This is a pre-production deployment of Warehouse, however changes made here WILL affect the production instance of PyPI.
Latest Version Dependencies status unknown Test status unknown Test coverage unknown
Project Description

cromlech.sqlalchemy provides utility functions and components to ease the use of SQLAlchemy.

Test setup

We get a temporary dir to place our databases:

>>> import tempfile
>>> working_dir = tempfile.mkdtemp('cromlech-sqlalchemy-test')

Configuration & registration

>>> from cromlech.sqlalchemy import create_and_register_engine
>>> user_db =  'sqlite:///%s/users.db' % working_dir
>>> store_db = 'sqlite:///%s/store.db' % working_dir
>>> engine = create_and_register_engine(user_db, 'Users')
>>> engine = create_and_register_engine(store_db, 'Store')


Let’s use declarative extension:

>>> from sqlalchemy.ext.declarative import declarative_base

>>> UserBase = declarative_base()

In SQLAlchemy you would declare wich engine the database is bound to. In our case the url will come in environ with the first connection so we deffer this to the moment when engine will be created. The package provides a method for that:

>>> from cromlech.sqlalchemy import deferred_bind
>>> deferred_bind(UserBase.metadata, 'Users')

We can define our SQLAlchemy model:

>>> from sqlalchemy import Column, Integer, String

>>> class User(UserBase):
...     __tablename__ = 'test_users'
...     id = Column('id', Integer, primary_key=True)
...     name = Column('name', String(50))
...     def __repr__(self):
...         return "User(%d, '%s')" % (,


To do any operation we will have use SQLAlchemySession context manager providing an engine name:

>>> from cromlech.sqlalchemy import SQLAlchemySession, get_session

Typically first request may create our tables:

>>> with SQLAlchemySession("Users") as session:
...     UserBase.metadata.create_all()
...     print session.query(User).all()

Transaction of database is ruled by the general transaction package:

>>> import transaction
>>> transaction.commit()

Inside the context you can also fetch the session by its configuration name, this is convenient from anywhere in the code, eg a fonction:

>>> def add_user(id, name):
...     session = get_session("Users")
...     session.add(User(id=id, name=name))

>>> with SQLAlchemySession("Users") as session:
...     add_user(1, 'bob')
...     print session.query(User).all()
[User(1, 'bob')]


For this chapter let’s assume we are inside a with SQLAlchemySession clause:

>>> ctxmanager = SQLAlchemySession("Users")
>>> session = ctxmanager.__enter__()

Transaction are linked to zope transaction using zope.sqlalchemy, all operations above are not yet commited. Let’s abort to see bob is not present no more:

>>> transaction.abort()

>>> print session.query(User).all()

Now add it again an commit so that it’s ok:

>>> transaction.commit()
>>> add_user(1, 'bob')
>>> transaction.commit()
>>> print session.query(User).all()
[User(1, 'bob')]

Now we really are in a new transaction:

>>> transaction.abort()
>>> print session.query(User).all()
[User(1, 'bob')]

Let’s end our session for now

>>> ctxmanager.__exit__(None, None, None)

More than one database

We can use more than one database, Let’s define another model in another database:

>>> StoreBase = declarative_base()

We can define our SQLAlchemy model:

>>> class Product(StoreBase):
...     __tablename__ = 'test_products'
...     id = Column('id', Integer, primary_key=True)
...     name = Column('name', String(50))
...     def __repr__(self):
...         return "Product(%d, '%s')" % (,

And bind:

>>> deferred_bind(StoreBase.metadata, 'Store')

Create tables:

>>> with transaction:
...     with SQLAlchemySession("Store") as session:
...         StoreBase.metadata.create_all()

and an product adder:

>>> def add_product(id, name):
...     session = get_session('Store')
...     session.add(Product(id=id, name=name))

We may do operations in databases using our objects:

>>> with transaction:
...   with SQLAlchemySession("Store") as session:
...     with SQLAlchemySession("Users") as session:
...         add_user(2, 'juan')
...         add_product(1, 'table')

It works :

>>> with SQLAlchemySession("Users") as session:
...     session.query(User).all()
[User(1, 'bob'), User(2, 'juan')]
>>> with SQLAlchemySession("Store") as session:
...     session.query(Product).all()
[Product(1, 'table')]

More than one metadata

Now we may also bind another metadata to same database, and this work even if it’s already initialised:

>>> GroupBase = declarative_base()

>>> class Group(GroupBase):
...     __tablename__ = 'test_groups'
...     id = Column('id', Integer, primary_key=True)
...     name = Column('name', String(50))
...     def __repr__(self):
...         return "Group(%d, '%s')" % (,

>>> deferred_bind(GroupBase.metadata, "Users")

>>> with transaction:
...     with SQLAlchemySession("Users") as session:
...         GroupBase.metadata.create_all()
...         session.add(Group(id=1, name='them'))
...         session.query(Group).all()
[Group(1, 'them')]


Calling an engine that does not exists without a config fails:

>>> from cromlech.sqlalchemy.components import query_engine
>>> print query_engine('spam_db')

Unique session

Whenever you know that your application will always associate same db name with same DBName, you can setup session once and only once, removing overhead to single request. For that use the SharedSQLAlchemySession context manager. Note that it is still a scoped session (no sharing between thread), and that you don’t have to care when it will be initialized, you just use the context manager

>>> from cromlech.sqlalchemy import SharedSQLAlchemySession
>>> with SharedSQLAlchemySession("Store") as session:
...     the_session = session
...     session.query(Product).all()
[Product(1, 'table')]

session will be reused:

>>> with SharedSQLAlchemySession("Store") as session:
...     print session is the_session

to phase or not two phase

You can pass a two_phase boolean argument to activate or not two_phase commit. see zope.sqlalchemy. If you do not pass it, two_phase will be set to true for mysql and postgre (known supporting database) and false otherwise.

You may force it however:

>>> with transaction:
...     with SQLAlchemySession("Store", two_phase=True):
...         add_product(10, 'chair')
Traceback (most recent call last):

as sqlite does not support it !


0.2a1 (2011-07-30)

  • The context manager and util functions no longer depend on an environ. This allows us to use this package outside of a WSGI stack.


  • Initial release
Release History

Release History


This version

History Node

TODO: Figure out how to actually get changelog content.

Changelog content for this version goes here.

Donec et mollis dolor. Praesent et diam eget libero egestas mattis sit amet vitae augue. Nam tincidunt congue enim, ut porta lorem lacinia consectetur. Donec ut libero sed arcu vehicula ultricies a non tortor. Lorem ipsum dolor sit amet, consectetur adipiscing elit.

Show More

Download Files

Download Files

TODO: Brief introduction on what you do with files - including link to relevant help section.

File Name & Checksum SHA256 Checksum Help Version File Type Upload Date
cromlech.sqlalchemy-0.2a1.tar.gz (7.8 kB) Copy SHA256 Checksum SHA256 Source Jul 30, 2011

Supported By

WebFaction WebFaction Technical Writing Elastic Elastic Search Pingdom Pingdom Monitoring Dyn Dyn DNS HPE HPE Development Sentry Sentry Error Logging CloudAMQP CloudAMQP RabbitMQ Heroku Heroku PaaS Kabu Creative Kabu Creative UX & Design Fastly Fastly CDN DigiCert DigiCert EV Certificate Rackspace Rackspace Cloud Servers DreamHost DreamHost Log Hosting