beanie-odm（豆豆 ODM）

Beanie ODM for MongoDB

This skill provides patterns for MongoDB integration using Beanie ODM with async Motor driver, optimized for FastAPI applications.

Database Initialization

Connection Setup

from beanie import init_beanie
from motor.motor_asyncio import AsyncIOMotorClient
from app.domains.users.models import User
from app.domains.products.models import Product

async def init_database(settings: Settings):
    client = AsyncIOMotorClient(settings.mongodb_url)

    await init_beanie(
        database=client[settings.database_name],
        document_models=[
            User,
            Product,
            # Add all document models
        ]
    )

Settings Configuration

from pydantic_settings import BaseSettings

class Settings(BaseSettings):
    mongodb_url: str = "mongodb://localhost:27017"
    database_name: str = "app_db"

    class Config:
        env_file = ".env"

Document Models

Basic Document

from beanie import Document, Indexed
from pydantic import Field, EmailStr
from datetime import datetime
from typing import Optional

class User(Document):
    email: Indexed(EmailStr, unique=True)
    name: str
    hashed_password: str
    is_active: bool = True
    created_at: datetime = Field(default_factory=datetime.utcnow)
    updated_at: datetime = Field(default_factory=datetime.utcnow)

    class Settings:
        name = "users"  # Collection name
        use_state_management = True

    class Config:
        json_schema_extra = {
            "example": {
                "email": "user@example.com",
                "name": "John Doe"
            }
        }

Document with Relationships

from beanie import Document, Link, BackLink
from typing import List, Optional

class Author(Document):
    name: str
    books: List[BackLink["Book"]] = Field(original_field="author")

    class Settings:
        name = "authors"

class Book(Document):
    title: str
    author: Link[Author]
    categories: List[Link["Category"]] = []

    class Settings:
        name = "books"

class Category(Document):
    name: str
    books: List[BackLink[Book]] = Field(original_field="categories")

    class Settings:
        name = "categories"

Embedded Documents

from beanie import Document
from pydantic import BaseModel
from typing import List

class Address(BaseModel):
    street: str
    city: str
    country: str
    postal_code: str

class Contact(BaseModel):
    type: str  # "email", "phone"
    value: str
    is_primary: bool = False

class Customer(Document):
    name: str
    addresses: List[Address] = []
    contacts: List[Contact] = []

    class Settings:
        name = "customers"

Query Patterns

Basic CRUD Operations

# Create
user = User(email="user@example.com", name="John")
await user.insert()

# Create with validation
user = await User.insert_one(
    User(email="user@example.com", name="John")
)

# Read by ID
user = await User.get(user_id)

# Read with filter
users = await User.find(User.is_active == True).to_list()

# Update
user.name = "Jane"
await user.save()

# Partial update
await user.set({User.name: "Jane", User.updated_at: datetime.utcnow()})

# Delete
await user.delete()

Advanced Queries

from beanie.operators import In, RegEx, And, Or

# Find with operators
active_users = await User.find(
    And(
        User.is_active == True,
        User.created_at >= start_date
    )
).to_list()

# Regex search
users = await User.find(
    RegEx(User.name, "^John", options="i")
).to_list()

# In operator
users = await User.find(
    In(User.email, ["a@test.com", "b@test.com"])
).to_list()

# Pagination
users = await User.find_all().skip(20).limit(10).to_list()

# Sorting
users = await User.find_all().sort(-User.created_at).to_list()

# Projection (select specific fields)
users = await User.find_all().project(UserSummary).to_list()

Aggregation Pipelines

from beanie import PydanticObjectId

class UserStats(BaseModel):
    total_users: int
    active_users: int
    avg_age: float

# Aggregation pipeline
pipeline = [
    {"$match": {"is_active": True}},
    {"$group": {
        "_id": None,
        "total": {"$sum": 1},
        "avg_age": {"$avg": "$age"}
    }}
]

result = await User.aggregate(pipeline).to_list()

# Using Beanie aggregation
from beanie.odm.queries.aggregation import AggregationQuery

stats = await User.find(User.is_active == True).aggregate([
    {"$group": {
        "_id": "$department",
        "count": {"$sum": 1}
    }}
]).to_list()

Indexes

from beanie import Document, Indexed
from pymongo import IndexModel, ASCENDING, DESCENDING, TEXT

class Product(Document):
    # Single field index
    sku: Indexed(str, unique=True)

    # Compound index defined in Settings
    name: str
    category: str
    price: float
    description: str

    class Settings:
        name = "products"
        indexes = [
            # Compound index
            IndexModel(
                [("category", ASCENDING), ("price", DESCENDING)],
                name="category_price_idx"
            ),
            # Text index
            IndexModel(
                [("name", TEXT), ("description", TEXT)],
                name="search_idx"
            ),
            # TTL index
            IndexModel(
                [("expires_at", ASCENDING)],
                expireAfterSeconds=0,
                name="ttl_idx"
            )
        ]

Transactions

from beanie import Document
from motor.motor_asyncio import AsyncIOMotorClientSession

async def transfer_funds(
    from_account_id: str,
    to_account_id: str,
    amount: float,
    session: AsyncIOMotorClientSession
):
    async with await session.start_transaction():
        from_account = await Account.get(from_account_id, session=session)
        to_account = await Account.get(to_account_id, session=session)

        if from_account.balance < amount:
            raise ValueError("Insufficient funds")

        await from_account.set(
            {Account.balance: from_account.balance - amount},
            session=session
        )
        await to_account.set(
            {Account.balance: to_account.balance + amount},
            session=session
        )

Service Layer Pattern

from typing import List, Optional
from beanie import PydanticObjectId

class UserService:
    async def get_by_id(self, user_id: str) -> Optional[User]:
        return await User.get(PydanticObjectId(user_id))

    async def get_by_email(self, email: str) -> Optional[User]:
        return await User.find_one(User.email == email)

    async def get_all(
        self,
        skip: int = 0,
        limit: int = 100,
        is_active: Optional[bool] = None
    ) -> List[User]:
        query = User.find_all()
        if is_active is not None:
            query = User.find(User.is_active == is_active)
        return await query.skip(skip).limit(limit).to_list()

    async def create(self, data: UserCreate) -> User:
        user = User(**data.model_dump())
        await user.insert()
        return user

    async def update(self, user_id: str, data: UserUpdate) -> Optional[User]:
        user = await self.get_by_id(user_id)
        if not user:
            return None
        update_data = data.model_dump(exclude_unset=True)
        update_data["updated_at"] = datetime.utcnow()
        await user.set(update_data)
        return user

Additional Resources

Reference Files

For detailed patterns and migration guides:

• references/migrations.md - Database migration strategies
• references/performance.md - Query optimization tips
• references/relationships.md - Link and BackLink patterns

Example Files

Working examples in examples/:

• examples/document_models.py - Complete document definitions
• examples/aggregations.py - Aggregation pipeline examples
• examples/service.py - Service layer implementation