dc-ml-emea-ar/utils/biz_utils.py

import re

def add_slash_to_text_as_regex(text: str):
    if text is None or len(text) == 0:
        return text
    special_char_iter = re.finditer("\W", text)
    for special_iter in special_char_iter:
        if len(special_iter.group().strip()) == 0:
            continue
        replace = r"\{0}".format(special_iter.group())
        if replace not in text:
            text = re.sub(replace, replace, text)
    text = re.sub(r"\s+", r"\\s+", text)
    return text


def clean_text(text: str) -> str:
    text = text.lower()
    # update the specical character which begin with \u, e.g \u2004 or \u00a0 to be space
    text = re.sub(r"\\u[0-9a-z]{4}", ' ', text)
    text = re.sub(r"( ){2,}", ' ', text.strip())
    return text