Create mocks with the mock function.
import { test, expect, mock } from "bun:test";
const random = mock(() => Math.random());
test("random", async () => {
const val = random();
expect(val).toBeGreaterThan(0);
expect(random).toHaveBeenCalled();
expect(random).toHaveBeenCalledTimes(1);
});
Alternatively, you can use the jest.fn() function, as in Jest. It behaves identically.
import { test, expect, jest } from "bun:test";
const random = jest.fn(() => Math.random());
test("random", async () => {
const val = random();
expect(val).toBeGreaterThan(0);
expect(random).toHaveBeenCalled();
expect(random).toHaveBeenCalledTimes(1);
});
The result of mock() is a new function that's been decorated with some additional properties.
import { mock } from "bun:test";
const random = mock((multiplier: number) => multiplier * Math.random());
random(2);
random(10);
random.mock.calls;
// [[ 2 ], [ 10 ]]
random.mock.results;
// [
// { type: "return", value: 0.6533907460954099 },
// { type: "return", value: 0.6452713933037312 }
// ]
The following properties and methods are implemented on mock functions.
.spyOn()
It's possible to track calls to a function without replacing it with a mock. Use spyOn() to create a spy; these spies can be passed to .toHaveBeenCalled() and .toHaveBeenCalledTimes().
import { test, expect, spyOn } from "bun:test";
const ringo = {
name: "Ringo",
sayHi() {
console.log(`Hello I'm ${this.name}`);
},
};
const spy = spyOn(ringo, "sayHi");
test("spyon", () => {
expect(spy).toHaveBeenCalledTimes(0);
ringo.sayHi();
expect(spy).toHaveBeenCalledTimes(1);
});
Module mocks with mock.module()
Module mocking lets you override the behavior of a module. Use mock.module(path: string, callback: () => Object) to mock a module.
import { test, expect, mock } from "bun:test";
mock.module("./module", () => {
return {
foo: "bar",
};
});
test("mock.module", async () => {
const esm = await import("./module");
expect(esm.foo).toBe("bar");
const cjs = require("./module");
expect(cjs.foo).toBe("bar");
});
Like the rest of Bun, module mocks support both import and require.
Overriding already imported modules
If you need to override a module that's already been imported, there's nothing special you need to do. Just call mock.module() and the module will be overridden.
import { test, expect, mock } from "bun:test";
// The module we're going to mock is here:
import { foo } from "./module";
test("mock.module", async () => {
const cjs = require("./module");
expect(foo).toBe("bar");
expect(cjs.foo).toBe("bar");
// We update it here:
mock.module("./module", () => {
return {
foo: "baz",
};
});
// And the live bindings are updated.
expect(foo).toBe("baz");
// The module is also updated for CJS.
expect(cjs.foo).toBe("baz");
});
Hoisting & preloading
If you need to ensure a module is mocked before it's imported, you should use --preload to load your mocks before your tests run.
// my-preload.ts
import { mock } from "bun:test";
mock.module("./module", () => {
return {
foo: "bar",
};
});
bun test --preload ./my-preload
To make your life easier, you can put preload in your bunfig.toml:
[test]
# Load these modules before running tests.
preload = ["./my-preload"]
What happens if I mock a module that's already been imported?
If you mock a module that's already been imported, the module will be updated in the module cache. This means that any modules that import the module will get the mocked version, BUT the original module will still have been evaluated. That means that any side effects from the original module will still have happened.
If you want to prevent the original module from being evaluated, you should use --preload to load your mocks before your tests run.
__mocks__ directory and auto-mocking
Auto-mocking is not supported yet. If this is blocking you from switching to Bun, please file an issue.
Implementation details
Module mocks have different implementations for ESM and CommonJS modules. For ES Modules, we've added patches to JavaScriptCore that allow Bun to override export values at runtime and update live bindings recursively.
As of Bun v1.0.19, Bun automatically resolves the specifier argument to mock.module() as though you did an import. If it successfully resolves, then the resolved specifier string is used as the key in the module cache. This means that you can use relative paths, absolute paths, and even module names. If the specifier doesn't resolve, then the original specifier is used as the key in the module cache.
After resolution, the mocked module is stored in the ES Module registry and the CommonJS require cache. This means that you can use import and require interchangeably for mocked modules.
The callback function is called lazily, only if the module is imported or required. This means that you can use mock.module() to mock modules that don't exist yet, and it means that you can use mock.module() to mock modules that are imported by other modules.