Custom Jest Matcher

Joist provides a toMatchEntity matcher for more pleasant assertions in Jest.

There are two main benefits:

Automatic loading of relations
Prettier actual vs. expected output

Automatic Loading of Relations

A potentially unwieldy pattern in tests is asserting against a “subtree” of data that was not initially loaded, i.e.:

const a1 = newAuthor(em);
// Invoke something that adds books with reviews
await addBooksAndReviews(a1);
// Because a1 is New we can access `books.get`, so this is easy...
expect(a1.books.get.length).toEqual(2);
// But beyond that, we can't drill into each book's reviews
// Compile error
expect(a1.books.get[0].reviews.get[0].title).toEqual("title");

And so test code would have to explicitly load what it wants to assert against, either with a separate await b1.reviews.load() for each individual relation (which can be tedious), or by declaring a “2nd version” of the entity with a populate load hint (which is better but also awkward):

const a1 = newAuthor(em);
// Invoke something that adds books with reviews
await addBooksAndReviews(a1);
// Preload the subtree of data we want to assert against
const a1_2 = await a1.populate({ books: "reviews" });
// Now we can use get
expect(a1_2.books.get.length).toEqual(2);
expect(a1_2.books.get[0].reviews.get[0].title).toEqual("title");

As a third option, toMatchEntity provides a toMatchObject-style API so that a test can idiomatically declare what the subtree of data should be:

const a1 = newAuthor(em);
// Invoke something that adds books with reviews
await addBooksAndReviews(a1);
expect(a1).toMatchEntity({
  books: [
    {
      title: "b1",
      reviews: [{ rating: 5 }],
    },
    {
      title: "b2",
      reviews: [{ rating: 4 }, { rating: -2 }],
    },
  ],
});

The upshot is that we get to assert against the entity “as if it’s JSON” or “just data”, and then toMatchEntity takes care of loading the various references and collections.

Prettier Output

Sometimes when entities are included in Jest failures, i.e. by Jest’s native toMatchObject, the Jest console output is ugly b/c Jest prints the internal implementation of the entity object (i.e. a failure for “expected a1” ends up printing the a1.books field, which is actually a OneToManyCollection with various internal flags/state, all of which are included in the output).

Even with ~3-4 entities in a native toMatchObject assertion, the output can get long and hard to visually parse.

Instead, toMatchEntity abbreviates each entity as simply its tagged id, so output for an assertion failure of “the collection expected two books of b:1 and b:2 but only had one b:2” will look like:

- Expected  - 0
+ Received  + 1

  Object {</>
    "books": Array [
+     "b:1",
      "b:2",
    ],
  }
`);

Note that if an entity is new, i.e. the test has not done em.flush (which is fine, tests should only em.flush if really necessary, to be as fast & lightweight as possible), the abbreviation for an unsaved Book will be a “test id” of b#1 where b is the entity’s tag, and the #1 is the index of that particular entity within the EntityManager’s entities of that type.

Installation

In your setupTests.ts, add:

import { toMatchEntity } from "joist-test-utils";

expect.extend({ toMatchEntity });