Problema do Jimmy Neutron

sequence_x_removal.py

from collections import deque
from itertools import cycle

import pytest

REPLACEMENTS = ["CCTTGG", "CTATGG", "CCACGG"]


def replace_gen_sub_seq(seq, replacements):
    seq = list(seq)
    replacements_len = len(replacements[0])
    x_pattern = deque('X' * replacements_len)
    current_substring = deque(maxlen=replacements_len)

    def append_and_return_exceeding_item(item_to_append):
        """Append item do current substring and returns exceding char if its maxlen is reached

        :param item_to_append:
        :return: item or None
        """
        exceeding_item = None
        if len(current_substring) == replacements_len:
            exceeding_item = current_substring[0]
        current_substring.append(item_to_append)
        return exceeding_item

    replacements_cycle = cycle(replacements)
    next(replacements_cycle) # advancing one
    for char, replacement in zip(seq, replacements_cycle):
        exceding_char = append_and_return_exceeding_item(char)
        if exceding_char:
            yield exceding_char
        if current_substring == x_pattern:
            yield from replacement
            current_substring.clear()
    yield from current_substring


@pytest.mark.parametrize(
    'seq,expected_seq',
    [

        ("XXXXXXXXXXXXTC", "CCTTGGCCTTGGTC"),
        ("CXXXXXXXXXXXXT", "CCTATGGCTATGGT"),
        ("CTXXXXXXXXXXXX", "CTCCACGGCCACGG"),
        ("CTXXXXXXXXXXXXXXXXXX", "CTCCACGGCCACGGCCACGG"),
        ("XXXXXXCXXXXXXTXXXXXX", "CCTTGGCCTATGGTCCACGG"),
    ]
)
def test_repeats_juntas(seq, expected_seq):
    seq_output = ''.join(replace_gen_sub_seq(seq, REPLACEMENTS))
    assert expected_seq == seq_output