Populating a StructuredProperty using the ndb.Model Constructor

This article is derived from a Stack Overflow question. The original question can be found here.

Question

I looked into the ndb GitHub sample code, but I couldn’t find any example which shows on how to create a ndb entity with a constructor that contains a StructuredProperty.

Here is the GitHub example.

What if I want to initialize a Contact entity with a list of phone numbers and this list of phone number is not a list of PhoneNumber objects. Instead it is a list of Python dictionaries.

So, given the following Model classes:

class PhoneNumber(ndb.Model):
    """A model representing a phone number."""
    phone_type = ndb.StringProperty(
        choices=('home', 'work', 'fax', 'mobile', 'other'))
    number = ndb.StringProperty()


class Contact(ndb.Model):
    """A Contact model that uses StructuredProperty for phone numbers."""
    # Basic info.
    name = ndb.StringProperty()
    birth_day = ndb.DateProperty()

    # Address info.
    address = ndb.StringProperty()

    phone_numbers = ndb.StructuredProperty(PhoneNumber, repeated=True)

I want to create a Contact using the following Python dictionaries:

phone_number_dicts = [{"phone_type" : "home", number = 122}, {"phone_type" : "work", number = 123}]

contact = Contact(name = "some name", birthday = "some day", phone_numbers = phone_number_dicts)

Am I required to convert a dict to a ndb entity explicitly?
Can I override ndb constructor which converts a dict to a ndb entity and assign?
Any other better approach?

Solution

Simply override the PhoneNumber constructor, so that you can pass in a dict as kwargs to its constructor via the Contact constructor.

class PhoneNumber(ndb.Model):
    phone_type = ndb.StringProperty(
        choices=('home', 'work', 'fax', 'mobile', 'other'))
    number = ndb.StringProperty()

    def __init__(self, *args, **kwargs):
        super(PhoneNumber, self).__init__(*args, **kwargs)
        self.__dict__.update(kwargs)


class Contact(ndb.Model):
    name = ndb.StringProperty()
    birth_day = ndb.DateProperty()
    address = ndb.StringProperty()
    phone_numbers = ndb.StructuredProperty(PhoneNumber, repeated=True)
    company_title = ndb.StringProperty()
    company_name = ndb.StringProperty()
    company_description = ndb.TextProperty()
    company_address = ndb.StringProperty()

    def __init__(self, *args, **kwargs):
        super(Contact, self).__init__(*args, **kwargs)
        if kwargs:
            self.phone_numbers = []
            for kwarg in kwargs.pop('phone_numbers'):
                if isinstance(kwarg, PhoneNumber):
                    self.phone_numbers.append(kwarg)
                else:
                    p = PhoneNumber(**kwarg)
                    self.phone_numbers.append(p)

In this way you can pass a dictionary representation of your PhoneNumber entities to the Contact constructor or a dictionary representation of the PhoneNumber properties to the PhoneNumber constructor.

Here are a few test cases I tried via the Interactive Console of the dev_appserver.py:

from google.appengine.ext import ndb

from models import Contact, PhoneNumber

kwargs = {
    'phone_numbers': [{
        'phone_type': 'home',
        'number': '123',
    }, {
        'phone_type': 'work',
        'number': '456',
    }, {
        'phone_type': 'fax',
        'number': '789',
    }]
}

c = Contact(**kwargs)
print 'Test Case 1:'
print c
print

kwargs = {
    'phone_numbers': [
        PhoneNumber(**{'phone_type': 'home','number': '123'}),
        PhoneNumber(**{'phone_type': 'work','number': '456'}),
        PhoneNumber(**{'phone_type': 'fax', 'number': '789'})
    ]
}

c = Contact(**kwargs)
print 'Test Case 2:'
print c
print

c = Contact(
    phone_numbers=[
        PhoneNumber(phone_type='home', number='123'),
        PhoneNumber(phone_type='work', number='456'),
        PhoneNumber(phone_type='fax', number='789')
    ]
)
print 'Test Case 3:'
print c
print

Output:

Test Case 1:
Contact(phone_numbers=[PhoneNumber(number='123', phone_type='home'),
PhoneNumber(number='456', phone_type='work'),
PhoneNumber(number='789', phone_type='fax')])

Test Case 2:
Contact(phone_numbers=[PhoneNumber(number='123', phone_type='home'),
PhoneNumber(number='456', phone_type='work'),
PhoneNumber(number='789', phone_type='fax')])

Test Case 3:
Contact(phone_numbers=[PhoneNumber(number='123', phone_type='home'),
PhoneNumber(number='456', phone_type='work'),
PhoneNumber(number='789', phone_type='fax')])

As expected, each case elicits the same Contact objects.

Nickolas Kraus

Populating a StructuredProperty using the ndb.Model Constructor

Question

Solution