Serialchemy

SQLAlchemy model serialization.

Motivation

Serialchemy was developed as a module of Flask-RESTAlchemy, a lib to create Restful APIs using Flask and SQLAlchemy. We first tried marshmallow-sqlalchemy, probably the most well-known lib for SQLAlchemy model serialization, but we faced issues related to nested models. We also think that is possible to build a simpler and more maintainable solution by having SQLAlchemy in mind from the ground up, as opposed to marshmallow-sqlalchemy that had to be designed and built on top of marshmallow.

How to Use it

Serializing Generic Types

Suppose we have an Employee SQLAlchemy model declared:

class Employee(Base):
    __tablename__ = "Employee"

    id = Column(Integer, primary_key=True)
    fullname = Column(String)
    admission = Column(DateTime, default=datetime(2000, 1, 1))
    company_id = Column(ForeignKey("Company.id"))
    company = relationship(Company)
    company_name = column_property(
        select([Company.name]).where(Company.id == company_id)
    )
    password = Column(String)

Generic Types are automatically serialized by `ModelSerializer`:

from serialchemy import ModelSerializer

emp = Employee(fullname="Roberto Silva", admission=datetime(2019, 4, 2))

serializer = ModelSerializer(Employee)
serializer.dump(emp)

# >>
{
    "id": None,
    "fullname": "Roberto Silva",
    "admission": "2019-04-02T00:00:00",
    "company_id": None,
    "company_name": None,
    "password": None,
}

New items can be deserialized by the same serializer:

new_employee = {"fullname": "Jobson Gomes", "admission": "2018-02-03"}
serializer.load(new_employee)

# >> <Employee object at 0x000001C119DE3940>

Serializers do not commit into the database. You must do this by yourself:

emp = serializer.load(new_employee)
session.add(emp)
session.commit()

Custom Serializers

For anything beyond Generic Types we must extend the ModelSerializer class:

class EmployeeSerializer(ModelSerializer):

    password = Field(load_only=True)  # passwords should be only deserialized
    company = NestedModelField(Company)  # dump company as nested object


serializer = EmployeeSerializer(Employee)
serializer.dump(emp)
# >>
{
    "id": 1,
    "fullname": "Roberto Silva",
    "admission": "2019-04-02T00:00:00",
    "company": {"id": 3, "name": "Acme Co"},
}

Extend Polymorphic Serializer

One of the possibilities is to serialize SQLalchemy joined table inheritance and it child tables as well. To do such it's necessary to set a variable with the desired model class name. Take this Employee class with for instance and let us assume it have a joined table inheritance:

class Employee(Base):
    ...
    type = Column(String(50))

    __mapper_args__ = {"polymorphic_identity": "employee", "polymorphic_on": type}


class Engineer(Employee):
    __tablename__ = "Engineer"
    id = Column(Integer, ForeignKey("employee.id"), primary_key=True)
    association = relationship(Association)

    __mapper_args__ = {
        "polymorphic_identity": "engineer",
    }

To use a extended ModelSerializer class on the Engineer class, you should create the serializer as it follows:

class EmployeeSerializer(
    PolymorphicModelSerializer
):  # Since this class will be polymorphic

    password = Field(load_only=True)
    company = NestedModelField(Company)


class EngineerSerializer(EmployeeSerializer):
    __model_class__ = Engineer  # This is the table Serialchemy will refer to
    association = NestedModelField(Association)

Contributing

For guidance on setting up a development environment and how to make a contribution to serialchemy, see the contributing guidelines.

Release

A reminder for the maintainers on how to make a new release.

Note that the VERSION should folow the semantic versioning as X.Y.Z Ex.: v1.0.5

1. Create a release-VERSION branch from upstream/master.
2. Update CHANGELOG.rst.
3. Push a branch with the changes.
4. Once all builds pass, push a VERSION tag to upstream.
5. Merge the PR.