encodes hash to words

encode_hash_to_words.py

# 235886 words
text_file = open("/usr/share/dict/words", "r")

lines = text_file.readlines()
words = []

for line in lines:
    words.append(line.strip())

import struct


# takes 12 hex chars
def encode(hash):
    binary = bytes.fromhex(hash)

    first_word = struct.unpack('>H',bytes([binary[0],  (binary[1] & 0b111111)]))[0]
    second_word =  struct.unpack('>H',bytes([(binary[1] & 0b11000000), binary[2]]))[0]
    third_word = struct.unpack('>H',bytes([binary[3],  (binary[4] & 0b111111)]))[0]
    forth_word =  struct.unpack('>H',bytes([(binary[4] & 0b11000000), binary[5]]))[0]

    return f"{words[first_word]}-{words[second_word]}-{words[third_word]}-{words[forth_word]}"

def decode(word1, word2, word3, word4):
    word1 = words.index(word1)
    word2 = words.index(word2)
    word3 = words.index(word3)
    word4 = words.index(word4)

    raw_bytes = struct.pack('>HHHH', word1, word2, word3, word4)
    
    output = bytes([raw_bytes[0]])
    output += bytes([(raw_bytes[1] & 0b111111) + (raw_bytes[2] & 0b11000000)])
    output += bytes([raw_bytes[3]])
    output += bytes([raw_bytes[4]])
    output += bytes([(raw_bytes[5] & 0b111111) + (raw_bytes[6] & 0b11000000)])
    output += bytes([raw_bytes[7]])

    return output.hex()


def test(example):
    words = encode(example)
    print(f"encoded {example} to {words}")
    print(f"decoded {words} to {decode(*words.split('-'))}")

print("====")
test("abcef1dbcec4")
print("====")
test("feedcafe1234")
print("====")
test("ffffffffffff")

# ====
# encoded abcef1dbcec4 to costocentral-decemstriate-disulphuret-decay
# decoded costocentral-decemstriate-disulphuret-decay to abcef1dbcec4
# ====
# encoded feedcafe1234 to ethanamide-decease-etacist-Abanic
# decoded ethanamide-decease-etacist-Abanic to feedcafe1234
# ====
# encoded ffffffffffff to euchromatin-decennial-euchromatin-decennial
# decoded euchromatin-decennial-euchromatin-decennial to ffffffffffff