SQLModel/SQLAlchemy ne convertit pas automatiquement ces objets en JSON

[cedvict]

First Check

• I added a very descriptive title here.
• I used the GitHub search to find a similar question and didn't find it.
• I searched the SQLModel documentation, with the integrated search.
• I already searched in Google "How to X in SQLModel" and didn't find any information.
• I already read and followed all the tutorial in the docs and didn't find an answer.
• I already checked if it is not related to SQLModel but to Pydantic.
• I already checked if it is not related to SQLModel but to SQLAlchemy.

Commit to Help

• I commit to help with one of those options 👆

Example Code

from sqlmodel import Field, Session, SQLModel, create_engine, Column, JSON



class Action(SQLModel):
    title: str
    location: str
    description: str | None = None

class Hero(SQLModel, table=True):
    id: int | None = Field(default=None, primary_key=True)
    name: str
    secret_name: str
    age: int | None = None
    actions: tuple[Action, ...] = Field(
        default_factory=tuple,
        sa_column=Column(JSON, nullable=True),
    )

Description

sqlalchemy.exc.StatementError: (builtins.TypeError) Object of type Action is not JSON serializable

Operating System

macOS

Operating System Details

No response

SQLModel Version

0.0.38

Python Version

3.14

Additional Context

No response

[delcenjo]

This error comes from SQLAlchemy's JSON column, not from SQLModel itself. A JSON column serializes its value with the standard library json module, and json.dumps doesn't know how to serialize a Pydantic / SQLModel object like Action — hence Object of type Action is not JSON serializable. A JSON column can only hold JSON-native types (dict, list, str, int, float, bool, None).

You have two clean options.

Option A — store plain dicts (simplest)

Type the column as JSON-native data and dump your models before assigning:

from typing import Any

from sqlmodel import Field, SQLModel, Column, JSON


class Action(SQLModel):
    title: str
    location: str
    description: str | None = None


class Hero(SQLModel, table=True):
    id: int | None = Field(default=None, primary_key=True)
    name: str
    actions: list[dict[str, Any]] = Field(default_factory=list, sa_column=Column(JSON))


hero = Hero(
    name="Deadpond",
    actions=[a.model_dump() for a in (Action(title="t", location="l"),)],
)

Read them back as models with Action.model_validate(d).

Option B — keep list[Action] and convert automatically with a TypeDecorator

If you want to keep the typed objects, wrap the JSON type so it serializes/deserializes the Pydantic models for you:

from pydantic import TypeAdapter
from sqlalchemy.types import TypeDecorator, JSON as SA_JSON
from sqlmodel import Field, SQLModel, Column

_actions_adapter = TypeAdapter(list[Action])


class ActionListType(TypeDecorator):
    impl = SA_JSON
    cache_ok = True

    def process_bind_param(self, value, dialect):
        if value is None:
            return None
        return _actions_adapter.dump_python(value, mode="json")

    def process_result_value(self, value, dialect):
        if value is None:
            return []
        return _actions_adapter.validate_python(value)


class Hero(SQLModel, table=True):
    id: int | None = Field(default=None, primary_key=True)
    name: str
    actions: list[Action] = Field(default_factory=list, sa_column=Column(ActionListType))

Two side notes

• Prefer list[Action] over tuple[Action, ...] for a JSON column — a JSON array maps naturally to a list.
• If you mutate actions in place (e.g. hero.actions.append(...)) and expect it to be persisted, wrap it with SQLAlchemy's MutableList so changes are tracked. Reassigning the whole list always works.