This is an SQLAlchemy dialect for the EXASOL database.

• EXASOL: http://www.exasol.com
• SQLAlchemy: http://www.sqlalchemy.org

We assume you have a good understanding of (unix)ODBC. If not, make sure you read their documentation carefully - there are lot's of traps to step into.

If you do not have access to an EXASolution database, download EXASolo for free from EXASOL: http://www.exasol.com/en/test-drive/

The database is a VM image. You will need VirtualBox, VMWare Player, or KVM to run the database. Start the database and make sure you can connect to it as described in the How-To from EXASOL.

On Linux/Unix like systems you need:

• the packages unixODBC and unixODBC-dev >= 2.2.14
• Python >= 2.7
• Download and install the ODBC client drivers from EXASOL >= 4.2
• configure ODBC.ini and ODBCINST.ini

> pip install sqlalchemy-exasol

from sqlalchemy import create_engine
e = create_engine("exa+pyodbc://A_USER:[email protected]:1234/my_schema")
r = e.execute("select 42 from dual").fetch_all()

The dialect supports two connection urls for create_engine. A DSN (Data Source Name) mode and a host mode:

Note:

• Schema name and parameters are optional for the host url string
• Always use all lower-case identifiers for schema, table and column names. SQLAlchemy treats all lower-case identifiers as case-insensitive, the dialect takes care of transforming the identifier into a case-insensitive representation of the specific database (in case of EXASol this is upper-case as for Oracle)
• As of EXASol client driver version 4.1.2 you can pass the flag 'INTTYPESINRESULTSIFPOSSIBLE=y' in the connection string (or configure it in your DSN). This will convert DECIMAL data types to Integer-like data types. Creating integers is a factor three faster in Python than creating Decimals.

The unixodbc Stack is not the most friendly for programmers. If you get strange errors from the driver mangager, you might have an issue with the names of the unixodbc libs. Have a look at https://github.com/blue-yonder/sqlalchemy_exasol/blob/master/fix_unixodbc_so.sh to find ideas on how to fix this on Ubuntu. Good luck!