Sohaib Farooqi Sohaib Farooqi - 2 months ago 11
Python Question

Serialize two nested schema with marshmallow

I am fairly new to python. I have two SQLAlchemy models as follows:

class listing(db.Model):
id = db.Integer(primary_key=True)
title = db.String()
location_id = db.Column(db.Integer, db.ForeignKey('location.id'))
location = db.relationship('Location', lazy='joined')

class location(db.Model):
id = db.Integer(primary_key=True)
title = db.String()


I have two Marshmallow schema classes for them:

class ListingSchema(Schema):
id = fields.Int()
title = fields.Str()
location_id = fields.Int()

class LocationSchema(Schema):
id = fields.Int()
title = fields.Str()


I have created a nested schema class like:

class NestedSchema(Schema):
listing = fields.Nested(ListingSchema)
location fields.Nested(LocationSchema)


I am doing the join query like :

listing,location = db.session.query(Listing,Location)\
.join(Location, and_(Listing.location_id == Location.id))\
.filter(Listing.id == listing_id).first()


data gets load in the objects i have checked. How can parse this schema?
I have tried

result,errors = nested_listing_Schema(listing,location)


This gives error: "Listing object is not iteratable."

Answer

The right is to use the class that you created NestedSchema and not nested_schema, do this:

result,errors = NestedSchema().dump({'listing':listing,'location':location})

And the result wiil be:

dict: {
       u'listing': {u'id': 8, u'title': u'foo'},
       u'location': {u'id': 30, u'title': u'bar'}
      }

But i do not understand why you wanted to make a "NestedSchema", i think you can do it another way.

First, forget the class "NestedSchema".

After, change your "ListingSchema", like this:

class ListingSchema(Schema):
    id = fields.Int()
    title = fields.Str()
    location_id = fields.Int()
    location = fields.Nested("LocationSchema") #The diff is here

And now you can do:

listing = db.session.query(Listing).get(listing_id) # Suppose listing_id = 8
result,errors = ListingSchema().dump(listing)
print result

The result will be:

dict: {
        u'id': 8, u'title': u'foo', u'location_id': 30, u'location': {u'id': 30, u'title': u'bar'}
      }

Note that now, "location" is a property of "listing".

And you can still make a Two-Way Nesting, just add a backref in Listing (model) and add ListingSchema as nested in LocationSchema

class Listing(db.Model):
    id = db.Column(db.Integer, primary_key=True)
    title = db.Column(db.String(45), nullable=False)
    location_id = db.Column(db.Integer, db.ForeignKey('location.id'))
    location = db.relationship('Location', lazy='joined', backref="listings") #the diff is here

class LocationSchema(Schema):
    id = fields.Int()
    title = fields.Str()
    listings = fields.Nested("ListingSchema", many=True, exclude=("location",)) #The property name is the same as in bakcref

The many=True is because we have a relationship One-to-Many. The exclude=("location") is to avoid recursion exception.

Now we can search by location too.

location = db.session.query(Location).get(location_id) # Suppose location_id = 30
result,errors = LocationSchema().dump(location)
print result

dict: {u'id': 30, u'title': u'bar', 
       u'listings': [
             {u'location_id': 30, u'id': 8, u'title': u'foo'},
             {u'location_id': 30, u'id': 9, u'title': u'foo bar baz'},
             ...
             ]
       }

You can see the docs about it here

Comments