validate_str.py

# coding=utf8

import re


def escape_emoji(txt):
    try:
        chars = re.compile(u'[\U00010000-\U0010ffff]')
    except re.error:
        chars = re.compile(u'[\uD800-\uDBFF][\uDC00-\uDFFF]')
    return chars.sub(u'', txt).encode("utf-8")


def validate(raw_str):
    try:
        raw_str = raw_str.decode('unicode-escape').encode('latin1').decode('utf8')
        new_str = escape_emoji(raw_str)
    except:
        raise
    return new_str.encode('utf8')