[struct.unpack_from("H", val, i)[0]] for i in range(0, len(val), 2)]


def dehumanize(val: typing.List[str], words_map: typing.Dict[str, int]) -> bytes:
    return b"".join((struct.pack("H", words_map[word]) for word in val))


if __name__ == "__main__":
    # Bring your own word list with length=65536
    with open("words.txt") as f:
        words = [w.strip() for w in f.readlines()]
    words_map = {word: i for i, word in enumerate(words)}

    # Warning: assumes buffers to encode have an even number of bytes
    encoded = "_".join(humanize(b"123456", words))
    decoded = dehumanize(encoded.split("_"), words_map)
    print(encoded)
    print(decoded)