Examples with SparkPlan org.apache.spark.sql.execution.SparkPlan used on opensource projects

Example 1 with SparkPlan

use of org.apache.spark.sql.execution.SparkPlan in project OpenLineage by OpenLineage.

the class SparkSQLExecutionContext method buildJob.

protected OpenLineage.JobBuilder buildJob(QueryExecution queryExecution) {
    SparkContext sparkContext = queryExecution.executedPlan().sparkContext();
    SparkPlan node = queryExecution.executedPlan();
    // Unwrap SparkPlan from WholeStageCodegen, as that's not a descriptive or helpful job name
    if (node instanceof WholeStageCodegenExec) {
        node = ((WholeStageCodegenExec) node).child();
    }
    return openLineage.newJobBuilder().namespace(this.eventEmitter.getJobNamespace()).name(sparkContext.appName().replaceAll(CAMEL_TO_SNAKE_CASE, "_$1").toLowerCase(Locale.ROOT) + "." + node.nodeName().replaceAll(CAMEL_TO_SNAKE_CASE, "_$1").toLowerCase(Locale.ROOT));
}

Example 2 with SparkPlan

use of org.apache.spark.sql.execution.SparkPlan in project OpenLineage by OpenLineage.

the class OpenLineageSparkListenerTest method testSqlEventWithJobEventEmitsOnce.

@Test
public void testSqlEventWithJobEventEmitsOnce() {
    SparkSession sparkSession = mock(SparkSession.class);
    SparkContext sparkContext = mock(SparkContext.class);
    EventEmitter emitter = mock(EventEmitter.class);
    QueryExecution qe = mock(QueryExecution.class);
    LogicalPlan query = UnresolvedRelation$.MODULE$.apply(TableIdentifier.apply("tableName"));
    SparkPlan plan = mock(SparkPlan.class);
    when(sparkSession.sparkContext()).thenReturn(sparkContext);
    when(sparkContext.appName()).thenReturn("appName");
    when(qe.optimizedPlan()).thenReturn(new InsertIntoHadoopFsRelationCommand(new Path("file:///tmp/dir"), null, false, Seq$.MODULE$.empty(), Option.empty(), null, Map$.MODULE$.empty(), query, SaveMode.Overwrite, Option.empty(), Option.empty(), Seq$.MODULE$.<String>empty()));
    when(qe.executedPlan()).thenReturn(plan);
    when(plan.sparkContext()).thenReturn(sparkContext);
    when(plan.nodeName()).thenReturn("execute");
    OpenLineageContext olContext = OpenLineageContext.builder().sparkSession(Optional.of(sparkSession)).sparkContext(sparkSession.sparkContext()).openLineage(new OpenLineage(OpenLineageClient.OPEN_LINEAGE_CLIENT_URI)).queryExecution(qe).build();
    olContext.getOutputDatasetQueryPlanVisitors().add(new InsertIntoHadoopFsRelationVisitor(olContext));
    ExecutionContext executionContext = new StaticExecutionContextFactory(emitter).createSparkSQLExecutionContext(1L, emitter, qe, olContext);
    executionContext.start(new SparkListenerSQLExecutionStart(1L, "", "", "", new SparkPlanInfo("name", "string", Seq$.MODULE$.empty(), Map$.MODULE$.empty(), Seq$.MODULE$.empty()), 1L));
    executionContext.start(new SparkListenerJobStart(0, 2L, Seq$.MODULE$.<StageInfo>empty(), new Properties()));
    ArgumentCaptor<OpenLineage.RunEvent> lineageEvent = ArgumentCaptor.forClass(OpenLineage.RunEvent.class);
    verify(emitter, times(2)).emit(lineageEvent.capture());
}