Well, I think it would be easier/better to just "outsource" the search to a search engine like nutch, which creates an index from the _output_ and not from the data source meaning that you could also index pages with content from the data module etc.
-will On 23.06.2010, at 12:35, Manuel Molaschi wrote: > > If so how would you query on two different location or repositories > > merging the hits ordered by their score? > To achieve this, you should go deep into jackrabbit / lucene api building a > MultiSearcher that reads multiple workspaces indicies and run query on it. > > Manuel > > 2010/6/22 Fabrizio Giustina <[email protected]> > > Hi Will, > we implemented jcr:score along with other nice features (spell check, > paging, lazy loading of contents) in the criteria project at > http://www.openmindlab.com/lab/products/mgnlcriteria.html > > We still didn't released a final version of this implementation (will > be version 2.0) although we have done at least a dozen of internal > beta releases, fully working on the real projects... we are cleaning > up the support for different jackrabbit versions (since at is some > features are tied to the jackrabbit version the project is compiled > with) before adding a proper documentation to the website and finalize > the release. > In the meanwhile anyway you can grab the source from svn for an > inspiration, it's well documented and with several tests. > > the only bad news: > > If so how would you query on two different location or repositories > > merging the hits ordered by their score? > > merging the score from different queries is not supported by Lucene, > the score number only has a value relative to a single query and > sorting different results using this value is not really meaningful > (other then definitively heavy from a performance side, since you will > have to get all the results for sorting) > > fabrizio > > > > On Sun, Jun 20, 2010 at 9:12 PM, Will Scheidegger > <[email protected]> wrote: > > > > Dear Magnolians > > > > So queries can be ordered by "jcr:score"... but how can I access the > > jcr:score? According to [1] jcr:score is a ColumnSpecifier, except that it > > seems the column cannot be accessed, because jcr:score > > - is not stored in the MetaData of the Content objects returned from the > > query > > - is not in the properties of the javax.jcr.Node object associated with the > > Content object > > > > Does this value only exist during query time and does not make it to the > > returned objects? If so how would you query on two different location or > > repositories merging the hits ordered by their score? Or to be more > > precise: How can I query the website and dms at the same time returning the > > hits in one combined list according to their score? > > > > Thanks! > > -will > > > > ---------------------------------------------------------------- > > For list details see > > http://www.magnolia-cms.com/home/community/mailing-lists.html > > To unsubscribe, E-mail to: <[email protected]> > > ---------------------------------------------------------------- > > > > > > ---------------------------------------------------------------- > For list details see > http://www.magnolia-cms.com/home/community/mailing-lists.html > To unsubscribe, E-mail to: <[email protected]> > ---------------------------------------------------------------- > > ---------------------------------------------------------------- For list details see http://www.magnolia-cms.com/home/community/mailing-lists.html To unsubscribe, E-mail to: <[email protected]> ----------------------------------------------------------------
