Slug + base64 identifier URL generation

slug.py

"""
slug.py

Contains operations and functions integral to slug processing and identifier extraction
"""

from math import ceil

from slugify import slugify
from typing import Optional
import re
import base64

use_little: bool = True  # Whether or not to use little or big endian. True for little, False for big.
use_sign: bool = False  # Whether to assume usage of signed values
use_128: bool = False  # If true, a 128-bit identifier will be assumed
general_limit: int = 50  # The general character limit for an identifier loaded slug
id_space: int = ceil((128 if use_128 else 64) / 6)  # The number of characters the identifier will occupy. 11 or 22

b64_alphabet = "-_0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz"
b64_pattern = rf"[{b64_alphabet}]+"


def asciify_identifier(n: int) -> str:
    """Converts a 64-bit or 128-bit integer to it's base64 ASCII counterpart string."""
    return base64.urlsafe_b64encode(int.to_bytes(n, 16 if use_128 else 8, 'little', signed=use_sign)).rstrip(b'=').decode('ascii')


def generate_slug(identifier: int, text: Optional[str]) -> str:
    """Generates a slug given an identifier, and optionally, some text."""
    ascii_id: str = asciify_identifier(identifier)
    if text is None:
        return ascii_id
    slug_text: str = slugify(text, replacements=[['\'', '']], max_length=20, word_boundary=True)
    return f'{slug_text}-{ascii_id}'


def extract_identifier(slug: str) -> Optional[int]:
    """Extracts an integer identifier from a loaded slug."""
    slug = slug.rstrip('/')
    if len(slug) < id_space:
        return None
    identifier = slug[-1 * id_space:]
    if re.fullmatch(b64_pattern, identifier) is not None:
        decoded: bytes = base64.urlsafe_b64decode(identifier.encode('ascii') + b'==')
        return int.from_bytes(decoded, 'little', signed=use_sign)


if __name__ == '__main__':
    import random

    # random.seed('f')
    n = int.from_bytes(random.randbytes((128 if use_128 else 64) // 8), 'little', signed=use_sign)
    n_encode = asciify_identifier(n)
    tests = [
        '22',
        '034y530habugaiusbgasigba',
        '------=+++===',
        ''
    ]

    for test in tests:
        test += n_encode
        n_decode = extract_identifier(test)
        print(f'{n_decode == n} | {test}')

s = generate_slug(random.getrandbits(64), 'Rowdy Bash 2017 Bloopers')
print(s)
print(extract_identifier(s))