afidgen 1.0.0rc0

An ID format designed to meet the trade-off between space and human readability.

afidgen

A Python library for generating afid ids.

Afids are compact, human-readable IDs made of lowercase letters, digits, and hyphens. They come in two variants — short (20 chars, 75 bits of randomness) and long (30 chars, 125 bits) — both shaped as prefix-tag-suffix:

res-2sed3-1p3dpw40ds              # short
evt-2sed3-1p3dpw40dsabckdvc7p1    # long

The prefix is yours to assign (typically the resource type); the tag and suffix are random, drawn from a Crockford Base32 alphabet that omits easily-confused characters. See the afid spec for the full grammar and analysis.

Requirements

Python 3.9 or later.

Installing

pip install afidgen

Usage

import afidgen

# Generate a short ID with the prefix "res".
resource_id = afidgen.random_short("res")

# Generate a long ID with the prefix "evt".
event_id = afidgen.random_long("evt")

# Create a generator. This is more efficient for generating multiple afids,
# because it only needs to validate the prefix once when you create the
# generator, instead of every time you generate an ID.
gen = afidgen.Generator.long("evt")
event_ids = [gen() for _ in range(1_000)]

# Generator is also iterable.
import itertools
more_event_ids = list(itertools.islice(gen, 1_000))

# By default, generators use secrets.token_bytes (cryptographically secure).
# You can override it when creating a generator to provide your own source of
# randomness — for example, random.randbytes if you need reproducibility or
# don't need security. The function must take a number of bytes and return
# that many random bytes.
import random
gen = afidgen.Generator.long("txi", randbytes=random.randbytes)
tx_id = gen()

Performance

For bulk generation, build a generator once with Generator.short or Generator.long and reuse it. The prefix is validated at construction, so each call only pays for the randomness draw and the encoding.

random_short / random_long are convenience helpers that build a fresh generator on every call. Use them for ad-hoc IDs; reach for the factories in tight loops.

Run benchmarks with uv run pytest bench/.

Concurrency

The default randomness source (secrets.token_bytes) is thread-safe, so a single Generator can be called from multiple threads. If you pass a custom randbytes callable, concurrent safety follows that callable.

Developing

From libs/python/:

uv run pytest                  # tests
uv run ruff format --check .   # format check
uv run ruff format .           # apply formatting
uv run ruff check .            # lint
uv run ty check                # typecheck
uv run pytest bench/           # benchmarks

Initial setup: uv sync.