Fix schema for set and frozenset with default value

[aminalaee]

Change Summary

Fixes JSON schema for set and frozenset when they have default values, the output json is not valid and cannot be serialized.

For example:

import json
from pydantic import BaseModel

class Test(BaseModel):
    x = {1}

print(Test.schema())

It will produce:

{
  "title": "Test",
  "type": "object",
  "properties": {
    "x": {
      "title": "X",
      "type": "array",
      "default": {1},
      "items": {
        "type": "integer"
      },
      "uniqueItems": true
    }
  },
  "required": [
    "x"
  ]
}

But instead should be:

{
  "title": "Test",
  "type": "object",
  "properties": {
    "x": {
      "title": "X",
      "default": [
        1,
      ],
      "type": "array",
      "items": {},
      "uniqueItems": true
    }
  }
}

Related issue number

Fixes #4136

Checklist

• Unit tests for the changes exist
• Tests pass on CI and coverage remains at 100%
• Documentation reflects the changes where applicable
• changes/<pull request or issue id>-<github username>.md file added describing change
  (see changes/README.md for details)
• My PR is ready to review, please add a comment including the phrase "please review" to assign reviewers

[aminalaee]

please review

[LordFckHelmchen]

I'm really not into the code base - so this is just a non-constructive critisism, but this change would narrow the scope compared to the original sequence_like check. What happens when these instances (set, frozenset, ...) are passed in?

[aminalaee]

I'm also very new to the codebase, but I think the sequence_like usage is not very accurate here.

If you look further, after these conditions it will try to use the pydantic_encoder for all unhandled types which in this case is set and frozenset.
And also there are tests for all cases and new ones for set and frozenset, so I think this is not narrowing down any type, just setting up different encoders for these types.

[LordFckHelmchen]

sounds reasonable! now we only need @samuelcolvin or @PrettyWood to approve :-)

[hramezani]

Thanks @aminalaee for this patch 👍 LGTM

I think it would be good to add frozenset to the documentation about The field schema mapping from Python / pydantic as well. it was missed there.

[hramezani]

please update

[aminalaee]

please review

[samuelcolvin]

Thanks so much @aminalaee.