Amazon Web Services 文档中描述的 Amazon Web Services 服务或功能可能因区域而异。要查看适用于中国区域的差异，请参阅 中国的 Amazon Web Services 服务入门 (PDF)。

本文属于机器翻译版本。若本译文内容与英语原文存在差异，则一律以英文原文为准。

订阅 Amazon Kinesis Data Streams

以下示例向您展示如何使用该subscribeToShard方法从 Amazon Kinesis 数据流中检索和处理数据。 Kinesis Data Streams 现在采用增强的扇出功能和低延迟 HTTP /2 数据检索API，使开发人员可以更轻松地在同一个数据流上运行多个低延迟、高性能的应用程序。 Kinesis

设置

首先，创建一个异步 Kinesis 客户端和一个SubscribeToShardRequest对象。以下每个示例都使用这些对象来订阅 Kinesis 事件。

导入

import java.util.concurrent.CompletableFuture;
import java.util.concurrent.atomic.AtomicInteger;
import org.reactivestreams.Subscriber;
import org.reactivestreams.Subscription;
import software.amazon.awssdk.core.async.SdkPublisher;
import software.amazon.awssdk.regions.Region;
import software.amazon.awssdk.services.kinesis.KinesisAsyncClient;
import software.amazon.awssdk.services.kinesis.model.ShardIteratorType;
import software.amazon.awssdk.services.kinesis.model.SubscribeToShardEvent;
import software.amazon.awssdk.services.kinesis.model.SubscribeToShardEventStream;
import software.amazon.awssdk.services.kinesis.model.SubscribeToShardRequest;
import software.amazon.awssdk.services.kinesis.model.SubscribeToShardResponse;
import software.amazon.awssdk.services.kinesis.model.SubscribeToShardResponseHandler;

代码

        Region region = Region.US_EAST_1;
        KinesisAsyncClient client = KinesisAsyncClient.builder()
        .region(region)
        .build();

        SubscribeToShardRequest request = SubscribeToShardRequest.builder()
                .consumerARN(CONSUMER_ARN)
                .shardId("arn:aws:kinesis:us-east-1:111122223333:stream/StockTradeStream")
                .startingPosition(s -> s.type(ShardIteratorType.LATEST)).build();

使用 Builder 接口

您可以使用该builder方法来简化创建SubscribeToShardResponseHandler。

使用生成器，您可以通过方法调用设置每个生命周期回调，而非实施完整的接口。

代码

    private static CompletableFuture<Void> responseHandlerBuilder(KinesisAsyncClient client, SubscribeToShardRequest request) {
        SubscribeToShardResponseHandler responseHandler = SubscribeToShardResponseHandler
                .builder()
                .onError(t -> System.err.println("Error during stream - " + t.getMessage()))
                .onComplete(() -> System.out.println("All records stream successfully"))
                // Must supply some type of subscriber
                .subscriber(e -> System.out.println("Received event - " + e))
                .build();
        return client.subscribeToShard(request, responseHandler);
    }

要对发布者进行更多控制，您可以使用 publisherTransformer 方法自定义发布者。

代码

    private static CompletableFuture<Void> responseHandlerBuilderPublisherTransformer(KinesisAsyncClient client, SubscribeToShardRequest request) {
        SubscribeToShardResponseHandler responseHandler = SubscribeToShardResponseHandler
                .builder()
                .onError(t -> System.err.println("Error during stream - " + t.getMessage()))
                .publisherTransformer(p -> p.filter(e -> e instanceof SubscribeToShardEvent).limit(100))
                .subscriber(e -> System.out.println("Received event - " + e))
                .build();
        return client.subscribeToShard(request, responseHandler);
    }

请参阅上的完整示例 GitHub。

使用自定义响应处理程序

要完全控制订阅用户和发布者，请实施 SubscribeToShardResponseHandler 接口。

在本示例中，您实施 onEventStream 方法，此方法允许您对发布者具有完全访问权限。此示例演示如何将发布者转换为事件记录以供订阅者打印。

代码

    private static CompletableFuture<Void> responseHandlerBuilderClassic(KinesisAsyncClient client, SubscribeToShardRequest request) {
        SubscribeToShardResponseHandler responseHandler = new SubscribeToShardResponseHandler() {

            @Override
            public void responseReceived(SubscribeToShardResponse response) {
                System.out.println("Receieved initial response");
            }

            @Override
            public void onEventStream(SdkPublisher<SubscribeToShardEventStream> publisher) {
                publisher
                        // Filter to only SubscribeToShardEvents
                        .filter(SubscribeToShardEvent.class)
                        // Flat map into a publisher of just records
                        .flatMapIterable(SubscribeToShardEvent::records)
                        // Limit to 1000 total records
                        .limit(1000)
                        // Batch records into lists of 25
                        .buffer(25)
                        // Print out each record batch
                        .subscribe(batch -> System.out.println("Record Batch - " + batch));
            }

            @Override
            public void complete() {
                System.out.println("All records stream successfully");
            }

            @Override
            public void exceptionOccurred(Throwable throwable) {
                System.err.println("Error during stream - " + throwable.getMessage());
            }
        };
        return client.subscribeToShard(request, responseHandler);
    }

请参阅上的完整示例 GitHub。

使用 Visitor 接口

您可以使用 Visitor 对象订阅您有兴趣观看的特定事件。

代码

    private static CompletableFuture<Void> responseHandlerBuilderVisitorBuilder(KinesisAsyncClient client, SubscribeToShardRequest request) {
        SubscribeToShardResponseHandler.Visitor visitor = SubscribeToShardResponseHandler.Visitor
                .builder()
                .onSubscribeToShardEvent(e -> System.out.println("Received subscribe to shard event " + e))
                .build();
        SubscribeToShardResponseHandler responseHandler = SubscribeToShardResponseHandler
                .builder()
                .onError(t -> System.err.println("Error during stream - " + t.getMessage()))
                .subscriber(visitor)
                .build();
        return client.subscribeToShard(request, responseHandler);
    }

请参阅上的完整示例 GitHub。

使用自定义订阅者

您也可以实施您自己的自定义订阅者以订阅流。

此代码段显示了一个示例订阅者。

代码

    private static class MySubscriber implements Subscriber<SubscribeToShardEventStream> {

        private Subscription subscription;
        private AtomicInteger eventCount = new AtomicInteger(0);

        @Override
        public void onSubscribe(Subscription subscription) {
            this.subscription = subscription;
            this.subscription.request(1);
        }

        @Override
        public void onNext(SubscribeToShardEventStream shardSubscriptionEventStream) {
            System.out.println("Received event " + shardSubscriptionEventStream);
            if (eventCount.incrementAndGet() >= 100) {
                // You can cancel the subscription at any time if you wish to stop receiving events.
                subscription.cancel();
            }
            subscription.request(1);
        }

        @Override
        public void onError(Throwable throwable) {
            System.err.println("Error occurred while stream - " + throwable.getMessage());
        }

        @Override
        public void onComplete() {
            System.out.println("Finished streaming all events");
        }
    }

您可以将自定义订阅者传递给 subscribe 方法，如以下代码片段所示。

代码

    private static CompletableFuture<Void> responseHandlerBuilderSubscriber(KinesisAsyncClient client, SubscribeToShardRequest request) {
        SubscribeToShardResponseHandler responseHandler = SubscribeToShardResponseHandler
                .builder()
                .onError(t -> System.err.println("Error during stream - " + t.getMessage()))
                .subscriber(MySubscriber::new)
                .build();
        return client.subscribeToShard(request, responseHandler);
    }

请参阅上的完整示例 GitHub。

将数据记录写入 Kinesis 数据流

您可以使用该KinesisClient对象通过putRecords方法将数据记录写入 Kinesis 数据流。要成功调用此方法，请创建一个PutRecordsRequest对象。将数据流的名称传递给 streamName 方法。此外，您必须使用 putRecords 方法传递数据（如下面的代码示例所示）。

导入

import software.amazon.awssdk.core.SdkBytes;
import software.amazon.awssdk.regions.Region;
import software.amazon.awssdk.services.kinesis.KinesisClient;
import software.amazon.awssdk.services.kinesis.model.PutRecordRequest;
import software.amazon.awssdk.services.kinesis.model.KinesisException;
import software.amazon.awssdk.services.kinesis.model.DescribeStreamRequest;
import software.amazon.awssdk.services.kinesis.model.DescribeStreamResponse;

在下面的 Java 代码示例中，请注意StockTrade对象用作写入 Kinesis 数据流的数据。在运行此示例之前，请确保已创建数据流。

代码

import software.amazon.awssdk.core.SdkBytes;
import software.amazon.awssdk.regions.Region;
import software.amazon.awssdk.services.kinesis.KinesisClient;
import software.amazon.awssdk.services.kinesis.model.PutRecordRequest;
import software.amazon.awssdk.services.kinesis.model.KinesisException;
import software.amazon.awssdk.services.kinesis.model.DescribeStreamRequest;
import software.amazon.awssdk.services.kinesis.model.DescribeStreamResponse;

/**
 * Before running this Java V2 code example, set up your development
 * environment, including your credentials.
 *
 * For more information, see the following documentation topic:
 *
 * https://docs.aws.amazon.com/sdk-for-java/latest/developer-guide/get-started.html
 */
public class StockTradesWriter {
    public static void main(String[] args) {
        final String usage = """

                Usage:
                    <streamName>

                Where:
                    streamName - The Amazon Kinesis data stream to which records are written (for example, StockTradeStream)
                """;

        if (args.length != 1) {
            System.out.println(usage);
            System.exit(1);
        }

        String streamName = args[0];
        Region region = Region.US_EAST_1;
        KinesisClient kinesisClient = KinesisClient.builder()
                .region(region)
                .build();

        // Ensure that the Kinesis Stream is valid.
        validateStream(kinesisClient, streamName);
        setStockData(kinesisClient, streamName);
        kinesisClient.close();
    }

    public static void setStockData(KinesisClient kinesisClient, String streamName) {
        try {
            // Repeatedly send stock trades with a 100 milliseconds wait in between.
            StockTradeGenerator stockTradeGenerator = new StockTradeGenerator();

            // Put in 50 Records for this example.
            int index = 50;
            for (int x = 0; x < index; x++) {
                StockTrade trade = stockTradeGenerator.getRandomTrade();
                sendStockTrade(trade, kinesisClient, streamName);
                Thread.sleep(100);
            }

        } catch (KinesisException | InterruptedException e) {
            System.err.println(e.getMessage());
            System.exit(1);
        }
        System.out.println("Done");
    }

    private static void sendStockTrade(StockTrade trade, KinesisClient kinesisClient,
            String streamName) {
        byte[] bytes = trade.toJsonAsBytes();

        // The bytes could be null if there is an issue with the JSON serialization by
        // the Jackson JSON library.
        if (bytes == null) {
            System.out.println("Could not get JSON bytes for stock trade");
            return;
        }

        System.out.println("Putting trade: " + trade);
        PutRecordRequest request = PutRecordRequest.builder()
                .partitionKey(trade.getTickerSymbol()) // We use the ticker symbol as the partition key, explained in
                                                       // the Supplemental Information section below.
                .streamName(streamName)
                .data(SdkBytes.fromByteArray(bytes))
                .build();

        try {
            kinesisClient.putRecord(request);
        } catch (KinesisException e) {
            System.err.println(e.getMessage());
        }
    }

    private static void validateStream(KinesisClient kinesisClient, String streamName) {
        try {
            DescribeStreamRequest describeStreamRequest = DescribeStreamRequest.builder()
                    .streamName(streamName)
                    .build();

            DescribeStreamResponse describeStreamResponse = kinesisClient.describeStream(describeStreamRequest);

            if (!describeStreamResponse.streamDescription().streamStatus().toString().equals("ACTIVE")) {
                System.err.println("Stream " + streamName + " is not active. Please wait a few moments and try again.");
                System.exit(1);
            }

        } catch (KinesisException e) {
            System.err.println("Error found while describing the stream " + streamName);
            System.err.println(e);
            System.exit(1);
        }
    }
}

请参阅上的完整示例 GitHub。

使用第三方库

您可以使用其他第三方库，而不是实现自定义订阅者。此示例演示了如何使用 RxJava 实现，但您可以使用任何实现 Reactive Streams 接口的库。有关该库的更多信息，请参阅 Github 上的RxJava 维基页面。

要使用该库，请将其作为依赖项添加。如果你使用的是 Maven，则该示例显示了要使用的POM代码片段。

POM参赛作品

<dependency>
 <groupId>io.reactivex.rxjava2</groupId>
 <artifactId>rxjava</artifactId>
 <version>2.2.21</version>
</dependency>

导入

import java.net.URI;
import java.util.concurrent.CompletableFuture;

import io.reactivex.Flowable;
import software.amazon.awssdk.auth.credentials.ProfileCredentialsProvider;
import software.amazon.awssdk.core.async.SdkPublisher;
import software.amazon.awssdk.http.Protocol;
import software.amazon.awssdk.http.SdkHttpConfigurationOption;
import software.amazon.awssdk.http.nio.netty.NettyNioAsyncHttpClient;
import software.amazon.awssdk.regions.Region;
import software.amazon.awssdk.services.kinesis.KinesisAsyncClient;
import software.amazon.awssdk.services.kinesis.model.ShardIteratorType;
import software.amazon.awssdk.services.kinesis.model.StartingPosition;
import software.amazon.awssdk.services.kinesis.model.SubscribeToShardEvent;
import software.amazon.awssdk.services.kinesis.model.SubscribeToShardRequest;
import software.amazon.awssdk.services.kinesis.model.SubscribeToShardResponseHandler;
import software.amazon.awssdk.utils.AttributeMap;

此示例 RxJava 在onEventStream生命周期方法中使用。这样，您就对发布者具有完全访问权限，这可用于创建 Rx Flowable。

代码

        SubscribeToShardResponseHandler responseHandler = SubscribeToShardResponseHandler
            .builder()
            .onError(t -> System.err.println("Error during stream - " + t.getMessage()))
            .onEventStream(p -> Flowable.fromPublisher(p)
                                        .ofType(SubscribeToShardEvent.class)
                                        .flatMapIterable(SubscribeToShardEvent::records)
                                        .limit(1000)
                                        .buffer(25)
                                        .subscribe(e -> System.out.println("Record batch = " + e)))
            .build();

您也可以使用带有 publisherTransformer 发布者的 Flowable 方法。您必须将Flowable发布者调整为 SdkPublisher，如以下示例所示。

代码

        SubscribeToShardResponseHandler responseHandler = SubscribeToShardResponseHandler
            .builder()
            .onError(t -> System.err.println("Error during stream - " + t.getMessage()))
            .publisherTransformer(p -> SdkPublisher.adapt(Flowable.fromPublisher(p).limit(100)))
            .build();

请参阅上的完整示例 GitHub。

更多信息