use of org.apache.samza.config.StreamConfig in project samza by apache.
the class TestOperatorImplGraph method testGetOutputToInputStreams.
@Test
public void testGetOutputToInputStreams() {
String inputStreamId1 = "input1";
String inputStreamId2 = "input2";
String inputStreamId3 = "input3";
String inputSystem = "input-system";
String outputStreamId1 = "output1";
String outputStreamId2 = "output2";
String outputSystem = "output-system";
String intStreamId1 = "test-app-1-partition_by-p1";
String intStreamId2 = "test-app-1-partition_by-p2";
String intSystem = "test-system";
HashMap<String, String> configs = new HashMap<>();
configs.put(JobConfig.JOB_NAME, "test-app");
configs.put(JobConfig.JOB_DEFAULT_SYSTEM, intSystem);
StreamTestUtils.addStreamConfigs(configs, inputStreamId1, inputSystem, inputStreamId1);
StreamTestUtils.addStreamConfigs(configs, inputStreamId2, inputSystem, inputStreamId2);
StreamTestUtils.addStreamConfigs(configs, inputStreamId3, inputSystem, inputStreamId3);
StreamTestUtils.addStreamConfigs(configs, outputStreamId1, outputSystem, outputStreamId1);
StreamTestUtils.addStreamConfigs(configs, outputStreamId2, outputSystem, outputStreamId2);
Config config = new MapConfig(configs);
when(this.context.getJobContext().getConfig()).thenReturn(config);
StreamApplicationDescriptorImpl graphSpec = new StreamApplicationDescriptorImpl(appDesc -> {
GenericSystemDescriptor isd = new GenericSystemDescriptor(inputSystem, "mockFactoryClass");
GenericInputDescriptor inputDescriptor1 = isd.getInputDescriptor(inputStreamId1, mock(Serde.class));
GenericInputDescriptor inputDescriptor2 = isd.getInputDescriptor(inputStreamId2, mock(Serde.class));
GenericInputDescriptor inputDescriptor3 = isd.getInputDescriptor(inputStreamId3, mock(Serde.class));
GenericSystemDescriptor osd = new GenericSystemDescriptor(outputSystem, "mockFactoryClass");
GenericOutputDescriptor outputDescriptor1 = osd.getOutputDescriptor(outputStreamId1, mock(Serde.class));
GenericOutputDescriptor outputDescriptor2 = osd.getOutputDescriptor(outputStreamId2, mock(Serde.class));
MessageStream messageStream1 = appDesc.getInputStream(inputDescriptor1).map(m -> m);
MessageStream messageStream2 = appDesc.getInputStream(inputDescriptor2).filter(m -> true);
MessageStream messageStream3 = appDesc.getInputStream(inputDescriptor3).filter(m -> true).partitionBy(m -> "m", m -> m, mock(KVSerde.class), "p1").map(m -> m);
OutputStream<Object> outputStream1 = appDesc.getOutputStream(outputDescriptor1);
OutputStream<Object> outputStream2 = appDesc.getOutputStream(outputDescriptor2);
messageStream1.join(messageStream2, mock(JoinFunction.class), mock(Serde.class), mock(Serde.class), mock(Serde.class), Duration.ofHours(2), "j1").partitionBy(m -> "m", m -> m, mock(KVSerde.class), "p2").sendTo(outputStream1);
messageStream3.join(messageStream2, mock(JoinFunction.class), mock(Serde.class), mock(Serde.class), mock(Serde.class), Duration.ofHours(1), "j2").sendTo(outputStream2);
}, config);
Multimap<SystemStream, SystemStream> outputToInput = OperatorImplGraph.getIntermediateToInputStreamsMap(graphSpec.getOperatorSpecGraph(), new StreamConfig(config));
Collection<SystemStream> inputs = outputToInput.get(new SystemStream(intSystem, intStreamId2));
assertEquals(inputs.size(), 2);
assertTrue(inputs.contains(new SystemStream(inputSystem, inputStreamId1)));
assertTrue(inputs.contains(new SystemStream(inputSystem, inputStreamId2)));
inputs = outputToInput.get(new SystemStream(intSystem, intStreamId1));
assertEquals(inputs.size(), 1);
assertEquals(inputs.iterator().next(), new SystemStream(inputSystem, inputStreamId3));
}
use of org.apache.samza.config.StreamConfig in project samza by apache.
the class TestQueryTranslator method testTranslate.
@Test
public void testTranslate() {
Map<String, String> config = SamzaSqlTestConfig.fetchStaticConfigsWithFactories(10);
config.put(SamzaSqlApplicationConfig.CFG_SQL_STMT, "Insert into testavro.outputTopic(id) select MyTest(id) from testavro.level1.level2.SIMPLE1 as s where s.id = 10");
Config samzaConfig = SamzaSqlApplicationRunner.computeSamzaConfigs(true, new MapConfig(config));
List<String> sqlStmts = fetchSqlFromConfig(config);
List<SamzaSqlQueryParser.QueryInfo> queryInfo = fetchQueryInfo(sqlStmts);
SamzaSqlApplicationConfig samzaSqlApplicationConfig = new SamzaSqlApplicationConfig(new MapConfig(config), queryInfo.stream().map(SamzaSqlQueryParser.QueryInfo::getSources).flatMap(Collection::stream).collect(Collectors.toList()), queryInfo.stream().map(SamzaSqlQueryParser.QueryInfo::getSink).collect(Collectors.toList()));
StreamApplicationDescriptorImpl appDesc = new StreamApplicationDescriptorImpl(streamApp -> {
}, samzaConfig);
QueryTranslator translator = new QueryTranslator(appDesc, samzaSqlApplicationConfig);
translator.translate(queryInfo.get(0), appDesc, 0);
OperatorSpecGraph specGraph = appDesc.getOperatorSpecGraph();
StreamConfig streamConfig = new StreamConfig(samzaConfig);
String inputStreamId = specGraph.getInputOperators().keySet().stream().findFirst().get();
String inputSystem = streamConfig.getSystem(inputStreamId);
String inputPhysicalName = streamConfig.getPhysicalName(inputStreamId);
String outputStreamId = specGraph.getOutputStreams().keySet().stream().findFirst().get();
String outputSystem = streamConfig.getSystem(outputStreamId);
String outputPhysicalName = streamConfig.getPhysicalName(outputStreamId);
Assert.assertEquals(1, specGraph.getOutputStreams().size());
Assert.assertEquals("testavro", outputSystem);
Assert.assertEquals("outputTopic", outputPhysicalName);
Assert.assertEquals(1, specGraph.getInputOperators().size());
Assert.assertEquals("testavro", inputSystem);
Assert.assertEquals("SIMPLE1", inputPhysicalName);
}
use of org.apache.samza.config.StreamConfig in project samza by apache.
the class TestQueryTranslator method testTranslateFanOut.
@Test
public void testTranslateFanOut() {
Map<String, String> config = SamzaSqlTestConfig.fetchStaticConfigsWithFactories(10);
String sql1 = "Insert into testavro.SIMPLE2 select * from testavro.SIMPLE1";
String sql2 = "Insert into testavro.SIMPLE3 select * from testavro.SIMPLE1";
List<String> sqlStmts = Arrays.asList(sql1, sql2);
config.put(SamzaSqlApplicationConfig.CFG_SQL_STMTS_JSON, JsonUtil.toJson(sqlStmts));
Config samzaConfig = SamzaSqlApplicationRunner.computeSamzaConfigs(true, new MapConfig(config));
List<SamzaSqlQueryParser.QueryInfo> queryInfo = fetchQueryInfo(sqlStmts);
SamzaSqlApplicationConfig samzaSqlApplicationConfig = new SamzaSqlApplicationConfig(new MapConfig(config), queryInfo.stream().map(SamzaSqlQueryParser.QueryInfo::getSources).flatMap(Collection::stream).collect(Collectors.toList()), queryInfo.stream().map(SamzaSqlQueryParser.QueryInfo::getSink).collect(Collectors.toList()));
StreamApplicationDescriptorImpl appDesc = new StreamApplicationDescriptorImpl(streamApp -> {
}, samzaConfig);
QueryTranslator translator = new QueryTranslator(appDesc, samzaSqlApplicationConfig);
translator.translate(queryInfo.get(0), appDesc, 0);
translator.translate(queryInfo.get(1), appDesc, 1);
OperatorSpecGraph specGraph = appDesc.getOperatorSpecGraph();
StreamConfig streamConfig = new StreamConfig(samzaConfig);
String inputStreamId = specGraph.getInputOperators().keySet().stream().findFirst().get();
String inputSystem = streamConfig.getSystem(inputStreamId);
String inputPhysicalName = streamConfig.getPhysicalName(inputStreamId);
String outputStreamId1 = specGraph.getOutputStreams().keySet().stream().findFirst().get();
String outputSystem1 = streamConfig.getSystem(outputStreamId1);
String outputPhysicalName1 = streamConfig.getPhysicalName(outputStreamId1);
String outputStreamId2 = specGraph.getOutputStreams().keySet().stream().skip(1).findFirst().get();
String outputSystem2 = streamConfig.getSystem(outputStreamId2);
String outputPhysicalName2 = streamConfig.getPhysicalName(outputStreamId2);
Assert.assertEquals(2, specGraph.getOutputStreams().size());
Assert.assertEquals("testavro", outputSystem1);
Assert.assertEquals("SIMPLE2", outputPhysicalName1);
Assert.assertEquals("testavro", outputSystem2);
Assert.assertEquals("SIMPLE3", outputPhysicalName2);
Assert.assertEquals(1, specGraph.getInputOperators().size());
Assert.assertEquals("testavro", inputSystem);
Assert.assertEquals("SIMPLE1", inputPhysicalName);
}
use of org.apache.samza.config.StreamConfig in project samza by apache.
the class TestQueryTranslator method testTranslateStreamTableLeftJoin.
@Test
public void testTranslateStreamTableLeftJoin() {
Map<String, String> config = SamzaSqlTestConfig.fetchStaticConfigsWithFactories(configs, 10);
String sql = "Insert into testavro.enrichedPageViewTopic(profileName, pageKey)" + " select p.name as profileName, pv.pageKey" + " from testavro.PAGEVIEW as pv" + " left join testavro.PROFILE.`$table` as p" + " on p.id = pv.profileId";
config.put(SamzaSqlApplicationConfig.CFG_SQL_STMT, sql);
Config samzaConfig = SamzaSqlApplicationRunner.computeSamzaConfigs(true, new MapConfig(config));
List<String> sqlStmts = fetchSqlFromConfig(config);
List<SamzaSqlQueryParser.QueryInfo> queryInfo = fetchQueryInfo(sqlStmts);
SamzaSqlApplicationConfig samzaSqlApplicationConfig = new SamzaSqlApplicationConfig(new MapConfig(config), queryInfo.stream().map(SamzaSqlQueryParser.QueryInfo::getSources).flatMap(Collection::stream).collect(Collectors.toList()), queryInfo.stream().map(SamzaSqlQueryParser.QueryInfo::getSink).collect(Collectors.toList()));
StreamApplicationDescriptorImpl streamAppDesc = new StreamApplicationDescriptorImpl(streamApp -> {
}, samzaConfig);
QueryTranslator translator = new QueryTranslator(streamAppDesc, samzaSqlApplicationConfig);
translator.translate(queryInfo.get(0), streamAppDesc, 0);
OperatorSpecGraph specGraph = streamAppDesc.getOperatorSpecGraph();
StreamConfig streamConfig = new StreamConfig(samzaConfig);
String input1StreamId = specGraph.getInputOperators().keySet().stream().findFirst().get();
String input1System = streamConfig.getSystem(input1StreamId);
String input1PhysicalName = streamConfig.getPhysicalName(input1StreamId);
String input2StreamId = specGraph.getInputOperators().keySet().stream().skip(1).findFirst().get();
String input2System = streamConfig.getSystem(input2StreamId);
String input2PhysicalName = streamConfig.getPhysicalName(input2StreamId);
String input3StreamId = specGraph.getInputOperators().keySet().stream().skip(2).findFirst().get();
String input3System = streamConfig.getSystem(input3StreamId);
String input3PhysicalName = streamConfig.getPhysicalName(input3StreamId);
String input4StreamId = specGraph.getInputOperators().keySet().stream().skip(3).findFirst().get();
String input4System = streamConfig.getSystem(input4StreamId);
String input4PhysicalName = streamConfig.getPhysicalName(input4StreamId);
String output1StreamId = specGraph.getOutputStreams().keySet().stream().findFirst().get();
String output1System = streamConfig.getSystem(output1StreamId);
String output1PhysicalName = streamConfig.getPhysicalName(output1StreamId);
String output2StreamId = specGraph.getOutputStreams().keySet().stream().skip(1).findFirst().get();
String output2System = streamConfig.getSystem(output2StreamId);
String output2PhysicalName = streamConfig.getPhysicalName(output2StreamId);
String output3StreamId = specGraph.getOutputStreams().keySet().stream().skip(2).findFirst().get();
String output3System = streamConfig.getSystem(output3StreamId);
String output3PhysicalName = streamConfig.getPhysicalName(output3StreamId);
Assert.assertEquals(3, specGraph.getOutputStreams().size());
Assert.assertEquals("kafka", output1System);
Assert.assertEquals("sql-job-1-partition_by-table_sql_0_join_2", output1PhysicalName);
Assert.assertEquals("kafka", output2System);
Assert.assertEquals("sql-job-1-partition_by-stream_sql_0_join_2", output2PhysicalName);
Assert.assertEquals("testavro", output3System);
Assert.assertEquals("enrichedPageViewTopic", output3PhysicalName);
Assert.assertEquals(4, specGraph.getInputOperators().size());
Assert.assertEquals("testavro", input1System);
Assert.assertEquals("PAGEVIEW", input1PhysicalName);
Assert.assertEquals("testavro", input2System);
Assert.assertEquals("PROFILE", input2PhysicalName);
Assert.assertEquals("kafka", input3System);
Assert.assertEquals("sql-job-1-partition_by-table_sql_0_join_2", input3PhysicalName);
Assert.assertEquals("kafka", input4System);
Assert.assertEquals("sql-job-1-partition_by-stream_sql_0_join_2", input4PhysicalName);
}
use of org.apache.samza.config.StreamConfig in project samza by apache.
the class TestQueryTranslator method testTranslateStreamTableRightJoin.
@Test
public void testTranslateStreamTableRightJoin() {
Map<String, String> config = SamzaSqlTestConfig.fetchStaticConfigsWithFactories(configs, 10);
String sql = "Insert into testavro.enrichedPageViewTopic(profileName, pageKey)" + " select p.name as profileName, pv.pageKey" + " from testavro.PROFILE.`$table` as p" + " right join testavro.PAGEVIEW as pv" + " on p.id = pv.profileId";
config.put(SamzaSqlApplicationConfig.CFG_SQL_STMT, sql);
Config samzaConfig = SamzaSqlApplicationRunner.computeSamzaConfigs(true, new MapConfig(config));
List<String> sqlStmts = fetchSqlFromConfig(config);
List<SamzaSqlQueryParser.QueryInfo> queryInfo = fetchQueryInfo(sqlStmts);
SamzaSqlApplicationConfig samzaSqlApplicationConfig = new SamzaSqlApplicationConfig(new MapConfig(config), queryInfo.stream().map(SamzaSqlQueryParser.QueryInfo::getSources).flatMap(Collection::stream).collect(Collectors.toList()), queryInfo.stream().map(SamzaSqlQueryParser.QueryInfo::getSink).collect(Collectors.toList()));
StreamApplicationDescriptorImpl streamAppDesc = new StreamApplicationDescriptorImpl(streamApp -> {
}, samzaConfig);
QueryTranslator translator = new QueryTranslator(streamAppDesc, samzaSqlApplicationConfig);
translator.translate(queryInfo.get(0), streamAppDesc, 0);
OperatorSpecGraph specGraph = streamAppDesc.getOperatorSpecGraph();
StreamConfig streamConfig = new StreamConfig(samzaConfig);
String input1StreamId = specGraph.getInputOperators().keySet().stream().findFirst().get();
String input1System = streamConfig.getSystem(input1StreamId);
String input1PhysicalName = streamConfig.getPhysicalName(input1StreamId);
String input2StreamId = specGraph.getInputOperators().keySet().stream().skip(1).findFirst().get();
String input2System = streamConfig.getSystem(input2StreamId);
String input2PhysicalName = streamConfig.getPhysicalName(input2StreamId);
String input3StreamId = specGraph.getInputOperators().keySet().stream().skip(2).findFirst().get();
String input3System = streamConfig.getSystem(input3StreamId);
String input3PhysicalName = streamConfig.getPhysicalName(input3StreamId);
String input4StreamId = specGraph.getInputOperators().keySet().stream().skip(3).findFirst().get();
String input4System = streamConfig.getSystem(input4StreamId);
String input4PhysicalName = streamConfig.getPhysicalName(input4StreamId);
String output1StreamId = specGraph.getOutputStreams().keySet().stream().findFirst().get();
String output1System = streamConfig.getSystem(output1StreamId);
String output1PhysicalName = streamConfig.getPhysicalName(output1StreamId);
String output2StreamId = specGraph.getOutputStreams().keySet().stream().skip(1).findFirst().get();
String output2System = streamConfig.getSystem(output2StreamId);
String output2PhysicalName = streamConfig.getPhysicalName(output2StreamId);
String output3StreamId = specGraph.getOutputStreams().keySet().stream().skip(2).findFirst().get();
String output3System = streamConfig.getSystem(output3StreamId);
String output3PhysicalName = streamConfig.getPhysicalName(output3StreamId);
Assert.assertEquals(3, specGraph.getOutputStreams().size());
Assert.assertEquals("kafka", output1System);
Assert.assertEquals("sql-job-1-partition_by-table_sql_0_join_2", output1PhysicalName);
Assert.assertEquals("kafka", output2System);
Assert.assertEquals("sql-job-1-partition_by-stream_sql_0_join_2", output2PhysicalName);
Assert.assertEquals("testavro", output3System);
Assert.assertEquals("enrichedPageViewTopic", output3PhysicalName);
Assert.assertEquals(4, specGraph.getInputOperators().size());
Assert.assertEquals("testavro", input1System);
Assert.assertEquals("PROFILE", input1PhysicalName);
Assert.assertEquals("testavro", input2System);
Assert.assertEquals("PAGEVIEW", input2PhysicalName);
Assert.assertEquals("kafka", input3System);
Assert.assertEquals("sql-job-1-partition_by-table_sql_0_join_2", input3PhysicalName);
Assert.assertEquals("kafka", input4System);
Assert.assertEquals("sql-job-1-partition_by-stream_sql_0_join_2", input4PhysicalName);
}
Aggregations