toMatchObject pass when comparing any object to a empty object ({})

🐛 Bug Report

it:

  expect({a:[{}]}).toMatchObject({a:{}})
  expect({ a: undefined }).toMatchObject({ a: {} });
  expect({ a: null }).toMatchObject({});
  expect({ a: "string" }).toMatchObject({});
  expect({ a: 1 }).toMatchObject({});

should fail but pass

To Reproduce

expect({a:[{}]}).toMatchObject({a:{}})

Expected behavior

fail

Link to repl or repo (highly encouraged)

Please provide either a repl.it demo or a minimal repository on GitHub.

Issues without a reproduction link are likely to stall.

Run npx envinfo --preset jest

Paste the results here:

  System:
    OS: macOS 10.14.4
    CPU: (8) x64 Intel(R) Core(TM) i7-7700HQ CPU @ 2.80GHz
  Binaries:
    Node: 10.15.1 - ~/.nvm/versions/node/v10.15.1/bin/node
    Yarn: 1.13.0 - /usr/local/bin/yarn
    npm: 6.9.0 - ~/.nvm/versions/node/v10.15.1/bin/npm

[pedrottimark]

Yeah, I agree we can do better, because the reason is debatable in the words of Tim https://github.com/facebook/jest/issues/8308#issuecomment-482348139

A similar positive test passes (bad) for non-object received value when expected object is empty:

  test('toMatchObject without key', () => {
    const expected = {a: {}};
    const received = {a: 13};
    expect(received).toMatchObject(expected);
  });

Because Array.prototype.every passes if expected object has no keys and the logic seems to assume an object as received value at:

https://github.com/facebook/jest/blob/master/packages/expect/src/utils.ts#L267-L281

A similar negative test passes (good) if the object contains a property:

  test('toMatchObject with key', () => {
    const expected = {a: {key: 'value'}};
    const received = {a: [{key: 'value'}]};
    expect(received).not.toMatchObject(expected);
  });

Here is a question about timing: Do y’all think that a change for the better is breaking?

[mario-petrovic]

@pedrottimark a bit late on this discussion, but as i see in the PR that was the product of this issue still accepts {} to be equal to any object. In that PR circular dependencies are handled.

Just pointing to the fact that people should expect this still to happen. For me it was not clear that closing this issue did not make test with mentioned assertion fail.