What is the proper way to capture a HTTP response with Puppeteer?

I believe you should do something along those lines. Note the callback function done.

What the code does, is that it attaches a listener for responses, then clicks the submit button. When a response is received it checks the status code, asserts it, and terminates the test by calling done.

You might want to have an if-statement that checks that it is the actual response from your form that you are checking in the callback, as the response handler might emit events for other concurrent requests.

it.only('returns a 400 response if email is taken', () => {  await page.goto(`${process.env.DOMAIN}/sign-up`)  await page.waitFor('input[id="Full Name"]')  await page.type('input[id="Full Name"]', 'Luke Skywalker')  await page.type('input[id="Email"]', 'LukeSkywalker@voyage.com')  await page.type('input[id="Password"]', 'LukeSkywalker123', {delay: 100})  page.on('response', (response) => {    if (      response.request().method === 'POST' &&       response.url === `${process.env.USERS_API_DOMAIN}/sessions`)     {      expect(response.status).toEqual(400)    }  })  await page.click('input[type="submit"]', {delay: 1000})})

I have not tested the code, but it should give you the right idea.

Edit: Adjusted to reflect what worked out in the end.

If you need to manipulate the request/response, use page.setRequestInterception(true) and page.on/page.once (as documented).

However, if all you need is to assert something about the response, the simplest and most idiomatic way to do so is with page.waitForResponse:

const updateDashboardResponse = await page.waitForResponse(response =>  response.url().includes('updateDashboard'));expect(updateDashboardResponse.status()).toBe(200);

This allows test flow to remain linear and avoids ambiguity around closing a test before a page.on handler receives a response event.

The accepted answer (which was also edited into the question) is incorrect. It introduces a race condition due to a 1 second delay added to the click call. At best, this slows down the test suite unnecessarily, and at worst it generates false failures should the request take longer than a second to resolve (unlikely if it's mocked, but it doesn't change the fact that the code is unsafe).

Whenever there's a callback in a Jest test case, the correct way to ensure it's been executed and all assertions depending on it firing have been made without adding artificial delays is to call done() from the callback. If there is a throw in the callback that makes done unreachable, call done(error) in the error handler to report the test case failure to Jest.

To do this, you'll need to add done as the parameter to the callback passed to the it, test or only function so that it's available in the block. This allows Jest's test runner to treat the test as asynchronous and not to resolve it until done is called. Without done, the test suite ignores the callback's assertions. async/await doesn't help because it's a separate asynchronous chain than the callback.

You only need to specify done as a parameter or return a promise (async implicitly returns a promise), never both. However, you'd still likely want to use await for Puppeteer library calls rather than then. You can use an async IIFE that eventually fires the done() call when all assertions have fired to get the best of both worlds.

For example,

it.only('returns a 400 response if email is taken', done => {  (async () => {    page.on('response', response => {      if (response.request().method === 'POST' &&           response.url === `${process.env.USERS_API_DOMAIN}/sessions`) {        try { /* try-catch pattern shown for illustration */          expect(response.status).toEqual(400);          done();        }         catch (err) {          done(err);        }      }    });        await page.goto(`${process.env.DOMAIN}/sign-up`);    await page.waitFor('input[id="Full Name"]');    await page.type('input[id="Full Name"]', 'Luke Skywalker');    await page.type('input[id="Email"]', 'LukeSkywalker@voyage.com');    await page.type('input[id="Password"]', 'LukeSkywalker123', {delay: 100});    await page.click('input[type="submit"]');  })();});

With this in mind, this answer shows a likely better approach using waitForResponse which lets you skip the callback and done entirely. The callback to waitForResponse is a string URL or function predicate that should return true for the target response that's being waited on:

it.only('returns a 400 response if email is taken', async () => {  await page.goto(`${process.env.DOMAIN}/sign-up`);  await page.waitFor('input[id="Full Name"]');  await page.type('input[id="Full Name"]', 'Luke Skywalker');  await page.type('input[id="Email"]', 'LukeSkywalker@voyage.com');  await page.type('input[id="Password"]', 'LukeSkywalker123', {delay: 100});  await page.click('input[type="submit"]');  const response = await page.waitForResponse(response =>    response.request().method === 'POST' &&     response.url === `${process.env.USERS_API_DOMAIN}/sessions`  );  expect(response.status).toEqual(400);});

I should also mention waitFor is deprecated in favor of waitForSelector in the above snippets and that .url and .method are functions. I haven't verified the above code; it's there to relate to the original post and show the high-level patterns.

Minimal example

index.html

This is the web page we're testing.

<!DOCTYPE html><html lang="en">  <head>    <meta charset="utf-8">  </head>  <body>    <button>Post</button>    <script>      document        .querySelector("button")        .addEventListener("click", e =>          fetch("https://jsonplaceholder.typicode.com/posts", {              method: "POST",              body: JSON.stringify({                title: "foo",                body: "bar",                userId: 1,              }),              headers: {                "Content-type": "application/json; charset=UTF-8",              },            })            .then(response => response.json())            .then(json => console.log(json))        )      ;    </script>  </body></html>