Is there a chance that the .idx is a compressed (b/g/zipped or rar-ed) .index?
Freedict used the DICT(d) platform without direct modifications. I am not sure what exactly the history of stardict was, but I think it started as a kind of overlay on DICT, so there is a chance that .idx is just a minor modification -- but I'm speculating.
I'm afraid that unless Sebastian has investigated this when working on extending Freedict export range, the exact answer may be hidden in stardict source code (or documentation).
But, as a first step, try decompressing .idx with various tools.
On 20/06/18 09:31, François Patte wrote:
For a dictionary, dictd protocol (and freedict) uses a .index file which
is human readable. stardict, goldendict... use a .idx file (which is not
human readable). Both stardict and goldendict are able to use a
dictionary with a .index file, but dictd is unable to use a dictionary
with a .idx file.
Is there a way to convert a .idx file into a .index file?