[Python] Ignore some default values in pydantic during JSON schema generation

(At the time of writing, this is on pydantic version 2.11.7) I have the following MRE:

import json
import time
from pydantic import BaseModel, Field

class SeededModel(BaseModel):
seed: int = Field(default_factory=lambda _: int(time.time() * 1000))
sensible_default: bool = False

class Base(BaseModel):
settings: SeededModel = SeededModel()

config_schema = Base.model_json_schema()
print(json.dumps(config_schema, indent=2))


When I go to run this, I get the following JSON schema:

{
"$defs": {
"SeededModel": {
"properties": {
"seed": {
"title": "Seed",
"type": "integer"
},
"sensible_default": {
"default": false,
"title": "Sensible Default",
"type": "boolean"
}
},
"title": "SeededModel",
"type": "object"
}
},
"properties": {
"settings": {
"$ref": "#/$defs/SeededModel",
"default": {
"seed": 1752596882987,
"sensible_default": false
}
}
},
"title": "Base",
"type": "object"
}


Is there a nice way to provide a default for Base#settings that preserves the default value for SeededModel#sensible_default but not for SeededModel#seed?

Continue reading...