Search in sources :

Example 51 with ETLStage

use of co.cask.cdap.etl.proto.v2.ETLStage in project cdap by caskdata.

the class PipelineSpecGeneratorTest method testCycle.

@Test(expected = IllegalStateException.class)
public void testCycle() {
    ETLBatchConfig etlConfig = ETLBatchConfig.builder("* * * * *").addStage(new ETLStage("source", MOCK_SOURCE)).addStage(new ETLStage("sink", MOCK_SINK)).addStage(new ETLStage("t1", MOCK_TRANSFORM_A)).addStage(new ETLStage("t2", MOCK_TRANSFORM_A)).addConnection("source", "t1").addConnection("t1", "t2").addConnection("t2", "t1").addConnection("t2", "sink").build();
    specGenerator.generateSpec(etlConfig);
}
Also used : ETLBatchConfig(co.cask.cdap.etl.proto.v2.ETLBatchConfig) ETLStage(co.cask.cdap.etl.proto.v2.ETLStage) Test(org.junit.Test)

Example 52 with ETLStage

use of co.cask.cdap.etl.proto.v2.ETLStage in project cdap by caskdata.

the class PipelineSpecGeneratorTest method testGenerateSpec.

@Test
public void testGenerateSpec() {
    /*
     *           ---- t1 ------------
     *           |            |      |
     * source ---             |      |--- t3 --- sink1
     *           |            |      |
     *           ------------ t2 --------------- sink2
     *           |                        |
     *           |                        |
     *           -------------------------
     */
    ETLBatchConfig etlConfig = ETLBatchConfig.builder("* * * * *").addStage(new ETLStage("source", MOCK_SOURCE)).addStage(new ETLStage("sink1", MOCK_SINK)).addStage(new ETLStage("sink2", MOCK_SINK)).addStage(new ETLStage("t1", MOCK_TRANSFORM_A)).addStage(new ETLStage("t2", MOCK_TRANSFORM_A)).addStage(new ETLStage("t3", MOCK_TRANSFORM_B)).addConnection("source", "t1").addConnection("source", "t2").addConnection("source", "sink2").addConnection("t1", "t2").addConnection("t1", "t3").addConnection("t1", "sink2").addConnection("t2", "sink2").addConnection("t2", "t3").addConnection("t3", "sink1").build();
    // test the spec generated is correct, with the right input and output schemas and artifact information.
    BatchPipelineSpec actual = specGenerator.generateSpec(etlConfig);
    Map<String, String> emptyMap = ImmutableMap.of();
    PipelineSpec expected = BatchPipelineSpec.builder().addStage(StageSpec.builder("source", new PluginSpec(BatchSource.PLUGIN_TYPE, "mocksource", emptyMap, ARTIFACT_ID)).addOutputSchema(SCHEMA_A, "t1", "t2", "sink2").build()).addStage(StageSpec.builder("sink1", new PluginSpec(BatchSink.PLUGIN_TYPE, "mocksink", emptyMap, ARTIFACT_ID)).addInputSchema("t3", SCHEMA_B).setErrorSchema(SCHEMA_B).build()).addStage(StageSpec.builder("sink2", new PluginSpec(BatchSink.PLUGIN_TYPE, "mocksink", emptyMap, ARTIFACT_ID)).addInputSchemas(ImmutableMap.of("t1", SCHEMA_A, "t2", SCHEMA_A, "source", SCHEMA_A)).setErrorSchema(SCHEMA_A).build()).addStage(StageSpec.builder("t1", new PluginSpec(Transform.PLUGIN_TYPE, "mockA", emptyMap, ARTIFACT_ID)).addInputSchema("source", SCHEMA_A).addOutputSchema(SCHEMA_A, "t2", "t3", "sink2").setErrorSchema(SCHEMA_B).build()).addStage(StageSpec.builder("t2", new PluginSpec(Transform.PLUGIN_TYPE, "mockA", emptyMap, ARTIFACT_ID)).addInputSchemas(ImmutableMap.of("source", SCHEMA_A, "t1", SCHEMA_A)).addOutputSchema(SCHEMA_A, "t3", "sink2").setErrorSchema(SCHEMA_B).build()).addStage(StageSpec.builder("t3", new PluginSpec(Transform.PLUGIN_TYPE, "mockB", emptyMap, ARTIFACT_ID)).addInputSchemas(ImmutableMap.of("t1", SCHEMA_A, "t2", SCHEMA_A)).addOutputSchema(SCHEMA_B, "sink1").setErrorSchema(SCHEMA_A).build()).addConnections(etlConfig.getConnections()).setResources(etlConfig.getResources()).setDriverResources(new Resources(1024, 1)).setClientResources(new Resources(1024, 1)).setStageLoggingEnabled(etlConfig.isStageLoggingEnabled()).build();
    Assert.assertEquals(expected, actual);
}
Also used : ETLBatchConfig(co.cask.cdap.etl.proto.v2.ETLBatchConfig) BatchPipelineSpec(co.cask.cdap.etl.batch.BatchPipelineSpec) ETLStage(co.cask.cdap.etl.proto.v2.ETLStage) BatchPipelineSpec(co.cask.cdap.etl.batch.BatchPipelineSpec) Resources(co.cask.cdap.api.Resources) Test(org.junit.Test)

Example 53 with ETLStage

use of co.cask.cdap.etl.proto.v2.ETLStage in project cdap by caskdata.

the class PipelineSpecGeneratorTest method testUnreachableStage.

@Test(expected = IllegalArgumentException.class)
public void testUnreachableStage() {
    ETLBatchConfig etlConfig = ETLBatchConfig.builder("* * * * *").addStage(new ETLStage("source", MOCK_SOURCE)).addStage(new ETLStage("sink", MOCK_SINK)).addStage(new ETLStage("transform", MOCK_TRANSFORM_A)).addConnection("source", "sink").addConnection("transform", "sink").build();
    specGenerator.generateSpec(etlConfig);
}
Also used : ETLBatchConfig(co.cask.cdap.etl.proto.v2.ETLBatchConfig) ETLStage(co.cask.cdap.etl.proto.v2.ETLStage) Test(org.junit.Test)

Example 54 with ETLStage

use of co.cask.cdap.etl.proto.v2.ETLStage in project cdap by caskdata.

the class PipelineSpecGeneratorTest method testConnectionOutOfSink.

@Test(expected = IllegalArgumentException.class)
public void testConnectionOutOfSink() {
    ETLBatchConfig etlConfig = ETLBatchConfig.builder("* * * * *").addStage(new ETLStage("source", MOCK_SOURCE)).addStage(new ETLStage("sink", MOCK_SINK)).addStage(new ETLStage("transform", MOCK_TRANSFORM_A)).addConnection("source", "sink").addConnection("sink", "transform").build();
    specGenerator.generateSpec(etlConfig);
}
Also used : ETLBatchConfig(co.cask.cdap.etl.proto.v2.ETLBatchConfig) ETLStage(co.cask.cdap.etl.proto.v2.ETLStage) Test(org.junit.Test)

Example 55 with ETLStage

use of co.cask.cdap.etl.proto.v2.ETLStage in project cdap by caskdata.

the class PipelineSpecGeneratorTest method testConflictingInputSchemasCondition.

@Test(expected = IllegalArgumentException.class)
public void testConflictingInputSchemasCondition() {
    /*
     *           ---- transformA ----
     *           |                  |
     * source ---                   |--- condition -- sink
     *           |                  |
     *           ---- transformB ----
     *
     * sink gets schema A and schema B as input, should fail
     */
    ETLBatchConfig etlConfig = ETLBatchConfig.builder("* * * * *").addStage(new ETLStage("source", MOCK_SOURCE)).addStage(new ETLStage("sink", MOCK_SINK)).addStage(new ETLStage("tA", MOCK_TRANSFORM_A)).addStage(new ETLStage("tB", MOCK_TRANSFORM_B)).addStage(new ETLStage("cond", MOCK_CONDITION)).addConnection("source", "tA").addConnection("source", "tB").addConnection("tA", "cond").addConnection("tB", "cond").addConnection("cond", "sink", true).build();
    specGenerator.generateSpec(etlConfig);
}
Also used : ETLBatchConfig(co.cask.cdap.etl.proto.v2.ETLBatchConfig) ETLStage(co.cask.cdap.etl.proto.v2.ETLStage) Test(org.junit.Test)

Aggregations

ETLStage (co.cask.cdap.etl.proto.v2.ETLStage)94 ETLBatchConfig (co.cask.cdap.etl.proto.v2.ETLBatchConfig)75 Test (org.junit.Test)64 ApplicationId (co.cask.cdap.proto.id.ApplicationId)62 ApplicationManager (co.cask.cdap.test.ApplicationManager)58 AppRequest (co.cask.cdap.proto.artifact.AppRequest)57 Schema (co.cask.cdap.api.data.schema.Schema)51 StructuredRecord (co.cask.cdap.api.data.format.StructuredRecord)50 Table (co.cask.cdap.api.dataset.table.Table)49 WorkflowManager (co.cask.cdap.test.WorkflowManager)44 KeyValueTable (co.cask.cdap.api.dataset.lib.KeyValueTable)39 HashSet (java.util.HashSet)16 ArrayList (java.util.ArrayList)15 ETLPlugin (co.cask.cdap.etl.proto.v2.ETLPlugin)14 HashMap (java.util.HashMap)14 DataStreamsConfig (co.cask.cdap.etl.proto.v2.DataStreamsConfig)11 TimeoutException (java.util.concurrent.TimeoutException)11 TopicNotFoundException (co.cask.cdap.api.messaging.TopicNotFoundException)7 SparkManager (co.cask.cdap.test.SparkManager)7 BatchPipelineSpec (co.cask.cdap.etl.batch.BatchPipelineSpec)6