Is it possible to use enums/constants for reserved words?

[liordon]

I'm a new user and am somewhat vexed by the need to use explicit strings everywhere. seems like a likely case for bugs to hide in typos and the like. I would much rather have constants or str enums to use for all the reserved words of schemas.

I looked at the main repo and the docs for examples and I'm still seeing string literals. is that really the case that the code uses literals all the way down?
if so, is there a reason not to adopt static definitions for reserved words? if not, can anyone point me in the right direction?

many thanks!

[LehiSalvador]

For avoiding typos in Python-authored schemas, a project-local string enum works with jsonschema and with JSON serialization:

from enum import Enum
import json
from jsonschema import Draft202012Validator

class Keyword(str, Enum):
    TYPE = "type"
    PROPERTIES = "properties"
    REQUIRED = "required"

class JsonType(str, Enum):
    OBJECT = "object"
    INTEGER = "integer"

schema = {
    Keyword.TYPE: JsonType.OBJECT,
    Keyword.PROPERTIES: {"count": {Keyword.TYPE: JsonType.INTEGER}},
    Keyword.REQUIRED: ["count"],
}

Draft202012Validator.check_schema(schema)
validator = Draft202012Validator(schema)
assert validator.is_valid({"count": 1})
assert not validator.is_valid({"count": "1"})
assert not validator.is_valid({})
print(json.dumps(schema))  # ordinary JSON string keys/values

Keyword.TPYE raises AttributeError, so member-name typos fail immediately. On Python 3.11+, StrEnum is another option; the str, Enum form above also works on older Python versions supported by the library.

Keep keyword names separate from type values, and include the vocabulary for the draft(s) your application supports. This is an authoring convention: imported JSON still needs validation and tests. In particular, check_schema does not reject every unrecognized keyword, so it does not guarantee typo detection in raw string keys.

This demonstrates an application-level approach; it does not explain the maintainers' internal design rationale or claim an official complete keyword enum.

References: Python string enums, jsonschema validator API.

AI-assisted response. Example executed with Python 3.14 and jsonschema 4.25.1, including serialization and the positive/negative assertions above.