@vankyle/storage-kysely
v0.2.7
Published
Kysely-based SQL metadata store for vankyle-storage.
Readme
@vankyle-hub/storage-kysely
IMetadataStore implementation backed by Kysely, supporting any Kysely-compatible SQL database.
Supported databases
- PostgreSQL
- MySQL
- SQL Server (MSSQL)
- SQLite
- Cloudflare D1 (via
D1Dialectfrom@vankyle-hub/storage-cloudflare)
Installation
pnpm add @vankyle-hub/storage-kysely kysely @vankyle-hub/storage-core @vankyle-hub/storage-sharedAdd your database driver (e.g. pg, mysql2, better-sqlite3) as needed.
Usage
import { KyselyMetadataStore } from "@vankyle-hub/storage-kysely";
import { Kysely, PostgresDialect } from "kysely";
import { Pool } from "pg";
import type { StorageDatabase } from "@vankyle-hub/storage-kysely";
const db = new Kysely<StorageDatabase>({
dialect: new PostgresDialect({
pool: new Pool({ connectionString: process.env.DATABASE_URL }),
}),
});
const metadata = new KyselyMetadataStore(db);Pass metadata to DefaultStorageService:
import { DefaultStorageService } from "@vankyle-hub/storage-core";
const service = new DefaultStorageService({ storage, metadata, bucket: "..." });Database schema
Six tables are created by the included migration:
| Table | Description |
|---|---|
| upload_sessions | Upload session state |
| uploaded_parts | Completed multipart upload parts |
| blobs | Physical objects in object storage |
| blob_references | Tracks which entities reference a blob |
| files | User-facing logical files |
| file_versions | Immutable snapshots linking a file to a blob |
Indexes
| Index | Columns | Type |
|---|---|---|
| idx_upload_sessions_status | status | Non-unique |
| idx_uploaded_parts_session | (session_id, part_number) | Unique |
| idx_blobs_sha256 | sha256 | Non-unique |
| idx_blobs_locator | (provider, bucket, object_key) | Unique |
Running migrations
@vankyle-hub/storage-kysely exports a migrations object compatible with the Kysely Migrator:
import { Migrator } from "kysely";
import { migrations } from "@vankyle-hub/storage-kysely";
const migrator = new Migrator({
db,
provider: {
getMigrations: async () => migrations,
},
});
const { error, results } = await migrator.migrateToLatest();
for (const result of results ?? []) {
console.log(`Migration "${result.migrationName}": ${result.status}`);
}
if (error) {
throw error;
}TypeScript schema type
StorageDatabase is the typed Kysely database interface. Import it for type inference:
import type { StorageDatabase } from "@vankyle-hub/storage-kysely";
const db = new Kysely<StorageDatabase>({ ... });Cloudflare D1
Use the D1Dialect from @vankyle-hub/storage-cloudflare to run KyselyMetadataStore on D1:
import { D1Dialect } from "@vankyle-hub/storage-cloudflare";
import { KyselyMetadataStore } from "@vankyle-hub/storage-kysely";
import { Kysely } from "kysely";
import type { StorageDatabase } from "@vankyle-hub/storage-kysely";
const db = new Kysely<StorageDatabase>({
dialect: new D1Dialect({ database: env.DB }),
});
const metadata = new KyselyMetadataStore(db);The SQL emitted by Kysely is SQLite-compatible, which D1 is based on. All six stores and the migration work without modification.
JSON fields
Some columns store JSON as text (metadata, providerSessionData, providerPartData). Row mappers in this package handle serialization (JSON.stringify) and deserialization (JSON.parse) automatically. Application code always receives typed JsonObject | undefined values.
SQLite date handling
SQLite returns DATE / DATETIME columns as strings. Row mappers convert them to Date objects via an ensureDate() helper. This is handled transparently — application code always receives Date values.
