Fastapi-SQLA

Fastapi-SQLA is an SQLAlchemy extension for FastAPI easy to setup with support for pagination, asyncio, SQLModel and pytest. It supports SQLAlchemy>=1.3 and is fully compliant with SQLAlchemy 2.0. It is developped, maintained and used on production by the team at @dialoguemd with love from Montreal 🇨🇦.

Installing

Using pip:

pip install fastapi-sqla

Quick Example

Assuming it runs against a DB with a table user with 3 columns, id, name and unique email:

# main.py
from contextlib import asynccontextmanager
from fastapi import FastAPI, HTTPException
from fastapi_sqla import Base, Item, Page, Paginate, Session, setup_middlewares, startup
from pydantic import BaseModel, EmailStr
from sqlalchemy import select
from sqlalchemy.exc import IntegrityError


@asynccontextmanager
async def lifespan(app: FastAPI):
    await startup()
    yield


app = FastAPI(lifespan=lifespan)
setup_middlewares(app)


class User(Base):
    __tablename__ = "user"


class UserIn(BaseModel):
    name: str
    email: EmailStr


class UserModel(UserIn):
    id: int

    class Config:
        orm_mode = True


@app.get("/users", response_model=Page[UserModel])
def list_users(paginate: Paginate):
    return paginate(select(User))


@app.get("/users/{user_id}", response_model=Item[UserModel])
def get_user(user_id: int, session: Session):
    user = session.get(User, user_id)
    if user is None:
        raise HTTPException(404)
    return {"data": user}


@app.post("/users", response_model=Item[UserModel])
def create_user(new_user: UserIn, session: Session):
    user = User(**new_user.model_dump())
    session.add(user)
    try:
        session.flush()
    except IntegrityError:
        raise HTTPException(409, "Email is already taken.")
    return {"data": user}

Creating a db using sqlite3:

sqlite3 db.sqlite <<EOF
CREATE TABLE user (
    id    INTEGER PRIMARY KEY AUTOINCREMENT,
    email TEXT NOT NULL,
    name  TEXT NOT NULL
);
CREATE UNIQUE INDEX user_email_idx ON user (email);
EOF

Running the app:

sqlalchemy_url=sqlite:///db.sqlite?check_same_thread=false uvicorn main:app

Configuration

Environment variables:

The keys of interest in os.environ are prefixed with sqlalchemy_. Each matching key (after the prefix is stripped) is treated as though it were the corresponding keyword argument to sqlalchemy.create_engine call.

The only required key is sqlalchemy_url, which provides the database URL, example:

export sqlalchemy_url=postgresql://postgres@localhost

Multi-session support

In order to configure multiple sessions for the application, set the environment variables with this prefix format: fastapi_sqla__MY_KEY__.

As with the default session, each matching key (after the prefix is stripped) is treated as though it were the corresponding keyword argument to sqlalchemy.create_engine call.

For example, to configure a session with the read_only key:

export fastapi_sqla__read_only__sqlalchemy_url=postgresql://postgres@localhost

asyncio support using asyncpg

SQLAlchemy >= 1.4 supports asyncio. To enable asyncio support against a Postgres DB, install asyncpg:

pip install asyncpg

And define the environment variable sqlalchemy_url with postgres+asyncpg scheme:

export sqlalchemy_url=postgresql+asyncpg://postgres@localhost

Setup the app AsyncContextManager (recommended):

import fastapi_sqla
from fastapi import FastAPI

@asynccontextmanager
async def lifespan(app: FastAPI):
    await fastapi_sqla.startup()
    yield


app = FastAPI(lifespan=lifespan)
fastapi_sqla.setup_middlewares(app)

Setup the app using startup/shutdown events (deprecated):

import fastapi_sqla
from fastapi import FastAPI

app = FastAPI()
fastapi_sqla.setup(app)

SQLAlchemy

Adding a new entity class:

from fastapi_sqla import Base


class Entity(Base):
    __tablename__ = "table-name-in-db"

Getting an sqla session

Using dependency injection

Use FastAPI dependency injection to get a session as a parameter of a path operation function.

The SQLAlchemy session is committed before the response is returned or rollbacked if any exception occurred:

from fastapi import APIRouter
from fastapi_sqla import AsyncSession, Session

router = APIRouter()


@router.get("/example")
def example(session: Session):
    return session.execute("SELECT now()").scalar()


@router.get("/async_example")
async def async_example(session: AsyncSession):
    return await session.scalar("SELECT now()")

In order to get a session configured with a custom key:

from typing import Annotated

from fastapi import APIRouter, Depends
from fastapi_sqla import (
    AsyncSessionDependency,
    SessionDependency,
    SqlaAsyncSession,
    SqlaSession,
)

router = APIRouter()


# Preferred

ReadOnlySession = Annotated[SqlaSession, Depends(SessionDependency(key="read_only"))]
AsyncReadOnlySession = Annotated[
    SqlaAsyncSession, Depends(AsyncSessionDependency(key="read_only"))
]

@router.get("/example")
def example(session: ReadOnlySession):
    return session.execute("SELECT now()").scalar()


@router.get("/async_example")
async def async_example(session: AsyncReadOnlySession):
    return await session.scalar("SELECT now()")


# Alternative

@router.get("/example/alt")
def example_alt(session: SqlaSession = Depends(SessionDependency(key="read_only"))):
    return session.execute("SELECT now()").scalar()


@router.get("/async_example/alt")
async def async_example_alt(
    session: SqlaAsyncSession = Depends(AsyncSessionDependency(key="read_only")),
):
    return await session.scalar("SELECT now()")

Using a context manager

When needing a session outside of a path operation, like when using FastAPI background tasks, use fastapi_sqla.open_session context manager. The SQLAlchemy session is committed when exiting context or rollbacked if any exception occurred:

from fastapi import APIRouter, BackgroundTasks
from fastapi_sqla import open_async_session, open_session

router = APIRouter()


@router.get("/example")
def example(bg: BackgroundTasks):
    bg.add_task(run_bg)
    bg.add_task(run_async_bg)


def run_bg():
    with open_session() as session:
        session.execute("SELECT now()").scalar()

def run_bg_with_key():
    with open_session(key="read_only") as session:
        session.execute("SELECT now()").scalar()

async def run_async_bg():
    async with open_async_session() as session:
        await session.scalar("SELECT now()")

async def run_async_bg_with_key():
    async with open_async_session(key="read_only") as session:
        await session.scalar("SELECT now()")

Pagination

from fastapi import APIRouter
from fastapi_sqla import Base, Page, Paginate
from pydantic import BaseModel
from sqlalchemy import select

router = APIRouter()


class User(Base):
    __tablename__ = "user"


class UserModel(BaseModel):
    id: int
    name: str

    class Config:
        orm_mode = True


@router.get("/users", response_model=Page[UserModel])
def all_users(paginate: Paginate):
    return paginate(select(User))

By default:

• It returns pages of 10 items, up to 100 items;

• Total number of items in the collection is queried using Query.count.

• Response example for /users?offset=40&limit=10:

  {
      "data": [
          {
              "id": 41,
              "name": "Pat Thomas"
          },
          {
              "id": 42,
              "name": "Mulatu Astatke"
          }
      ],
      "meta": {
          "offset": 40,
          "total_items": 42,
          "total_pages": 5,
          "page_number": 5
      }
  }
  

Paginating non-scalar results

To paginate a query which doesn't return scalars, specify scalars=False when invoking paginate:

from fastapi import APIRouter
from fastapi_sqla import Base, Page, Paginate
from pydantic import BaseModel
from sqlalchemy import func, select
from sqlalchemy.orm import relationship

router = APIRouter()


class User(Base):
    __tablename__ = "user"
    notes = relationship("Note")


class Note(Base):
    __tablename__ = "note"


class UserModel(BaseModel):
    id: int
    name: str
    notes_count: int


@router.get("/users", response_model=Page[UserModel])
def all_users(paginate: Paginate):
    query = (
        select(User.id, User.name, func.count(Note.id).label("notes_count"))
        .join(Note)
        .group_by(User)
    )
    return paginate(query, scalars=False)

Customize pagination

You can customize:

• Minimum and maximum number of items per pages;
• How the total number of items in the collection is queried;

To customize pagination, create a dependency using fastapi_sqla.Pagination:

from fastapi import APIRouter, Depends
from fastapi_sqla import Base, Page, Pagination, Session
from pydantic import BaseModel
from sqlalchemy import func, select

router = APIRouter()


class User(Base):
    __tablename__ = "user"


class UserModel(BaseModel):
    id: int
    name: str


def query_count(session: Session) -> int:
    return session.execute(select(func.count()).select_from(User)).scalar()


CustomPaginate = Pagination(min_page_size=5, max_page_size=500, query_count=query_count)


@router.get("/users", response_model=Page[UserModel])
def all_users(paginate: CustomPaginate = Depends()):
    return paginate(select(User))

Async pagination

When using the asyncio support, use the AsyncPaginate dependency:

from fastapi import APIRouter
from fastapi_sqla import Base, Page, AsyncPaginate
from pydantic import BaseModel
from sqlalchemy import select

router = APIRouter()


class User(Base):
    __tablename__ = "user"


class UserModel(BaseModel):
    id: int
    name: str

    class Config:
        orm_mode = True


@router.get("/users", response_model=Page[UserModel])
async def all_users(paginate: AsyncPaginate):
    return await paginate(select(User))

Customize pagination by creating a dependency using fastapi_sqla.AsyncPagination:

from fastapi import APIRouter, Depends
from fastapi_sqla import Base, Page, AsyncPagination, AsyncSession
from pydantic import BaseModel
from sqlalchemy import func, select

router = APIRouter()


class User(Base):
    __tablename__ = "user"


class UserModel(BaseModel):
    id: int
    name: str


async def query_count(session: AsyncSession) -> int:
    result = await session.execute(select(func.count()).select_from(User))
    return result.scalar()


CustomPaginate = AsyncPagination(min_page_size=5, max_page_size=500, query_count=query_count)


@router.get("/users", response_model=Page[UserModel])
async def all_users(paginate: CustomPaginate = Depends()):
    return await paginate(select(User))

Multi-session support

Pagination supports multiple sessions as well. To paginate using a session configured with a custom key:

from typing import Annotated

from fastapi import APIRouter, Depends
from fastapi_sqla import (
    AsyncPaginateSignature,
    AsyncPagination,
    Base,
    Page,
    PaginateSignature,
    Pagination,
)
from pydantic import BaseModel
from sqlalchemy import func, select

router = APIRouter()


class User(Base):
    __tablename__ = "user"


class UserModel(BaseModel):
    id: int
    name: str


# Preferred

ReadOnlyPaginate = Annotated[
    PaginateSignature, Depends(Pagination(session_key="read_only"))
]
AsyncReadOnlyPaginate = Annotated[
    AsyncPaginateSignature, Depends(AsyncPagination(session_key="read_only"))
]

@router.get("/users", response_model=Page[UserModel])
def all_users(paginate: ReadOnlyPaginate):
    return paginate(select(User))

@router.get("/async_users", response_model=Page[UserModel])
async def async_all_users(paginate: AsyncReadOnlyPaginate):
    return await paginate(select(User))


# Alternative

@router.get("/users/alt", response_model=Page[UserModel])
def all_users_alt(
    paginate: PaginateSignature = Depends(
        Pagination(session_key="read_only")
    ),
):
    return paginate(select(User))

@router.get("/async_users/alt", response_model=Page[UserModel])
async def async_all_users_alt(
    paginate: AsyncPaginateSignature = Depends(
        AsyncPagination(session_key="read_only")
    ),
):
    return await paginate(select(User))

SQLModel support 🎉

If your project uses SQLModel, then Session dependency is an SQLModel session::

    from http import HTTPStatus

    from fastapi import FastAPI, HTTPException
    from fastapi_sqla import Item, Page, Paginate, Session, setup
    from sqlmodel import Field, SQLModel, select

    class Hero(SQLModel, table=True):
        id: int | None = Field(default=None, primary_key=True)
        name: str
        secret_name: str
        age: int | None = None


    app = FastAPI()
    setup(app)

    @app.get("/heros", response_model=Page[Hero])
    def list_hero(paginate: Paginate) -> Page[Hero]:
        return paginate(select(Hero))


    @app.get("/heros/{hero_id}", response_model=Item[Hero])
    def get_hero(hero_id: int, session: Session) -> Item[Hero]:
        hero = session.get(Hero, hero_id)
        if hero is None:
            raise HTTPException(HTTPStatus.NOT_FOUND)
        return {"data": hero}

Pytest fixtures

This library provides a set of utility fixtures, through its PyTest plugin, which is automatically installed with the library.

By default, no records are actually written to the database when running tests. There currently is no way to change this behaviour.

sqla_modules

You must define this fixture, in order for the plugin to reflect table metadata in your SQLAlchemy entities. It should just import all of the application's modules which contain SQLAlchemy models.

Example:

# tests/conftest.py
from pytest import fixture


@fixture
def sqla_modules():
    from app import sqla  # noqa

db_url

The DB url to use.

When CI key is set in environment variables, it defaults to using postgres as the host name:

postgresql://postgres@postgres/postgres

In other cases, the host is set to localhost:

postgresql://postgres@localhost/postgres

Of course, you can override it by overloading the fixture:

from pytest import fixture


@fixture(scope="session")
def db_url():
    return "postgresql://postgres@localhost/test_database"

async_sqlalchemy_url

DB url to use when using asyncio support. Defaults to db_url fixture with postgresql+asyncpg:// scheme.

session & async_session

Sqla sessions to create db fixture:

• All changes done at test setup or during the test are rollbacked at test tear down;
• No record will actually be written in the database;
• Changes in one regular session need to be committed to be available from other regular sessions;
• Changes in one async session need to be committed to be available from other async sessions;
• Changes from regular sessions are not available from async session and vice-versa even when committed;

Example:

from pytest import fixture


@fixture
def patient(session):
    from er.sqla import Patient
    patient = Patient(first_name="Bob", last_name="David")
    session.add(patient)
    session.commit()
    return patient


@fixture
async def doctor(async_session):
    from er.sqla import Doctor
    doctor = Doctor(name="who")
    async_session.add(doctor)
    await async_session.commit()
    return doctor

db_migration

A session scope fixture that runs alembic upgrade at test session setup and alembic downgrade at tear down.

It depends on alembic_ini_path fixture to get the path of alembic.ini file.

To use in a test or test module:

from pytest import mark

pytestmark = mark.usefixtures("db_migration")

To use globally, add to pytest options:

[pytest]
usefixtures =
    db_migration

Or depends on it in top-level conftest.py and mark it as auto-used:

from pytest import fixture


@fixture(scope="session", autouse=True)
def db_migration(db_migration):
    pass

alembic_ini_path

It returns the path of alembic.ini configuration file. By default, it returns ./alembic.ini.

Development

Prerequisites

• Python >=3.9
• Poetry to install package dependencies.
• A postgres DB reachable at postgresql://postgres@localhost/postgres

Setup

$ poetry install --extras tests --extras asyncpg --extras aws_rds_iam

Running tests

$ poetry run pytest

Runing tests on multiple environments

$ poetry run tox