Java Code Examples for org.apache.samza.application.descriptors.StreamApplicationDescriptor#withDefaultSystem()

@Override
public void describe(StreamApplicationDescriptor appDescriptor) {
  KafkaSystemDescriptor trackingSystem = new KafkaSystemDescriptor("tracking");

  KafkaInputDescriptor<PageViewEvent> inputStreamDescriptor =
      trackingSystem.getInputDescriptor("pageViewEvent", new JsonSerdeV2<>(PageViewEvent.class));

  KafkaOutputDescriptor<KV<String, MyStreamOutput>> outputStreamDescriptor =
      trackingSystem.getOutputDescriptor("pageViewEventPerMember",
          KVSerde.of(new StringSerde(), new JsonSerdeV2<>(MyStreamOutput.class)));

  appDescriptor.withDefaultSystem(trackingSystem);
  MessageStream<PageViewEvent> pageViewEvents = appDescriptor.getInputStream(inputStreamDescriptor);
  OutputStream<KV<String, MyStreamOutput>> pageViewEventPerMember = appDescriptor.getOutputStream(outputStreamDescriptor);

  pageViewEvents
      .partitionBy(pve -> pve.getMemberId(), pve -> pve,
          KVSerde.of(new StringSerde(), new JsonSerdeV2<>(PageViewEvent.class)), "partitionBy")
      .window(Windows.keyedTumblingWindow(
          KV::getKey, Duration.ofMinutes(5), () -> 0, (m, c) -> c + 1, null, null), "window")
      .map(windowPane -> KV.of(windowPane.getKey().getKey(), new MyStreamOutput(windowPane)))
      .sendTo(pageViewEventPerMember);
}
 

@Override
public void describe(StreamApplicationDescriptor appDescriptor) {
  KafkaSystemDescriptor trackingSystem = new KafkaSystemDescriptor("tracking");

  KafkaInputDescriptor<PageViewEvent> inputStreamDescriptor =
      trackingSystem.getInputDescriptor("pageViewEvent", new JsonSerdeV2<>(PageViewEvent.class));

  KafkaOutputDescriptor<KV<String, StatsOutput>> outputStreamDescriptor =
      trackingSystem.getOutputDescriptor("pageViewEventPerMember",
          KVSerde.of(new StringSerde(), new JsonSerdeV2<>(StatsOutput.class)));

  appDescriptor.withDefaultSystem(trackingSystem);
  MessageStream<PageViewEvent> pageViewEvents = appDescriptor.getInputStream(inputStreamDescriptor);
  OutputStream<KV<String, StatsOutput>> pageViewEventPerMember = appDescriptor.getOutputStream(outputStreamDescriptor);

  pageViewEvents
      .partitionBy(pve -> pve.getMemberId(), pve -> pve,
          KVSerde.of(new StringSerde(), new JsonSerdeV2<>(PageViewEvent.class)), "partitionBy")
      .map(KV::getValue)
      .flatMap(new MyStatsCounter())
      .map(stats -> KV.of(stats.memberId, stats))
      .sendTo(pageViewEventPerMember);
}
 

@Override
public void describe(StreamApplicationDescriptor appDescriptor) {
  KafkaSystemDescriptor kafkaSystemDescriptor = new KafkaSystemDescriptor(KAFKA_SYSTEM_NAME)
      .withConsumerZkConnect(KAFKA_CONSUMER_ZK_CONNECT)
      .withProducerBootstrapServers(KAFKA_PRODUCER_BOOTSTRAP_SERVERS)
      .withDefaultStreamConfigs(KAFKA_DEFAULT_STREAM_CONFIGS);

  KVSerde<String, PageView> serde = KVSerde.of(new StringSerde(), new JsonSerdeV2<>(PageView.class));
  KafkaInputDescriptor<KV<String, PageView>> inputDescriptor =
      kafkaSystemDescriptor.getInputDescriptor(INPUT_STREAM_ID, serde);
  KafkaOutputDescriptor<KV<String, PageView>> outputDescriptor =
      kafkaSystemDescriptor.getOutputDescriptor(OUTPUT_STREAM_ID, serde);

  appDescriptor.withDefaultSystem(kafkaSystemDescriptor);

  MessageStream<KV<String, PageView>> pageViews = appDescriptor.getInputStream(inputDescriptor);
  OutputStream<KV<String, PageView>> filteredPageViews = appDescriptor.getOutputStream(outputDescriptor);

  pageViews
      .filter(kv -> !INVALID_USER_ID.equals(kv.value.userId))
      .sendTo(filteredPageViews);
}
 

@Override
public void describe(StreamApplicationDescriptor appDescriptor) {
  KafkaSystemDescriptor kafkaSystemDescriptor = new KafkaSystemDescriptor(KAFKA_SYSTEM_NAME)
      .withConsumerZkConnect(KAFKA_CONSUMER_ZK_CONNECT)
      .withProducerBootstrapServers(KAFKA_PRODUCER_BOOTSTRAP_SERVERS)
      .withDefaultStreamConfigs(KAFKA_DEFAULT_STREAM_CONFIGS);

  KafkaInputDescriptor<String> stockSymbolInputDescriptor =
      kafkaSystemDescriptor.getInputDescriptor(INPUT_STREAM_ID, new StringSerde());
  KafkaOutputDescriptor<StockPrice> stockPriceOutputDescriptor =
      kafkaSystemDescriptor.getOutputDescriptor(OUTPUT_STREAM_ID, new JsonSerdeV2<>(StockPrice.class));
  appDescriptor.withDefaultSystem(kafkaSystemDescriptor);
  MessageStream<String> stockSymbolStream = appDescriptor.getInputStream(stockSymbolInputDescriptor);
  OutputStream<StockPrice> stockPriceStream = appDescriptor.getOutputStream(stockPriceOutputDescriptor);

  RemoteTableDescriptor<String, Double> remoteTableDescriptor =
      new RemoteTableDescriptor("remote-table")
          .withReadRateLimit(10)
          .withReadFunction(new StockPriceReadFunction());
  CachingTableDescriptor<String, Double> cachedRemoteTableDescriptor =
      new CachingTableDescriptor<>("cached-remote-table", remoteTableDescriptor)
          .withReadTtl(Duration.ofSeconds(5));
  Table<KV<String, Double>> cachedRemoteTable = appDescriptor.getTable(cachedRemoteTableDescriptor);

  stockSymbolStream
      .map(symbol -> new KV<String, Void>(symbol, null))
      .join(cachedRemoteTable, new JoinFn())
      .sendTo(stockPriceStream);

}
 

@Override
public void describe(StreamApplicationDescriptor appDescriptor) {
  Serde<Profile> profileSerde = new JsonSerdeV2<>(Profile.class);
  Serde<PageView> pageViewSerde = new JsonSerdeV2<>(PageView.class);
  Serde<EnrichedPageView> joinResultSerde = new JsonSerdeV2<>(EnrichedPageView.class);

  KafkaSystemDescriptor kafkaSystemDescriptor = new KafkaSystemDescriptor(KAFKA_SYSTEM_NAME)
      .withConsumerZkConnect(KAFKA_CONSUMER_ZK_CONNECT)
      .withProducerBootstrapServers(KAFKA_PRODUCER_BOOTSTRAP_SERVERS)
      .withDefaultStreamConfigs(KAFKA_DEFAULT_STREAM_CONFIGS);

  KafkaInputDescriptor<Profile> profileInputDescriptor =
      kafkaSystemDescriptor.getInputDescriptor(PROFILE_STREAM_ID, profileSerde);
  KafkaInputDescriptor<PageView> pageViewInputDescriptor =
      kafkaSystemDescriptor.getInputDescriptor(PAGEVIEW_STREAM_ID, pageViewSerde);
  KafkaOutputDescriptor<EnrichedPageView> joinResultOutputDescriptor =
      kafkaSystemDescriptor.getOutputDescriptor(OUTPUT_TOPIC, joinResultSerde);

  RocksDbTableDescriptor<String, Profile> profileTableDescriptor =
      new RocksDbTableDescriptor<String, Profile>("profile-table", KVSerde.of(new StringSerde(), profileSerde));

  appDescriptor.withDefaultSystem(kafkaSystemDescriptor);

  MessageStream<Profile> profileStream = appDescriptor.getInputStream(profileInputDescriptor);
  MessageStream<PageView> pageViewStream = appDescriptor.getInputStream(pageViewInputDescriptor);
  OutputStream<EnrichedPageView> joinResultStream = appDescriptor.getOutputStream(joinResultOutputDescriptor);
  Table<KV<String, Profile>> profileTable = appDescriptor.getTable(profileTableDescriptor);

  profileStream
      .map(profile -> KV.of(profile.userId, profile))
      .sendTo(profileTable);

  pageViewStream
      .partitionBy(pv -> pv.userId, pv -> pv, KVSerde.of(new StringSerde(), pageViewSerde), "join")
      .join(profileTable, new JoinFn())
      .sendTo(joinResultStream);
}
 

@Override
public void describe(StreamApplicationDescriptor appDescriptor) {
  // Define a system descriptor for Kafka
  KafkaSystemDescriptor kafkaSystemDescriptor =
      new KafkaSystemDescriptor("kafka").withConsumerZkConnect(KAFKA_CONSUMER_ZK_CONNECT)
          .withProducerBootstrapServers(KAFKA_PRODUCER_BOOTSTRAP_SERVERS)
          .withDefaultStreamConfigs(KAFKA_DEFAULT_STREAM_CONFIGS);

  KafkaInputDescriptor<PageView> pageViewInputDescriptor =
      kafkaSystemDescriptor.getInputDescriptor(INPUT_PAGEVIEW_STREAM_ID, new JsonSerdeV2<>(PageView.class));

  // Define a system descriptor for Azure Blob Storage
  GenericSystemDescriptor azureBlobSystemDescriptor =
      new GenericSystemDescriptor(OUTPUT_SYSTEM, "org.apache.samza.system.azureblob.AzureBlobSystemFactory");

  GenericOutputDescriptor<PageViewAvroRecord> azureBlobOuputDescriptor =
      azureBlobSystemDescriptor.getOutputDescriptor(OUTPUT_STREAM, new NoOpSerde<>());

  // Set Kafka as the default system for the job
  appDescriptor.withDefaultSystem(kafkaSystemDescriptor);

  // Define the input and output streams with descriptors
  MessageStream<PageView> pageViewInput = appDescriptor.getInputStream(pageViewInputDescriptor);
  OutputStream<PageViewAvroRecord> pageViewAvroRecordOutputStream = appDescriptor.getOutputStream(azureBlobOuputDescriptor);

  // Define the execution flow with the high-level API
  pageViewInput
      .map((message) -> {
        LOG.info("Sending: Received PageViewEvent with pageId: " + message.pageId);
        return PageViewAvroRecord.buildPageViewRecord(message);
      })
      .sendTo(pageViewAvroRecordOutputStream);
}
 

@Override
public void describe(StreamApplicationDescriptor appDescriptor) {

  Duration windowDuration =
      appDescriptor.getConfig().containsKey("deploy.test") ? Duration.ofMillis(10) : Duration.ofSeconds(10);
  // Define a SystemDescriptor for Wikipedia data
  WikipediaSystemDescriptor wikipediaSystemDescriptor = new WikipediaSystemDescriptor("irc.wikimedia.org", 6667);

  // Define InputDescriptors for consuming wikipedia data
  WikipediaInputDescriptor wikipediaInputDescriptor = wikipediaSystemDescriptor
      .getInputDescriptor("en-wikipedia")
      .withChannel(WIKIPEDIA_CHANNEL);
  WikipediaInputDescriptor wiktionaryInputDescriptor = wikipediaSystemDescriptor
      .getInputDescriptor("en-wiktionary")
      .withChannel(WIKTIONARY_CHANNEL);
  WikipediaInputDescriptor wikiNewsInputDescriptor = wikipediaSystemDescriptor
      .getInputDescriptor("en-wikinews")
      .withChannel(WIKINEWS_CHANNEL);

  // Define a system descriptor for Kafka
  KafkaSystemDescriptor kafkaSystemDescriptor = new KafkaSystemDescriptor("kafka")
      .withConsumerZkConnect(KAFKA_CONSUMER_ZK_CONNECT)
      .withProducerBootstrapServers(KAFKA_PRODUCER_BOOTSTRAP_SERVERS)
      .withDefaultStreamConfigs(KAFKA_DEFAULT_STREAM_CONFIGS);

  // Define an output descriptor
  KafkaOutputDescriptor<WikipediaStatsOutput> statsOutputDescriptor =
      kafkaSystemDescriptor.getOutputDescriptor("wikipedia-stats", new JsonSerdeV2<>(WikipediaStatsOutput.class));


  // Set the default system descriptor to Kafka, so that it is used for all
  // internal resources, e.g., kafka topic for checkpointing, coordinator stream.
  appDescriptor.withDefaultSystem(kafkaSystemDescriptor);
  MessageStream<WikipediaFeedEvent> wikipediaEvents = appDescriptor.getInputStream(wikipediaInputDescriptor);
  MessageStream<WikipediaFeedEvent> wiktionaryEvents = appDescriptor.getInputStream(wiktionaryInputDescriptor);
  MessageStream<WikipediaFeedEvent> wikiNewsEvents = appDescriptor.getInputStream(wikiNewsInputDescriptor);
  OutputStream<WikipediaStatsOutput> wikipediaStats = appDescriptor.getOutputStream(statsOutputDescriptor);

  // Merge inputs
  MessageStream<WikipediaFeedEvent> allWikipediaEvents =
      MessageStream.mergeAll(ImmutableList.of(wikipediaEvents, wiktionaryEvents, wikiNewsEvents));


  // Parse, update stats, prepare output, and send
  allWikipediaEvents
      .map(WikipediaParser::parseEvent)
      .window(Windows.tumblingWindow(windowDuration,
          WikipediaStats::new, new WikipediaStatsAggregator(), WikipediaStats.serde()), "statsWindow")
      .map(this::formatOutput)
      .sendTo(wikipediaStats);
}
 

@Override
public void describe(StreamApplicationDescriptor app) {

  KafkaSystemDescriptor kafkaSystemDescriptor = new KafkaSystemDescriptor(KAFKA_SYSTEM_NAME)
      .withConsumerZkConnect(KAFKA_CONSUMER_ZK_CONNECT)
      .withProducerBootstrapServers(KAFKA_PRODUCER_BOOTSTRAP_SERVERS)
      .withDefaultStreamConfigs(KAFKA_DEFAULT_STREAM_CONFIGS);

  KafkaInputDescriptor<String> wordInputDescriptor =
      kafkaSystemDescriptor.getInputDescriptor(INPUT_STREAM_ID, new StringSerde());

  KafkaOutputDescriptor<String> countOutputDescriptor =
      kafkaSystemDescriptor.getOutputDescriptor(OUTPUT_STREAM_ID, new StringSerde());

  MyCouchbaseTableWriteFunction writeFn = new MyCouchbaseTableWriteFunction(BUCKET_NAME, CLUSTER_NODES)
      .withBootstrapCarrierDirectPort(COUCHBASE_PORT)
      .withUsernameAndPassword(BUCKET_NAME, BUCKET_PASSWORD)
      .withTimeout(Duration.ofSeconds(5));

  TableRetryPolicy retryPolicy = new TableRetryPolicy()
      .withFixedBackoff(Duration.ofSeconds(1))
      .withStopAfterAttempts(3);

  RemoteTableDescriptor couchbaseTableDescriptor = new RemoteTableDescriptor("couchbase-table")
      .withReadFunction(new NoOpTableReadFunction())
      .withReadRateLimiterDisabled()
      .withWriteFunction(writeFn)
      .withWriteRetryPolicy(retryPolicy)
      .withWriteRateLimit(4);

  app.withDefaultSystem(kafkaSystemDescriptor);
  MessageStream<String> wordStream = app.getInputStream(wordInputDescriptor);
  OutputStream<String> countStream = app.getOutputStream(countOutputDescriptor);
  app.getTable(couchbaseTableDescriptor);

  wordStream
      .flatMap(m -> Arrays.asList(m.split(" ")))
      .filter(word -> word != null && word.length() > 0)
      .map(new MyCountFunction())
      .map(countString -> currentTime() + " " + countString)
      .sendTo(countStream);
}