I am working with a database containing Unicode data in NVARCHAR2 fields. Selecting rows that contain those fields just works with current cx_Oracle and SQLAlchemy as long as I don’t actually try to limit my result set by one of those in the query (like ‘%ü%’). I tried it with PL/SQL Developer and like unistr(‘%\00fc%’) [...]

The module to access an Oracle database with Python is cx_Oracle. This works for all versions from 8i to 11g. To compile you need to have Oracle libraries, but luckily you do not need the full installation — the “instant client” with OCI is enough. (Thanks for pointing that out in this article.)
When using the [...]