toContainEqual, toHaveProperty, toBeOneOf and return matchers treat any two Maps or Sets as equal

[harshit-d3v]

Describe the bug

Nine matchers that take a value to compare against compare Maps and Sets without iterableEquality, so any two of them match. All of these pass on 5.0.1:

const a = new Map([['a', 1]])
const b = new Map([['b', 2]])

expect([a]).toContainEqual(b)
expect([new Set([1])]).toContainEqual(new Set([2]))
expect({ m: a }).toHaveProperty('m', b)
expect(a).toBeOneOf([b])

const fn = vi.fn(() => a); fn()
expect(fn).toHaveReturnedWith(b)
expect(fn).toHaveLastReturnedWith(b)
expect(fn).toHaveNthReturnedWith(1, b)
// and the three *ResolvedWith variants

Controls fail correctly: expect(a).toEqual(b) and expect(fn).toHaveBeenCalledWith(b) both reject.

The six return/resolve matchers also drop customTesters, so a tester registered with expect.addEqualityTesters applies to toHaveBeenCalledWith but not to toHaveReturnedWith.

Cause

packages/expect/src/jest-expect.ts: toContainEqual and toHaveProperty call jestEquals(..., customTesters) without appending iterableEquality, unlike toEqual, toHaveBeenCalledWith and every other matcher in the file. The six return/resolve matchers call bare jestEquals(a, b) with no testers at all. toBeOneOf in custom-matchers.ts has the same gap.

Without iterableEquality, equals() falls through to own-enumerable-property comparison, and a Map or Set has none, so they compare equal.

Reproduction

The snippet above in any project with vitest@5.0.1. Every expect(...) line passes; each should fail.

System Info

vitest 5.0.1
node 22.18.0
win32 x64

Used Package Manager

npm

Validations

• Follow our Code of Conduct
• Read the Contributing Guidelines
• Read the docs
• Check that there isn't already an issue that reports the same bug
• Check that this is a concrete bug
• The provided reproduction is a minimal reproducible example of the bug

I have a fix with tests on a branch. I'll open the PR once #11295 is resolved, since outside contributors get one open PR at a time. Claude helped me find and verify this.

[sahiljadhav7]

@hi-ogawa can you review my pull request