El vie, 28-03-2008 a las 08:57 +0200, Jukka Zitting escribió:
> Hi,
> 
> On Fri, Mar 28, 2008 at 8:43 AM, Paco Avila <[EMAIL PROTECTED]> wrote:
> > El vie, 28-03-2008 a las 08:26 +0200, Jukka Zitting escribió:
> >  > Or just a normal string property with the text to be indexed.
> >
> >  But, in this case, the query can't be:
> >
> >   /jcr:root//element(*,my:document)[jcr:contains(nt:resource,'hola
> >  mundo')]
> >
> >  and should be something like (if I store the text in my:docText
> >  property:
> >
> >   /jcr:root//element(*,my:document)[jcr:contains(my:docText,'hola
> >  mundo')]
> >
> >  because Lucene is not indexing the "document text version".
> 
> You could use jcr:contains(., 'hola mundo') that looks in all
> properties of a node.

But nt:resource is a subnode and my:docText is a property,
jcr:contains(., 'hola mundo') will search in all properties and
subnodes?

> Alternatively, you could also put the text in a TIFF comment and
> implement a custom TextExtractor class that pulls that comment for
> Jackrabbit to index as the text version of the TIFF file.
> 
> >  By the way, can I get the text generated by text-extractors or
> >  it is only used by Lucene engine?
> 
> No, it's only used for Lucene. But of course you can instantiate and
> run the text extractors manually on any binary property you like.

Thanks.

Reply via email to