fastapi（FastAPI 开发）

FastAPI Skill

Production-tested patterns for FastAPI with Pydantic v2, SQLAlchemy 2.0 async, and JWT authentication.

Latest Versions (verified January 2026):

• FastAPI: 0.128.0
• Pydantic: 2.11.7
• SQLAlchemy: 2.0.30
• Uvicorn: 0.35.0
• python-jose: 3.3.0

Requirements:

• Python 3.9+ (Python 3.8 support dropped in FastAPI 0.125.0)
• Pydantic v2.7.0+ (Pydantic v1 support completely removed in FastAPI 0.128.0)

Quick Start

Project Setup with uv

# Create project
uv init my-api
cd my-api

# Add dependencies
uv add fastapi[standard] sqlalchemy[asyncio] aiosqlite python-jose[cryptography] passlib[bcrypt]

# Run development server
uv run fastapi dev src/main.py

Minimal Working Example

# src/main.py
from fastapi import FastAPI
from pydantic import BaseModel

app = FastAPI(title="My API")

class Item(BaseModel):
    name: str
    price: float

@app.get("/")
async def root():
    return {"message": "Hello World"}

@app.post("/items")
async def create_item(item: Item):
    return item

Run: uv run fastapi dev src/main.py

Docs available at: http://127.0.0.1:8000/docs

Project Structure (Domain-Based)

For maintainable projects, organize by domain not file type:

my-api/
├── pyproject.toml
├── src/
│   ├── __init__.py
│   ├── main.py              # FastAPI app initialization
│   ├── config.py            # Global settings
│   ├── database.py          # Database connection
│   │
│   ├── auth/                # Auth domain
│   │   ├── __init__.py
│   │   ├── router.py        # Auth endpoints
│   │   ├── schemas.py       # Pydantic models
│   │   ├── models.py        # SQLAlchemy models
│   │   ├── service.py       # Business logic
│   │   └── dependencies.py  # Auth dependencies
│   │
│   ├── items/               # Items domain
│   │   ├── __init__.py
│   │   ├── router.py
│   │   ├── schemas.py
│   │   ├── models.py
│   │   └── service.py
│   │
│   └── shared/              # Shared utilities
│       ├── __init__.py
│       └── exceptions.py
└── tests/
    └── test_main.py

Core Patterns

Pydantic Schemas (Validation)

# src/items/schemas.py
from pydantic import BaseModel, Field, ConfigDict
from datetime import datetime
from enum import Enum

class ItemStatus(str, Enum):
    DRAFT = "draft"
    PUBLISHED = "published"
    ARCHIVED = "archived"

class ItemBase(BaseModel):
    name: str = Field(..., min_length=1, max_length=100)
    description: str | None = Field(None, max_length=500)
    price: float = Field(..., gt=0, description="Price must be positive")
    status: ItemStatus = ItemStatus.DRAFT

class ItemCreate(ItemBase):
    pass

class ItemUpdate(BaseModel):
    name: str | None = Field(None, min_length=1, max_length=100)
    description: str | None = None
    price: float | None = Field(None, gt=0)
    status: ItemStatus | None = None

class ItemResponse(ItemBase):
    id: int
    created_at: datetime

    model_config = ConfigDict(from_attributes=True)

Key Points:

• Use Field() for validation constraints
• Separate Create/Update/Response schemas
• from_attributes=True enables SQLAlchemy model conversion
• Use str | None (Python 3.10+) not Optional[str]

SQLAlchemy Models (Database)

# src/items/models.py
from sqlalchemy import String, Float, DateTime, Enum as SQLEnum
from sqlalchemy.orm import Mapped, mapped_column
from datetime import datetime
from src.database import Base
from src.items.schemas import ItemStatus

class Item(Base):
    __tablename__ = "items"

    id: Mapped[int] = mapped_column(primary_key=True)
    name: Mapped[str] = mapped_column(String(100))
    description: Mapped[str | None] = mapped_column(String(500), nullable=True)
    price: Mapped[float] = mapped_column(Float)
    status: Mapped[ItemStatus] = mapped_column(
        SQLEnum(ItemStatus), default=ItemStatus.DRAFT
    )
    created_at: Mapped[datetime] = mapped_column(
        DateTime, default=datetime.utcnow
    )

Database Setup (Async SQLAlchemy 2.0)

# src/database.py
from sqlalchemy.ext.asyncio import create_async_engine, AsyncSession, async_sessionmaker
from sqlalchemy.orm import DeclarativeBase

DATABASE_URL = "sqlite+aiosqlite:///./database.db"

engine = create_async_engine(DATABASE_URL, echo=True)
async_session = async_sessionmaker(engine, expire_on_commit=False)

class Base(DeclarativeBase):
    pass

async def get_db():
    async with async_session() as session:
        try:
            yield session
            await session.commit()
        except Exception:
            await session.rollback()
            raise

Router Pattern

# src/items/router.py
from fastapi import APIRouter, Depends, HTTPException, status
from sqlalchemy.ext.asyncio import AsyncSession
from sqlalchemy import select

from src.database import get_db
from src.items import schemas, models

router = APIRouter(prefix="/items", tags=["items"])

@router.get("", response_model=list[schemas.ItemResponse])
async def list_items(
    skip: int = 0,
    limit: int = 100,
    db: AsyncSession = Depends(get_db)
):
    result = await db.execute(
        select(models.Item).offset(skip).limit(limit)
    )
    return result.scalars().all()

@router.get("/{item_id}", response_model=schemas.ItemResponse)
async def get_item(item_id: int, db: AsyncSession = Depends(get_db)):
    result = await db.execute(
        select(models.Item).where(models.Item.id == item_id)
    )
    item = result.scalar_one_or_none()
    if not item:
        raise HTTPException(status_code=404, detail="Item not found")
    return item

@router.post("", response_model=schemas.ItemResponse, status_code=status.HTTP_201_CREATED)
async def create_item(
    item_in: schemas.ItemCreate,
    db: AsyncSession = Depends(get_db)
):
    item = models.Item(**item_in.model_dump())
    db.add(item)
    await db.commit()
    await db.refresh(item)
    return item

Main App

# src/main.py
from contextlib import asynccontextmanager
from fastapi import FastAPI
from fastapi.middleware.cors import CORSMiddleware

from src.database import engine, Base
from src.items.router import router as items_router
from src.auth.router import router as auth_router

@asynccontextmanager
async def lifespan(app: FastAPI):
    # Startup: Create tables
    async with engine.begin() as conn:
        await conn.run_sync(Base.metadata.create_all)
    yield
    # Shutdown: cleanup if needed

app = FastAPI(title="My API", lifespan=lifespan)

# CORS middleware
app.add_middleware(
    CORSMiddleware,
    allow_origins=["http://localhost:3000"],  # Your frontend
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

# Include routers
app.include_router(auth_router)
app.include_router(items_router)

JWT Authentication

Auth Schemas

# src/auth/schemas.py
from pydantic import BaseModel, EmailStr

class UserCreate(BaseModel):
    email: EmailStr
    password: str

class UserResponse(BaseModel):
    id: int
    email: str

    model_config = ConfigDict(from_attributes=True)

class Token(BaseModel):
    access_token: str
    token_type: str = "bearer"

class TokenData(BaseModel):
    user_id: int | None = None

Auth Service

# src/auth/service.py
from datetime import datetime, timedelta
from jose import JWTError, jwt
from passlib.context import CryptContext
from src.config import settings

pwd_context = CryptContext(schemes=["bcrypt"], deprecated="auto")

def hash_password(password: str) -> str:
    return pwd_context.hash(password)

def verify_password(plain: str, hashed: str) -> bool:
    return pwd_context.verify(plain, hashed)

def create_access_token(data: dict, expires_delta: timedelta | None = None) -> str:
    to_encode = data.copy()
    expire = datetime.utcnow() + (expires_delta or timedelta(minutes=15))
    to_encode.update({"exp": expire})
    return jwt.encode(to_encode, settings.SECRET_KEY, algorithm="HS256")

def decode_token(token: str) -> dict | None:
    try:
        return jwt.decode(token, settings.SECRET_KEY, algorithms=["HS256"])
    except JWTError:
        return None

Auth Dependencies

# src/auth/dependencies.py
from fastapi import Depends, HTTPException, status
from fastapi.security import OAuth2PasswordBearer
from sqlalchemy.ext.asyncio import AsyncSession
from sqlalchemy import select

from src.database import get_db
from src.auth import service, models, schemas

oauth2_scheme = OAuth2PasswordBearer(tokenUrl="/auth/login")

async def get_current_user(
    token: str = Depends(oauth2_scheme),
    db: AsyncSession = Depends(get_db)
) -> models.User:
    credentials_exception = HTTPException(
        status_code=status.HTTP_401_UNAUTHORIZED,
        detail="Could not validate credentials",
        headers={"WWW-Authenticate": "Bearer"},
    )

    payload = service.decode_token(token)
    if payload is None:
        raise credentials_exception

    user_id = payload.get("sub")
    if user_id is None:
        raise credentials_exception

    result = await db.execute(
        select(models.User).where(models.User.id == int(user_id))
    )
    user = result.scalar_one_or_none()

    if user is None:
        raise credentials_exception

    return user

Auth Router

# src/auth/router.py
from fastapi import APIRouter, Depends, HTTPException, status
from fastapi.security import OAuth2PasswordRequestForm
from sqlalchemy.ext.asyncio import AsyncSession
from sqlalchemy import select

from src.database import get_db
from src.auth import schemas, models, service
from src.auth.dependencies import get_current_user

router = APIRouter(prefix="/auth", tags=["auth"])

@router.post("/register", response_model=schemas.UserResponse)
async def register(
    user_in: schemas.UserCreate,
    db: AsyncSession = Depends(get_db)
):
    # Check existing
    result = await db.execute(
        select(models.User).where(models.User.email == user_in.email)
    )
    if result.scalar_one_or_none():
        raise HTTPException(status_code=400, detail="Email already registered")

    user = models.User(
        email=user_in.email,
        hashed_password=service.hash_password(user_in.password)
    )
    db.add(user)
    await db.commit()
    await db.refresh(user)
    return user

@router.post("/login", response_model=schemas.Token)
async def login(
    form_data: OAuth2PasswordRequestForm = Depends(),
    db: AsyncSession = Depends(get_db)
):
    result = await db.execute(
        select(models.User).where(models.User.email == form_data.username)
    )
    user = result.scalar_one_or_none()

    if not user or not service.verify_password(form_data.password, user.hashed_password):
        raise HTTPException(
            status_code=status.HTTP_401_UNAUTHORIZED,
            detail="Incorrect email or password"
        )

    access_token = service.create_access_token(data={"sub": str(user.id)})
    return schemas.Token(access_token=access_token)

@router.get("/me", response_model=schemas.UserResponse)
async def get_me(current_user: models.User = Depends(get_current_user)):
    return current_user

Protect Routes

# In any router
from src.auth.dependencies import get_current_user
from src.auth.models import User

@router.post("/items")
async def create_item(
    item_in: schemas.ItemCreate,
    current_user: User = Depends(get_current_user),  # Requires auth
    db: AsyncSession = Depends(get_db)
):
    item = models.Item(**item_in.model_dump(), user_id=current_user.id)
    # ...

Configuration

# src/config.py
from pydantic_settings import BaseSettings

class Settings(BaseSettings):
    DATABASE_URL: str = "sqlite+aiosqlite:///./database.db"
    SECRET_KEY: str = "your-secret-key-change-in-production"
    ACCESS_TOKEN_EXPIRE_MINUTES: int = 30

    class Config:
        env_file = ".env"

settings = Settings()

Create .env:

DATABASE_URL=sqlite+aiosqlite:///./database.db
SECRET_KEY=your-super-secret-key-here
ACCESS_TOKEN_EXPIRE_MINUTES=30

Critical Rules

Always Do

1. Separate Pydantic schemas from SQLAlchemy models - Different jobs, different files
2. Use async for I/O operations - Database, HTTP calls, file access
3. Validate with Pydantic Field() - Constraints, defaults, descriptions
4. Use dependency injection - Depends() for database, auth, validation
5. Return proper status codes - 201 for create, 204 for delete, etc.

Never Do

1. Never use blocking calls in async routes - No time.sleep(), use asyncio.sleep()
2. Never put business logic in routes - Use service layer
3. Never hardcode secrets - Use environment variables
4. Never skip validation - Always use Pydantic schemas
5. **Never use * in CORS origins for production** - Specify exact origins

Known Issues Prevention

This skill prevents 7 documented issues from official FastAPI GitHub and release notes.

Issue #1: Form Data Loses Field Set Metadata

Error: model.model_fields_set includes default values when using Form() Source: GitHub Issue #13399 Why It Happens: Form data parsing preloads default values and passes them to the validator, making it impossible to distinguish between fields explicitly set by the user and fields using defaults. This bug ONLY affects Form data, not JSON body data.

Prevention:

# AVOID: Pydantic model with Form when you need field_set metadata
from typing import Annotated
from fastapi import Form

@app.post("/form")
async def endpoint(model: Annotated[MyModel, Form()]):
    fields = model.model_fields_set  # Unreliable!

# USE: Individual form fields or JSON body instead
@app.post("/form-individual")
async def endpoint(
    field_1: Annotated[bool, Form()] = True,
    field_2: Annotated[str | None, Form()] = None
):
    # You know exactly what was provided

# OR: Use JSON body when metadata matters
@app.post("/json")
async def endpoint(model: MyModel):
    fields = model.model_fields_set  # Works correctly

Issue #2: BackgroundTasks Silently Overwritten by Custom Response

Error: Background tasks added via BackgroundTasks dependency don't run Source: GitHub Issue #11215 Why It Happens: When you return a custom Response with a background parameter, it overwrites all tasks added to the injected BackgroundTasks dependency.

Prevention:

# WRONG: Mixing both mechanisms
from fastapi import BackgroundTasks
from starlette.responses import Response, BackgroundTask

@app.get("/")
async def endpoint(tasks: BackgroundTasks):
    tasks.add_task(send_email)  # This will be lost!
    return Response(
        content="Done",
        background=BackgroundTask(log_event)  # Only this runs
    )

# RIGHT: Use only BackgroundTasks dependency
@app.get("/")
async def endpoint(tasks: BackgroundTasks):
    tasks.add_task(send_email)
    tasks.add_task(log_event)
    return {"status": "done"}  # All tasks run

Rule: Pick ONE mechanism and stick with it. Don't mix injected BackgroundTasks with Response(background=...).

Issue #3: Optional Form Fields Break with TestClient (Regression)

Error: 422: "Input should be 'abc' or 'def'" for optional Literal fields Source: GitHub Issue #12245 Why It Happens: Starting in FastAPI 0.114.0, optional form fields with Literal types fail validation when passed None via TestClient.

Prevention:

from typing import Annotated, Literal, Optional
from fastapi import Form
from fastapi.testclient import TestClient

# PROBLEMATIC: Optional Literal with Form (breaks in 0.114.0+)
@app.post("/")
async def endpoint(
    attribute: Annotated[Optional[Literal["abc", "def"]], Form()]
):
    return {"attribute": attribute}

client = TestClient(app)
data = {"attribute": None}  # or omit the field
response = client.post("/", data=data)  # Returns 422

# WORKAROUND 1: Don't pass None explicitly, omit the field
data = {}  # Omit instead of None
response = client.post("/", data=data)  # Works

# WORKAROUND 2: Avoid Literal types with optional form fields
@app.post("/")
async def endpoint(attribute: Annotated[str | None, Form()] = None):
    # Validate in application logic instead
    if attribute and attribute not in ["abc", "def"]:
        raise HTTPException(400, "Invalid attribute")

Issue #4: Pydantic Json Type Doesn't Work with Form Data

Error: "JSON object must be str, bytes or bytearray" Source: GitHub Issue #10997 Why It Happens: Using Pydantic's Json type directly with Form() fails.

Prevention:

from typing import Annotated
from fastapi import Form
from pydantic import Json, BaseModel

# WRONG: Json type directly with Form
@app.post("/broken")
async def broken(json_list: Annotated[Json[list[str]], Form()]) -> list[str]:
    return json_list  # Returns 422

# RIGHT: Accept as str, parse with Pydantic
class JsonListModel(BaseModel):
    json_list: Json[list[str]]

@app.post("/working")
async def working(json_list: Annotated[str, Form()]) -> list[str]:
    model = JsonListModel(json_list=json_list)  # Pydantic parses here
    return model.json_list  # Works

Issue #5: Annotated with ForwardRef Breaks OpenAPI Generation

Error: Missing or incorrect OpenAPI schema for dependency types Source: GitHub Issue #13056 Why It Happens: When using Annotated with Depends() and a forward reference (from __future__ import annotations), OpenAPI schema generation fails.

Prevention:

# PROBLEMATIC: Forward reference with Depends
from __future__ import annotations
from dataclasses import dataclass
from typing import Annotated
from fastapi import Depends, FastAPI

app = FastAPI()

def get_potato() -> Potato:  # Forward reference
    return Potato(color='red', size=10)

@app.get('/')
async def read_root(potato: Annotated[Potato, Depends(get_potato)]):
    return {'Hello': 'World'}
# OpenAPI schema doesn't include Potato definition correctly

@dataclass
class Potato:
    color: str
    size: int

# WORKAROUND 1: Don't use __future__ annotations in route files
# Remove: from __future__ import annotations

# WORKAROUND 2: Use string literals for type hints
def get_potato() -> "Potato":
    return Potato(color='red', size=10)

Issue #6: Pydantic v2 Path Parameter Union Type Breaking Change

Error: Path parameters with int | str always parse as str in Pydantic v2 Source: GitHub Issue #11251 Why It Happens: Major breaking change when migrating from Pydantic v1 to v2.

Prevention:

from uuid import UUID

# PROBLEMATIC: Union with str in path parameter
@app.get("/int/{path}")
async def int_path(path: int | str):
    return str(type(path))
    # Pydantic v1: returns <class 'int'> for "123"
    # Pydantic v2: returns <class 'str'> for "123"

# RIGHT: Avoid union types with str in path/query parameters
@app.get("/int/{path}")
async def int_path(path: int):
    return str(type(path))  # Works correctly

# ALTERNATIVE: Use validators if type coercion needed
from pydantic import field_validator

class PathParams(BaseModel):
    path: int | str

    @field_validator('path')
    def coerce_to_int(cls, v):
        if isinstance(v, str) and v.isdigit():
            return int(v)
        return v

Issue #7: ValueError in field_validator Returns 500 Instead of 422

Error: 500 Internal Server Error when raising ValueError in custom validators Source: GitHub Discussion #10779 Why It Happens: When raising ValueError inside a Pydantic @field_validator with Form fields, FastAPI returns 500 instead of 422.

Prevention:

from typing import Annotated
from fastapi import Form
from pydantic import BaseModel, field_validator, ValidationError, Field

# WRONG: ValueError in validator
class MyForm(BaseModel):
    value: int

    @field_validator('value')
    def validate_value(cls, v):
        if v < 0:
            raise ValueError("Value must be positive")  # Returns 500!
        return v

# RIGHT 1: Raise ValidationError instead
class MyForm(BaseModel):
    value: int

    @field_validator('value')
    def validate_value(cls, v):
        if v < 0:
            raise ValidationError("Value must be positive")  # Returns 422
        return v

# RIGHT 2: Use Pydantic's built-in constraints
class MyForm(BaseModel):
    value: Annotated[int, Field(gt=0)]  # Built-in validation, returns 422

Common Errors & Fixes

422 Unprocessable Entity

Cause: Request body doesn't match Pydantic schema

Debug:

1. Check /docs endpoint - test there first
2. Verify JSON structure matches schema
3. Check required vs optional fields

Fix: Add custom validation error handler:

from fastapi.exceptions import RequestValidationError
from fastapi.responses import JSONResponse

@app.exception_handler(RequestValidationError)
async def validation_exception_handler(request, exc):
    return JSONResponse(
        status_code=422,
        content={"detail": exc.errors(), "body": exc.body}
    )

CORS Errors

Cause: Missing or misconfigured CORS middleware

Fix:

app.add_middleware(
    CORSMiddleware,
    allow_origins=["http://localhost:3000"],  # Not "*" in production
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

Async Blocking Event Loop

Cause: Blocking call in async route (e.g., time.sleep(), sync database client, CPU-bound operations)

Symptoms (production-scale):

• Throughput plateaus far earlier than expected
• Latency "balloons" as concurrency increases
• Request pattern looks almost serial under load

Fix: Use async alternatives:

# WRONG: Blocks event loop
import time
from sqlalchemy import create_engine  # Sync client

@app.get("/users")
async def get_users():
    time.sleep(0.1)  # Even small blocking adds up at scale!
    result = sync_db_client.query("SELECT * FROM users")  # Blocks!
    return result

# RIGHT 1: Use async database driver
from sqlalchemy.ext.asyncio import AsyncSession
from sqlalchemy import select

@app.get("/users")
async def get_users(db: AsyncSession = Depends(get_db)):
    await asyncio.sleep(0.1)  # Non-blocking
    result = await db.execute(select(User))
    return result.scalars().all()

# RIGHT 2: Use def (not async def) for CPU-bound routes
# FastAPI runs def routes in thread pool automatically
@app.get("/cpu-heavy")
def cpu_heavy_task():  # Note: def not async def
    return expensive_cpu_work()  # Runs in thread pool

"Field required" for Optional Fields

Cause: Using Optional[str] without default

Fix:

# Wrong
description: Optional[str]  # Still required!

# Right
description: str | None = None  # Optional with default

Testing

# tests/test_main.py
import pytest
from httpx import AsyncClient, ASGITransport
from src.main import app

@pytest.fixture
async def client():
    async with AsyncClient(
        transport=ASGITransport(app=app),
        base_url="http://test"
    ) as ac:
        yield ac

@pytest.mark.asyncio
async def test_root(client):
    response = await client.get("/")
    assert response.status_code == 200

@pytest.mark.asyncio
async def test_create_item(client):
    response = await client.post(
        "/items",
        json={"name": "Test", "price": 9.99}
    )
    assert response.status_code == 201
    assert response.json()["name"] == "Test"

Run: uv run pytest

Deployment

Uvicorn (Development)

uv run fastapi dev src/main.py

Uvicorn (Production)

uv run uvicorn src.main:app --host 0.0.0.0 --port 8000

Gunicorn + Uvicorn (Production with workers)

uv add gunicorn
uv run gunicorn src.main:app -w 4 -k uvicorn.workers.UvicornWorker --bind 0.0.0.0:8000

Docker

FROM python:3.12-slim

WORKDIR /app
COPY . .

RUN pip install uv && uv sync

EXPOSE 8000
CMD ["uv", "run", "uvicorn", "src.main:app", "--host", "0.0.0.0", "--port", "8000"]

References

• FastAPI Documentation
• FastAPI Best Practices
• Pydantic v2 Documentation
• SQLAlchemy 2.0 Async
• uv Package Manager

Last verified: 2026-01-21 | Skill version: 1.1.0 | Changes: Added 7 known issues (form data bugs, background tasks, Pydantic v2 migration gotchas), expanded async blocking guidance with production patterns Maintainer: Jezweb | jeremy@jezweb.net