mrinal-scio/clean.py

## clean.py
def clean_address(s):
    s1 = re.sub(r'^[A-Z0-9]{4}-', '', s) # T-Mobile specific
    s2 = re.sub(r'[^\s,a-zA-Z]', ' ', s1) # drop all numbers and non-comma punct
    s3 = re.sub('Remote', '', s2) # drop "Remote"
    s4 = re.sub(r'\s+', ' ', s3) # normalize whitespace
    return s4.lstrip().rstrip()
	def clean_address(s):
	s1 = re.sub(r'^[A-Z0-9]{4}-', '', s) # T-Mobile specific
	s2 = re.sub(r'[^\s,a-zA-Z]', ' ', s1) # drop all numbers and non-comma punct
	s3 = re.sub('Remote', '', s2) # drop "Remote"
	s4 = re.sub(r'\s+', ' ', s3) # normalize whitespace
	return s4.lstrip().rstrip()