A PynamoDB integration for Graphene.

For instaling graphene, just run this command in your shell

pip install graphene-pynamodb

Here is a simple PynamoDB model:

from uuid import uuid4
from pynamodb.attributes import UnicodeAttribute
from pynamodb.models import Model


class User(Model):
    class Meta:
        table_name = "my_users"
        host = "http://localhost:8000"

    id = UnicodeAttribute(hash_key=True)
    name = UnicodeAttribute(null=False)


if not User.exists():
    User.create_table(read_capacity_units=1, write_capacity_units=1, wait=True)
    User(id=str(uuid4()), name="John Snow").save()
    User(id=str(uuid4()), name="Daenerys Targaryen").save()

To create a GraphQL schema for it you simply have to write the following:

import graphene
from graphene_pynamodb import PynamoObjectType


class UserNode(PynamoObjectType):
    class Meta:
        model = User
        interfaces = (graphene.Node,)


class Query(graphene.ObjectType):
    users = graphene.List(UserNode)

    def resolve_users(self, args, context, info):
        return list(User.scan())


schema = graphene.Schema(query=Query)

Then you can simply query the schema:

query = '''
    query {
      users {
        name
      }
    }
'''
result = schema.execute(query)

To learn more check out the following examples:

• Full example: Flask PynamoDB example

graphene-pynamodb includes a basic implementation of relationships using lists. OneToOne and OneToMany relationships are serialized as a List of the ids and unserialized lazyly. The limit for an item's size in DynamoDB is 400KB (see http://docs.aws.amazon.com/amazondynamodb/latest/developerguide/Limits.html) This means the total "row" size including the serialized relationship needs to fit within 400KB so make sure to use this accordingly.

In addition, scan operations on DynamoDB are unsorted by design. This means that there is no reliable way to get a paginated result (Cursor support) on a root PynamoConnectionField.

This means that if you need to paginate items, it is best to have them as a OneToMany relationship inside another Field (usually viewer or node).

After cloning this repo, ensure dependencies are installed by running:

python setup.py install

After developing, the full test suite can be evaluated by running:

python setup.py test # Use --pytest-args="-v -s" for verbose mode