Optimize output of non-required properties

[414owen]

For this schema:

{
  "properties": {
    "a": {
      "const": "a"
    },
    "b": {
      "const": "b"
    },
    "c": {
      "const": "c"
    }
  }
}

We currently produce the following regex (simplified, and line-broken)

\{("a":"a"(,"b":"b")?(,"c":"c")?
 |("a":"a",)?"b":"b"(,"c":"c")?
 |("a":"a",)?("b":"b",)?"c":"c")?\}

This works perfectly well, but contains redundancy. This is seen by the
fact that all three alternatives would match JSON with all three fields.

The different alternatives currently model which field is mandatory.

I propose that we make the alternatives model the choice of last field.
This will produce a regex like this:

\{("a":"a"
 |("a":"a",)?"b":"b"
 |("a":"a",)?("b":"b",)?"c":"c")?\}

This will give us a shorter, but 100% equivalent regex.

---

Both of the test cases I've changed have been run through this regex equivalence tester

[torymur]

That's great @414owen and thanks for sharing regex equivalence checker tool too! 🌟