TsparqlSparqlConnection

TrackerSparqlConnection holds a connection to a RDF triple store.

This triple store may be of three types:

• Local to the process, created through tracker_sparql_connection_new().
• A HTTP SPARQL endpoint over the network, created through tracker_sparql_connection_remote_new()
• A DBus SPARQL endpoint owned by another process in the same machine, created through tracker_sparql_connection_bus_new()

When creating a local triple store, it is required to give details about its structure. This is done by passing a location to an ontology, see more on how are ontologies defined. A local database may be stored in a filesystem location, or it may reside in memory.

A TrackerSparqlConnection is private to the calling process, it can be exposed to other hosts/processes via a TrackerEndpoint, see tracker_endpoint_dbus_new() and tracker_endpoint_http_new().

When issuing SPARQL queries and updates, it is recommended that these are created through TrackerSparqlStatement to avoid the SPARQL injection class of bugs, see tracker_sparql_connection_query_statement() and tracker_sparql_connection_update_statement(). For SPARQL updates it is also possible to use a “builder” approach to generate RDF data, see TrackerResource. It is also possible to create TrackerSparqlStatement objects for SPARQL queries and updates from SPARQL strings embedded in a GResource, see tracker_sparql_connection_load_statement_from_gresource().

To get the best performance, it is recommended that SPARQL updates are clustered through TrackerBatch.

TrackerSparqlConnection also offers a number of methods for the simple cases, tracker_sparql_connection_query() may be used when there is a SPARQL query string directly available, and the tracker_sparql_connection_update() family of functions may be used for one-off updates. All functions have asynchronous variants.

When a SPARQL query is executed, a TrackerSparqlCursor will be obtained to iterate over the query results.

Depending on the ontology definition, TrackerSparqlConnection may emit notifications whenever resources of certain types get insert, modified or deleted from the triple store (see nrl:notify. These notifications can be handled via a TrackerNotifier obtained with tracker_sparql_connection_create_notifier().

After done with a connection, it is recommended to call tracker_sparql_connection_close() or tracker_sparql_connection_close_async() explicitly to cleanly close the connection and prevent consistency checks on future runs. The triple store connection will be implicitly closed when the TrackerSparqlConnection object is disposed.

A TrackerSparqlConnection may be used from multiple threads, asynchronous updates are executed sequentially on arrival order, asynchronous queries are dispatched in a thread pool.

If you ever have the need to procedurally compose SPARQL query strings, consider the use of tracker_sparql_escape_string() for literal strings and the tracker_sparql_escape_uri() family of functions for URIs.