utils.py

import re

def normalize_text(s):
 s = str(s)
 # if not s.isupper():
 # s = re.sub(r"(\w)([A-Z])", r"\1 \2", s) # Respace
 s = re.sub(r'\（.*\）', '', s) # Remove japanese brackets
 s = re.sub(r'\(.*\)', '', s) # Remove english brackets
 s = s.strip()
 s = s.replace(' ', '_')
 s = s.upper()
 return s