* Benct Philip Jonsson said on 2009-01-07 16:26:17 +0100
> taliesin the storyteller wrote:
> >As for the actual lookup, there are so called lookup "strategies":
> >exact, prefix, substring, regular expression, soundex, levenshtein etc.
> >How easy it easy to use a specific strategy to search varies with
> >clients.
> I looked at this a bit last time it was up and if
> I understood correctly (which may well not be the case)
> the 'dictionary' is essentially a key--value hash.
> Is there any means to peep into the values when searching?
> (Even though that'd be slow of course!)

kdict (for KDE) has a mode where it lists all possible candidates for a
search and then lets you choose, so it's client-dependent. Will that do? 

> As it happens I've got several tsv dictionary files which
> could easily be converted to a key--value format, but would
> it be reasonably easy to populate a DICT‑dictionary from
> such a file, withouot any retyping of actual entries?

If they are regular, and not just pretend tsv, it is easy enough to
programmatically change them into something that is easily indexable by
the dict-tools (to build the key-value pairs that the server uses).

dictfmt(1) is one way of making the index:
It supports several formats

For TEI, there's

Dictdlib (no good online docs but there's some in the package) is a
python library to read and write dictd-files.

t., who has the flu and thus not so quick to answer :(