Hello,
indeed I was able to make the article example (search on paragraph nodes and
used "toto -titi") work with using index aggregates.
But unfortunately it does not work with the "multi language" model we have
chosen, because each node also stores multilingual properties in a subnode,
so we do not only have subnodes like:
article/paragraph/subparagraph
but on each level, we also have translation subnodes for different languages
article/translation
article/paragraph/translation
article/paragraph/subparagraph/translation
So what we would really need is:
aggregating article/paragraph and article/paragraph/subparagraph into
article
and
aggregating article/paragraph/translation and
article/paragraph/subparagraph/translation into article/translation
Do you think that this could easily be solved (perhaps with a patch) or is
it too difficult to do ?
Regards,
Benjamin
2009/10/19 khaled tlili <
[hidden email]>
> Hi,
>
> Thanks 4 your answer.
>
> If I understand well, If I add :
>
> <aggregate primaryType="jnt:article">
> <include>*</include>
> <include>*/*</include>
> <include>*/*/*</include>
> <include>*/*/*/*</include>
> </aggregate>
>
> in the indexing configuration file, the following SQL-2 query
>
> select * from jnt:article as a where contain(a.*, 'toto -titi')
>
> would search even on children of jnt:article ?
>
> Regards.
>
> KT.
>
> On Fri, Oct 16, 2009 at 1:06 PM, Ard Schrijvers
> <
[hidden email]>wrote:
>
> > You can achieve this with indexing aggregate
> >
> >
http://wiki.apache.org/jackrabbit/IndexingConfiguration> >
> > Regards Ard
> >
> > On Fri, Oct 16, 2009 at 12:15 PM, khaled tlili <
[hidden email]>
> > wrote:
> > > Hi,
> > >
> > > In the jackrappib api (i.e
> > >
> >
>
http://jackrabbit.apache.org/api/1.4/org/apache/jackrabbit/core/query/jsr283/qom/FullTextSearch.html> > ),
> > > it's written that:
> > >
> > > "The full-text search expression is evaluated against the set of
> > full-text
> > > indexed properties within the full-text search scope. If property is
> > > specified, the full-text search scope is the property of that name on
> the
> > > selector node in the node-tuple; otherwise the full-text search scope
> is
> > all
> > > properties of the selector node (*or, in some implementations, all
> > > properties in the node subtree*)."
> > >
> > > In the jackrabbit implementation, does the implementation of the
> > full-text
> > > search look into all the subtree properties? If not, how can I "force"
> > the
> > > full-text search to look into all properties of node subtree ?
> > >
> > > I need this feature to make queries like " select * from jnt:article
> as
> > a
> > > where contain(a.*, 'toto -titi') "
> > >
> > > Example:
> > > Suupose that we have the following nodes structure:
> > > [jnt:paragraph] > ...
> > > bigText insertText
> > >
> > >
> > > [jnt:article] > ...
> > > smallText title
> > > bigText intro
> > > + * (jnt:paragraph)
> > >
> > > What would be the SQL-2 query that allows to retrieve all articles that
> > > contain "toto" but not "titi" ( = "titi" not present in all paragraph
> of
> > the
> > > arcticle) ?
> > >
> > > Regards.
> > > KT
> > >
> >
>
