How do you deal with circular imports ?

[askz]

For example I have models like this :

# modules/patients/models.py
class Patient(db.Model, OwnerMixin):	
	__owner_backref_name__ = "patients"

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

	phone = db.Column(db.String)
	email = db.Column(db.String)
	first_name = db.Column(db.String)
	last_name = db.Column(db.String)
	birthday = db.Column(db.DateTime)
	zip_code = db.Column(db.String)
	address = db.Column(db.String)

        # I would like to do this, but there's circular import....
       @aggregated('claims', db.Column(db.Integer))
	def unpaid_amount(self):
		return db.func.count('1')

# modules/accounting/models.py
class Claim(db.Model, OwnerMixin):
	__owner_backref_name__ = 'claims'

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

	date = db.Column(db.DateTime, nullable=False)
	label = db.Column(db.String, nullable=False)
	amount = db.Column(db.Float, nullable=False)
	ref = db.Column(db.String)	

	patient_id = db.Column(db.Integer, db.ForeignKey('patient.id'), nullable=False)
	patient = db.relationship(
				Patient,
				backref=db.backref('claims', cascade="delete, delete-orphan")
	)

Patient is in dedicated module "patient"
and claim model is in 'accounting' modules with other models related to accounting.
The problem is I can't import claim model and I can't access to claim model in patient model in order to, for example, make the sum of claims amount.
I think this is some design issue, but I want to be sure if there's any workaround before rewriting....

[khorolets]

Hi!

In the example of @aggregated models are in the same file.

I'd try something like this http://docs.sqlalchemy.org/en/latest/orm/query.html#sqlalchemy.orm.query.Query.count in the query.

Probably @frol could suggest something more useful.

[frol]

I am not completely sure I got your question, but here are some ideas:

1. You can pass a string as a first argument to the db.relationship (instead of the class object)
2. The general approach to workaround circular dependency is to import inside a function.

[askz]

Thanks for the quick answers.

Sorry for the missing examples.
Here it is :

@aggregated('unpaid_consultations', db.Column(db.Integer))
	def unpaid_consultations_count(self):
		return db.func.count(Patient.unpaid_consultations)
	
	unpaid_consultations = db.relationship(
		'Claim',
		primaryjoin=
				'and_(Claim.patient_id == Patient.id, Claim.invoice_id.is_(None))',
		viewonly=True
	)

This is actually working really fine.

I would like to be able to have maybe aggregate like this :

	@aggregated('unpaid_consultations', db.Column(db.Integer))
	def unpaid_sum(self):
		return db.func.sum('amount').label('unpaid_sum')

To return the sum of the amount column select by the above relationship...
I know I'm mixing things here but this is for the example.

(I know this isn't related to this project directly, I appreciate your help!)

[askz]

In fact my query is something like :

 SELECT SUM(amount) FROM claim WHERE patient_id = %d AND invoice_id IS NOT NULL;

[khorolets]

@askz And what about @frol's advice about providing related model as string without importing it? Didn't it help?

patient = db.relationship(
    'Patient',
    backref=db.backref('claims', cascade="delete, delete-orphan")
)

@frol should it work?

[askz]

Yes this is what I've been up these last minutes.
Many thanks. I resolved it with this solution plus :

	@aggregated('claims', db.Column(db.Integer))
	def unpaid_sum(self):
		return db.func.sum(Claim.amount)