A SQLite extension for generating and working with ULIDs. Built on top of sqlite-loadable-rs and ulid-rs.
If your company or organization finds this library useful, consider supporting my work!
.load ./ulid0
select ulid(); -- '01gqr4j69cc7w1xdbarkcbpq17'
select ulid_bytes(); -- X'0185310899dd7662b8f1e5adf9a5e7c0'
select ulid_with_prefix('invoice'); -- 'invoice_01gqr4jmhxhc92x1kqkpxb8j16'
select ulid_with_datetime('2023-01-26 22:53:20.556); -- '01gqr4j69cc7w1xdbarkcbpq17'
select ulid_datetime('01gqr4j69cc7w1xdbarkcbpq17') -- '2023-01-26 22:53:20.556'
Use as a PRIMARY KEY
for a table.
create table log_events(
id ulid primary key,
data any
);
insert into log_events(id, data) values (ulid(), 1);
insert into log_events(id, data) values (ulid(), 2);
insert into log_events(id, data) values (ulid(), 3);
select * from log_events;
/*
┌────────────────────────────┬──────┐
│ id │ data │
├────────────────────────────┼──────┤
│ 01gqr4vr487bytsf10ktfmheg4 │ 1 │
│ 01gqr4vr4dfcfk80m2yp6j866z │ 2 │
│ 01gqr4vrjxg0yex9jr0f100v1c │ 3 │
└────────────────────────────┴──────┘
*/
Consider using ulid_bytes()
for speed and smaller IDs. They generate about 1.6x faster than ulid()
, and take up 16 bytes instead of 26 bytes. You can use ulid()
to create a text representation of a BLOB ULID.
create table log_events(
id ulid primary key,
data any
);
insert into log_events(id, data) values (ulid_bytes(), 1);
insert into log_events(id, data) values (ulid_bytes(), 2);
insert into log_events(id, data) values (ulid_bytes(), 3);
select hex(id), ulid(id), data from log_events;
/*
┌──────────────────────────────────┬────────────────────────────┬──────┐
│ hex(id) │ ulid(id) │ data │
├──────────────────────────────────┼────────────────────────────┼──────┤
│ 0185F0539EBF286DA9F56BA4D9981783 │ 01gqr577nz51ptkxbbmkcsg5w3 │ 1 │
│ 0185F0539EC54F85745C1ECB64DF3A97 │ 01gqr577p59y2q8q0ysdjdyemq │ 2 │
│ 0185F0539ED48113F6F67BF3F6A4BFF7 │ 01gqr577pmg49zdxkvyfva9fzq │ 3 │
└──────────────────────────────────┴────────────────────────────┴──────┘
*/
Extract the timestamp component of a ULID with ulid_datetime()
.
select ulid_datetime(ulid()); -- '2023-01-26 23:07:36.508'
select unixepoch(ulid_datetime(ulid())); -- 1674774499
select strftime('%Y-%m-%d', ulid_datetime(ulid())); -- '2023-01-26''
Consider using ulid_with_prefix()
to generate a text ULID with a given prefix, to differentiate between different ID types.
select ulid_with_prefix('customer'); -- 'customer_01gqr5j1ebk31wv30wgp8ebehj'
select ulid_with_prefix('product'); -- 'product_01gqr5prjgsa77dhrxf2dt1dgv'
select ulid_with_prefix('order'); -- 'order_01gqr5q35n68jk0sycy1ntr083'
Not definitive, hastily ran on a Macbook, not representative of real-life usecases. The uuid()
SQL function comes from the official uuid.c
extension.
Test case | Time |
---|---|
generate_series() to generate 1 million rows |
28.5 ms ± 0.8 ms (1x ) |
Calling ulid_bytes() 1 million times |
88.4 ms ± 2.8 ms ,, 3.10 ± 0.13 slower |
Calling uuid() 1 million times |
141.6 ms ± 1.5 ms , or 4.97 ± 0.15 slower |
Calling ulid() 1 million times |
344.3 ms ± 11.9 ms , or 12.07 ± 0.53 slower |
So ulid_bytes()
is pretty fast, but returns an unreadable blob instead of a nicely formatted text ID. The ulid()
function does that, but is more than twice as slow than uuid()
.
However, generating 1 million ulid()
IDs in ~350ms is most likely "good enough" for most SQLite usecases.
Language | Install | |
---|---|---|
Python | pip install sqlite-ulid |
|
Datasette | datasette install datasette-sqlite-ulid |
|
Node.js | npm install sqlite-ulid |
|
Deno | deno.land/x/sqlite_ulid |
|
Ruby | gem install sqlite-ulid |
|
Rust | cargo add sqlite-ulid |
|
Github Release |
The Releases page contains pre-built binaries for Linux x86_64, MacOS, and Windows.
For Python developers, install the sqlite-ulid
package with:
pip install sqlite-ulid
import sqlite3
import sqlite_ulid
db = sqlite3.connect(':memory:')
db.enable_load_extension(True)
sqlite_ulid.load(db)
db.execute('select ulid()').fetchone()
# ('01gr7gwc5aq22ycea6j8kxq4s9',)
See python/sqlite_ulid
for more details.
For Node.js developers, install the sqlite-ulid
npm package with:
npm install sqlite-ulid
import Database from "better-sqlite3";
import * as sqlite_ulid from "sqlite-ulid";
const db = new Database(":memory:");
db.loadExtension(sqlite_ulid.getLoadablePath());
See npm/sqlite-ulid/README.md
for more details.
For Deno developers, use the deno.land/x/sqlite_ulid module:
import { Database } from "https://deno.land/x/sqlite3@0.8.0/mod.ts";
import * as sqlite_ulid from "https://deno.land/x/sqlite_ulid@v${VERSION}/mod.ts";
const db = new Database(":memory:");
db.enableLoadExtension = true;
sqlite_ulid.load(db);
const [version] = db.prepare("select ulid_version()").value<[string]>()!;
console.log(version);
And for Datasette, install the datasette-sqlite-ulid
plugin with:
datasette install datasette-sqlite-ulid
See python/datasette_sqlite_ulid
for more details.
If you want to use sqlite-ulid
as a Runtime-loadable extension, Download the ulid0.dylib
(for MacOS), ulid0.so
(Linux), or ulid0.dll
(Windows) file from a release and load it into your SQLite environment.
Note: The
0
in the filename (ulid0.dylib
/ulid0.so
/ulid0.dll
) denotes the major version ofsqlite-ulid
. Currentlysqlite-ulid
is pre v1, so expect breaking changes in future versions.
For example, if you are using the SQLite CLI, you can load the library like so:
.load ./ulid0
select ulid_version();
-- v0.1.0
In Python, you should prefer the sqlite-ulid
Python package. However, you can manually load a pre-compiled extension with the builtin sqlite3 module:
import sqlite3
con = sqlite3.connect(":memory:")
con.enable_load_extension(True)
con.load_extension("./ulid0")
print(con.execute("select ulid_version()").fetchone())
# ('v0.1.0',)
Or in Node.js using better-sqlite3:
const Database = require("better-sqlite3");
const db = new Database(":memory:");
db.loadExtension("./ulid0");
console.log(db.prepare("select ulid_version()").get());
// { 'ulid_version()': 'v0.1.0' }
With Datasette, you should prefer the datasette-sqlite-ulid
Datasette plugin. However, you can manually load a pre-compiled extension into a Datasette instance like so:
datasette data.db --load-extension ./ulid0
I (Alex 👋🏼) spent a lot of time and energy on this project and many other open source projects. If your company or organization uses this library (or you're feeling generous), then please consider supporting my work, or share this project with a friend!