Ascetic, a lightweight Python datamapper ORM

Ascetic exists as a super-lightweight datamapper ORM (Object-Relational Mapper) for Python.


Ascetic ORM based on “Data Mapper” pattern. It also supports “Active Record” pattern, but only as a wrapper, the model class is fully free from any service logic. Ascetic ORM follows the KISS principle. Has automatic population of fields from database (see the example below) and minimal size. You do not have to specify the columns in the class. This follows the DRY principle. Ascetic ORM as small as possible.

Inside ascetic.contrib (currently under development) you can find the next solutions:

All extensions support composite primary/foreign keys.

Identity Map” has SERIALIZABLE isolation level by default.

What Ascetic ORM does not? Ascetic ORM does not make any data type conversions (use connection features like this), and does not has “Unit of Work”. I recommend using a Storm ORM, if you need these features.

Ascetic ORM is released under the MIT License (see LICENSE file for details).

This project is currently under development, and not stable. If you are looking for stable KISS-style ORM, pay attention to Storm ORM.


PostgreSQL Example

Using these tables:

CREATE TABLE ascetic_tests_models_author (
    id serial NOT NULL PRIMARY KEY,
    first_name VARCHAR(40) NOT NULL,
    last_name VARCHAR(40) NOT NULL,
    bio TEXT
    id serial NOT NULL PRIMARY KEY,
    title VARCHAR(255),
    author_id integer REFERENCES ascetic_tests_models_author(id) ON DELETE CASCADE


You can configure in one the following ways:

1. Put in your PYTHONPATH file with your settings. See file ascetic/ for more details.

2. Define settings module in environment variable ASCETIC_SETTINGS.

3. Call ascetic.settings.configure(), for example:

import ascetic.settings.configure
    'DATABASES': {
        'default': {
            'engine': "postgresql",
            'user': "devel",
            'database': "devel_ascetic",
            'password': "devel",
            'debug': True,
            'initial_sql': "SET NAMES 'UTF8';",

Model declaration

There is two way to declare models as DataMapper or ActiveRecord.

Datamapper way

class Author(object):
    def __init__(self, id=None, first_name=None, last_name=None, bio=None): = id
        self.first_name = first_name
        self.last_name = last_name = bio

class AuthorMapper(Mapper):
        defaults = {'bio': 'No bio available'}
        validations = {'first_name': (
            lambda v: len(v) > 1 or "Too short first name",
            lambda self, key, value: value != self.last_name or "Please, enter another first name",


class Book(object):
    def __init__(self, id=None, title=None, author_id=None): = id
        self.title = title
        self.author_id = author_id

class BookMapper(Mapper):
    db_table = 'books'
    relationships = {
        'author': ForeignKey(Author, related_name='books')


ActiveRecord way

Indeed, it’s not an ActiveRecord, - it’s just a wrapper over DataMapper.

from ascetic.model import Model
from ascetic.mappers import get_mapper
from ascetic.relations import ForeignKey, OneToMany

class Author(Model):
    class Mapper(object):
        defaults = {'bio': 'No bio available'}
        validations = {'first_name': (
            lambda v: len(v) > 1 or "Too short first name",
            lambda self, key, value: value != self.last_name or "Please, enter another first name",

class Book(Model):
    author = ForeignKey(Author, related_name='books')

    class Mapper(object):
        db_table = 'books'

Now we can create, retrieve, update and delete entries in our database. Creation

james = Author(first_name='James', last_name='Joyce')
get_mapper(Author).save(james)  # Datamapper way

u = Book(title='Ulysses',  # Use ActiveRecord wrapper


a = Author.get(1)
a.first_name # James
a.books      # Returns list of author's books

# Returns a list, using LIMIT based on slice
a = Author.q[:10]   # LIMIT 0, 10
a = Author.q[20:30] # LIMIT 20, 10


a = Author.get(1) = 'What a crazy guy! Hard to read but... wow!'



SQLBuilder integration

object_list = Book.q.tables(
    (Book.s & Author.s).on(Book.s.author_id ==
    (Author.s.first_name != 'James') & (Author.s.last_name != 'Joyce')

Query object based on sqlbuilder.smartsql, see more info.

Signals support

  • pre_init
  • post_init
  • pre_save
  • post_save
  • pre_delete
  • post_delete
  • class_prepared


You can use Ascetic ORM with lightweight web-frameworks, like wheezy.web, Bottle, Tornado, pysi, etc.

Other projects

See also: