Projects/Nepomuk/QueryingMethods: Difference between revisions

    From KDE TechBase
    (Code examples)
    Line 3: Line 3:
    = QueryServiceClient =
    = QueryServiceClient =


    The Nepomuk <code>QueryServiceClient</code> is an asynchronous method of running queries in Nepomuk. It accepts both Nepomuk Queries and direct SPARQL queries. It provides convenient signal which your application can connect to in order to receive the results.
    The Nepomuk [http://api.kde.org/4.x-api/kdelibs-apidocs/nepomuk-core/html/classNepomuk2_1_1Query_1_1QueryServiceClient.html <code>QueryServiceClient</code>] is an asynchronous method of running queries in Nepomuk. It accepts both Nepomuk Queries and direct SPARQL queries. It provides convenient signal which your application can connect to in order to receive the results.


    Please note that the asynchronous nature of the QueryServiceClient is achieved by transmitting the query over dbus to the storage service, where a new thread is spawned, the query is run on that thread, and the results are returned over dbus. This is fine for casual use, but when you care about high performance, it is recommended that you avoid the QueryServiceClient.
    Please note that the asynchronous nature of the [http://api.kde.org/4.x-api/kdelibs-apidocs/nepomuk-core/html/classNepomuk2_1_1Query_1_1QueryServiceClient.html QueryServiceClient] is achieved by transmitting the query over dbus to the storage service, where a new thread is spawned, the query is run on that thread, and the results are returned over dbus. This is fine for casual use, but when you care about high performance, it is recommended that you avoid the QueryServiceClient.


    Example -
    Example -
    Line 28: Line 28:
             }
             }
         }
         }
    </syntaxhighlight>
    The [http://api.kde.org/4.x-api/kdelibs-apidocs/nepomuk-core/html/classNepomuk2_1_1Query_1_1QueryServiceClient.html QueryServiceClient] also offers ways of updating the results via [http://api.kde.org/4.x-api/kdelibs-apidocs/nepomuk-core/html/classNepomuk2_1_1Query_1_1QueryServiceClient.html#a1b843e5bf89ddb1a0d5c66eb0193375b entriesRemoved] and [http://api.kde.org/4.x-api/kdelibs-apidocs/nepomuk-core/html/classNepomuk2_1_1Query_1_1QueryServiceClient.html#a6ed2be5579a00e9b8de1f5f23faf1fb6  newEntries]. While these methods may seem very attractive, their implementation is not. They work by running the entire query when some data in the Nepomuk database changes.
    In the future the queries will only be re-run when data related to the query changes (heuristics), but till then the entire query is re-run every time ANY data changes. Therefore it is recommended that you delete the QueryServiceClient when it has finished execution as follows -
    <syntaxhighlight lang="cpp-qt">
        Nepomuk2::Query::QueryServiceClient* client = new Nepomuk2::Query::QueryServiceClient( this );
        connect( client, SIGNAL(finishedListing()), client, SLOT(deleteLater()) );
    </syntaxhighlight>
    </syntaxhighlight>



    Revision as of 11:52, 23 August 2012

    There isn't much point of pushing all your data into Nepomuk, if you cannot query it. This article mentions the different ways you can query Nepomuk and the advantages and disadvantages of each.

    QueryServiceClient

    The Nepomuk QueryServiceClient is an asynchronous method of running queries in Nepomuk. It accepts both Nepomuk Queries and direct SPARQL queries. It provides convenient signal which your application can connect to in order to receive the results.

    Please note that the asynchronous nature of the QueryServiceClient is achieved by transmitting the query over dbus to the storage service, where a new thread is spawned, the query is run on that thread, and the results are returned over dbus. This is fine for casual use, but when you care about high performance, it is recommended that you avoid the QueryServiceClient.

    Example -

        Nepomuk2::Query::Query query( .. build Query .. );
        Nepomuk2::Query::QueryServiceClient* client = new Nepomuk2::Query::QueryServiceClient( this );
        client->query( query );
        connect( client, SIGNAL(newEntries(QList<Nepomuk2::Query::Result>)),
                 this, SLOT(newEntries(QList<Nepomuk2::Query::Result>)) );
        
        void TestObject::newEntries(const QList< Nepomuk2::Query::Result >& list)
        {
            foreach( const Nepomuk2::Query::Result &r, list ) {
                Nepomuk2::Resource res = r.resource();
                QHash<QUrl, Nepomuk2::Variant> prop = res.properties();
                QHashIterator<QUrl, Nepomuk2::Variant> it( prop );
                while( it.hasNext() ) {
                    it.next();
                    kDebug() << it.key() << " " << it.value();
                }
                kDebug() << "--------------------------";
            }
        }
    

    The QueryServiceClient also offers ways of updating the results via entriesRemoved and newEntries. While these methods may seem very attractive, their implementation is not. They work by running the entire query when some data in the Nepomuk database changes.

    In the future the queries will only be re-run when data related to the query changes (heuristics), but till then the entire query is re-run every time ANY data changes. Therefore it is recommended that you delete the QueryServiceClient when it has finished execution as follows -

        Nepomuk2::Query::QueryServiceClient* client = new Nepomuk2::Query::QueryServiceClient( this );
        connect( client, SIGNAL(finishedListing()), client, SLOT(deleteLater()) );
    

    Result Iterator

    The ResultIterator can take either a Nepomuk or SPARQL query and provides an iterator which you can use to get the results. The iterator is obviously blocking, so you may want to put it another thread, depending on the kind of query.

    Example -

        Nepomuk2::Query::Query query( .. build Query .. );
        Nepomuk::Query::ResultIterator it( query );
        while( it.next() ) {
                Nepomuk2::Resource res = it.current().resource();
                QHash<QUrl, Nepomuk2::Variant> prop = res.properties();
                QHashIterator<QUrl, Nepomuk2::Variant> it( prop );
    
                while( it.hasNext() ) {
                    it.next();
                    kDebug() << it.key() << " " << it.value();
                }
                kDebug() << "--------------------------";
            }
        }
    

    Soprano::Model

    The underlying Soprano model can also be used to directly run the queries. You can only run SPARQL queries on the model. Use this method when you're writing a custom query. The underlying Model also provides an iterator similar to that of the ResultIterator.

    Example -

        QString query = QString::fromLatin1("select ?r where { ?r a nco:Contact . }");
        Soprano::Model* model = Nepomuk2::ResourceManager::instance()->mainModel();
        Soprano::QueryResultIterator it = model->executeQuery( query, Soprano::Query::QueryLanguageSparql );
    
        while( it.next() ) {
            kDebug() << it["r"].uri();
        }