Sanitize text

sanitize_text.py

import re

def sanitize_text(text): 

    # remove extra white space
    text = ' '.join(text.split())
    
    # remove white space at start/end of text
    text = text.lstrip()
    text = text.rstrip()

    # capitalize only first letter without changing the others
    text = text[0].upper() + text[1:]
    
    # replace invalid punctuation at end of text
    if text[-1] == ";":
        text = text[:-1]
    if text[-1] == ",":
        text = text[:-1]

    # add full stop if the text does not end with valid punctuation
    if re.match('[\w\s]+[?.!]$', text) is None:
        text = text + "."

    return(text)