view content/Coding/024-m209.rst @ 4:7ce6393e6d30

Adding converted blog posts from old blog.
author Brian Neal <bgneal@gmail.com>
date Thu, 30 Jan 2014 21:45:03 -0600
parents
children 6e0d4799796d
line wrap: on
line source
Introducing m209
################

:date: 2013-08-01 20:05
:tags: Python, m209, Enigma
:slug: introducing-m209
:author: Brian Neal

I'm very pleased to announce yet another M-209_ simulator written in Python,
creatively called m209_. Last summer I worked on Enigma_ simulators in both
`Python <http://py-enigma.readthedocs.org/en/latest/>`_ and `C++
<https://bitbucket.org/bgneal/cpp-enigma>`_, and I thought it would be fun to
try another World War II-era crypto device. m209_ is a Python 3 library and
command-line utility for encrypting and decrypting text by simulating the
operation of an actual M-209_ device.

One fun part about doing something like this is researching the original
device. It seems like there are more resources online about the M-209_ than the
Enigma_.  I even found an actual 1940's War Department training film on YouTube
that explains how to operate the M-209_, including the procedure for encrypting
and decrypting messages!  I want to thank `Mark J. Blair`_ for his very
informative pages on the M-209_ which were very helpful to me. Check out the
`m209 references section <https://m209.readthedocs.org/en/latest/#references>`_
for these and other useful links.

The M-209_ isn't as complex as the Enigma_. That isn't meant to knock it. The
M-209_, while cryptographically not as secure as the Enigma_, is a remarkable
piece of mechanical engineering.  It is much more portable and easier to
operate compared to the Enigma_. It has user-friendly features like printing to
paper tape and a letter counter for backing up when mistakes are made.
According to Wikipedia, about 140,000 of these machines were produced. They
even come up on eBay a few times a year, and seem to go for between $1000
- $2000 USD. Maybe someday I can score an actual unit!

Coding the actual simulator wasn't all that hard.  I spent much more time on
the unit tests, documentation, and creating an application to generate key
lists. Writing the documentation gave me some good practice with Sphinx_, an
awesome Python based documentation tool that uses the `reStructured Text`_
markup language.

Writing the key list generator was actually the hardest part. The procedure for
creating key lists is spelled out in a M-209 manual from 1944 (which exists
online as a series of photos). The procedure is kind of loosely specified, and
a lot is left up to the soldier creating the key list. I came up with an
ad-hoc, heuristic-based algorithm that works most of the time. If it got stuck
it simply started over, and retried up to a certain number of attempts.

While researching the procedure, I noticed what appears to be a typo in the
data tables in the manual that are used when developing a key list. On top of
that I found several sets of initial numbers that I could not generate a key
list from. In other words, using these starting numbers, my algorithm could not
generate M-209 settings that satisfied the exit criteria for the procedure in
the manual.  After a while, I just removed those troublesome initial conditions
as possible inputs. It would be interesting to return to this some day and
write a program to search the solution space exhaustively to see if there
really was a solution for these numbers. It could just be that my
trial-and-error algorithm could not find a solution, even after tens of
thousands of attempts. However this doesn't seem likely. I wonder if these
initial settings caused lots of head scratching for the poor officer trying to
create a key list.

In any event, if you are into this kind of thing, I hope you check out m209_.
Doing a project like this is a lot of fun. I enjoy doing the research, creating
the code, and working on the test suite. I also get some practice with Python
packaging and writing documentation with Sphinx.

Future enhancements include adding the ability to read Mark Blair's key lists
that he created for his C++ simulator. This would make it easier for our two
simulators to interoperate.

Links:

* `m209 documentation <http://m209.readthedocs.org>`_
* `m209 on PyPi <https://pypi.python.org/pypi/m209>`_
* `m209 source code repository on Bitbucket <https://bitbucket.org/bgneal/m209/>`_

.. _M-209: http://en.wikipedia.org/wiki/M-209
.. _m209: http://m209.readthedocs.org/
.. _Enigma: https://en.wikipedia.org/wiki/Enigma_machine
.. _Mark J. Blair: http://www.nf6x.net/groups/m209group/
.. _Sphinx: http://sphinx-doc.org/index.html
.. _reStructured Text: http://docutils.sf.net/rst.html