test_utilities.py

import random
from typing import Set

class InvalidNucleotideError(Exception):
    """Raised when an invalid nucleotide is encountered."""
    pass

STANDARD_NUCLEOTIDES: Set[str] = {'A', 'T', 'G', 'C'}
INVALID_CHARS: Set[str] = {'Z', 'U', 'X', 'L'}


def generate_random_sequence(length: int, nucleotides: Set[str] = STANDARD_NUCLEOTIDES) -> str:
    """Generates a random sequence string from a given set of nucleotides."""
    return "".join(random.choices(list(nucleotides), k=length))


def generate_invalid_sequence(length: int) -> str:
    """Generates a sequence containing valid DNA bases and invalid characters."""
    valid_part_len = length // 2
    invalid_part_len = length - valid_part_len

    valid_part = generate_random_sequence(valid_part_len, STANDARD_NUCLEOTIDES)
    invalid_part = generate_random_sequence(invalid_part_len, INVALID_CHARS)

    # Shuffle and join the parts
    full_list = list(valid_part + invalid_part)
    random.shuffle(full_list)
    return "".join(full_list)