使用SQLAlchemy为这种继承建模的正确方法是什么？

Question

I have a pair of PostgreSQL tables that look somewhat like this (simplified example):

CREATE TABLE people (
    id SERIAL PRIMARY KEY,
    created timestamp with time zone DEFAULT now() NOT NULL,
    modified timestamp with time zone NULL,
    email varchar NOT NULL UNIQUE,
    inactive boolean NOT NULL DEFAULT False
);

CREATE TABLE engineers (
    id integer NOT NULL REFERENCES people(id) ON DELETE CASCADE ON UPDATE CASCADE UNIQUE,
    created timestamp with time zone DEFAULT now() NOT NULL,
    modified timestamp with time zone NULL,
    login_name varchar NOT NULL UNIQUE,
    PRIMARY KEY(id)
);

engineers inherits from people because the "id" column people passes through to engineers as a foreign key. This column is also defined as a primary key in both tables. If I wanted query the login_name using someone's email address, it could be done like this via SQL:

SELECT p.email FROM engineers e 
  JOIN people p ON p.id = e.id 
  WHERE p.inactive = False AND e.login_name = 'john.smith';

How do I model this relationship using SQLAlchemy's declarative style and perform a similar query? It seems like " Joined Table Inheritance " describes my usage scenario, but I don't have a discriminator column in my tables. I've also been trying to use " Concrete Table Inheritance " but am just managing to confuse myself.

I'd appreciate any suggestions. Thanks!

Edited

The reason this table structure is the way it is is because a "person" might be an "engineer", "accountant", or "tester" (or some combination of the three). The people table contains the attributes that are common to everyone, such as a name, phone, number, date of hire, etc.

The "created" and "modified" columns on the engineers table are not shared between people ; the two columns are distinct to each table. These aren't absolutely necessary, they're just added by default to each table definition in the database and are used to track changes for audit/logging purposes.

Answer 1

What you have here is not clearly either type because you duplicate some columns but not others.

From the docs :

SQLAlchemy supports three forms of inheritance: single table inheritance , where several types of classes are represented by a single table, concrete table inheritance , where each type of class is represented by independent tables, and joined table inheritance , where the class hierarchy is broken up among dependent tables, each class represented by its own table that only includes those attributes local to that class.

Because you duplicate the created and modified columns, this is a candidate for concrete inheritance. However, because you expect Employee to be an incomplete entity without columns from Person , this is also joined table inheritance.

I suggest you try to make this fit into joined table inheritance, but I'm not entirely sure how the mapper will handle the presence of duplicated columns. It is possible in joined table inheritance to manipulate the subclasses separately as relations, but I'm not sure how they will be merged, if they can be merged, or what mapper configuration is available to handle which one is used. If there is a way, this section will probably show you how.

However here is something to get you started. polymorphic_on can also be any sql expression (scroll down to the polymorphic_on argument) --it doesn't have to be a column. If your only subclass is the engineers table, you use an EXISTS subquery in your discriminator.

Below is some untested code that will probably not work without messing with it a bit--it just shows you the pattern you'll have to use.

people_table = Table('people', metadata,
    Column('id', Integer, primary_key=True),
    Column('email', String, unique=True),
    Column('inactive', Boolean, default=False),
)

engineers_table = Table('engineers', metadata,
    Column('id', Integer, ForeignKey('people.id'), primary_key=True),
    Column('engineer_info', String),
)

class Person(object):
    pass

class Engineer(Person):
    pass


discriminator = case(
    [
        (exists().where(people_table.c.id==engineers_table.c.id), 'engineer'),
    ], else_='person')

mapper(Person, people_table, polymorphic_identity='person', polymorphic_on=discriminator)
mapper(Engineer, engineer_table, polymorphic_identity='engineer')

Needless to say, if it's at all possible you should, for your own sanity:

1. Add a discriminator column to your person table.
2. Get rid of the created and modified columns in the engineer table.