Heya,

Your issues stem from three things: (1) dict[key] raises KeyError on unknown codes, (2) split(" ") produces empty strings when there are consecutive spaces, and (3) you’re not distinguishing letter separators (single space) from word separators (typically / or multiple spaces).

## The fix

### 1. Use dict.get() instead of dict[...]

dict.get(key, default) returns the default instead of raising KeyError. This alone solves the crash on @.

### 2. Standardize the word separator

The conventional Morse convention is:

• 1 space between letters
• 3 spaces (or a /) between words

Normalize the input first so the parser only deals with one format.

### 3. Parse in two levels: words → letters

Split on the word separator first, then on the letter separator. This makes empty tokens disappear naturally and makes the intent explicit.

## Working example

MORSE_CODE_DICT = {
    '.-':   'A', '-...': 'B', '-.-.': 'C', '-..':  'D', '.':    'E',
    '..-.': 'F', '--.':  'G', '....': 'H', '..':   'I', '.---': 'J',
    '-.-':  'K', '.-..': 'L', '--':   'M', '-.':   'N', '---':  'O',
    '.--.': 'P', '--.-': 'Q', '.-.':  'R', '...':  'S', '-':    'T',
    '..-':  'U', '...-': 'V', '.--':  'W', '-..-': 'X', '-.--': 'Y',
    '--..': 'Z',
    '-----':'0', '.----':'1', '..---':'2', '...--':'3', '....-':'4',
    '.....':'5', '-....':'6', '--...':'7', '---..':'8', '----.':'9',
}

def decode_morse(morse: str, unknown: str = '?') -> str:
    # Normalize: treat "/" as a word separator, collapse runs of spaces
    # into the canonical 3-space word break.
    normalized = morse.replace('/', '   ').strip()

    words = [w for w in normalized.split('   ') if w]
    decoded_words = []

    for word in words:
        letters = [MORSE_CODE_DICT.get(code, unknown)
                   for code in word.split(' ') if code]
        decoded_words.append(''.join(letters))

    return ' '.join(decoded_words)

### Test

>>> decode_morse('.... . .-.. .-.. ---   .-- --- .-. .-.. -..')
'HELLO WORLD'

>>> decode_morse('.... . .-.. .-.. ---   .-- --- .-. .-.. -.. @')
'HELLO WORLD?'

>>> decode_morse('.... . .-.. .-.. --- / .-- --- .-. .-.. -..')
'HELLO WORLD'

## Why this works

## A couple of further notes

• Strict mode: if you’d rather reject invalid input than substitute ?, raise your own ValueError with the offending token — it’s friendlier than a bare KeyError:

  if code not in MORSE_CODE_DICT:
      raise ValueError(f"Invalid Morse token: {code!r}")
  

• Encoder symmetry: when building the encoder, emit ' ' between letters and ' ' (three spaces) between words so a round-trip decode(encode(x)) == x.upper() holds for supported characters.
• Regex alternative: re.split(r' {3,}', morse.strip()) for words and re.split(r' +', word) for letters also collapses runs gracefully if you prefer not to normalize first.

This pattern — normalize input → split coarse → split fine → lookup with default — generalizes well to any token-stream decoder, not just Morse.