Trying to unit test your Next.js API route handlers? Want to avoid mucking around with custom servers and writing boring test infra just to get some unit tests working? Want your handlers to receive actual NextApiRequest and NextApiResponse objects rather than having to hack something together with express? Then look no further! This package allows you to test your Next.js API routes/handlers in an isolated Next.js-like environment simply, quickly, and without hassle.
One day, Next.js might expose an API for testing our handlers (or at least
generating proper NextApiRequest and NextApiResponse objects). Until then,
there's next-test-api-route-handler
! 🎉🎉🎉
This package uses Next.js's internal API resolver to precisely emulate API route handling. Since this is not a public or documented interface, the next dependency is locked to
. This package is automatically tested for compatibility with each full release of Next.js:
. Any incompatibilities, albeit rare, will be fixed as they are detected.
If you're looking for a version of next-test-api-route-handler
that works with
some long-ago previous version of Next.js, consult CHANGELOG.md.
npm install next-test-api-route-handler
import { testApiHandler } from 'next-test-api-route-handler'
The interface for testApiHandler
looks like this:
async function testApiHandler({ requestPatcher, responsePatcher, params, handler, test }: {
requestPatcher?: (req: IncomingMessage) => void,
responsePatcher?: (res: ServerResponse) => void,
params?: Record<string, unknown>,
handler: (req: NextApiRequest, res: NextApiResponse) => Promise<void>
test: (obj: { fetch: (init?: RequestInit) => ReturnType<typeof fetch> }) => Promise<void>,
})
requestPatcher
is a function that receives an
IncomingMessage.
Use this function to modify the request before it's injected into Next.js's
resolver.
responsePatcher
is a function that receives an
ServerResponse.
Use this function to modify the response before it's injected into Next.js's
resolver.
params
is an object representing "processed" dynamic routes, e.g. testing a
handler that expects /api/user/:id
requires params: { id: ...}
. This should
not be confused with requiring query string parameters, which are parsed out
from the url and added to the params object automatically.
handler
is the actual route handler under test (usually imported from
pages/api/*
). It should be an async function that accepts
NextApiRequest
and
NextApiResponse
objects as its two parameters.
test
is a function that returns a promise (or async) where test assertions can
be run. This function receives one parameter: fetch
, which is a simple
unfetch instance (note
that the url parameter, i.e. the first parameter in
fetch(...)
, is
omitted). Use this to send HTTP requests to the handler under test.
Suppose we have an API endpoint we use to test our application's error handling.
The endpoint responds with status code HTTP 200
for every request except the
10th, where status code HTTP 555
is returned instead.
How might we test that this endpoint responds with HTTP 555
once for every
nine HTTP 200
responses?
import * as UnreliableHandler from '../pages/api/unreliable'
import { testApiHandler } from 'next-test-endpoint'
import { shuffle } from 'fast-shuffle'
import type { WithConfig } from '@ergodark/next-types'
// Import the handler under test from the pages/api directory and respect the
// Next.js config object if it's exported
const unreliableHandler: WithConfig<typeof UnreliableHandler.default> = UnreliableHandler.default;
unreliableHandler.config = UnreliableHandler.config;
it('injects contrived errors at the required rate', async () => {
expect.hasAssertions();
// Signal to the endpoint (which is configurable) that there should be 1
// error among every 10 requests
process.env.REQUESTS_PER_CONTRIVED_ERROR = '10';
const expectedReqPerError = parseInt(process.env.REQUESTS_PER_CONTRIVED_ERROR);
// Returns one of ['GET', 'POST', 'PUT', 'DELETE'] at random
const getMethod = () => shuffle(['GET', 'POST', 'PUT', 'DELETE'])[0];
// Returns the status code from a response object
const getStatus = async (res: Promise<Response>) => (await res).status;
await testApiHandler({
handler: unreliableHandler,
test: async ({ fetch }) => {
// Run 20 requests with REQUESTS_PER_CONTRIVED_ERROR = '10' and
// record the results
const results1 = await Promise.all([
...[...Array(expectedReqPerError - 1)].map(_ => getStatus(fetch({ method: getMethod() }))),
getStatus(fetch({ method: getMethod() })),
...[...Array(expectedReqPerError - 1)].map(_ => getStatus(fetch({ method: getMethod() }))),
getStatus(fetch({ method: getMethod() }))
].map(p => p.then(s => s, _ => null)));
process.env.REQUESTS_PER_CONTRIVED_ERROR = '0';
// Run 10 requests with REQUESTS_PER_CONTRIVED_ERROR = '0' and
// record the results
const results2 = await Promise.all([
...[...Array(expectedReqPerError)].map(_ => getStatus(fetch({ method: getMethod() }))),
].map(p => p.then(s => s, _ => null)));
// We expect results1 to be an array with eighteen `200`s and two
// `555`s in any order
// https://github.com/jest-community/jest-extended#toincludesamemembersmembers
// because responses could be received out of order
expect(results1).toIncludeSameMembers([
...[...Array(expectedReqPerError - 1)].map(_ => 200),
555,
...[...Array(expectedReqPerError - 1)].map(_ => 200),
555
]);
// We expect results2 to be an array with ten `200`s
expect(results2).toStrictEqual([
...[...Array(expectedReqPerError)].map(_ => 200),
]);
}
});
});
Suppose we have an authenticated API endpoint our application uses to search for flights. The endpoint responds with an array of flights satisfying the query.
How might we test that this endpoint returns flights in our database as expected?
import * as V3FlightsSearchHandler from '../pages/api/v3/flights/search'
import { testApiHandler } from 'next-test-endpoint'
import { DUMMY_API_KEY as KEY, getFlightData, RESULT_SIZE } from '../backend'
import type { WithConfig } from '@ergodark/next-types'
// Import the handler under test from the pages/api directory and respect the
// Next.js config object if it's exported
const v3FlightsSearchHandler: WithConfig<typeof V3FlightsSearchHandler.default> = V3FlightsSearchHandler.default;
v3FlightsSearchHandler.config = V3FlightsSearchHandler.config;
it('returns expected public flights with respect to match', async () => {
expect.hasAssertions();
// Get the flight data currently in the test database
const flights = getFlightData();
// Take any JSON object and stringify it into a URL-ready string
const encode = (o: Record<string, unknown>) => encodeURIComponent(JSON.stringify(o));
// This function will return in order the URIs we're interested in testing
// against our handler. Query strings are parsed automatically, though we
// also could have used `params` or `fetch({ ... })` itself instead.
//
// Example URI for `https://google.com/search?params=yes` would be
// `/search?params=yes`
const genUrl = function*() {
yield `/?match=${encode({ airline: 'Spirit' })}`; // i.e. we want all the flights matching Spirit airlines!
yield `/?match=${encode({ type: 'departure' })}`;
yield `/?match=${encode({ landingAt: 'F1A' })}`;
yield `/?match=${encode({ seatPrice: 500 })}`;
yield `/?match=${encode({ seatPrice: { $gt: 500 }})}`;
yield `/?match=${encode({ seatPrice: { $gte: 500 }})}`;
yield `/?match=${encode({ seatPrice: { $lt: 500 }})}`;
yield `/?match=${encode({ seatPrice: { $lte: 500 }})}`;
}();
await testApiHandler({
// Patch the request object to include our dummy URI
requestPatcher: req => {
req.url = genUrl.next().value || undefined;
// Could have done this instead of fetch({ headers: { KEY }}) below:
// req.headers = { KEY };
},
handler: v3FlightsSearchHandler,
test: async ({ fetch }) => {
// 8 URLS from genUrl means 8 calls to fetch:
const responses = await Promise.all([...Array(8)].map(_ => {
return fetch({ headers: { KEY }}).then(r => r.ok ? r.json() : r.status);
}));
// We expect all of the responses to be 200
expect(responses.some(o => !o?.success)).toBe(false);
// We expect the array of flights returned to match our
// expectations given we already know what dummy data will be
// returned:
// https://github.com/jest-community/jest-extended#toincludesamemembersmembers
// because responses could be received out of order
expect(responses.map(r => r.flights)).toIncludeSameMembers([
flights.filter(f => f.airline == 'Spirit').slice(0, RESULT_SIZE),
flights.filter(f => f.type == 'departure').slice(0, RESULT_SIZE),
flights.filter(f => f.landingAt == 'F1A').slice(0, RESULT_SIZE),
flights.filter(f => f.seatPrice == 500).slice(0, RESULT_SIZE),
flights.filter(f => f.seatPrice > 500).slice(0, RESULT_SIZE),
flights.filter(f => f.seatPrice >= 500).slice(0, RESULT_SIZE),
flights.filter(f => f.seatPrice < 500).slice(0, RESULT_SIZE),
flights.filter(f => f.seatPrice <= 500).slice(0, RESULT_SIZE),
]);
}
});
// We expect these two to fail with 400 errors
await testApiHandler({
handler: v3FlightsSearchHandler,
requestPatcher: req => { req.url = `/?match=${encode({ ffms: { $eq: 500 }})}` },
test: async ({ fetch }) => expect((await fetch({ headers: { KEY }})).status).toBe(400)
});
await testApiHandler({
handler: v3FlightsSearchHandler,
requestPatcher: req => { req.url = `/?match=${encode({ bad: 500 })}` },
test: async ({ fetch }) => expect((await fetch({ headers: { KEY }})).status).toBe(400)
});
});
See test/index.test.ts for more examples.
Documentation can be found under docs/
and can be built with
npm run build-docs
.
New issues and pull requests are always welcome and greatly appreciated! If you submit a pull request, take care to maintain the existing coding style and add unit tests for any new or changed functionality. Please lint and test your code, of course!
Use npm run build
to compile src/
into dist/
, which is what makes it into
the published package. Use npm run build-docs
to re-build the documentation.
Use npm test
to run the unit tests, npm run check-build
to run the e2e
tests, and check-types
to run a type check. Use npm run list-tasks
to list
all available run scripts.
Note that using the NPM run scripts to build the documentation and distributables requires a linux-like development environment. None of the run scripts are likely to work on non-POSIX environments. If you're on Windows, use WSL.
This package is published using
publish-please via npx publish-please
.
- 1.0.x Initial release