Connecting to Microsoft SQL Server with python in OS X Yosemite

Due in part to my technology ADD, and desire to be a polyglot developer; it is by no means unusual to find myself in a position where I am working with a mixed technology stack.

In this case I have a number of python micro-services and command line applications running on a Mac Mini loaded with OS X Yosemite that need to retrieve data from a Microsoft SQL Sever 2008 R2 instance on Windows Server 2008 R2.

While SQL Server is a great Relational Database Management System (RDBMS), it isn’t exactly known for it’s ease of use when it comes to cross-platform communication. A quick Google Search will return a range of results (and horror stories) on how to get this all to work.

Getting Started

It is assumed that you already have a working Python installation (Python 2.7+ or 3.3+) and Pip.

Additionally, we will need:

• Homebrew or Macports
• FreeTDS
• pymssql

Decisions and Rationale

I personally prefer to use Homebrew due to it’s simplicity; though I do have a MacBook with Macports installed.

The pymssql vs pyodbc argument is not one I really care to worry about. Though I generally choose to use pymssql due to better support for MS SQL Stored Procedures, and a less troublesome install. I have had issues in the past installing pyodbc via Pip.

Installing FreeTDS

FreeTDS is a set of *nix libraries, that allow applications to talk to Microsoft SQL Server.

In your typical *nix environment FreeTDS sources configuration from two difference places:

${HOME}/.freetds.conf
/etc/freetds.conf

This is slightly different under OS X, and will vary depending on the package manager used to install FreeTDS.

Homebrew Install

brew install freetds

Homebrew symlinks the global freetds.conf file to:

/usr/local/etc/freetds.conf

Macports Install

sudo port install freetds

Macports installs the global freetds.conf file to:

/opt/local/etc/freetds/freetds.conf

FreeTDS Configuration

FreeTDS reads the ${HOME}/.freetds.conf before resorting to the global freetds.conf. You can find out what the location of your global freetds.conf by executing the following command:

tsql -C

Your result will be something along these lines:

$ tsql -C
Compile-time settings (established with the "configure" script)
                            Version: freetds v0.91
             freetds.conf directory: /usr/local/Cellar/freetds/0.91_2/etc
     MS db-lib source compatibility: no
        Sybase binary compatibility: no
                      Thread safety: yes
                      iconv library: yes
                        TDS version: 7.1
                              iODBC: no
                           unixodbc: no
              SSPI "trusted" logins: no
                           Kerberos: no

Full details about the FreeTDS configuration files can be found here.

Testing FreeTDS

tsql -H <host> -p <port> -U <username> -P <password>

If everything is working you will see something along the lines of:

locale is "en_AU.UTF-8"
locale charset is "UTF-8"
using default charset "UTF-8"

My Macports machine presented the following errors:

locale is "en_AU.UTF-8"
locale charset is "UTF-8"
using default charset "UTF-8"
Error 20018 (severity 9):
    Unexpected EOF from the server
    OS error 36, "Operation now in progress"
Error 20002 (severity 9):
    Adaptive Server connection failed
There was a problem connecting to the server.

This was overcome by specifying the TDS Version as an Environment Variable like so:

TDSVER=7.0 tsql -H <host> -p <1433 or custom port> -U <username> -P <password>

Your other option for overcoming this type of error is to specify DNS names within your local freetds.conf file, for example:

[myhost]
	host = dbdev.mydomain.com
	port = 1433
	tds version = 7.0

You can this test this by using the following command:

tsql -S myhost -U <USER> -P <PASSWORD>

Installing pymssql

pymssql is easily installed on your machine via Pip.

pip install pymssql

Once installed you can test the package in a fairly simple manner via Python interactive.

$ python3 -i
Python 3.4.2 (default, Oct 17 2014, 20:25:14) 
[GCC 4.2.1 Compatible Apple LLVM 6.0 (clang-600.0.51)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import pymssql
>>> conn = pymssql.connect("HOST","USER","PASSWORD","DATABASE")
>>> cursor = conn.cursor()
>>> cursor.execute("select * from dbo.[TABLE]")
>>> for row in cursor:
...     print("row = %r" % (row, ))
... 
row = (1, 'LITERAL', Decimal('123.123'))
row = (2, 'LITERAL', Decimal('123.123'))
row = (3, 'LITERAL', Decimal('123.123'))
row = (4, 'LITERAL', Decimal('123.123'))
row = (5, 'LITERAL', Decimal('123.123'))
row = (6, 'LITERAL', Decimal('123.123'))
>>> conn.close()
>>> exit()

pymssql adheres to the Python DB-API 2.0 specification. Meaning that all of the standard cursor methods are available. It also allows you to use ORM packages like SQLAlchemy. pymssql also allows you to call MS SQL Stored procedures via the cursor.execproc method, which for me is a must have feature, particularly when dealing with odd inflexible (EVERYTHING MUST BE A SP) DBA.