Best association approach to connect Sensor model to others

问题

I'm trying to implement the following with Flask+SQLAlchemy: I have two database models containing information about bee apiaries and bee hives. I would like to add feature to somehow connect both of them to Sensor model. Those can be attached to one of apiaries or one of bee hives. Here is what I have.

class Apiary(db.Model):
    __tablename__ = 'apiary'
    # ... fields ...
    beehives = db.relationship("BeeHive", backref=db.backref('apiary', lazy='dynamic'), lazy='dynamic')


class BeeHive(db.Model)
    __tablename__ = 'beehive'
    # ... fields ...
    apiary_id = db.Column(db.Integer(), db.ForeignKey('apiary.id'))


class SensorType(db.Model):

    __tablename__ = 'sensor_type'

    id = db.Column(db.Integer(), primary_key=True)
    title = db.Column(db.Unicode(32), unique=True)
    sensors = db.relationship('Sensor', backref=db.backref('sensor_type', lazy='dynamic'), lazy='dynamic')


class Sensor(db.Model):

    __tablename__ = 'sensor'

    id = db.Column(db.Integer(), primary_key=True)
    serial = db.Column(UUID(), unique=True)
    sensor_type_id = db.Column(db.Integer(), db.ForeignKey('sensor_type.id'))
    readings = db.relationship('SensorReading', backref=db.backref('sensor', lazy='dynamic'), lazy='dynamic')


class SensorReading(db.Model):

    __tablename__ = 'sensor_reading'

    id = db.Column(db.BigInteger(), primary_key=True)
    value = # TODO
    timestamp = db.Column(db.DateTime(), default=db.func.now())

I was surfing through the internet, reading SQLAlchemy documentation and found something about "polymorphic loading". I have good feeling that this is probably what I was searching for, but don't know how to implement it in my case. I have seen similar thing in "Django world" and they call it "GenericForeignKey".

UPDATE: I have found SQLAlchemy examples about this type of association. Can anyone advice me which of those would be optimum approach? discriminator_on_related, generic_fk, table_per_association or table_per_related? Which of those will be the least headache in further expanding application? Cascading delete?

回答1:

After two days of experiments I have came to final conclusion. Examples have been taken from this source.

"Discriminator on association" (candidate for answer):
- (+) has backward reference
- (?) can have 1 parent object
- (-) complexity
"Generic Foreign Key":
- (+) low complexity
- (+) has backward reference
- (?) can have 1 parent object
- (-) programmer's code must take care of cascading actions
"table per association":
- (+) multiple parents
- (+) shared table stays intact
- (?) number of tables raises with number of associated tables
- (-) no backward reference
"table per related" (candidate for answer):
- (+) every associated object is in same table
- (+) can have multiple tables
- (-) associated objects are somehow separated for each foreign table

Answer: "Discriminator on association" as Sensor do not have ability of superposition and therefore no need for multiple parents.

来源：https://stackoverflow.com/questions/28261954/best-association-approach-to-connect-sensor-model-to-others

标签

sqlalchemy

flask-sqlalchemy