# 전처리 함수
def preprocess(df):
    url_pattern = r"https?://\S+|www\.\S+"
    df["comment_text"] = df["comment_text"].str.replace(url_pattern, " ")

    # apply unidecode
    df["comment_text"] = df["comment_text"].map(unidecode.unidecode)

    # apply lower
    df["comment_text"] = df["comment_text"].str.lower()

    return df