Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!pacific.mps.ohio-state.edu!linac!mp.cs.niu.edu!rickert
From: rickert@mp.cs.niu.edu (Neil Rickert)
Newsgroups: comp.compression
Subject: Re: How to code to a text file.
Keywords: text file arithmetic coding hard coded wired range printable characters
Message-ID: <1991Mar23.132656.32451@mp.cs.niu.edu>
Date: 23 Mar 91 13:26:56 GMT
References: <550@spam.ua.oz>
Organization: Northern Illinois University
Lines: 31

In article <550@spam.ua.oz> ross@spam.ua.oz.au (Ross Williams) writes:
>
>(Abstract: This short article describes a simple way to code a stream
>of random bits into a stream of text characters.)
>
 I think you would do better with Huffman coding.  Or, more precisely, with
Huffman decoding.

 Take your 94 printable characters (any other number of characters will work).
Determine a fixed Huffman encoding for data that uses these characters
uniformly.  Announce the encoding.

 To encode a bit stream, do the appropriate Huffman decoding, resulting in
a byte stream using these 94 characters.  To reproduce the bit stream, do the
Huffman encoding.  The nice part is that it allows bit streams to be
transparently sent from an ASCII site to an EBCDIC site.  The Huffman table
at each site reflect the character set actually used.  (Of course use only
characters for which the ASCII to EBCDIC conversion is universally agreed to).

 Some care is probably needed to add padding bits if needbe, and remove them
on reconstruction of the bitstream, but a leading bit count could handle
this.

 Of course, this method only works if the Huffman table to be used is
agreed to by sender and receiver.

-- 
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
  Neil W. Rickert, Computer Science               <rickert@cs.niu.edu>
  Northern Illinois Univ.
  DeKalb, IL 60115                                   +1-815-753-6940