SpoilerOverflow

util.py

import re


def sanitize_body(text):
    body = text.replace("\\\'", "\'")
    body = " ".join(body.split())
    body = re.sub('<[^<]+?>', '', body)
    body = body.strip()

    return body


def extract_content(text):
    pattern = r"<blockquote class=\"spoiler\">.+<\/blockquote>"
    spoiler = re.findall(pattern, text, flags=re.DOTALL)
    for i in range(len(spoiler)):
        spoiler[i] = sanitize_body(spoiler[i])

    body = re.sub(pattern, "", text, flags=re.DOTALL)
    body = sanitize_body(body)

    return {"normal": body, "spoiler": spoiler}


def count_word(text):
    return len(text.split())


def count_character(text):
    characters = 0
    words = text.split()
    for word in words:
        characters += len(word)

    return characters


def calculate_percentage(value, total):
    return 100.0 * value / total


def calculate_spoiler_percentage(text):
    content = extract_content(text)

    normal = content["normal"]
    normal_chars = count_character(normal)
    normal_words = count_word(normal)

    spoiler_chars = 0
    spoiler_words = 0
    for item in content["spoiler"]:
        spoiler_chars += count_character(item)
        spoiler_words += count_word(item)

    spoiler_word_percentage = calculate_percentage(spoiler_words, normal_words + spoiler_words)
    spoiler_char_percentage = calculate_percentage(spoiler_chars, normal_chars + spoiler_chars)

    return {"word": spoiler_word_percentage, "characters": spoiler_char_percentage}