UUID Generator From String: Deterministic Identifiers

Use v5, not a hash you format yourself

v5 hashes a namespace UUID plus your string with SHA-1 and formats the result with correct version and variant bits:

import uuid
NS = uuid.UUID('6ba7b810-9dad-11d1-80b4-00c04fd430c8')   # your own constant
uuid.uuid5(NS, 'user@example.com')
# same input -> same UUID, every time, in every language

Taking the first 32 hex characters of a SHA-256 and inserting hyphens produces something that looks like a UUID and fails strict validation, because the version and variant nibbles are wrong.

Pick and freeze a namespace

The namespace separates contexts, so the same string in two systems does not collide. RFC 9562 defines DNS, URL, OID and X.500; for application data, generate one v4 and hard-code it.

Treat that constant as part of your schema. Changing it changes every derived identifier, which is a data migration rather than a config tweak.

Where it pays off

Idempotent imports — re-running a load updates existing rows instead of duplicating them, with no lookup table.

Cross-service agreement — two systems derive the same id for the same entity without coordinating.

Content addressing — deriving an id from a normalised file path or a canonical URL.

Normalise the input first

The hash is over exact bytes, so User@Example.com and user@example.com produce different UUIDs. Lowercase, trim, and apply Unicode normalisation (NFC) before hashing, and document the rule — anyone recomputing the id must apply the same steps.

And remember the identifier is not secret: anyone knowing the namespace and the input can compute it. Never use a v5 UUID as an access token.