Re: [aseek-users] Several different indexes? (was: Re: Variant ofFAQ 11: How can I restrict search to different *subsets* of sites?)

[Kir Kolyshkin]

Axel Beckert - ecos gmbh wrote:

Hi again!

Thanks for your fast reply, but we already figured out, that the ul
parameter won't solve our problem. That way I noticed, that I missed
something in my scenario description, which makes things a little bit
more complicated:

We have a bunch of normal multilingual web servers (each language in
it's own directory branch) plus a http download server for all sites.

We want to offer searching on the language subset of each site plus
always the appropriate directories/files of the download server.

The directories or files on the download server which should be be
searchable together with subset X should be exactly those, which are
linked inside subset X. Each subsets represents one language of the
website of a part of the company and all the downloadable documents
linked from that part should be searched if someone searches this
subset.

So determining which files on the download server should be
searchable together with which subset can only be determined when
indexing them.
I suspect, that we need either different indexes for each subset
(with recursing into external sites at depth 1) or some kind of subset
marker for URLs in the index which show from which subset the URL is
linked.

Ans so my new question is: Is it somehow possible to search over
different indexes or merge indexes and mark the URLs from which index
run or subset they came?

Or do we need to crawl the sites with a different tool and insert the
appropriate URLs into a file read with index -f or with index -u?
Would that work at all? How are successive -u and -f parameters
treated? Are they ORed or ANDed?

Probably the best solution to your problem would be custom code developed, which will use data from aspseek mysql db to form subsets you need. I will
tell you the secret - ASPseek's subset is just a vector of URL_IDs (result of SQL query), sorted and written into file. So, you just add your own variant of subset, which will include URLs selected by SQL wildcard, and those linked directly from those URLs.

This seems to be not a complex task, so find a person who knows C++ and SQL, and (s)he will implement in in a matter of days. You can even when contribute the resulting patch back to ASPseek ;)

Alternatively, change your web site layout ;)
--
== kir_at_asplinux.ru == 7551596_at_ICQ == 6722750_at_sms.beemail.ru ==

Stuckness shouldn't be avoided. It's the psychic predecessor of all
real understanding. An egoless acceptance of stuckness is a key to an
understanding of all Quality, in mechanical work as in other endeavors.
-- R. Pirsig, "Zen and the Art of Motorcycle Maintenance"