[jira] [Commented] (MARMOTTA-552) Sparql Query for Dataset View fails

["Sebastian Schaffert (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/MARMOTTA-552?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14172462#comment-14172462 ] 

Sebastian Schaffert commented on MARMOTTA-552:
----------------------------------------------

The problem here is the SELECT in the second UNION part (which is actually not really needed anyways). The SQL generated looks like this:
{code}
 SELECT U1.U2 AS V4, COUNT(U1_V2.id) AS V5
 FROM ((SELECT NULL AS U4, NULL AS U3, P1.object AS U2, P1.subject AS U1
 FROM triples P1
    INNER JOIN nodes AS P1_subject_V1 ON P1.subject = P1_subject_V1.id 
    INNER JOIN nodes AS P1_object_V2 ON P1.object = P1_object_V2.id 
 WHERE P1.deleted = false
      AND P1.predicate = 520225347082346498
 
 ) UNION (SELECT NULL AS U2, S1.V1 AS U1
 FROM (SELECT P1.subject AS V1
 FROM triples P1
    INNER JOIN nodes AS P1_subject_V1 ON P1.subject = P1_subject_V1.id 
    INNER JOIN nodes AS P1_predicate_V2 ON P1.predicate = P1_predicate_V2.id 
    INNER JOIN nodes AS P1_object_V3 ON P1.object = P1_object_V3.id 
 WHERE P1.deleted = false
       AND NOT (EXISTS (SELECT _P1.object AS V1, P1.subject AS V1
 FROM triples _P1
    INNER JOIN nodes AS _P1_subject_V1 ON _P1.subject = _P1_subject_V1.id 
    INNER JOIN nodes AS _P1_object_V1 ON _P1.object = _P1_object_V1.id 
 WHERE _P1.deleted = false
      AND P1.subject = _P1.subject
      AND _P1.predicate = 520225347082346498
 
 ))
 
 GROUP BY P1.subject 

 ) AS S1
 )) AS U1 LEFT JOIN nodes AS U1_V2 ON U1.U1 = U1_V2.id 
 GROUP BY U1.U2 
{code}

The problem is that the second part of the UNION has a different number of columns. This is caused by the SQLUnion implementation collecting all variables contained in the subquery instead of only considering those of the projection.

> Sparql Query for Dataset View fails
> -----------------------------------
>
>                 Key: MARMOTTA-552
>                 URL: https://issues.apache.org/jira/browse/MARMOTTA-552
>             Project: Marmotta
>          Issue Type: Bug
>            Reporter: Thomas Kurz
>              Labels: SPARQL
>
> The dataset view tries to list classes with a SPARQL query that fails:
> {code}
> SELECT  ?class (count(?s) AS ?count)
>    WHERE
>      {   { ?s <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> ?class }
>        UNION
>          { SELECT  ?s
>            WHERE
>              { ?s ?a ?b
>                FILTER NOT EXISTS {?s <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> ?class }
>              }
>           GROUP BY ?s
>         }
>     }
>   GROUP BY ?class
> {code}



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)