This data was obtained from the US National Archives Catalog:

For 1950 - 2005: https://catalog.archives.gov/id/4734832

For 2006: https://catalog.archives.gov/id/4734834

The allpuf.dat.gz file is just the concatenation of the two main DAT files.

preproc.py does all the preprocessing for allpuf.dat.gz to prepare the output for the bot.

bot.py needs your API credentials for Twitter but should be ready to run other than that.

For more details, see the blog post:

http://aadrake.com/remembering-the-fallen.html