sticky-marshmallow

sticky-marshmallow is a Python library that provides mongoDB persistence for marshmallow schema based data. It dereferences nested entities and stores them in their own collections. sticky-marshmallow uses a few practices from DDD (repositories, entities, value objects).

This project follows the convention-over-configuration philosophy and makes the following assumptions:

• Root-level objects as well as nested objects (using fields.Nested) that need to be stored in their own collection require an id = fields.Str() field.
• The schema name excluding Schema is used as its collection name (except when explicitly provided through its Meta class).
• Schemas have a @post_load decorated method that returns a Python class that allows its id attribute to be set.

Defining repositories

There are two ways of defining repositories to store objects. The first and easiest is to instantiate sticky-marshmallow's Repository object directly with your schema as its only argument: Repository(BookSchema). The second method provides more configuration options through its Meta class and requires subclassing Repository:

from sticky_marshmallow import Repository

class BookRepository(Repository):
    class Meta:
        schema = BookSchema
        collection = 'book'

Database connection

sticky-marshmallow needs to be aware of your database connection. You have two options here:

1. Use sticky-marshmallow's connect() pymongo wrapper to connect to a database, which automatically registers the database:

from sticky_marshmallow import connect

db = connect('test', host='localhost')

Or:

2. Use your own code to connect to a database, then register the database:

import pymongo
from sticky_marshmallow import register_db

db = pymongo.MongoClient(**connection_settings)['test']
register_db(db)

Example

Notes

This library is not built with performance in mind.

TODO

✓ Implement Cursor that lazily loads MongoDB documents ✓ Implement delete() ✓ Return object from save()

• Implement save() with iterables as nested objects
• Use collection from Repository.Meta
• Prevent infinite loops