rdfapi-php-interest

[Rdfapi-php-interest] Nested SPARQL Graph Patterns

[<pl...@do...>]

Dear all,

I've recently been digging around inside the SPARQL engine, and have =20
come across some behaviour which I *believe* to be contrary to the =20
SPARQL spec (although I'm very much open to being corrected :)

Given a query WHERE clause such as

{
     { ?s :p1 ?o }
   OPTIONAL
     { ?s :p2 ?o2 .
       ?s :p3 ?o3 .
     }
}

RAP parses it into a Query object as expected; two GraphPattern =20
objects, with the latter being marked as the OPTIONAL clause part of =20
the former. However, consider:

{
     { ?s :p1 ?o }
   OPTIONAL
     {
       { ?s :p2 ?o2 . }
       { ?s :p3 ?o3 . }
     }
}

This would appear to be a valid SPARQL construct; the group graph =20
pattern of the right-hand-side of the OPTIONAL clause contains two =20
basic graph patterns within it, each containing a single query triple =20
pattern. However, RAP parses this into a Query opbject equivalent to:

{
     { ?s :p1 ?o }
   OPTIONAL
     { }
     { ?s :p2 ?o2 . }
     { ?s :p3 ?o3 . }
}

...effectively saying "?s :p1 ?o; :p2 ?o2; :p3 ?o3, and optionally =20
nothing else". Is this the desired behaviour for the parser? Please =20
feel free to correct me if I've made a mistake; but it appears as if =20
the advent of a new BGP is causing the GGP to be terminated (the spec =20
only provides for the advent of a new BGP to close any previously open =20
BGP, not GGP, as far as I'm aware :)

Thanks very much for any clarification you may be able to offer!

Cheers,
Peter

Re: [Rdfapi-php-interest] Nested SPARQL Graph Patterns

[Christian W. <cw...@cw...>]

Peter,

> triple pattern. However, RAP parses this into a Query opbject
> equivalent to:
>=20
> {
>      { ?s :p1 ?o }
>    OPTIONAL
>      { }
>      { ?s :p2 ?o2 . }
>      { ?s :p3 ?o3 . }
> }

This is a behavior of SparqlParser that is weird, but cannot be taken
away simply due to the internal working.
In my SparqlEngineDb, I check if the pattern is empty and skip it in
that case.


--=20
Regards/Mit freundlichen Gr=FC=DFen
Christian Weiske

Re: [Rdfapi-php-interest] Nested SPARQL Graph Patterns

[<pl...@do...>]

Christian (or whoever may be able to answer :),

Unfortunately (unless you've spotted something I've not?) I don't see =20
any way that you can disambiguate this case. How can we know from the =20
Query object if the query was
  {
       { ?s :p1 ?o }
     OPTIONAL
       { ?s :p2 ?o2 . }
       { ?s :p3 ?o3 . }
  }

or

  {
       { ?s :p1 ?o }
     OPTIONAL
       { { ?s :p2 ?o2 . }
         { ?s :p3 ?o3 . }
       }
  }

...as the Query object contains nothing linking those BGPs (?s :p2 ?o2 =20
and ?s :p3 ?o3) as the right hand side of the OPTIONAL? They are =20
simply graph patterns in the Array, the same as ?s :p1 ?o; the SPARQL =20
engine will simply join them onto the result, not left-join them. Am I =20
missing something?

Cheers,
Peter

Quoting Christian Weiske <cw...@cw...>:

> Peter,
>
>> triple pattern. However, RAP parses this into a Query opbject
>> equivalent to:
>>
>> {
>>      { ?s :p1 ?o }
>>    OPTIONAL
>>      { }
>>      { ?s :p2 ?o2 . }
>>      { ?s :p3 ?o3 . }
>> }
>
> This is a behavior of SparqlParser that is weird, but cannot be taken
> away simply due to the internal working.
> In my SparqlEngineDb, I check if the pattern is empty and skip it in
> that case.
>
>
> --
> Regards/Mit freundlichen Gr=FC=DFen
> Christian Weiske
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2005.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> Rdfapi-php-interest mailing list
> Rdf...@li...
> https://lists.sourceforge.net/lists/listinfo/rdfapi-php-interest
>