如何组织SQLAlchemy关系以进行类似JSON的字典访问？

Question

I have an application where each user can have many different clothing sizes associated with their account. I have modeled these associations with association tables, and linked them through relationship() s in my models. Before I began trying to shove the relationships into a dictionary, they were linked at the top level, like so:

class User(db.Model):

    id = db.Column(db.Integer, primary_key=True)

    # user many-to-many size associations (using link tables)
    sz_shirt_dress_sleeve = db.relationship(
        'SizeKeyShirtDressSleeve',
        secondary=LinkUserSizeShirtDressSleeve,
        backref=db.backref('users', lazy='dynamic'))
    sz_shirt_dress_neck = db.relationship(
        'SizeKeyShirtDressNeck',
        secondary=LinkUserSizeShirtDressNeck,
        backref=db.backref('users', lazy='dynamic'))
    sz_shirt_casual = db.relationship(
        'SizeKeyShirtCasual',
        secondary=LinkUserSizeShirtCasual,
        backref=db.backref('users', lazy='dynamic'))

And I could access them like so:

>>> from app.models import User
>>> u1 = User.query.first()
>>> u1.sz_shirt_dress_sleeve
[Dress shirt sleeve size: 3000]

Because I will eventually have ~20 of these relationships, and because I want to access them more programmatically elsewhere, I would like to store them in a JSON looking dict like this:

class User(db.Model):

    id = db.Column(db.Integer, primary_key=True)

    sizes = {
        'shirt-sleeve': {
            'key': 'shirt-sleeve',
            'values': db.relationship(
                'SizeKeyShirtDressSleeve',
                secondary=LinkUserSizeShirtDressSleeve,
                backref=db.backref('users', lazy='dynamic'))
        }
    }

By doing that, I would be able to access the list of sizes like this:

>>> <some_user>.sizes['shirt-sleeve']['values']
[3000, 3050, 3100]

The above attempt at fitting these associations inside a dict doesn't work, as I've found, because "[SQLAlchemy's use of metaclasses] doesn't pick up these objects inside something else like a dictionary." Well crap.

In a roundabout way to I tried to appease the metaclass top level sensing with this:

class User(db.Model):

    id = db.Column(db.Integer, primary_key=True)

    sz_shirt_dress_neck = db.relationship(
        'SizeKeyShirtDressNeck',
        secondary=LinkUserSizeShirtDressNeck,
        backref=db.backref('users', lazy='dynamic'))

    sizes = {
        'shirt-neck': {
            'key': 'shirt-neck',
            'values': lambda self: self.sz_shirt_dress_neck
        }
    }

That didn't work (though it didn't error either):

>>> from app.models import User
>>> u1 = User.query.first()
>>> u1.sizes['shirt-neck']['values']
<function User.<lambda> at 0x102aedb70>

How can I store these relationships in a JSON like dictionary? (and a small side question: where can I learn design principles to teach me if I should do it this way? I'm self taught and these larger design concepts aren't really covered by documentation/tutorials)

Answer 1

Define a method on your model and modify User object then return it:

class User(db.Model):
    # your codes...
    def as_dict(self):
        self.sizes = {
            'shirt-neck': {
                'key': 'shirt-neck',
                'values': self.sz_shirt_dress_neck
            },
            # another one
        }
        return self


>>> u1 = User.query.first()
>>> if u1:
...     user = u1.as_dict()
...     user.sizes