This gist parses 99% twitter-like hashtags in unicode

python_parse_hashtags.py

# -*- coding: utf-8 -*-
import re

# Because 're' module do not support \p{L} unicode class
# need to use simplistic regex with \w instead \p{L}. But such regexp
# also match digits like #123, #003 and so on.
# So it need to further filter out digit-only tags.
# See https://regex101.com/r/cK5oJ0/
HASHTAG_EXP = r'(?:^|_|[^\w&/]+)(?:#|＃)([\wÀ-ÖØ-öø-ÿ]+)'
HASHTAG_REGEX = re.compile(HASHTAG_EXP, re.UNICODE | re.IGNORECASE)


def parse_hashtags_from_text(text, to_lower_case=True):
    return (
        t.lower() if to_lower_case else t
        for t in HASHTAG_REGEX.findall(unicode(text))
        if not t.isdigit()
    )