Morse bruteforce

morse_bruteforce.py

'''
Brute-force a morse code message where the spaces between letters have been lost,
but the spaces between words still exist
Big thanks to this Stackoverflow post: https://stackoverflow.com/a/8375498/

Instructions:
    1. Put your "broken" morse message into the morseString variable below
    2. Put a newline seperated dictionary into the same folder as this script and name it dict.txt
        e.g. from https://github.com/oprogramador/most-common-words-by-language
    3. Run the script
'''

morseString = "-......... ..... .- -....-"

dictFileName = "dict.txt"

import sys

MORSE_TO_ABC = dict(zip([
    '.-', '-...', '-.-.', '-..', '.', '..-.', '--.', '....',
    '..', '.---', '-.-', '.-..', '--', '-.', '---', '.--.',
    '--.-', '.-.', '...', '-', '..-', '...-', '.--', '-..-',
    '-.--', '--..'
], 'ABCDEFGHIJKLMNOPQRSTUVWXYZ'))

replaceWords = {
    "Ä": "AE",
    "Ö": "OE",
    "Ü": "UE",
    "ß": "SS"
}

WORDS = set()
for word in open(dictFileName).readlines():
    word = word.strip().upper()
    for a,b in replaceWords.items():
        word = word.replace(a, b)
    WORDS.add(word)

possibleAbcWordsForMorseWord = {}

def decodeMorseWord(morseWord, thorough=False):
    if (morseWord in possibleAbcWordsForMorseWord):
        return possibleAbcWordsForMorseWord[morseWord]

    possibleLetters = {}
    for i in range(0, len(morseWord)):
        possibleLetters[i] = {}
        # print(f"\nCurrent: {morseSign} {i:02d}", end=" ")
        currentSign = ""
        for j, additionalMorseSign in enumerate(morseWord[i:]):
            currentSign += additionalMorseSign
            # print(additionalMorseSign, end="")
            if (currentSign in MORSE_TO_ABC):
                abcLetter = MORSE_TO_ABC[currentSign]
                # print(abcLetter, end="")
                possibleLetters[i][abcLetter] = i + j + 1

    possibleAbcWordsForMorseWord[morseWord] = []
    def findWords(index, word):
        if (index == len(morseWord)):
            if (not word in possibleAbcWordsForMorseWord[morseWord]):
                if (word in WORDS):
                    possibleAbcWordsForMorseWord[morseWord].append(word)
                elif (thorough):
                    for dictWord in WORDS:
                        if word in dictWord:
                            possibleAbcWordsForMorseWord[morseWord].append(word + " (" + dictWord + ")")
        else:
            for letter, nextIndex in possibleLetters[index].items():
                findWords(nextIndex, word + letter)

    for letter, nextIndex in possibleLetters[0].items():
        findWords(0, "")

    return possibleAbcWordsForMorseWord[morseWord]

morseWords = morseString.split(" ")
longestWordLen = len(max(morseWords, key=len))
for morseWord in morseWords:
    print(f"Morse: {morseWord}", end="")
    print(" " * (longestWordLen - len(morseWord)), end=" ")
    sys.stdout.flush()
    abcWords = decodeMorseWord(morseWord)
    print(abcWords)

print("== DONE ==")