@auth/kysely-adapter
Official Kysely adapter for Auth.js / NextAuth.js.
Installationβ
- npm
- Yarn
- pnpm
npm install kysely @auth/kysely-adapter
yarn add kysely @auth/kysely-adapter
pnpm add kysely @auth/kysely-adapter
KyselyAdapter()β
KyselyAdapter(
db
):Adapter
Setupβ
This adapter supports the same first party dialects that Kysely (as of v0.24.2) supports: PostgreSQL, MySQL, and SQLite. The examples below use PostgreSQL with the pg client.
- npm
- Yarn
- pnpm
npm install pg
npm install --save-dev @types/pg
yarn add pg
yarn add --dev @types/pg
pnpm add pg
pnpm add --save-dev @types/pg
import NextAuth from "next-auth"
import GoogleProvider from "next-auth/providers/google"
import { KyselyAdapter } from "@auth/kysely-adapter"
import { db } from "../../../db"
export default NextAuth({
adapter: KyselyAdapter(db),
providers: [
GoogleProvider({
clientId: process.env.GOOGLE_CLIENT_ID,
clientSecret: process.env.GOOGLE_CLIENT_SECRET,
}),
],
})
Kysely's constructor requires a database interface that contains an entry with an interface for each of your tables. You can define these types manually, or use kysely-codegen
/ prisma-kysely
to automatically generate them. Check out the default models required by Auth.js.
import { PostgresDialect } from "kysely"
import { Pool } from "pg"
// This adapter exports a wrapper of the original `Kysely` class called `KyselyAuth`,
// that can be used to provide additional type-safety.
// While using it isn't required, it is recommended as it will verify
// that the database interface has all the fields that Auth.js expects.
import { KyselyAuth } from "@auth/kysely-adapter"
import type { GeneratedAlways } from "kysely"
interface Database {
User: {
id: GeneratedAlways<string>
name: string | null
email: string
emailVerified: Date | null
image: string | null
}
Account: {
id: GeneratedAlways<string>
userId: string
type: string
provider: string
providerAccountId: string
refresh_token: string | null
access_token: string | null
expires_at: number | null
token_type: string | null
scope: string | null
id_token: string | null
session_state: string | null
}
Session: {
id: GeneratedAlways<string>
userId: string
sessionToken: string
expires: Date
}
VerificationToken: {
identifier: string
token: string
expires: Date
}
}
export const db = new KyselyAuth<Database>({
dialect: new PostgresDialect({
pool: new Pool({
host: process.env.DATABASE_HOST,
database: process.env.DATABASE_NAME,
user: process.env.DATABASE_USER,
password: process.env.DATABASE_PASSWORD,
}),
}),
})
An alternative to manually defining types is generating them from the database schema using kysely-codegen, or from Prisma schemas using prisma-kysely. When using generated types with KyselyAuth
, import Codegen
and pass it as the second generic arg:
import type { Codegen } from "@auth/kysely-adapter"
new KyselyAuth<Database, Codegen>(...)
Schemaβ
import { Kysely, sql } from "kysely"
export async function up(db: Kysely<any>): Promise<void> {
await db.schema
.createTable("User")
.addColumn("id", "uuid", (col) =>
col.primaryKey().defaultTo(sql`gen_random_uuid()`)
)
.addColumn("name", "text")
.addColumn("email", "text", (col) => col.unique().notNull())
.addColumn("emailVerified", "timestamptz")
.addColumn("image", "text")
.execute()
await db.schema
.createTable("Account")
.addColumn("id", "uuid", (col) =>
col.primaryKey().defaultTo(sql`gen_random_uuid()`)
)
.addColumn("userId", "uuid", (col) =>
col.references("User.id").onDelete("cascade").notNull()
)
.addColumn("type", "text", (col) => col.notNull())
.addColumn("provider", "text", (col) => col.notNull())
.addColumn("providerAccountId", "text", (col) => col.notNull())
.addColumn("refresh_token", "text")
.addColumn("access_token", "text")
.addColumn("expires_at", "bigint")
.addColumn("token_type", "text")
.addColumn("scope", "text")
.addColumn("id_token", "text")
.addColumn("session_state", "text")
.execute()
await db.schema
.createTable("Session")
.addColumn("id", "uuid", (col) =>
col.primaryKey().defaultTo(sql`gen_random_uuid()`)
)
.addColumn("userId", "uuid", (col) =>
col.references("User.id").onDelete("cascade").notNull()
)
.addColumn("sessionToken", "text", (col) => col.notNull().unique())
.addColumn("expires", "timestamptz", (col) => col.notNull())
.execute()
await db.schema
.createTable("VerificationToken")
.addColumn("identifier", "text", (col) => col.notNull())
.addColumn("token", "text", (col) => col.notNull().unique())
.addColumn("expires", "timestamptz", (col) => col.notNull())
.execute()
await db.schema
.createIndex("Account_userId_index")
.on("Account")
.column("userId")
.execute()
await db.schema
.createIndex("Session_userId_index")
.on("Session")
.column("userId")
.execute()
}
export async function down(db: Kysely<any>): Promise<void> {
await db.schema.dropTable("Account").ifExists().execute()
await db.schema.dropTable("Session").ifExists().execute()
await db.schema.dropTable("User").ifExists().execute()
await db.schema.dropTable("VerificationToken").ifExists().execute()
}
This schema is adapted for use in Kysely and is based upon our main schema.
For more information about creating and running migrations with Kysely, refer to the Kysely migrations documentation.
Naming conventionsβ
If mixed snake_case and camelCase column names is an issue for you and/or your underlying database system, we recommend using Kysely's CamelCasePlugin
(see the documentation here) feature to change the field names. This won't affect NextAuth.js, but will allow you to have consistent casing when using Kysely.
Parametersβ
βͺ db: Kysely
<Database
>
Returnsβ
Adapter
KyselyAuth<DB, T>
β
Wrapper over the original Kysely
class in order to validate the passed in
database interface. A regular Kysely instance may also be used, but wrapping
it ensures the database interface implements the fields that Auth.js
requires. When used with kysely-codegen
, the Codegen
type can be passed as
the second generic argument. The generated types will be used, and
KyselyAuth
will only verify that the correct fields exist.
Extendsβ
Kysely
<DB
>
Type parametersβ
βͺ DB extends T
βͺ T = Database