encode ascii text as DNA sequences

ascii-to-dna.py

#!/usr/bin/env python3

from itertools import product

from typing import Dict, Tuple
import sys


def generate_codon_mapping(
    dna_alphabet="GATC",
    alphabet="|abcdefghijklmnopqrstuvwxyz.!?,;() 01234567890",
    codon_length=3,
) -> Dict[str, str]:
    """
    Returns a dictionary mapping codon -> character. | is preserved as a
    special character meaning START so you can decode DNA sequences that have
    been concatenated together
    """
    codons = ["".join(x) for x in list(product(dna_alphabet, repeat=codon_length))]
    assert len(codons) >= len(alphabet), "We require more codons!"
    return {codon: character for codon, character in zip(codons, alphabet)}


def encode(input_string: str) -> str:
    character_to_codon = {character: codon for codon, character in generate_codon_mapping().items()}
    output = []
    for character in input_string:
        output.append(character_to_codon[character])
    return "".join(output)


def decode(input_string: str) -> str:
    """
    Raw transcribe function. Does not care about start codons.
    """
    codon_to_character = generate_codon_mapping()
    output = []
    for i in range(0, len(input_string))[::3]:
        codon = input_string[i : i + 3]
        character = codon_to_character[codon]
        output.append(character)
    return "".join(output)


def test_encode_decode():
    test_string = "|the (quick?) fox jumped over; the lazy! d0g."
    assert decode(encode(test_string)) == test_string