Messenger_1Million_Stats/million/analyze/count_analysis.py

import million.analyze.message_evaluation as msg_val

def check_extra_or_missing_letter(word, reference):
    len_word = len(word)
    len_ref = len(reference)

    if abs(len_word - len_ref) != 1:
        return False

    shortest = word if len_word < len_ref else reference
    longest = word if len_word > len_ref else reference

    for i in range(len(shortest)):
        if shortest[i] != longest[i]:
            return shortest[i:] == longest[i + 1 :]

    return True


def check_single_letter_differ(word, reference):
    return sum(1 for x, y in zip(reference, word) if x != y) == 1


def check_letter_swap(word, reference):
    if len(word) != len(reference):
        return False

    for i in range(len(word) - 1):
        if word[i] != reference[i]:
            return word[i + 1] + word[i] + word[i + 2 :] == reference[i:]

    return False


def check_typo(word, reference):
    if len(reference) == len(word):
        return check_single_letter_differ(word, reference) or \
            check_letter_swap(word, reference)
    else:
        return check_extra_or_missing_letter(word, reference)


def find_value_around_index(messages, value, idx, amplitude) -> int:
    check_value = lambda x: msg_val.get(messages[x]) == value

    if check_value(idx):
        return idx

    for offset in range(1, amplitude):
        o_idx = idx + offset * +1
        if check_value(o_idx):
            return o_idx

        o_idx = idx + offset * -1
        if check_value(o_idx):
            return o_idx

    return -1