The way PynamoDB is implemented is that it looks to a specific single DynamoDB table:
class UserModel(Model):
class Meta:
# Specific table.
table_name = 'dynamodb-user'
region = 'us-west-1'
The way my infrastructure works is that it has as many dynamodb tables as I have clients, so a single Lambda function has to deal with any amount of separate tables that are identical in structure e.g. represent "UserModel". I can't specify a concrete one.
How would I make this model definition dynamic?
Thanks!
Open-sourced a solution that is tested and works.
Read README.md for more details.
Code:
from typing import TypeVar, Generic, Type
from pynamodb.models import Model
T = TypeVar('T')
class ModelTypeFactory(Generic[T]):
def __init__(self, model_type: Type[T]):
self.__model_type = model_type
# Ensure that given generic belongs to pynamodb.Model class.
if not issubclass(model_type, Model):
raise TypeError('Given model type must inherit from pynamodb.Model class!')
def create(self, custom_table_name: str, custom_region: str) -> Type[T]:
parent_class = self.__model_type
class InnerModel(parent_class):
class Meta:
table_name = custom_table_name
region = custom_region
return InnerModel