Hi Fred,
Fred Eaker wrote:
> I am interested in performing fine grain full-text search on text
> documents stored in a Jackrabbit repository.
>
> The results I would like to get from the search include the nodes that
> the documents are stored in, along with specific information about
> where the searched text appears in the document itself.
JSR 170 does not define any methods that deal with highlighting search results.
Therefore jackrabbit does not implement such a functionality. But I'm not sure
if that's what you are looking for.
> For example, if my search for the word "security" hits an XML document
> stored in an nt:resource node, I would like to know the xml:id and
> xml:lang of of <p> element that the hit resides in.
if your XML document is exploded as document view under the nt:resource node as
jcr:data child node you could use the following query:
//p[jcr:contains(jcr:xmltext, 'security')](@xml:id, @xml:lang)
> To get this kind of functionality, would a custom QueryHandler need to
> be written that indexes and searches that kind of information?
if your XML document is stored as a binary property jcr:data, then you'd
probably have to write your own query handler or do some post processing to get
the id and lang information.
> I have written a Lucene-based program that indexes, searches and
> returns the kind of results I am interested in. Would I now need to
> incorporate this into Jackrabbit, or is this kind of functionality
> possible already?
see above.
regards
marcel
|