🃏 prisma-test-utils

⚠️ This project is temporarily unmaintained. Please reach out in the #prisma2-preview channel on the Prisma Slack if you're interested in collaborating on it. We're planning to pick up development soon again (see this issue).

In testing workflows, generating seed data usually includes a lot of boilerplate and hardcoded fixtures that need to be migrated with changing code.

prisma-test-utils solves this by generating test util functions based on your Prisma Schema. As your application evolves, the generated data also evolves deterministically.

Features

• 🙈 Data model agnostic: Optimised for you datamodel.
• 🦑 Flexible: Cherry picked default settings.
• 🐶 Out-of-the-box usage: Plug-in generator for your Prisma Schema.
• 🐠 Seeds mock data: Populates your database with mock data.
• 🦋 Per-test database: Creates an isolated database for each test.

Installation

TBD

Configuration

1generator testutils {
2  provider = "prisma-test-utils"
3  output = "node_modules/@generated/prisma-test-utils"
4}

Usage

prisma-test-utils packs two incredibly useful functions. The first one, seed, helps you populate your data with vast amount of data. The second one, pool, can be used to create a pool of databases that you can use during testing, and are wiped after you've finished.

Seeding

1import Photon from '@generated/photon'
2import seed from '@generated/test-utils/seed'
3
4test('test with seed data', async () => {
5  await seed({
6    client,
7    models: kit => ({
8      _: {
9        /* Default number of instances. */
10        amount: 500,
11      },
12      Blog: {
13        factory: {
14          /* Use functions from the kit. */
15          name: kit.faker.sentence,
16          /* Define custom mocks. */
17          description: 'My custom blog description',
18          /* Define custom mock functions. */
19          entry: () => {
20            return `A generated entry from the function.`
21          },
22          /* Manage relations. */
23          posts: {
24            max: 100,
25          },
26        },
27      },
28    }),
29  })
30
31  const blogs = await client.blogs()
32})

Options

It is possible to selectively override the seed generation making the seeding workflow very flexible.

All options are autogenerated and checked at compile time. You'll be warned about any relation constraints that your datamodel presents.

1beforeAll(async () => {
2  const data = await seed(
3    photon,
4    bag => ({
5      Post: {
6        amount: 5,
7        factory: {
8          published: 'false',
9        },
10      },
11    }),
12    {
13      seed: 42,
14      silent: false,
15      instances: 5,
16    },
17  )
18})

Database Pools

We can configure our pool requirements before running any test cases.

1import SQLitePool, { Pool } from '@generated/prisma-test-utils'
2
3let pool: Pool
4
5beforeAll(async () => {
6  pool = new SQLitePool({
7    pool: {
8      min: 3,
9      max: 5,
10    },
11  })
12})

This allows us to request an isolated database per test case

1test('one of my parallel tests', async () => {
2  /* Acquire new db instance. */
3  const db = await pool.getDBInstance()
4
5  // Write the test case logic
6  const client = new Photon({
7    datasources: {
8      db: db.url,
9    },
10  })
11
12  /* Release the instance. */
13  client.disconnect()
14  pool.releaseDBInstance(db)
15})

API

1/* All pool instances. */
2
3class Pool {
4  async getDBInstance(): Promise<DBInstance>
5  async releaseDBInstance(db: DBInstance): Promise<void>
6  async run<T>(fn: (db: DBInstance) => Promise<T>): Promise<T>
7  async drain(): Promise<void>
8}
9
10/* PostgreSQL */
11
12interface PostgreSQLConnection {
13  host: string
14  port: number
15  user: string
16  password?: string
17  database: string
18  schema: string
19}
20
21interface PostgreSQLPoolOptions {
22  connection: (id: string) => PostgreSQLConnection
23  pool?: {
24    max?: number
25  }
26}
27
28/* MySQL */
29
30interface MySQLConnection {
31  host: string
32  port: string
33  user: string
34  password?: string
35  database: string
36}
37
38interface MySQLPoolOptions {
39  connection: (id string) => MySQLConnection
40  pool?: {
41    max?: number
42  }
43}
44
45/* SQLite */
46
47interface SQLitePoolOptions {
48  databasePath: (id: string) => string
49  pool?: {
50    max?: number
51  }
52}

Local development

:construction: NOTE: Please comment your work and read the comments that are already in there.

I didn't want to remove half the files of this library - the pool part - and that's why there's more files than you'll usually need for developing seed utils. Please don't remove the extra files as this work very nicely the way it is.

The most important file for seeding is src/static/seed.ts and src/intellisense/seed.ts. The first one is the logic and the second one provides customized types.

Furthermore:

• To create a new DB instance: Spin up the docker-compose up -d and use TablePlus or alternative to import the sql.

• To examine the behaviour of the library: Uncomment src/__test file and start the debugger. src/__test file references files in the tests/seed folder. Read on about that!

• To setup tests/seed folder: Navigate to that directory and use yarn prisma2 <cmd> to setup everything that you need. I usually use one of these functions:

  • yarn prisma2 introspect yarn prisma2 introspect --url="postgresql://prisma:prisma@127.0.0.1/ruma"
  • yarn prisma2 migrate up --experimental
  • yarn prisma2 migrate save --name "init" --experimental
  • yarn prisma2 generate. I have also added the README.md file in there with missing generator definitions from introspection. Copy and paste them to the top.

• To push changes: Preferablly do a PR, and don't forget to comment out src/__test file.

• To publish a new version: I use npx np --no-tests.

• To apply changes in the intellisense: Run yarn build.

• To test the utils outside debugger: Run yarn build:runtime.

• To get new VSCode type definitions after changing the schema: Reload VSCode :slightly_smiling_face:

LICENSE

MIT @ Prisma