On Thu, 26 Jun 2003 at 10:21:10 +0200, Martin Sj?gren wrote: > I'm looking around at indexing&querying engines and aspseek seems very > impressive. However, our interest is not in web indexing, but having a > full text search in our workflow application (a client/server system). > Thus, I wonder, how is the support for > * Indexing non-web sources, like strings in a database, or having an API > (preferrably well-documented :) for indexing so I can write a program to > drag the strings out of the database and put them in the index?
Not at present (nor probably in the near future). However it is pretty simple to create [dynamic] pseudo web pages representing the data in your tables and then index these. Of course it is also possible to pack specific values/data into say keywords meta data for further processing on the client side. > * Querying this from something else than a cgi script? I realize that > since there's searchd, I can ask it from another program, but how does > that work? ASPseek's [stable] API does not cater for this very well at all. However I rewrote the API some months back to incorporate an appropriate framework for development of a PHP module. A side effect is that the new API can quite easily be used to develop other clients (or to integrate support into other software). However it is undocumented :~/ Oh, and not available in the current stable release. There are patches however for stable and it is incorporated into the current 1.3.0 development tree (see CVS section of http://www.aspseek.org/). You'll find patches for stable here: http://aspseek.unixatwork.com/ The PHP module, also available at the above location, is partially documented and most PHP functions are more or less calls straight into the API so the PHP module documentation (and or source code) will give you some clues. Not knowing what your workflow system is developed in I can't offer much in terms of supported development languages. However the API suits integration using C/C++ and the PHP module obviously provides support for PHP. I've not bothered to develop a Perl XS module for the API but it would actually be very simple to do. > Also, since we have multiple text sources, I would like to index them > separately, is it simple to have several indices and when querying, > specify which index to search? Not too difficult using subsets/spaces however I've tended to take the approach of using sites and constraining the client using &st= (or rather in my case the PHP module call 'aspseek_setoption(ASEEKOPT_SITEID, $st)'. The latter approach however means separate sites (domain names) for each resource you want to build an index for. Another alternative would be running completely separate engines/indexers which is also pretty straight forward. > Does aspseek have these things? If not, can someone give me an idea of > how much work it would be to implement these things? We'd rather spend > time on improving an existing, free, engine than writing our own hack. No, but they are not too difficult to fudge ;) Matt.
pgp00000.pgp
Description: PGP signature
