The C source is pretty messed up. You will probably not want to read it.
You don't need to either; the theory is quite simple. The data is first
packed with LZ77. Furthermore it is made lower-case only to reduce the
number of different characters. Then the most common characters are given
a 4-bit code, others an 8-bit, and the data is bitpacked. It ends up at
422b, and the decompressor at 130b, giving a total of 552b.

Max Bennedich  06/98
  max-b@algonet.se
