Skip to content


Subversion checkout URL

You can clone with
Download ZIP
Integration SQLAlchemy in Django projects for effective query building purposes
Branch: master

Merge pull request #34 from vmarkovtsev/master

Add support for BigAutoField
latest commit 287489aa2d
@Deepwalker authored




Small package for integration SQLAlchemy into an existent Django project. The primary use case of this package is building complex queries that are not possible with the Django ORM.

You need to include aldjemy at the end of INSTALLED_APPS. When models are imported, aldjemy will read all models and contribute sa attribute to them. sa attribute is a class, mapped to Table class.

Internally, aldjemy generates tables from Django models. This is an important distinction from the standard decision of using SQLAlchemy reflection.

Code example:'Brubeck')

M2M sample:"GROUP_NAME")

Explicit joins is part of SQLAlchemy philosophy, so aldjemy can't get you exactly the same experience as Django. But aldjemy is not positioned as Django ORM drop-in replacement. It's a helper for special situations.

We have some stuff in the aldjemy cache too:

from aldjemy import core
core.Cache.models # All generated models
core.get_tables() # All tables, and M2M tables too

You can use this stuff if you need - maybe you want to build queries with tables, or something like this.


You can add your own field types to map django types to sqlalchemy ones with ALDJEMY_DATA_TYPES settings parameter. Parameter must be a dict, key is result of field.get_internal_type(), value must be a one arg function. You can get idea from aldjemy.types.

Also it is possible to extend/override list of supported SQLALCHEMY engines using ALDJEMY_ENGINES settings parameter. Parameter should be a dict, key is substring after last dot from Django database engine setting (e.g. sqlite3 from django.db.backends.sqlite3), value is SQLAlchemy driver which will be used for connection (e.g. sqlite, sqlite+pysqlite). It could be helpful if you want to use django-postgrespool.


Often django models have helper function and properties that helps to represent the model's data (__unicode__), or represent some model based logic.

To integrate it with aldjemy models you can put these methods into a separate mixin:

class TaskMixin(object):
    def __unicode__(self):
        return self.code

class Task(TaskMixin, models.Model):
    aldjemy_mixin = TaskMixin
    code = models.CharField(_('code'), max_length=32, unique=True)

Voilà! You can use unicode on aldjemy classes, because this mixin will be mixed into generated aldjemy model.

If you want to expose all methods and properties without creating a separate mixin class, you can use the aldjemy.meta.AldjemyMeta metaclass:

from aldjemy.meta import AldjemyMeta

class Task(models.Model):
    code = models.CharField(_('code'), max_length=32, unique=True)

    def __unicode__(self):
        return self.code

    __metaclass__ = AldjemyMeta

The result is same as with the example above, only you didn't need to create the mixin class at all.

Also note that with Python 3, the syntax is a bit different:

class Task(models.Model, metaclass=AldjemyMeta):
    code = models.CharField(_('code'), max_length=32, unique=True)

    def __str__(self):
        return self.code
Something went wrong with that request. Please try again.