barcode_regex.py

import re


BARCODE_PATTERN = r"""
                   ^ [\W_] *    # Ignore any leading non-alphanumerics
                   ([\d\s]+)    # Capture any digits, or spaces
                   [\W_]*$      # Ignore any trailing non-alphanumerics
                   """

BARCODE_REGEX = re.compile(BARCODE_PATTERN, re.VERBOSE)


def sanitize_barcode(barcode):
    """ Remove non-alphanumerics from the start and end of the barcode string,
    then remove any space characters which may be remaining inside the
    barcode.
    """
    return ''.join(BARCODE_REGEX.sub(r'\1', barcode).split())


################
# Normal input #
################

sanitize_barcode("'0123456789  ")
# '0123456789'

sanitize_barcode("'0123456789")
# '0123456789'

sanitize_barcode("0123456789")
# '0123456789'


#####################
# Leading bad chars #
#####################

sanitize_barcode(" $,)_*  $'0123456789")
# '0123456789'

# Trailing bad chars
sanitize_barcode("'0123456789$ ,$)_*  ")
# '0123456789'


#######################
# Whitespace inserted #
#######################

sanitize_barcode("'01 23 45 67 89")
# '0123456789'


################
# Bad Barcodes #
################

sanitize_barcode("'X0123456789")
# "'X0123456789"
sanitize_barcode("'01234Y56789")
# "'01234Y56789"
sanitize_barcode("'0123456789Z")
# "'0123456789Z"
sanitize_barcode("  'X '0123456789$', '")
# "'X'0123456789$','"


######################
# Everything at once #
######################

sanitize_barcode(" $!@#$ '01 \r 23 \n 45 \t 67 \f 89 \v )(*&^%  ")
# '0123456789'