Search in sources :

Example 11 with GenericOutputDescriptor

use of org.apache.samza.system.descriptors.GenericOutputDescriptor in project samza by apache.

the class TestStreamApplicationDescriptorImpl method testGetOutputStreamWithNullSerde.

@Test(expected = IllegalArgumentException.class)
public void testGetOutputStreamWithNullSerde() {
    String streamId = "test-stream-1";
    GenericSystemDescriptor sd = new GenericSystemDescriptor("mockSystem", "mockSystemFactoryClass");
    GenericOutputDescriptor osd = sd.getOutputDescriptor(streamId, null);
    new StreamApplicationDescriptorImpl(appDesc -> {
        appDesc.getOutputStream(osd);
    }, getConfig());
}
Also used : GenericOutputDescriptor(org.apache.samza.system.descriptors.GenericOutputDescriptor) GenericSystemDescriptor(org.apache.samza.system.descriptors.GenericSystemDescriptor) Test(org.junit.Test)

Example 12 with GenericOutputDescriptor

use of org.apache.samza.system.descriptors.GenericOutputDescriptor in project samza by apache.

the class TestJobGraphJsonGenerator method testRepartitionedWindowStreamApplication.

@Test
public void testRepartitionedWindowStreamApplication() throws Exception {
    Map<String, String> configMap = new HashMap<>();
    configMap.put(JobConfig.JOB_NAME, "test-app");
    configMap.put(JobConfig.JOB_DEFAULT_SYSTEM, "test-system");
    StreamTestUtils.addStreamConfigs(configMap, "PageView", "hdfs", "hdfs:/user/dummy/PageViewEvent");
    StreamTestUtils.addStreamConfigs(configMap, "PageViewCount", "kafka", "PageViewCount");
    Config config = new MapConfig(configMap);
    // set up external partition count
    Map<String, Integer> system1Map = new HashMap<>();
    system1Map.put("hdfs:/user/dummy/PageViewEvent", 512);
    Map<String, Integer> system2Map = new HashMap<>();
    system2Map.put("PageViewCount", 16);
    SystemAdmin systemAdmin1 = createSystemAdmin(system1Map);
    SystemAdmin systemAdmin2 = createSystemAdmin(system2Map);
    SystemAdmins systemAdmins = mock(SystemAdmins.class);
    when(systemAdmins.getSystemAdmin("hdfs")).thenReturn(systemAdmin1);
    when(systemAdmins.getSystemAdmin("kafka")).thenReturn(systemAdmin2);
    StreamManager streamManager = new StreamManager(systemAdmins);
    StreamApplicationDescriptorImpl graphSpec = new StreamApplicationDescriptorImpl(appDesc -> {
        KVSerde<String, PageViewEvent> pvSerde = KVSerde.of(new StringSerde(), new JsonSerdeV2<>(PageViewEvent.class));
        GenericSystemDescriptor isd = new GenericSystemDescriptor("hdfs", "mockSystemFactoryClass");
        GenericInputDescriptor<KV<String, PageViewEvent>> pageView = isd.getInputDescriptor("PageView", pvSerde);
        KVSerde<String, Long> pvcSerde = KVSerde.of(new StringSerde(), new LongSerde());
        GenericSystemDescriptor osd = new GenericSystemDescriptor("kafka", "mockSystemFactoryClass");
        GenericOutputDescriptor<KV<String, Long>> pageViewCount = osd.getOutputDescriptor("PageViewCount", pvcSerde);
        MessageStream<KV<String, PageViewEvent>> inputStream = appDesc.getInputStream(pageView);
        OutputStream<KV<String, Long>> outputStream = appDesc.getOutputStream(pageViewCount);
        inputStream.partitionBy(kv -> kv.getValue().getCountry(), kv -> kv.getValue(), pvSerde, "keyed-by-country").window(Windows.keyedTumblingWindow(kv -> kv.getValue().getCountry(), Duration.ofSeconds(10L), () -> 0L, (m, c) -> c + 1L, new StringSerde(), new LongSerde()), "count-by-country").map(pane -> new KV<>(pane.getKey().getKey(), pane.getMessage())).sendTo(outputStream);
    }, config);
    ExecutionPlanner planner = new ExecutionPlanner(config, streamManager);
    ExecutionPlan plan = planner.plan(graphSpec);
    String json = plan.getPlanAsJson();
    System.out.println(json);
    // deserialize
    ObjectMapper mapper = new ObjectMapper();
    JobGraphJsonGenerator.JobGraphJson nodes = mapper.readValue(json, JobGraphJsonGenerator.JobGraphJson.class);
    JobGraphJsonGenerator.OperatorGraphJson operatorGraphJson = nodes.jobs.get(0).operatorGraph;
    assertEquals(2, operatorGraphJson.inputStreams.size());
    assertEquals(4, operatorGraphJson.operators.size());
    assertEquals(1, nodes.sourceStreams.size());
    assertEquals(1, nodes.sinkStreams.size());
    assertEquals(1, nodes.intermediateStreams.size());
    // verify partitionBy op output to the intermdiate stream of the same id
    assertEquals(operatorGraphJson.operators.get("test-app-1-partition_by-keyed-by-country").get("outputStreamId"), "test-app-1-partition_by-keyed-by-country");
    assertEquals(operatorGraphJson.operators.get("test-app-1-send_to-5").get("outputStreamId"), "PageViewCount");
}
Also used : LongSerde(org.apache.samza.serializers.LongSerde) StreamApplicationDescriptorImpl(org.apache.samza.application.descriptors.StreamApplicationDescriptorImpl) TableDescriptor(org.apache.samza.table.descriptors.TableDescriptor) GenericSystemDescriptor(org.apache.samza.system.descriptors.GenericSystemDescriptor) JobConfig(org.apache.samza.config.JobConfig) HashMap(java.util.HashMap) Serde(org.apache.samza.serializers.Serde) TestLocalTableDescriptor(org.apache.samza.table.descriptors.TestLocalTableDescriptor) GenericInputDescriptor(org.apache.samza.system.descriptors.GenericInputDescriptor) SendToTableOperatorSpec(org.apache.samza.operators.spec.SendToTableOperatorSpec) StringSerde(org.apache.samza.serializers.StringSerde) HashSet(java.util.HashSet) OperatorSpec(org.apache.samza.operators.spec.OperatorSpec) StreamTestUtils(org.apache.samza.testUtils.StreamTestUtils) TestExecutionPlanner(org.apache.samza.execution.TestExecutionPlanner) Duration(java.time.Duration) Map(java.util.Map) ApplicationConfig(org.apache.samza.config.ApplicationConfig) MapConfig(org.apache.samza.config.MapConfig) KV(org.apache.samza.operators.KV) NoOpSerde(org.apache.samza.serializers.NoOpSerde) MessageStream(org.apache.samza.operators.MessageStream) LongSerde(org.apache.samza.serializers.LongSerde) Before(org.junit.Before) Windows(org.apache.samza.operators.windows.Windows) StreamTableJoinFunction(org.apache.samza.operators.functions.StreamTableJoinFunction) GenericOutputDescriptor(org.apache.samza.system.descriptors.GenericOutputDescriptor) ObjectMapper(com.fasterxml.jackson.databind.ObjectMapper) Set(java.util.Set) Matchers(org.hamcrest.Matchers) StreamSpec(org.apache.samza.system.StreamSpec) Test(org.junit.Test) JoinFunction(org.apache.samza.operators.functions.JoinFunction) Collectors(java.util.stream.Collectors) Mockito(org.mockito.Mockito) OperatorSpecs(org.apache.samza.operators.spec.OperatorSpecs) StreamTableJoinOperatorSpec(org.apache.samza.operators.spec.StreamTableJoinOperatorSpec) SystemAdmin(org.apache.samza.system.SystemAdmin) Config(org.apache.samza.config.Config) JsonSerdeV2(org.apache.samza.serializers.JsonSerdeV2) KVSerde(org.apache.samza.serializers.KVSerde) Assert(org.junit.Assert) Collections(java.util.Collections) OutputStream(org.apache.samza.operators.OutputStream) SystemAdmins(org.apache.samza.system.SystemAdmins) StringSerde(org.apache.samza.serializers.StringSerde) HashMap(java.util.HashMap) JobConfig(org.apache.samza.config.JobConfig) ApplicationConfig(org.apache.samza.config.ApplicationConfig) MapConfig(org.apache.samza.config.MapConfig) Config(org.apache.samza.config.Config) TestExecutionPlanner(org.apache.samza.execution.TestExecutionPlanner) StreamApplicationDescriptorImpl(org.apache.samza.application.descriptors.StreamApplicationDescriptorImpl) MapConfig(org.apache.samza.config.MapConfig) SystemAdmins(org.apache.samza.system.SystemAdmins) ObjectMapper(com.fasterxml.jackson.databind.ObjectMapper) KV(org.apache.samza.operators.KV) SystemAdmin(org.apache.samza.system.SystemAdmin) GenericSystemDescriptor(org.apache.samza.system.descriptors.GenericSystemDescriptor) Test(org.junit.Test)

Example 13 with GenericOutputDescriptor

use of org.apache.samza.system.descriptors.GenericOutputDescriptor in project samza by apache.

the class TestJobGraphJsonGenerator method testRepartitionedJoinStreamApplication.

@Test
public void testRepartitionedJoinStreamApplication() throws Exception {
    /**
     * the graph looks like the following.
     * number in parentheses () indicates number of stream partitions.
     * number in parentheses in quotes ("") indicates expected partition count.
     * number in square brackets [] indicates operator ID.
     *
     * input3 (32) -> filter [7] -> partitionBy [8] ("64") -> map [10] -> join [14] -> sendTo(output2) [15] (16)
     *                                                                   |
     *              input2 (16) -> partitionBy [3] ("64") -> filter [5] -| -> sink [13]
     *                                                                   |
     *                                         input1 (64) -> map [1] -> join [11] -> sendTo(output1) [12] (8)
     */
    Map<String, String> configMap = new HashMap<>();
    configMap.put(JobConfig.JOB_NAME, "test-app");
    configMap.put(JobConfig.JOB_DEFAULT_SYSTEM, "test-system");
    StreamTestUtils.addStreamConfigs(configMap, "input1", "system1", "input1");
    StreamTestUtils.addStreamConfigs(configMap, "input2", "system2", "input2");
    StreamTestUtils.addStreamConfigs(configMap, "input3", "system2", "input3");
    StreamTestUtils.addStreamConfigs(configMap, "output1", "system1", "output1");
    StreamTestUtils.addStreamConfigs(configMap, "output2", "system2", "output2");
    Config config = new MapConfig(configMap);
    // set up external partition count
    Map<String, Integer> system1Map = new HashMap<>();
    system1Map.put("input1", 64);
    system1Map.put("output1", 8);
    Map<String, Integer> system2Map = new HashMap<>();
    system2Map.put("input2", 16);
    system2Map.put("input3", 32);
    system2Map.put("output2", 16);
    SystemAdmin systemAdmin1 = createSystemAdmin(system1Map);
    SystemAdmin systemAdmin2 = createSystemAdmin(system2Map);
    SystemAdmins systemAdmins = mock(SystemAdmins.class);
    when(systemAdmins.getSystemAdmin("system1")).thenReturn(systemAdmin1);
    when(systemAdmins.getSystemAdmin("system2")).thenReturn(systemAdmin2);
    StreamManager streamManager = new StreamManager(systemAdmins);
    StreamApplicationDescriptorImpl graphSpec = new StreamApplicationDescriptorImpl(appDesc -> {
        KVSerde<Object, Object> kvSerde = new KVSerde<>(new NoOpSerde(), new NoOpSerde());
        String mockSystemFactoryClass = "factory.class.name";
        GenericSystemDescriptor system1 = new GenericSystemDescriptor("system1", mockSystemFactoryClass);
        GenericSystemDescriptor system2 = new GenericSystemDescriptor("system2", mockSystemFactoryClass);
        GenericInputDescriptor<KV<Object, Object>> input1Descriptor = system1.getInputDescriptor("input1", kvSerde);
        GenericInputDescriptor<KV<Object, Object>> input2Descriptor = system2.getInputDescriptor("input2", kvSerde);
        GenericInputDescriptor<KV<Object, Object>> input3Descriptor = system2.getInputDescriptor("input3", kvSerde);
        GenericOutputDescriptor<KV<Object, Object>> output1Descriptor = system1.getOutputDescriptor("output1", kvSerde);
        GenericOutputDescriptor<KV<Object, Object>> output2Descriptor = system2.getOutputDescriptor("output2", kvSerde);
        MessageStream<KV<Object, Object>> messageStream1 = appDesc.getInputStream(input1Descriptor).map(m -> m);
        MessageStream<KV<Object, Object>> messageStream2 = appDesc.getInputStream(input2Descriptor).partitionBy(m -> m.key, m -> m.value, mock(KVSerde.class), "p1").filter(m -> true);
        MessageStream<KV<Object, Object>> messageStream3 = appDesc.getInputStream(input3Descriptor).filter(m -> true).partitionBy(m -> m.key, m -> m.value, mock(KVSerde.class), "p2").map(m -> m);
        OutputStream<KV<Object, Object>> outputStream1 = appDesc.getOutputStream(output1Descriptor);
        OutputStream<KV<Object, Object>> outputStream2 = appDesc.getOutputStream(output2Descriptor);
        messageStream1.join(messageStream2, (JoinFunction<Object, KV<Object, Object>, KV<Object, Object>, KV<Object, Object>>) mock(JoinFunction.class), mock(Serde.class), mock(Serde.class), mock(Serde.class), Duration.ofHours(2), "j1").sendTo(outputStream1);
        messageStream2.sink((message, collector, coordinator) -> {
        });
        messageStream3.join(messageStream2, (JoinFunction<Object, KV<Object, Object>, KV<Object, Object>, KV<Object, Object>>) mock(JoinFunction.class), mock(Serde.class), mock(Serde.class), mock(Serde.class), Duration.ofHours(1), "j2").sendTo(outputStream2);
    }, config);
    ExecutionPlanner planner = new ExecutionPlanner(config, streamManager);
    ExecutionPlan plan = planner.plan(graphSpec);
    String json = plan.getPlanAsJson();
    System.out.println(json);
    // deserialize
    ObjectMapper mapper = new ObjectMapper();
    JobGraphJsonGenerator.JobGraphJson nodes = mapper.readValue(json, JobGraphJsonGenerator.JobGraphJson.class);
    assertEquals(5, nodes.jobs.get(0).operatorGraph.inputStreams.size());
    assertEquals(11, nodes.jobs.get(0).operatorGraph.operators.size());
    assertEquals(3, nodes.sourceStreams.size());
    assertEquals(2, nodes.sinkStreams.size());
    assertEquals(2, nodes.intermediateStreams.size());
}
Also used : StreamApplicationDescriptorImpl(org.apache.samza.application.descriptors.StreamApplicationDescriptorImpl) TableDescriptor(org.apache.samza.table.descriptors.TableDescriptor) GenericSystemDescriptor(org.apache.samza.system.descriptors.GenericSystemDescriptor) JobConfig(org.apache.samza.config.JobConfig) HashMap(java.util.HashMap) Serde(org.apache.samza.serializers.Serde) TestLocalTableDescriptor(org.apache.samza.table.descriptors.TestLocalTableDescriptor) GenericInputDescriptor(org.apache.samza.system.descriptors.GenericInputDescriptor) SendToTableOperatorSpec(org.apache.samza.operators.spec.SendToTableOperatorSpec) StringSerde(org.apache.samza.serializers.StringSerde) HashSet(java.util.HashSet) OperatorSpec(org.apache.samza.operators.spec.OperatorSpec) StreamTestUtils(org.apache.samza.testUtils.StreamTestUtils) TestExecutionPlanner(org.apache.samza.execution.TestExecutionPlanner) Duration(java.time.Duration) Map(java.util.Map) ApplicationConfig(org.apache.samza.config.ApplicationConfig) MapConfig(org.apache.samza.config.MapConfig) KV(org.apache.samza.operators.KV) NoOpSerde(org.apache.samza.serializers.NoOpSerde) MessageStream(org.apache.samza.operators.MessageStream) LongSerde(org.apache.samza.serializers.LongSerde) Before(org.junit.Before) Windows(org.apache.samza.operators.windows.Windows) StreamTableJoinFunction(org.apache.samza.operators.functions.StreamTableJoinFunction) GenericOutputDescriptor(org.apache.samza.system.descriptors.GenericOutputDescriptor) ObjectMapper(com.fasterxml.jackson.databind.ObjectMapper) Set(java.util.Set) Matchers(org.hamcrest.Matchers) StreamSpec(org.apache.samza.system.StreamSpec) Test(org.junit.Test) JoinFunction(org.apache.samza.operators.functions.JoinFunction) Collectors(java.util.stream.Collectors) Mockito(org.mockito.Mockito) OperatorSpecs(org.apache.samza.operators.spec.OperatorSpecs) StreamTableJoinOperatorSpec(org.apache.samza.operators.spec.StreamTableJoinOperatorSpec) SystemAdmin(org.apache.samza.system.SystemAdmin) Config(org.apache.samza.config.Config) JsonSerdeV2(org.apache.samza.serializers.JsonSerdeV2) KVSerde(org.apache.samza.serializers.KVSerde) Assert(org.junit.Assert) Collections(java.util.Collections) OutputStream(org.apache.samza.operators.OutputStream) SystemAdmins(org.apache.samza.system.SystemAdmins) KVSerde(org.apache.samza.serializers.KVSerde) HashMap(java.util.HashMap) JobConfig(org.apache.samza.config.JobConfig) ApplicationConfig(org.apache.samza.config.ApplicationConfig) MapConfig(org.apache.samza.config.MapConfig) Config(org.apache.samza.config.Config) TestExecutionPlanner(org.apache.samza.execution.TestExecutionPlanner) StreamApplicationDescriptorImpl(org.apache.samza.application.descriptors.StreamApplicationDescriptorImpl) MapConfig(org.apache.samza.config.MapConfig) SystemAdmins(org.apache.samza.system.SystemAdmins) ObjectMapper(com.fasterxml.jackson.databind.ObjectMapper) KV(org.apache.samza.operators.KV) StreamTableJoinFunction(org.apache.samza.operators.functions.StreamTableJoinFunction) JoinFunction(org.apache.samza.operators.functions.JoinFunction) NoOpSerde(org.apache.samza.serializers.NoOpSerde) SystemAdmin(org.apache.samza.system.SystemAdmin) GenericSystemDescriptor(org.apache.samza.system.descriptors.GenericSystemDescriptor) Test(org.junit.Test)

Aggregations

GenericOutputDescriptor (org.apache.samza.system.descriptors.GenericOutputDescriptor)13 KVSerde (org.apache.samza.serializers.KVSerde)12 GenericSystemDescriptor (org.apache.samza.system.descriptors.GenericSystemDescriptor)12 Test (org.junit.Test)12 Serde (org.apache.samza.serializers.Serde)11 NoOpSerde (org.apache.samza.serializers.NoOpSerde)9 IntegerSerde (org.apache.samza.serializers.IntegerSerde)8 HashMap (java.util.HashMap)7 GenericInputDescriptor (org.apache.samza.system.descriptors.GenericInputDescriptor)7 StreamApplicationDescriptorImpl (org.apache.samza.application.descriptors.StreamApplicationDescriptorImpl)6 Config (org.apache.samza.config.Config)6 JobConfig (org.apache.samza.config.JobConfig)6 MapConfig (org.apache.samza.config.MapConfig)6 KV (org.apache.samza.operators.KV)6 StringSerde (org.apache.samza.serializers.StringSerde)6 Map (java.util.Map)5 MessageStream (org.apache.samza.operators.MessageStream)5 OutputStream (org.apache.samza.operators.OutputStream)5 Duration (java.time.Duration)4 Collections (java.util.Collections)4