Django Smart Manager

Django Smart Manager provides a simple framework for representing and managing Django models from serializable templates.

Problem Overview

Oftentimes what we model in Django spans multiple objects and tables. Managing a single object that is represented by multiple models can be quite cumbersome through the shell or through basic Django administration. This app provides a framework such that a user can write templates that represent many models and complex relationships.

For example, assume that you model a person. The Person model contains a unique identifier for that person, multiple PhoneNumber models that point to it, and multiple Address models. With Django Smart Manager, one can construct a template in the following manner:

{
    'unique_id': 'person_unique_id':
    'phone_numbers': ['865-123-4985', '956-345-5678'],
    'addresses': [{
        'street': 'my street address1',
        'city': 'my city',
    }],
}

Using the framework (as shown soon), we can construct a parser for this template that maintains the appropriate model representation underneath while also providing a much simpler way to manage all of those underlying models. This management includes updates to the data in the template and deletions to objects in the template.

Building the Person Example

Before we show how to build the example just illustrated in the problem overview, the models in the example are laid out below:

class Person(models.Model):
    unique_id = models.CharField(max_length=64, unique=True)


class PhoneNumber(models.Model):
    person = models.ForeignKey(Person)
    number = models.CharField(max_length=32)


class Address(models.Model):
    person = models.ForeignKey(Person)
    street = models.CharField(max_length=128)
    city = models.CharField(max_length=128)

In order to achieve the ability of specifying the Person model (and its related models) via a template as shown above, the user must create a model template class that inherits BaseSmartManager. This class will be responsible for taking in the template and managing the object(s) represented by the template.

For the sake of example, let's assume we're going to build three model template classes: one to manage addresses, one to manage phone numbers, and one to manage a person and its associated addresses and phone numbers. We'll begin from the bottom up by building the Address model template. The code for this is shown below:

from smart_manager import BaseSmartManager


class AddressSmartManager(BaseSmartManager):
    def build(self):
        self.build_obj(
            person_id=self._template['person'],
            street=self._template['street'],
            city=self._template['city'],
        )

Now that we have this class, it can be called in the following way to build its associated object:

AddressSmartManager({
    'person': 1,  # The pk of a Person object
    'street': 'my street',
    'city': 'my city',
}).build()

Underneath the hood, it is passing the parameters of build_obj to the upsert function Django Manager Utils and also internally maintaining all of the objects that have been built.

Now that the model template class has been created, a SmartManager model can be constructed as follows:

from smart_manager import SmartManager


mt = SmartManager.objects.create(smart_manager_class='path.to.AddressSmartManager', template={
    'person': 1,
    'street': 'my street',
    'city': 'my city',
})

Once this model is created, it manages all of the objects associated with the template. If the user was to change the template and save the mt variable from the example, the underlying Address model would be updated. Similarly, the underlying Address model will also be deleted when mt is deleted. The deletion behavior can be turned off by specifying manages_deletions=False in the creation of the model template.

While this example is trivial, the power of Django Smart Manager is unleashed when you start to build more and more complex objects that need ot be managed. Let's assume that the user can now build the associated PhoneNumberSmartManager class for creating PhoneNumber objects and move on to creating the PersonSmartManager model template class:

class PersonSmartManager(BaseSmartManager):
    def build(self):
        # Build the parent person object
        person = self.build_obj(unique_id=self._template['unique_id'])

        # Build its child phone number objects using the PhoneNumberSmartManager
        for phone_number in self._template['phone_numbers']:
            self.build_obj_using(PhoneSmartManager, {
                'person': person.id,
                'phone_number': phone_number
            })

        # Build its child address objects using the AddressSmartManager
        for address in self._template['addresses']:
            self.build_obj_using(AddressSmartManager, {
                'person': person.id,
                'street': address['street'],
                'city': address['city'],
            })

Note that the PersonSmartManager uses the build_obj_using function to build an object using another model template. This ensures that the objects managed by that model template are also managed by the calling model template.

Similarly, one can now make a SmartManager object using this model template class to manage a complete Person object.

Using Smart Managers Directly with Django Models and Managers

Smart mangers can be directly used by models and managers in Django. All they need to do is inherit the SmartModelMixin or SmartManagerMixin. If a model inherits SmartModelMixin, it is provided a smart_upsert and smart_delete function. These function both take a smart manager class and template. The smart_upsert function upserts the template using the smart manager that linked to it (or creating a smart manager if it doesn't exist). The smart_delete function will delete the smart manager that is associated with the model, which in turn deletes all other objects managed by that smart manager.

If a Django model manager inherits SmartManagerMixin, it is provided a smart_create function that takes a smart manager class and template. The objects are created using the template and a smart manager is returned.

These methods are meant as convenience methods so that a user can still interact with their models and not have to directly query smart managers.

Note that all functions return a smart manager, and the mixins can be imported directly from smart_manager as so:

from smart_manager import SmartModelMixin, SmartManagerMixin

Caveats with Smart Managers

It is up to the programmer to ultimately define how a template manages its underlying objects. By default, Django Smart Manager will manage deletions of every object built using the build_obj function. This, however, can cause undesired side effects for some objects that simply should not be deleted if the template is deleted. If this is the case, a is_deletable kwarg can be passed to the build_obj function to override the default behavior of managing its deletion.