ODMantic

Documentation: https://art049.github.io/odmantic/

Sync and Async ODM (Object Document Mapper) for MongoDB based on standard python type hints. Built on top of pydantic for model definition and validation.

Core features:

• Simple: define your model by typing your fields using python types, build queries using python comparison operators

• Developer experience: field/method autocompletion, type hints, data validation, perform database operations with a functional API

• Fully typed: leverage static analysis to reduce runtime issues

• AsyncIO support: works well with ASGI frameworks (FastAPI, quart, sanic, Starlette, ...) but works also perfectly in synchronous environments

• Serialization: built in JSON serialization and JSON schema generation

Python: 3.7 and later (tested against 3.7, 3.8, 3.9 and 3.10)

MongoDB: 4.0 and later

pip install odmantic

To enjoy an async context without any code boilerplate, you can reproduce the following steps using the AsyncIO REPL (only for Python 3.8+).

python3.8 -m asyncio

If you are using an earlier version of Python you can use IPython which provide an Autoawait feature (starting from Python 3.6).

from typing import Optional

from odmantic import Field, Model


class Publisher(Model):
    name: str
    founded: int = Field(ge=1440)
    location: Optional[str] = None

By defining the Publisher class, we've just created an ODMantic model 🎉. In this example, the model will represent book publishers.

This model contains three fields:

• name: This is the name of the Publisher. This is a simple string field without any specific validation but it will be required to build a new Publisher.

• founded: This is the year of foundation of the Publisher. Since the printing press has been invented in 1440, it would be handy to allow only values above 1440. The ge keyword argument passed to the Field is exactly doing this. The model will require a founded value greater or equal than 1440.

• location: This field will contain the country code of the Publisher. Defining this field as Optional with a None default value makes it a non required field that will be set automatically when not specified.

The collection name has been defined by ODMantic as well. In this case it will be publisher.

instances = [
    Publisher(name="HarperCollins", founded=1989, location="US"),
    Publisher(name="Hachette Livre", founded=1826, location="FR"),
    Publisher(name="Lulu", founded=2002)
]

We defined three instances of the Publisher model. They all have a name property as it was required. All the foundations years are later than 1440. The last publisher has no location specified so by default this field is set to None (it will be stored as null in the database).

For now, those instances only exists locally. We will persist them in a database in the next step.

For the next steps, you'll need to start a local MongoDB server.The easiest way is to use docker. Simply run the next command in a terminal (closing the terminal will terminate the MongoDB instance and remove the container).

docker run --rm -p 27017:27017 mongo

First, let's connect to the database using the engine. In ODMantic, every database operation is performed using the engine object.

from odmantic import AIOEngine

engine = AIOEngine()

By default, the AIOEngine (stands for AsyncIOEngine) automatically tries to connect to a MongoDB instance running locally (on port 27017). Since we didn't provide any database name, it will use the database named test by default.

The next step is to persist the instances we created before. We can perform this operation using the AIOEngine.save_all method.

await engine.save_all(instances)

Most of the engine I/O methods are asynchronous, hence the await keyword used here. Once the operation is complete, we should be able to see our created documents in the database. You can use Compass or RoboMongo if you'd like to have a graphical interface.

Another possibility is to use mongo CLI directly:

mongo --eval "db.publisher.find({})"

Output:

connecting to: mongodb://127.0.0.1:27017
{
  "_id": ObjectId("5f67b331514d6855bc5c54c9"),
  "founded": 1989,
  "location": "US",
  "name": "HarperCollins"
},
{
  "_id": ObjectId("5f67b331514d6855bc5c54ca"),
  "founded":1826,
  "location": "FR",
  "name": "Hachette Livre"
},
{
  "_id": ObjectId("5f67b331514d6855bc5c54cb"),
  "founded": 2002,
  "location": null,
  "name": "Lulu"
}

The created instances are stored in the test database under the publisher collection.

We can see that an _id field has been added to each document. MongoDB need this field to act as a primary key. Actually, this field is added by ODMantic and you can access it under the name id.

print(instances[0].id)
#> ObjectId("5f67b331514d6855bc5c54c9")

Since we now have some documents in the database, we can start building some queries.

First, let's find publishers created before the 2000s:

early_publishers = await engine.find(Publisher, Publisher.founded <= 2000)
print(early_publishers)
#> [Publisher(name="HarperCollins", founded=1989, location="US),
#>  Publisher(name="Hachette Livre", founded=1826, location="FR")]

Here, we called the engine.find method. The first argument we need to specify is the Model class we want to query on (in our case Publisher). The second argument is the actual query. Similarly to SQLAlchemy, you can build ODMantic queries using the regular python operators.

When awaited, the engine.find method will return the list of matching instances stored in the database.

Another possibility is to query for at most one instance. For example, if we want to retrieve a publisher from Canada (CA):

ca_publisher = await engine.find_one(Publisher, Publisher.location == "CA")
print(ca_publisher)
#> None

Here the result is None because no matching instances have been found in the database. The engine.find_one method returns an instance if one exists in the database otherwise, it will return None.

Finally, let's edit some instances. For example, we can set the location for the publisher named Lulu. First, we need to gather the instance from the database:

lulu = await engine.find_one(Publisher, Publisher.name == "Lulu")
print(lulu)
#> Publisher(name="Lulu", founded=2002, location=None)

We still have the same instance, with no location set. We can change this field:

lulu.location = "US"
print(lulu)
#> Publisher(name="Lulu", founded=2002, location="US)

The location has been changed locally but the last step to persist this change is to save the document:

await engine.save(lulu)

We can now check the database state:

mongo --eval "db.publisher.find({name: 'Lulu'})"

Output:

connecting to: mongodb://127.0.0.1:27017
{
  "_id": ObjectId("5f67b331514d6855bc5c54cb"),
  "founded": 2002,
  "location": "US",
  "name": "Lulu"
}

The document have been successfully updated !

Now, what if we would like to change the foundation date with an invalid one (before 1440) ?

lulu.founded = 1000
#> ValidationError: 1 validation error for Publisher
#> founded
#>   ensure this value is greater than 1440
#>   (type=value_error.number.not_gt; limit_value=1440)

This will raise an exception as it's not matching the model definition. The raised exception is actually a ValidationError created by from pydantic.

If you already have experience with Pydantic and FastAPI, the Usage with FastAPI example sould be interesting for you to get kickstarted.

Otherwise, to get started on more advanced practices like relations and building more advanced queries, you can directly check the other sections of the documentation.

If you wish to contribute to the project (Thank you! 😃), you can have a look to the Contributing section of the documentation.

This project is licensed under the terms of the ISC license.