JavaによるApache Flinkの紹介

DataSet amounts = env.fromElements(1, 29, 40, 50);

int threshold = 30;
List collect = amounts
  .filter(a -> a > threshold)
  .reduce((integer, t1) -> integer + t1)
  .collect();

assertThat(collect.get(0)).isEqualTo(90);

private static class Person {
    private int age;
    private String name;

    // standard constructors/getters/setters
}

DataSet personDataSource = env.fromCollection(
  Arrays.asList(
    new Person(23, "Tom"),
    new Person(75, "Michael")));

List ages = personDataSource
  .map(p -> p.age)
  .collect();

assertThat(ages).hasSize(2);
assertThat(ages).contains(23, 75);

Tuple3 address
  = new Tuple3<>(1, "5th Avenue", "London");
DataSet> addresses
  = env.fromElements(address);

Tuple2 firstTransaction
  = new Tuple2<>(1, "Transaction_1");
DataSet> transactions
  = env.fromElements(firstTransaction, new Tuple2<>(12, "Transaction_2"));

private static class IdKeySelectorTransaction
  implements KeySelector, Integer> {
    @Override
    public Integer getKey(Tuple2 value) {
        return value.f0;
    }
}

private static class IdKeySelectorAddress
  implements KeySelector, Integer> {
    @Override
    public Integer getKey(Tuple3 value) {
        return value.f0;
    }
}

List, Tuple3>>
  joined = transactions.join(addresses)
  .where(new IdKeySelectorTransaction())
  .equalTo(new IdKeySelectorAddress())
  .collect();

assertThat(joined).hasSize(1);
assertThat(joined).contains(new Tuple2<>(firstTransaction, address));

Tuple2 secondPerson = new Tuple2<>(4, "Tom");
Tuple2 thirdPerson = new Tuple2<>(5, "Scott");
Tuple2 fourthPerson = new Tuple2<>(200, "Michael");
Tuple2 firstPerson = new Tuple2<>(1, "Jack");
DataSet> transactions = env.fromElements(
  fourthPerson, secondPerson, thirdPerson, firstPerson);

List> sorted = transactions
  .sortPartition(new IdKeySelectorTransaction(), Order.ASCENDING)
  .collect();

assertThat(sorted)
  .containsExactly(firstPerson, secondPerson, thirdPerson, fourthPerson);

StreamExecutionEnvironment executionEnvironment
 = StreamExecutionEnvironment.getExecutionEnvironment();

DataStream dataStream = executionEnvironment.fromElements(
  "This is a first sentence",
  "This is a second sentence with a one word");

SingleOutputStreamOperator upperCase = text.map(String::toUpperCase);

upperCase.print();
env.execute();

1> THIS IS A FIRST SENTENCE
2> THIS IS A SECOND SENTENCE WITH A ONE WORD

SingleOutputStreamOperator> windowed
  = env.fromElements(
  new Tuple2<>(16, ZonedDateTime.now().plusMinutes(25).toInstant().getEpochSecond()),
  new Tuple2<>(15, ZonedDateTime.now().plusMinutes(2).toInstant().getEpochSecond()))
  .assignTimestampsAndWatermarks(
    new BoundedOutOfOrdernessTimestampExtractor
      >(Time.seconds(20)) {

        @Override
        public long extractTimestamp(Tuple2 element) {
          return element.f1 * 1000;
        }
    });

SingleOutputStreamOperator> reduced = windowed
  .windowAll(TumblingEventTimeWindows.of(Time.seconds(5)))
  .maxBy(0, true);
reduced.print();

1> (15,1491221519)