Kafka Streams开发入门(5)

1. 背景

上一篇演示了split操作算子的用法。今天展示一下split的逆操作：merge。Merge算子的作用是把多股实时消息流合并到一个单一的流中。

2. 功能演示说明

假设我们有多个Kafka topic，每个topic表示某类特定音乐类型的歌曲，比如有摇滚乐、古典乐等。本例中我们演示如何使用Kafka Streams将这些歌曲合并到一个Kafka topic中。我们依然使用Protocol Buffer对歌曲进行序列化和反序列化。你大概可以认为歌曲可以用下面的格式来表示：

{"artist": "Metallica", "title": "Fade to Black"}

{"artist": "Smashing Pumpkins", "title": "Today"}

3. 初始化项目

首先创建项目目录：

$ mkdir merge-streams
$ cd merge-streams/

4. 配置项目

创建Gradle项目配置文件build.gradle：

buildscript {

    repositories {
        jcenter()
    }
    dependencies {
        classpath 'com.github.jengelman.gradle.plugins:shadow:4.0.2'
    }
}

plugins {
    id 'java'
    id "com.google.protobuf" version "0.8.10"
}
apply plugin: 'com.github.johnrengelman.shadow'

repositories {
    mavenCentral()
    jcenter()

    maven {
        url 'http://packages.confluent.io/maven'
    }
}

group 'huxihx.kafkastreams'

sourceCompatibility = 1.8
targetCompatibility = '1.8'
version = '0.0.1'

dependencies {
    implementation 'com.google.protobuf:protobuf-java:3.0.0'
    implementation 'org.slf4j:slf4j-simple:1.7.26'
    implementation 'org.apache.kafka:kafka-streams:2.3.0'
    implementation 'com.google.protobuf:protobuf-java:3.9.1'

    testCompile group: 'junit', name: 'junit', version: '4.12'
}

protobuf {
    generatedFilesBaseDir = "$projectDir/src/"
    protoc {
        artifact = 'com.google.protobuf:protoc:3.0.0'
    }
}

jar {
    manifest {
        attributes(
                'Class-Path': configurations.compile.collect { it.getName() }.join(' '),
                'Main-Class': 'huxihx.kafkastreams.MergeStreams'
        )
    }
}

shadowJar {
    archiveName = "kstreams-transform-standalone-${version}.${extension}"
}

保存上面的文件，然后执行下列命令下载Gradle的wrapper套件：

$ gradle wrapper

之后在merge-streams目录下创建一个名为configuration的文件夹用于保存我们的参数配置文件dev.properties：

application.id=merging-app
bootstrap.servers=localhost:9092

input.rock.topic.name=rock-song-events
input.rock.topic.partitions=1
input.rock.topic.replication.factor=1

input.classical.topic.name=classical-song-events
input.classical.topic.partitions=1
input.classical.topic.replication.factor=1

output.topic.name=all-song-events
output.topic.partitions=1
output.topic.replication.factor=1

这里我们配置了两个输入topic，分别表示摇滚乐和古典乐歌曲。同时我们还创建了一个输出topic，用于保存merge之后的歌曲流。

5. 创建消息Schema

接下来创建用到的topic的schema。在merge-streams下执行命令创建保存schema的文件夹：

$ mkdir -p src/main/proto

之后在proto文件夹下创建名为song_event.proto文件，内容如下：

syntax = "proto3";

package huxihx.kafkastreams.proto;

message SongEvent {
    string name = 1;
    string title = 2;
}

保存之后在merge-stream下运行gradlew命令：

$ ./gradlew build

此时，你应该可以在merge-streams/src/main/java/huxihx/kafkastreams/proto下看到生成的Java类：SongEventOuterClass。

6. 创建Serdes

这一步我们为所需的topic消息创建各自的Serdes。首先执行下面的命令创建对应的文件夹目录：

$ mkdir -p src/main/java/huxihx/kafkastreams/serdes

之后在新创建的serdes文件夹下创建ProtobufSerializer.java：

package huxihx.kafkastreams.serdes;

import com.google.protobuf.MessageLite;
import org.apache.kafka.common.serialization.Serializer;

public class ProtobufSerializer<T extends MessageLite> implements Serializer<T> {
    @Override
    public byte[] serialize(String topic, T data) {
        return data == null ? new byte[0] : data.toByteArray();
    }
}

然后是ProtobufDeserializer.java：

package huxihx.kafkastreams.serdes;

import com.google.protobuf.InvalidProtocolBufferException;
import com.google.protobuf.MessageLite;
import com.google.protobuf.Parser;
import org.apache.kafka.common.errors.SerializationException;
import org.apache.kafka.common.serialization.Deserializer;

import java.util.Map;

public class ProtobufDeserializer<T extends MessageLite> implements Deserializer<T> {

    private Parser<T> parser;

    @Override
    public void configure(Map<String, ?> configs, boolean isKey) {
        parser = (Parser<T>) configs.get("parser");
    }

    @Override
    public T deserialize(String topic, byte[] data) {
        try {
            return parser.parseFrom(data);
        } catch (InvalidProtocolBufferException e) {
            throw new SerializationException("Failed to deserialize from a protobuf byte array.", e);
        }
    }
}

最后是ProtobufSerdes.java：

package huxihx.kafkastreams.serdes;

import com.google.protobuf.MessageLite;
import com.google.protobuf.Parser;
import org.apache.kafka.common.serialization.Deserializer;
import org.apache.kafka.common.serialization.Serde;
import org.apache.kafka.common.serialization.Serializer;

import java.util.HashMap;
import java.util.Map;

public class ProtobufSerdes<T extends MessageLite> implements Serde<T> {

    private final Serializer<T> serializer;
    private final Deserializer<T> deserializer;

    public ProtobufSerdes(Parser<T> parser) {
        serializer = new ProtobufSerializer<>();
        deserializer = new ProtobufDeserializer<>();
        Map<String, Parser<T>> config = new HashMap<>();
        config.put("parser", parser);
        deserializer.configure(config, false);
    }

    @Override
    public Serializer<T> serializer() {
        return serializer;
    }

    @Override
    public Deserializer<T> deserializer() {
        return deserializer;
    }
}

7. 开发主流程

在src/main/java/huxihx/kafkastreams下创建MergeStreams.java文件：

package huxihx.kafkastreams;

import huxihx.kafkastreams.proto.SongEventOuterClass;
import huxihx.kafkastreams.serdes.ProtobufSerdes;
import org.apache.kafka.clients.admin.AdminClient;
import org.apache.kafka.clients.admin.AdminClientConfig;
import org.apache.kafka.clients.admin.NewTopic;
import org.apache.kafka.clients.admin.TopicListing;
import org.apache.kafka.common.serialization.Serdes;
import org.apache.kafka.streams.KafkaStreams;
import org.apache.kafka.streams.StreamsBuilder;
import org.apache.kafka.streams.StreamsConfig;
import org.apache.kafka.streams.Topology;
import org.apache.kafka.streams.kstream.Consumed;
import org.apache.kafka.streams.kstream.KStream;

import java.io.FileInputStream;
import java.io.IOException;
import java.util.ArrayList;
import java.util.Collection;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import java.util.Properties;
import java.util.concurrent.CountDownLatch;
import java.util.stream.Collectors;

public class MergeStreams {
    public static void main(String[] args) throws Exception {
        if (args.length < 1) {
            throw new IllegalArgumentException("Config file path must be specified.");
        }

        MergeStreams app = new MergeStreams();
        Properties envProps = app.loadEnvProperties(args[0]);
        Properties streamProps = app.createStreamsProperties(envProps);
        Topology topology = app.buildTopology(envProps);

        app.preCreateTopics(envProps);

        final KafkaStreams streams = new KafkaStreams(topology, streamProps);
        final CountDownLatch latch = new CountDownLatch(1);

        Runtime.getRuntime().addShutdownHook(new Thread("streams-shutdown-hook") {
            @Override
            public void run() {
                streams.close();
                latch.countDown();
            }
        });

        try {
            streams.start();
            latch.await();
        } catch (Exception e) {
            System.exit(1);
        }
        System.exit(0);
    }

    private Topology buildTopology(Properties envProps) {
        final StreamsBuilder builder = new StreamsBuilder();
        final String rockEvents = envProps.getProperty("input.rock.topic.name");
        final String classicalEvents = envProps.getProperty("input.classical.topic.name");
        final String allEvents = envProps.getProperty("output.topic.name");

        KStream<String, SongEventOuterClass.SongEvent> rockStreams =
                builder.stream(rockEvents, Consumed.with(Serdes.String(), songEventProtobufSerdes()));
        KStream<String, SongEventOuterClass.SongEvent> classicalStreams =
                builder.stream(classicalEvents, Consumed.with(Serdes.String(), songEventProtobufSerdes()));
        KStream<String, SongEventOuterClass.SongEvent> allStreams = rockStreams.merge(classicalStreams);

        allStreams.to(allEvents, Produced.with(Serdes.String(), songEventProtobufSerdes()));
        return builder.build();
    }

    private Properties createStreamsProperties(Properties envProps) {
        Properties props = new Properties();
        props.put(StreamsConfig.APPLICATION_ID_CONFIG, envProps.getProperty("application.id"));
        props.put(StreamsConfig.BOOTSTRAP_SERVERS_CONFIG, envProps.getProperty("bootstrap.servers"));
        props.put(StreamsConfig.DEFAULT_KEY_SERDE_CLASS_CONFIG, Serdes.String().getClass());
        props.put(StreamsConfig.DEFAULT_VALUE_SERDE_CLASS_CONFIG, Serdes.String().getClass());
        return props;
    }

    private void preCreateTopics(Properties envProps) throws Exception {
        Map<String, Object> config = new HashMap<>();
        config.put(AdminClientConfig.BOOTSTRAP_SERVERS_CONFIG, envProps.getProperty("bootstrap.servers"));
        String inputTopic1 = envProps.getProperty("input.rock.topic.name");
        String inputTopic2 = envProps.getProperty("input.classical.topic.name");
        String outputTopic = envProps.getProperty("output.topic.name");
        try (AdminClient client = AdminClient.create(config)) {
            Collection<TopicListing> existingTopics = client.listTopics().listings().get();

            List<NewTopic> topics = new ArrayList<>();
            List<String> topicNames = existingTopics.stream().map(TopicListing::name).collect(Collectors.toList());
            if (!topicNames.contains(inputTopic1))
                topics.add(new NewTopic(
                        envProps.getProperty("input.rock.topic.name"),
                        Integer.parseInt(envProps.getProperty("input.rock.topic.partitions")),
                        Short.parseShort(envProps.getProperty("input.rock.topic.replication.factor"))));

            if (!topicNames.contains(inputTopic2))
                topics.add(new NewTopic(
                        envProps.getProperty("input.classical.topic.name"),
                        Integer.parseInt(envProps.getProperty("input.classical.topic.partitions")),
                        Short.parseShort(envProps.getProperty("input.classical.topic.replication.factor"))));

            if (!topicNames.contains(outputTopic))
                topics.add(new NewTopic(
                        envProps.getProperty("output.topic.name"),
                        Integer.parseInt(envProps.getProperty("output.topic.partitions")),
                        Short.parseShort(envProps.getProperty("output.topic.replication.factor"))));

            if (!topics.isEmpty())
                client.createTopics(topics).all().get();
        }
    }

    private Properties loadEnvProperties(String fileName) throws IOException {
        Properties envProps = new Properties();
        try (FileInputStream input = new FileInputStream(fileName)) {
            envProps.load(input);
        }
        return envProps;
    }

    private static ProtobufSerdes<SongEventOuterClass.SongEvent> songEventProtobufSerdes() {
        return new ProtobufSerdes<>(SongEventOuterClass.SongEvent.parser());
    }
}

主要的逻辑在buildTopology方法中，我们调用KStream的merge方法将两个输入消息流合并到一个输出消息流中。

8. 编写测试Producer和Consumer

和之前的入门系列一样，我们编写TestProducer和TestConsumer类。在src/main/java/huxihx/kafkastreams/tests/TestProducer.java和TestConsumer.java，内容分别如下：

TestProducer.java

package huxihx.kafkastreams.tests;

import huxihx.kafkastreams.proto.SongEventOuterClass;
import huxihx.kafkastreams.serdes.ProtobufSerializer;
import org.apache.kafka.clients.producer.KafkaProducer;
import org.apache.kafka.clients.producer.Producer;
import org.apache.kafka.clients.producer.ProducerConfig;
import org.apache.kafka.clients.producer.ProducerRecord;

import java.util.Arrays;
import java.util.List;
import java.util.Properties;

public class TestProducer {
    // 测试输入消息集1
    private static final List<SongEventOuterClass.SongEvent> TEST_SONG_EVENTS1 = Arrays.asList(
            SongEventOuterClass.SongEvent.newBuilder().setName("Metallica").setTitle("Fade to Black").build(),
            SongEventOuterClass.SongEvent.newBuilder().setName("Smashing Pumpkins").setTitle("Today").build(),
            SongEventOuterClass.SongEvent.newBuilder().setName("Pink Floyd").setTitle("Another Brick in the Wall").build(),
            SongEventOuterClass.SongEvent.newBuilder().setName("Van Halen").setTitle("Jump").build(),
            SongEventOuterClass.SongEvent.newBuilder().setName("Led Zeppelin").setTitle("Kashmir").build()
    );

    // 测试输入消息集2
    private static final List<SongEventOuterClass.SongEvent> TEST_SONG_EVENTS2 = Arrays.asList(
            SongEventOuterClass.SongEvent.newBuilder().setName("Wolfgang Amadeus Mozart").setTitle("The Magic Flute").build(),
            SongEventOuterClass.SongEvent.newBuilder().setName("Johann Pachelbel").setTitle("Canon").build(),
            SongEventOuterClass.SongEvent.newBuilder().setName("Ludwig van Beethoven").setTitle("Symphony No. 5").build(),
            SongEventOuterClass.SongEvent.newBuilder().setName("Edward Elgar").setTitle("Pomp and Circumstance").build()
    );

    public static void main(String[] args) {
        if (args.length < 1) {
            throw new IllegalArgumentException("Must specify a test set (1 or 2).");
        }
        int choice = Integer.parseInt(args[0]);
        if (choice != 1 && choice != 2) {
            throw new IllegalArgumentException("Must specify a test set (1 or 2).");
        }

        Properties props = new Properties();
        props.put(ProducerConfig.BOOTSTRAP_SERVERS_CONFIG, "localhost:9092");
        props.put(ProducerConfig.ACKS_CONFIG, "all");
        props.put(ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG, "org.apache.kafka.common.serialization.StringSerializer");
        props.put(ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG, new ProtobufSerializer<SongEventOuterClass.SongEvent>().getClass());

        try (final Producer<String, SongEventOuterClass.SongEvent> producer = new KafkaProducer<>(props)) {
            if (choice == 1) {
                TEST_SONG_EVENTS1.stream().map(song ->
                        new ProducerRecord<String, SongEventOuterClass.SongEvent>("rock-song-events", song))
                        .forEach(producer::send);
            } else {
                TEST_SONG_EVENTS2.stream().map(song ->
                        new ProducerRecord<String, SongEventOuterClass.SongEvent>("classical-song-events", song))
                        .forEach(producer::send);
            }
        }

    }
}

TestConsumer.java

package huxihx.kafkastreams.tests;

import com.google.protobuf.Parser;
import huxihx.kafkastreams.proto.SongEventOuterClass;
import huxihx.kafkastreams.serdes.ProtobufDeserializer;
import org.apache.kafka.clients.consumer.Consumer;
import org.apache.kafka.clients.consumer.ConsumerConfig;
import org.apache.kafka.clients.consumer.ConsumerRecord;
import org.apache.kafka.clients.consumer.ConsumerRecords;
import org.apache.kafka.clients.consumer.KafkaConsumer;
import org.apache.kafka.common.serialization.Deserializer;
import org.apache.kafka.common.serialization.StringDeserializer;

import java.time.Duration;
import java.util.Arrays;
import java.util.HashMap;
import java.util.Map;
import java.util.Properties;

public class TestConsumer {

    public static void main(String[] args) {
        if (args.length < 1) {
            throw new IllegalStateException("Must specify an output topic name.");
        }

        Deserializer<SongEventOuterClass.SongEvent> deserializer = new ProtobufDeserializer<>();
        Map<String, Parser<SongEventOuterClass.SongEvent>> config = new HashMap<>();
        config.put("parser", SongEventOuterClass.SongEvent.parser());
        deserializer.configure(config, false);

        Properties props = new Properties();
        props.put(ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG, "localhost:9092");
        props.put(ConsumerConfig.GROUP_ID_CONFIG, "test-group");
        props.put(ConsumerConfig.AUTO_COMMIT_INTERVAL_MS_CONFIG, "1000");
        props.put(ConsumerConfig.AUTO_OFFSET_RESET_CONFIG, "earliest");

        try (final Consumer<String, SongEventOuterClass.SongEvent> consumer = new KafkaConsumer<>(props, new StringDeserializer(), deserializer)) {
            consumer.subscribe(Arrays.asList(args[0]));
            while (true) {
                ConsumerRecords<String, SongEventOuterClass.SongEvent> records = consumer.poll(Duration.ofMillis(500));
                for (ConsumerRecord<String, SongEventOuterClass.SongEvent> record : records) {
                    System.out.printf("offset = %d, key = %s, value = %s%n", record.offset(), record.key(), record.value());
                }
            }
        }
    }
}

9. 测试　　

首先我们运行下列命令构建项目：

$ ./gradlew shadowJar

然后启动Kafka集群，之后运行Kafka Streams应用：

$ java -jar build/libs/kstreams-transform-standalone-0.0.1.jar configuration/dev.properties

现在启动两个终端分别测试两组TestProducer发送测试事件：　　

$ java -cp build/libs/kstreams-transform-standalone-0.0.1.jar huxihx.kafkastreams.tests.TestProducer 1

$ java -cp build/libs/kstreams-transform-standalone-0.0.1.jar huxihx.kafkastreams.tests.TestProducer 2

最后启动TestConsumer验证Kafka Streams将两个topic输入消息流合并到一个输出消息流：

$ java -cp build/libs/kstreams-transform-standalone-0.0.1.jar huxihx.kafkastreams.tests.TestConsumer all-song-events

offset = 0, key = null, value = name: "Metallica"
title: "Fade to Black"

offset = 1, key = null, value = name: "Metallica"
title: "Fade to Black"

offset = 2, key = null, value = name: "Smashing Pumpkins"
title: "Today"

offset = 3, key = null, value = name: "Smashing Pumpkins"
title: "Today"

offset = 4, key = null, value = name: "Pink Floyd"
title: "Another Brick in the Wall"

offset = 5, key = null, value = name: "Pink Floyd"
title: "Another Brick in the Wall"

offset = 6, key = null, value = name: "Van Halen"
title: "Jump"

offset = 7, key = null, value = name: "Van Halen"
title: "Jump"

offset = 8, key = null, value = name: "Led Zeppelin"
title: "Kashmir"

offset = 9, key = null, value = name: "Led Zeppelin"
title: "Kashmir"

offset = 10, key = null, value = name: "Wolfgang Amadeus Mozart"
title: "The Magic Flute"

offset = 11, key = null, value = name: "Johann Pachelbel"
title: "Canon"

offset = 12, key = null, value = name: "Ludwig van Beethoven"
title: "Symphony No. 5"

offset = 13, key = null, value = name: "Edward Elgar"
title: "Pomp and Circumstance"