Human Readable ID

Human Readable ID (hrid) generates short, human-readable, collision-aware, friendly IDs that are ideal for experiments, jobs, and filenames.

Why human-readable-id?

Traditional IDs (UUIDs, hashes) are:

• hard to read
• hard to remember
• hard to communicate
• unpleasant in logs, filenames, and UIs

human-readable-id produces identifiers like:

gentle_river_42
silent_orbit_a9f3c2d1

They are:

• readable
• easy to communicate
• deterministic when seeded
• configurable for collision safety

Features

• Predicate–object word structure for human readable IDs
• Friendly words only
• Deterministic generation from a seed
• Configurable number of words and suffix length
• Optional hash-based suffixes
• Collision-space analysis (--collision)
• Word trimming for predictable ID length
• Implementations in Bash and Python
• No sudo required, HPC-readable

Installation

Python

You can install human-readable-id via pip (also includes CLI commands):

pip install human-readable-id

Bash (CLI)

The pure Bash version installs the hrid command and wordlists locally (no sudo) and does not require Python:

curl -fsSL "https://raw.githubusercontent.com/Karol-G/human-readable-id/main/bash/install.sh" | bash

This installs:

• Binary: ~/.local/bin/hrid
• Wordlists: ~/.local/share/human-readable-id/ (copied from the Python package’s canonical lists)

Make sure ~/.local/bin is on your $PATH:

export PATH="$HOME/.local/bin:$PATH"

Usage

CLI (Both Python and Bash)

Generate a human readable ID:

hrid # elastic_jargon_503, medieval_chess_277, nine_crayfish_660

Deterministic ID from a seed:

hrid my_seed # calm_menu_496

Use more words and a hash suffix:

hrid -w 3 -n 8 --hash # various_elegant_museum_2a750add

With a seed, the hash suffix is deterministic; without a seed it is random.

Trim words to enforce predictable length:

hrid -t 4 # clim_figu_144

Analyze collision space instead of generating an ID:

hrid --collision # scalloped_wombat_617

# predicates: 1450
# objects:    3062
# words:      2 (predicates=1, objects=1)
# suffix:     digits length 3 (space=10^3)
# combinations_M: 4439900000
# n_for_Ecollision_1: 94233

hrid -n 8 --hash --collision # distinct_sphere_3f9b7140

# predicates: 1450
# objects:    3062
# words:      2 (predicates=1, objects=1)
# suffix:     hex hash length 8 (space=16^8)
# combinations_M: ≈ 1,91e16 (< 2^63-1)
# n_for_Ecollision_1: 195290683

Python API

Basic usage:

from human_readable_id import generate_hrid

generate_hrid() # alert_tarn_100

Deterministic generation:

generate_hrid(seed="my_seed") # calm_menu_496

Custom configuration:

generate_hrid(
    seed="experiment-001",
    words=3,
    numbers=8,
    use_hash_suffix=True,
    trim=4,
) # nerv_oval_batt_c8597bc1

Collision awareness

human-readable-id explicitly exposes the size of its ID space.

Using --collision (or the Python equivalent) reports:

• total number of possible IDs (exact)
• smallest number of generated IDs needed for an expected collision of 1 (exact)

The Python implementation computes these values with integer arithmetic (no rounding); the Bash CLI falls back to approximate formatting for extremely large spaces.

This helps choose safe parameters instead of guessing.

Wordlists

IDs are built from curated wordlists:

• predicates (verbs / adjectives)
• objects (nouns)

Canonical wordlists live in python/src/human_readable_id/words/ and ship with the Python package. The Bash installer copies those same files into ~/.local/share/human-readable-id/ for the CLI. Update the canonical files and rerun the installer (or copy them manually) if you want custom lists.

Contributing

Contributions are welcome! Please open a pull request with clear changes and add tests when appropriate.

Issues

Found a bug or have a request? Open an issue at https://github.com/Karol-G/human-readable-id/issues.

License

Distributed under the MIT license. See LICENSE for details.