I am trying to define a model using sqlalchemy such that one of the columns i only need in memory for processing, i do not have a corresponding database column for that. I need it such that when i save the in memory object, it should not try to persist that special column.

More context: I am mapping a json object to an sql alchemy object using python marshmallow library. In the json response i have additional fields which i do not need in the database but do need in memory for computation.

I searched for the sqlalchemy docs, but could not find a way to skip the persistence of certain columns.

class AdSchema(Schema):"""Schema Mapping class."""id = fields.Int(dump_only=True)available_formats = fields.List(fields.Str())@post_loaddef make_ad(self, data):return AdModel(**data)class AdModel(Base):"""Ad ORM class."""__tablename__ = 'ad_model'id = Column(Integer, primary_key=True)

Am i missing something here?

What I am essentially looking for is a "virtual" attribute, similar to what is available in Rails ORM How to add a virtual attribute to a model in Ruby on Rails?

I was able to get around this by doing the following:

class MyModel(Base):__tablename__ = 'my_model'id = Column(Integer, primary_key=True)def __init__(self, **kwargs):self.non_persisted_column = kwargs['non_persisted_column']kwargs.pop('non_peristed_column')super().__init__(kwargs)

Effectively, a class attribute non_persisted_column is added to the class in the __init__ method. Then, non_persisted_column is removed from kwargs before kwargs is passed into the SQLAlchemy Base super class's __init__ method, so SQLAlchemy will not raise an error and will not persist non_persisted_column to the database.