In SQLAlchemy, can I create an Engine from an existing ODBC connection?

Question

I am working in an environment where I am given an ODBC connection, which has been created using credentials to which I don't have access (for security reasons). However I would like to access the underlying database using SQLAlchemy - so my question is, can I pass this ODBC connection to something like create_engine, or alternatively, wrap it in such a way that it looks like a SQLAlchemy connection?

As a supplementary question (and working on the optimistic assumption that the first part can be satisfied) is there a way that I can tell SQLA what dialect to use for the underlying RDBMS?

Thanks

Answer 1

yes you can:

from sqlalchemy import create_engine
from sqlalchemy.pool import StaticPool
eng = create_engine("mssql+pyodbc://", poolclass=StaticPool, creator=lambda: my_odbc_connection)

however, if you truly have only one connection already created, as opposed to a callable that creates them, you must only use this engine in a single thread, one operation at a time. It is not threadsafe for use in a multithreaded application.

If OTOH you can actually get at a Python function that creates new connections whenever called, this is much more appopriate:

from sqlalchemy import create_engine
eng = create_engine("mssql+pyodbc://", creator=my_odbc_connection_function)

the above engine will pool connections normally and can be used freely as a source of connectivity.