SQLAlchemy 0.9 Documentation

Release: 0.9.4 | Release Date: March 28, 2014 | Download PDF

Deprecated Event Interfaces

This section describes the class-based core event interface introduced in SQLAlchemy 0.5. The ORM analogue is described at Deprecated ORM Event Interfaces.

Deprecated since version 0.7: The new event system described in Events replaces the extension/proxy/listener system, providing a consistent interface to all events without the need for subclassing.

Execution, Connection and Cursor Events

class sqlalchemy.interfaces.ConnectionProxy

Allows interception of statement execution by Connections.


ConnectionProxy is deprecated. Please refer to ConnectionEvents.

Either or both of the execute() and cursor_execute() may be implemented to intercept compiled statement and cursor level executions, e.g.:

class MyProxy(ConnectionProxy):
    def execute(self, conn, execute, clauseelement,
                *multiparams, **params):
        print "compiled statement:", clauseelement
        return execute(clauseelement, *multiparams, **params)

    def cursor_execute(self, execute, cursor, statement,
                       parameters, context, executemany):
        print "raw statement:", statement
        return execute(cursor, statement, parameters, context)

The execute argument is a function that will fulfill the default execution behavior for the operation. The signature illustrated in the example should be used.

The proxy is installed into an Engine via the proxy argument:

e = create_engine('someurl://', proxy=MyProxy())
begin(conn, begin)

Intercept begin() events.

begin_twophase(conn, begin_twophase, xid)

Intercept begin_twophase() events.

commit(conn, commit)

Intercept commit() events.

commit_twophase(conn, commit_twophase, xid, is_prepared)

Intercept commit_twophase() events.

cursor_execute(execute, cursor, statement, parameters, context, executemany)

Intercept low-level cursor execute() events.

execute(conn, execute, clauseelement, *multiparams, **params)

Intercept high level execute() events.

prepare_twophase(conn, prepare_twophase, xid)

Intercept prepare_twophase() events.

release_savepoint(conn, release_savepoint, name, context)

Intercept release_savepoint() events.

rollback(conn, rollback)

Intercept rollback() events.

rollback_savepoint(conn, rollback_savepoint, name, context)

Intercept rollback_savepoint() events.

rollback_twophase(conn, rollback_twophase, xid, is_prepared)

Intercept rollback_twophase() events.

savepoint(conn, savepoint, name=None)

Intercept savepoint() events.

Connection Pool Events

class sqlalchemy.interfaces.PoolListener

Hooks into the lifecycle of connections in a Pool.


PoolListener is deprecated. Please refer to PoolEvents.


class MyListener(PoolListener):
    def connect(self, dbapi_con, con_record):
        '''perform connect operations'''
    # etc.

# create a new pool with a listener
p = QueuePool(..., listeners=[MyListener()])

# add a listener after the fact

# usage with create_engine()
e = create_engine("url://", listeners=[MyListener()])

All of the standard connection Pool types can accept event listeners for key connection lifecycle events: creation, pool check-out and check-in. There are no events fired when a connection closes.

For any given DB-API connection, there will be one connect event, n number of checkout events, and either n or n - 1 checkin events. (If a Connection is detached from its pool via the detach() method, it won’t be checked back in.)

These are low-level events for low-level objects: raw Python DB-API connections, without the conveniences of the SQLAlchemy Connection wrapper, Dialect services or ClauseElement execution. If you execute SQL through the connection, explicitly closing all cursors and other resources is recommended.

Events also receive a _ConnectionRecord, a long-lived internal Pool object that basically represents a “slot” in the connection pool. _ConnectionRecord objects have one public attribute of note: info, a dictionary whose contents are scoped to the lifetime of the DB-API connection managed by the record. You can use this shared storage area however you like.

There is no need to subclass PoolListener to handle events. Any class that implements one or more of these methods can be used as a pool listener. The Pool will inspect the methods provided by a listener object and add the listener to one or more internal event queues based on its capabilities. In terms of efficiency and function call overhead, you’re much better off only providing implementations for the hooks you’ll be using.

checkin(dbapi_con, con_record)

Called when a connection returns to the pool.

Note that the connection may be closed, and may be None if the connection has been invalidated. checkin will not be called for detached connections. (They do not return to the pool.)

A raw DB-API connection
The _ConnectionRecord that persistently manages the connection
checkout(dbapi_con, con_record, con_proxy)

Called when a connection is retrieved from the Pool.

A raw DB-API connection
The _ConnectionRecord that persistently manages the connection
The _ConnectionFairy which manages the connection for the span of the current checkout.

If you raise an exc.DisconnectionError, the current connection will be disposed and a fresh connection retrieved. Processing of all checkout listeners will abort and restart using the new connection.

connect(dbapi_con, con_record)

Called once for each new DB-API connection or Pool’s creator().

A newly connected raw DB-API connection (not a SQLAlchemy Connection wrapper).
The _ConnectionRecord that persistently manages the connection
first_connect(dbapi_con, con_record)

Called exactly once for the first DB-API connection.

A newly connected raw DB-API connection (not a SQLAlchemy Connection wrapper).
The _ConnectionRecord that persistently manages the connection