I wrote a python script which takes the gimste and adds hyperlinks from the lujvo and gismu mentioned in the list to their entry. The output is here. It is about 500K, almost double the size of the original. It weeds out most English words which look like lujvo (such as "circle"), but there are probably a few that it misses, especially words like "do" and "no".
The source code which generated this consists of 4 parts. hlinks.py is the module that actually does the work. lujvo.py provides some lujvo services. gimste.dat and rafste.dat are the data files used to look up gismu and rafsi. If you want to run this yourself you'll have to change the name of the variable path in lujvo.py and either change or override the input and output filenames in hlinks.py
I am working on a larger module that will analyze lujvo in various ways.
Please mail comments, etc. to araizen@softhome.net. I will probably not have much time to do anything with this after November 9, however.