The problem is that CRA's default Jest setup automatically resets the mocks, which removes the mockResolvedValue
you set.
One way to solve this, which also gives you more control to have different values in different tests (e.g. to test error handling) and assert on what it was called with, is to expose the mock function from the module too:
export const mockListForOrg = jest.fn();
export const Octokit = jest.fn().mockImplementation(() => ({
repos: {
listForOrg: mockListForOrg,
},
}));
Then you configure the value you want in the test, after Jest would have reset it:
import { mockListForOrg } from "@octokit/rest";
import { foo } from "./foo";
test("foo should be true", async () => {
mockListForOrg.mockResolvedValueOnce([1, 2]);
expect(await foo()).toEqual([1, 2]);
});
Another option is to add the following into your package.json
to override that configuration, per this issue:
{
...
"jest": {
"resetMocks": false
}
}
This could lead to issues with mock state (calls received) being retained between tests, though, so you'll need to make sure they're getting cleared and/or reset somewhere.
Note that you generally shouldn't mock what you don't own, though - if the interface to @octokit/rest
changes your tests will continue to pass but your code won't work. To avoid this issue, I would recommend either or both of:
- Moving the assertions to the transport layer, using e.g. MSW to check that the right request gets made; or
- Writing a simple facade that wraps
@octokit/rest
, decoupling your code from the interface you don't own, and mocking that;
along with higher-level (end-to-end) tests to make sure everything works correctly with the real GitHub API.
In fact, deleting the mocks and writing such a test using MSW:
import { rest } from "msw";
import { setupServer } from "msw/node";
import { foo } from "./foo";
const server = setupServer(rest.get("https://api.github.com/orgs/octokit/repos", (req, res, ctx) => {
return res(ctx.status(200), ctx.json([1, 2]));
}));
beforeAll(() => server.listen());
afterAll(() => server.close());
test("foo should be true", async () => {
expect(await foo()).toEqual([1, 2]);
});
exposes that the current assumption about what octokit.repos.listForOrg
would return is inaccurate, because this test fails:
● foo should be true
expect(received).toEqual(expected) // deep equality
Expected: [1, 2]
Received: {"data": [1, 2], "headers": {"content-type": "application/json", "x-powered-by": "msw"}, "status": 200, "url": "https://api.github.com/orgs/octokit/repos?type=public"}
13 |
14 | test("foo should be true", async () => {
> 15 | expect(await foo()).toEqual([1, 2]);
| ^
16 | });
17 |
at Object.<anonymous> (src/foo.test.js:15:25)
Your implementation should actually look something more like:
export async function foo() {
const { data } = await octokit.repos.listForOrg({ org: "octokit", type: "public" });
return data;
}
or:
export function foo() {
return octokit.repos.listForOrg({ org: "octokit", type: "public" }).then(({ data }) => data);
}