Search in sources :

Example 16 with JsonSerdeV2

use of org.apache.samza.serializers.JsonSerdeV2 in project samza by apache.

the class BroadcastAssertApp method describe.

@Override
public void describe(StreamApplicationDescriptor appDescriptor) {
    Config config = appDescriptor.getConfig();
    String inputTopic = config.get(INPUT_TOPIC_NAME_PROP);
    final JsonSerdeV2<PageView> serde = new JsonSerdeV2<>(PageView.class);
    KafkaSystemDescriptor ksd = new KafkaSystemDescriptor(SYSTEM);
    KafkaInputDescriptor<PageView> isd = ksd.getInputDescriptor(inputTopic, serde);
    final MessageStream<PageView> broadcastPageViews = appDescriptor.getInputStream(isd).broadcast(serde, "pv");
    /**
     * Each task will see all the pageview events
     */
    MessageStreamAssert.that("Each task contains all broadcast PageView events", broadcastPageViews, serde).forEachTask().containsInAnyOrder(Arrays.asList(new PageView("v1", "p1", "u1"), new PageView("v2", "p2", "u1"), new PageView("v3", "p1", "u2"), new PageView("v4", "p3", "u2")));
}
Also used : PageView(org.apache.samza.test.operator.data.PageView) Config(org.apache.samza.config.Config) KafkaSystemDescriptor(org.apache.samza.system.kafka.descriptors.KafkaSystemDescriptor) JsonSerdeV2(org.apache.samza.serializers.JsonSerdeV2)

Example 17 with JsonSerdeV2

use of org.apache.samza.serializers.JsonSerdeV2 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)

Aggregations

JsonSerdeV2 (org.apache.samza.serializers.JsonSerdeV2)17 Config (org.apache.samza.config.Config)13 StringSerde (org.apache.samza.serializers.StringSerde)13 KafkaSystemDescriptor (org.apache.samza.system.kafka.descriptors.KafkaSystemDescriptor)13 StreamApplication (org.apache.samza.application.StreamApplication)11 StreamApplicationDescriptor (org.apache.samza.application.descriptors.StreamApplicationDescriptor)11 KV (org.apache.samza.operators.KV)11 KVSerde (org.apache.samza.serializers.KVSerde)11 KafkaInputDescriptor (org.apache.samza.system.kafka.descriptors.KafkaInputDescriptor)11 MessageStream (org.apache.samza.operators.MessageStream)10 ApplicationRunner (org.apache.samza.runtime.ApplicationRunner)10 ApplicationRunners (org.apache.samza.runtime.ApplicationRunners)10 KafkaOutputDescriptor (org.apache.samza.system.kafka.descriptors.KafkaOutputDescriptor)10 CommandLine (org.apache.samza.util.CommandLine)10 Duration (java.time.Duration)9 OutputStream (org.apache.samza.operators.OutputStream)8 Windows (org.apache.samza.operators.windows.Windows)8 PageViewEvent (org.apache.samza.example.models.PageViewEvent)6 HashMap (java.util.HashMap)4 WindowPane (org.apache.samza.operators.windows.WindowPane)4