UUIDv5 Generator — Namespace + Name via SHA-1

Real-world use cases

• Give a config entity a stable ID without a database

  You have a list of integrations keyed by their domain and you want an ID column that never drifts. Hash each domain in the DNS namespace and use the v5 UUID as the primary key. Re-import the same config a year later and every row keeps its original ID, because the ID is a pure function of the domain, not a counter or a random value handed out at insert time.

• Deduplicate records that arrive from two sources

  A user shows up in your CRM export and again in a billing feed, keyed only by email. Compute uuidV5(email, your-namespace) on both sides and the two rows collapse to one ID, so a join or an upsert merges them with no fuzzy matching. The namespace keeps these IDs from colliding with v5 IDs you derive from, say, order numbers.

• Build content-addressed keys for a cache or CDN path

  Instead of storing a random key next to each cached document, derive the key from the document's canonical URL with the URL namespace. The cache path becomes reproducible: any service that knows the URL can recompute the exact same v5 UUID and find the entry, with no shared lookup table and no coordination between workers.

• Generate deterministic test fixtures across a team

  Two engineers writing the same integration test need fixtures with identical IDs so assertions line up in CI and on each laptop. Derive every fixture ID with v5 from a fixed namespace plus a readable name like user/alice. The IDs are stable forever, readable in diffs, and you can paste this tool's shareable link into the PR so a reviewer reproduces them instantly.

Common pitfalls

• Hashing the name without a namespace. Skipping the namespace, or reusing one namespace for unrelated kinds of names, lets a DNS host and an order code that happen to share a string collide into the same UUID. Always pair each family of names with its own fixed namespace so the IDs stay disjoint.

• Expecting v5 to be unguessable like v4. Because v5 is a plain hash of public inputs, anyone who knows your namespace and naming scheme can recompute every ID. That is a feature for reproducibility but a trap if you treat the UUID as a secret. For tokens that must be unpredictable, use v4 instead.

• Mismatching the name encoding. v5 hashes the UTF-8 bytes of the name, so a name encoded as UTF-16 or with stray whitespace produces a different UUID than the one your other system computed. Normalize and trim the name to exactly the same bytes everywhere, or the IDs silently stop matching.

Privacy