/**
  * Query SPARQL endpoint with a CONSTRUCT query
  *
  * @param qExec QueryExecution encapsulating the query
  * @return model retrieved by querying the endpoint
  */
 private Model getConstructModel(QueryExecution qExec) {
   return qExec.execConstruct(ModelFactory.createDefaultModel());
 }
示例#2
0
 void runTestConstruct(Query query, QueryExecution qe) throws Exception {
   // Do the query!
   Model resultsActual = qe.execConstruct();
   compareGraphResults(resultsActual, query);
 }
  /**
   * Starts a harvester with predefined queries to synchronize with the changes from the SPARQL
   * endpoint
   */
  public boolean sync() {
    logger.info("Sync resources newer than {}", startTime);

    String rdfQueryTemplate =
        "PREFIX xsd:<http://www.w3.org/2001/XMLSchema#> "
            + "SELECT DISTINCT ?resource WHERE { "
            + " GRAPH ?graph { %s }"
            + " ?graph <%s> ?time .  %s "
            + " FILTER (?time > xsd:dateTime(\"%s\")) }";

    String queryStr =
        String.format(
            rdfQueryTemplate, syncConditions, syncTimeProp, graphSyncConditions, startTime);
    Set<String> syncUris = executeSyncQuery(queryStr, "resource");

    if (syncUris == null) {
      logger.error("Errors occurred during sync procedure. Aborting!");
      return false;
    }

    /**
     * If desired, query for old data that has the sync conditions modified
     *
     * <p>This option is useful in the case in which the application indexes resources that match
     * some conditions. In this case, if they are modified and no longer match the initial
     * conditions, they will not be synchronized. When syncOldData is True, the modified resources
     * that no longer match the conditions are deleted.
     */
    int deleted = 0;
    int count = 0;
    if (this.syncOldData) {
      SimpleDateFormat sdf = new SimpleDateFormat("yyyy-MM-dd'T'HH:mm:ss");
      queryStr =
          String.format(
              rdfQueryTemplate,
              syncConditions,
              syncTimeProp,
              graphSyncConditions,
              sdf.format(new Date(0)));

      HashSet<String> allIndexURIs = executeSyncQuery(queryStr, "resource");

      if (allIndexURIs == null) {
        logger.error("Errors occurred during modified content sync query. Aborting!");
        return false;
      }

      deleted = removeMissingUris(allIndexURIs);
    }

    /* Prepare a series of bulk uris to be described so we can make
     * a smaller number of calls to the SPARQL endpoint. */
    ArrayList<ArrayList<String>> bulks = new ArrayList<ArrayList<String>>();
    ArrayList<String> currentBulk = new ArrayList<String>();

    for (String uri : syncUris) {
      currentBulk.add(uri);

      if (currentBulk.size() == EEASettings.DEFAULT_BULK_SIZE) {
        bulks.add(currentBulk);
        currentBulk = new ArrayList<String>();
      }
    }

    if (currentBulk.size() > 0) {
      bulks.add(currentBulk);
    }

    /* Execute RDF queries for the resources in each bulk */
    for (ArrayList<String> bulk : bulks) {
      String syncQuery = getSyncQueryStr(bulk);

      try {
        Query query = QueryFactory.create(syncQuery);
        QueryExecution qExec = QueryExecutionFactory.sparqlService(rdfEndpoint, query);
        try {
          Model constructModel = ModelFactory.createDefaultModel();
          qExec.execConstruct(constructModel);
          BulkRequestBuilder bulkRequest = client.prepareBulk();

          /**
           * When adding the model to ES do not use toDescribeURIs as the query already returned the
           * correct labels.
           */
          addModelToES(constructModel, bulkRequest, false);
          count += bulk.size();
        } catch (Exception e) {
          logger.error("Error while querying for modified content. {}", e.getLocalizedMessage());
          return false;
        } finally {
          qExec.close();
        }
      } catch (QueryParseException qpe) {
        logger.warn(
            "Could not parse Sync query. Please provide a relevant query. {}",
            qpe.getLocalizedMessage());
        return false;
      }
    }
    logger.info(
        "Finished synchronisation: Deleted {}, Updated {}/{}", deleted, count, syncUris.size());
    return true;
  }