Content

• Fixtures
  • Base concepts
  • Test fixtures
  • Worker fixtures
• Annotations
  • Annotation API
  • Flaky tests
• Built-in fixtures
  • testWorkerIndex
  • testInfo
• Reporters
  • Reporter API
• Parameters
  • In the command line
  • Generating tests
• Parallelism and sharding
  • Workers
  • Shards
• Command line

Fixtures

Base concepts

Playwright test runner is based on the concept of the test fixtures. Test fixtures are used to establish environment for each test, giving the test everything it needs and nothing else. Here is how typical test environment setup differs between traditional BDD and the fixture-based one:

Without fixtures:

1describe('database', () => {
2  let database;
3  let table;
4
5  beforeAll(() => {
6    database = connect();
7  });
8
9  afterAll(() => {
10    database.dispose();
11  });
12
13  beforeEach(()=> {
14    table = database.createTable();
15  });
16
17  afterEach(()=> {
18    database.dropTable(table);
19  });
20
21  it('create user', () => {
22     table.insert();
23     // ...
24  });
25
26  it('update user', () => {
27     table.insert();
28     table.update();
29     // ...
30  });
31
32  it('delete user', () => {
33     table.insert();
34     table.delete();
35     // ...
36  });
37});

With fixtures:

1import { fixtures } from '@playwright/test-runner';
2
3const { it } = fixtures
4    .defineWorkerFixtures<{ database: Database }>({
5      database: async ({}, runTest) => {
6        const db = connect();
7        await runTest(db);
8        db.dispose();
9      }
10    })
11    .defineTestFixtures<{ table: Table }>({
12      table: async ({}, runTest) => {
13        const t = database.createTable();
14        await runTest(t);
15        database.dropTable(t);
16      }
17    });
18
19it('create user', ({ table }) => {
20    table.insert();
21    // ...
22});
23
24it('update user', ({ table }) => {
25    table.insert();
26    table.update();
27    // ...
28});
29
30it('delete user', ({ table }) => {
31    table.insert();
32    table.delete();
33    // ...
34});

You declare exact fixtures that the test needs and the runner initializes them for each test individually. Tests can use any combinations of the fixtures to tailor precise environment they need. You no longer need to wrap tests in describes that set up environment, everything is declarative and typed.

There are two types of fixtures: test and worker. Test fixtures are set up for each test and worker fixtures are set up for each process that runs test files.

Test fixtures

Test fixtures are set up for each test. Consider the following test file:

1// hello.spec.ts
2import { it, expect } from './hello.fixtures';
3
4it('hello world', ({ hello, world }) => {
5  expect(`${hello}, ${world}!`).toBe('Hello, World!');
6});
7
8it('hello test', ({ hello, test }) => {
9  expect(`${hello}, ${test}!`).toBe('Hello, Test!');
10});

It uses fixtures hello, world and test that are set up by the framework for each test run.

Here is how test fixtures are declared and defined:

1// hello.fixtures.ts
2import { fixtures as baseFixtures } from '@playwright/test-runner';
3export { expect } from '@playwright/test-runner';
4
5// Define test fixtures |hello|, |world| and |test|.
6
7type TestFixtures = {
8  hello: string;
9  world: string;
10  test: string;
11};
12
13const fixtures = baseFixtures.defineTestFixtures<TestFixtures>({
14  hello: async ({}, runTest) => {
15    // Set up fixture.
16    const value = 'Hello';
17    // Run the test with the fixture value.
18    await runTest(value);
19    // Clean up fixture.
20  },
21
22  world: async ({}, runTest) => {
23    await runTest('World');
24  },
25
26  test: async ({}, runTest) => {
27    await runTest('Test');
28  }
29});
30export const it = fixtures.it;

Fixtures can use other fixtures.

1  ...
2  helloWorld: async ({hello, world}, runTest) => {
3    await runTest(`${hello}, ${world}!`);
4  }
5  ...

With fixtures, test organization becomes flexible - you can put tests that make sense next to each other based on what they test, not based on the environment they need.

Worker fixtures

Playwright test runner uses worker processes to run test files. You can specify the maximum number of workers using --workers command line option. Similarly to how test fixtures are set up for individual test runs, worker fixtures are set up for each worker process. That's where you can set up services, run servers, etc. Playwright test runner will reuse the worker process for as many test files as it can, provided their worker fixtures match and hence environments are identical.

Here is how the test looks:

1// express.spec.ts
2import { it, expect } from './express.fixtures';
3import fetch from 'node-fetch';
4
5it('fetch 1', async ({ port }) => {
6  const result = await fetch(`http://localhost:${port}/1`);
7  expect(await result.text()).toBe('Hello World 1!');
8});
9
10it('fetch 2', async ({ port }) => {
11  const result = await fetch(`http://localhost:${port}/2`);
12  expect(await result.text()).toBe('Hello World 2!');
13});

And here is how fixtures are declared and defined:

1// express.fixtures.ts
2import { fixtures as baseFixtures } from '@playwright/test-runner';
3export { expect } from '@playwright/test-runner';
4import express from 'express';
5import type { Express } from 'express';
6
7// Declare worker fixtures.
8type ExpressWorkerFixtures = {
9  port: number;
10  express: Express;
11};
12const fixtures = baseFixtures.defineWorkerFixtures<ExpressWorkerFixtures>({
13  // Define |port| fixture that has unique value value of the worker process index.
14  port: async ({ testWorkerIndex }, runTest) => {
15    await runTest(3000 + testWorkerIndex);
16  },
17
18  // Define the express worker fixture, make it start automatically for every worker.
19  autoExpress: async ({ port }, runTest) => {
20    const app = express();
21    app.get('/1', (req, res) => {
22      res.send('Hello World 1!')
23    });
24    app.get('/2', (req, res) => {
25      res.send('Hello World 2!')
26    });
27    let server;
28    console.log('Starting server...');
29    await new Promise(f => {
30      server = app.listen(port, f);
31    });
32    console.log('Server ready');
33    await runTest(server);
34    console.log('Stopping server...');
35    await new Promise(f => server.close(f));
36    console.log('Server stopped');
37  },
38});
39export const it = fixtures.it;

Parameters

It is common to run tests in different configurations, for example when running web app tests against multiple browsers or testing two different versions of api endpoint. Playwright test runner supports this via parameters - define the parameter and start using it in a test or a fixture.

Consider the following test that uses an API url endpoint:

1// api.spec.ts
2import { it, expect } from './api.fixtures';
3import fetch from 'node-fetch';
4
5it('fetch 1', async ({ apiUrl }) => {
6  const result = await fetch(`${apiUrl}/hello`);
7  expect(await result.text()).toBe('Hello');
8});

Here is how to define the api version parameter:

1// api.fixtures.ts
2import { fixtures as baseFixtures } from '@playwright/test-runner';
3export { expect } from '@playwright/test-runner';
4
5const fixtures = baseFixtures
6    .defineParameter('version', 'API version', 'v1')
7    .defineWorkerFixtures<{ apiUrl: string }>({
8      apiUrl: async ({ version }, runTest) => {
9        const server = await startServer();
10        await runTest(`http://localhost/api/${version}`);
11        await server.close();
12      }
13    });
14export const it = fixtures.it;

In the command line

Given the example above, it is possible to run tests against the specific api version.

TODO: do not assume this is read top-bottom, each section should be self-contained

TODO: update the npx command.

1# Run against the default version (v1 in our case).
2npx test-runner tests
3# Run against the specified version.
4npx test-runner tests -p version=v2

Generating tests

TODO: do not assume this is read top-bottom, each section should be self-contained

It is also possible to run tests against multiple api versions.

1// api.fixtures.ts
2
3// Generate three versions of each test that directly or indirectly
4// depends on the |version| parameter.
5fixtures.generateParametrizedTests('version', ['v1', 'v2', 'v3']);

TODO: update the npx command.

1npx test-runner tests -p version=v1 -p version=v2