Validate GSA unique entity identifier (UEI)

validate_uei.py

from functools import reduce
from itertools import starmap
import re
from typing import Iterable

def checksum(uei: str) -> bool:
    """Gets a checksum of a UEI"""

    def reducer(chars: Iterable[int]) -> int:

        return reduce(
            lambda x, y: x+y,
            starmap(
                lambda index, char: (char * index) % 10,
                enumerate(chars, 1),
            ),
            0,
        )

    assert len(uei) == 12

    chars = map(ord, uei[:-1])
    sum_ = reducer(chars)

    while sum_ > 9:
        sum_ = reducer(tuple(int(d) for d in str(sum_)))

    return sum_ == int(uei[-1])

def uei_is_plausible(uei: str) -> bool:
    """Returns `True` if a UEI is a plausible UEI, else `False`.
    After running this method, an API should be called
    to check whether the UEI actually exists.

    A GSA unique entity identifier (UEI) has the following characteristics:
    - The Unique Entity ID is a 12-character, alphanumeric value.
    - The letters “O” and “I” are not used to avoid confusion with zero and one.
    - The first character is not zero to avoid cutting off digits that can occur
      during data imports, for example, when importing data into spreadsheet
      programs.
    - Nine-digit sequences are not used in the identifier to avoid collision
      with the nine-digit DUNS Number or Taxpayer Identification Number (TIN).
    - The first five characters are structured to avoid collision with the
      Commercial and Government Entity code formatting or CAGE code.
    - The Unique Entity ID is not case sensitive.
    - The final character is a checksum of the first 11 characters.
      Checksums are used to detect errors within data.

    Examples:
    uei_is_plausible('VN1AJFAD19J9')  # Valid
    uei_is_plausible('99999999f995')  # Valid
    uei_is_plausible('ABCDEF12345')  # Invalid: too few characters
    uei_is_plausible('io10io10io19')  # Invalid: contains 'I' and/or 'O'
    uei_is_plausible('123456789FF4')  # Invalid: too many consecutive digits
    uei_is_plausible('1A2B3C4D5F6G')  # Invalid: incorrect checksum

    See also: GSA implementation at https://github.com/GSA-TTS/uei-js/
    """

    # The Unique Entity ID is not case sensitive
    uei = str(uei).upper()

    valid = True

    # The Unique Entity ID is a 12-character, alphanumeric value
    if len(uei) != 12 or not (uei.isalnum() and uei.isascii()):
        valid = False

    # The letters “O” and “I” are not included
    elif ('O' in uei) or ('I' in uei):
        valid = False

    # The first character is not zero
    elif uei[0] == '0':
        valid = False

    # Nine-digit sequences are not used
    elif re.match(r'\d{9}', uei):
        valid = False

    # The final character is a checksum of the first 11 characters
    elif not uei[-1].isdigit() or not checksum(uei):
        valid = False

    return valid