lzw.py

[Robert Kern]

[posted and e-mailed]

"Dinu C. Gherman" wrote:
> 
> Hi,
> 
> I had some fun writing Python functions to encode and decode
> data using a variant of the Lempel-Ziv coding algorithm family,
> namely LZW, Lempel-Ziv-Welch.
> 
> This is something like a "taboo algorithm" if I understand the
> weird logic of companies and their mind-boggling interpretation
> of intellectual property rights. To me this sounds like saying
> some of Newton's algorithms were illegal to use without paying
> for it.

Except that Newton was never granted a patent on the method and apparati
for them.

> In any case I don't want to mess with Unisys and let them im-
> pose their view of the world on me, but I also don't want to
> end, somewhat early, in "what-happens-to-be-just-a-regular"
> car accident...
> 
> So, is there a way to make my rather limited brain understand,
> what I and you can do with such a code without going to all
> patent offices and legal departments of companies? Am I, my-
> self, allowed to use it? Is there a difference between using
> the coding and decoding part? Can I pass anything of this to
> somebody else? Is it illegal to send it, receive it, read it,
> download it, use it? -- All of it?

IIRC, you cannot use or distribute the patented algorithm without a
license.  See http://corp2.unisys.com/LeadStory/lxwfaq.html for more
information and the contact information for the people you need to talk
to to get a license.  Supposedly, they give out free (_gratis_) licenses
to many "freeware," closed source projects; they may be more sticky
about open source software.

> Of course, there is zlib, but who says there won't be another
> deep-pocket company popping up tomorrow just to say they have
> invented this "before" already?

I consider the possibility unlikely.  Besides, I assume that you want
the LZW for encoding/decoding binary data for PDF in addition to other
desires.  zlib won't help you read or write LZW data.  You can use it to
avoid writing it, but you are still stuck on reading it.

Just contact the people mentioned on the webpage and hope that you get
lucky.

> Regards,
> 
> Dinu
> 
> --
> Dinu C. Gherman

Robert Kern
kern at caltech.edu