PynamoDB Single Table

A Pydantic ORM built on top of PynamoDB.

Features

Provides a Django-inspired "Active Record"-style ORM using single-table design built on top of DynamoDB.

import abc
from datetime import datetime
from pydantic import Field, EmailStr
from pynamodb_single_table import SingleTableBaseModel

class BaseTableModel(SingleTableBaseModel, abc.ABC):
    class _PynamodbMeta:
        table_name = "MyDynamoDBTable"

class User(BaseTableModel):
    __table_name__ = "user"
    __str_id_field__ = "username"
    username: str
    email: EmailStr
    account_activated_on: datetime = Field(default_factory=datetime.now)

# Make sure the table exists in DynamoDB
BaseTableModel.ensure_table_exists(billing_mode="PAY_PER_REQUEST")

# Create a record
john, was_created = User.get_or_create(username="john_doe", email="john.doe@email.com")
assert was_created

# Retrieve
john_again = User.get_by_str("john_doe")
assert john_again.email == "john.doe@email.com"

# Update
now = datetime.now()
john_again.account_activated_on = now
john_again.save()

assert User.get_by_str("john_doe").account_activated_on == now

# Delete
john_again.delete()

Motivation

Many use cases need little more than structured CRUD operations with a table-like design (e.g., for storing users and groups), but figuring out how to host that efficiently in the cloud can be a pain.

DynamoDB is awesome for CRUD when you have clean keys. It's a truly serverless NoSQL database, including nice features like:

• High performance CRUD operations when you know your primary keys
• Scale-to-zero usage-based pricing available
• Official local testing capability
• Conditional CRUD operations to avoid race conditions
• Multiple methods of indexing into data
• Scalable with reliable performance

This project, in part, emerges from my frustration with the lack of many truly serverless SQL database services. By "truly serverless", I mean purely usage-based pricing (generally a combination of storage costs and query costs). Many small, startup applications use trivial amounts of query throughput and story trivial amounts of data, but finding a way to deploy such an application into the cloud without shelling out $10-$100's per month is tricky. In AWS, now that Aurora Serverless V1 is being replaced, there is no way to do this.

However, DynamoDB provides not just the basic functionality needed to do this, it's actually a really good option if your data usage patterns can fit within its constraints. That means, primarily, that you can always do key-based lookups, and that you can avoid changing your indexing strategy or database schema too much (e.g. modifying a table from having nullable columns into non-nullable). DynamoDB can do custom queries at tolerable rates, but you're going to get sub-par speed and cost efficiency if you're regularly doing searches across entire tables instead of direct hash key lookups.

Requirements

This project is built on the backs of Pydantic and Pynamodb. I am incredibly grateful to the developers and communities of both of those projects.

Installation

You can install PynamoDB Single Table via pip from PyPI:

$ pip install pynamodb_single_table

Contributing

Contributions are very welcome. To learn more, see the Contributor Guide.

License

Distributed under the terms of the MIT license, PynamoDB Single Table is free and open source software.

Issues

If you encounter any problems, please file an issue along with a detailed description.

Credits

This project was generated from @cjolowicz's Hypermodern Python Cookiecutter template.