I am new to all of this myself, so caveat emptor. However, I think know of at
least two ways to do this.
The first is to use two different buckets, but with the same key universe. One
bucket is for the portion of the records that you want indexed, and the other
is for everything else. Depending on the application, you may want there to be
some duplication between the data that is stored, but this way you can insure
that only the right bits are indexed. For example, I have an application where
some source HTML is stored in one bucket (along with other fields that never
change -- like creation date and source site), and another bucket stores the
stuff that should be indexed and that also periodically changes (i.e., the text
of the html that should be indexed and some user assigned tags).
The second approach is not one that I tried but is something that should work
in principle. You could modify the precommit hooks to selectively index based
on your criteria (in this case, MIME type).
-- GWF
On Jan 21, 2011, at 3:46 PM, Gordon Tillman wrote:
> Howdy Folks,
>
> Is there a way to configure Riak search so that only objects with certain
> content-types (e.g., application/json) are to be indexed?
>
> I have installed a schema on a test bucket that contains field definitions
> for the (JSON) fields that I wish to have indexed. At then end I include a
> dynamic_field definition that instructs search to skip everything else, like
> this:
>
> %% anything left over is not indexed
> {dynamic_field, [
> {name, "*"},
> {skip, true}
> ]}
>
> After populating my bucket with some test data and then dumping the contents
> of the corresponding index bucket (_rsid_test), I see that there are entries
> for every key that is in the test bucket, even for keys that correspond to
> objects that do not have a content-type of application/json and who do not
> contain any data that corresponds to one of the field definitions.
>
> I would really like to avoid having any indexing information stored for
> objects that are not application/json. Is that possible?
>
> Many thanks!
>
> --gordon
> _______________________________________________
> riak-users mailing list
> [email protected]
> http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com
_______________________________________________
riak-users mailing list
[email protected]
http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com
