Kafka Streams开发入门(1)

 背景

最近发现Confluent公司在官网上发布了Kafka Streams教程，共有10节课，每节课给出了Kafka Streams的一个功能介绍。这个系列教程对于我们了解Kafka Streams还是很有帮助的。为什么要了解Kafka Streams？其实我一直觉得国内对于Flink有点过于迷恋了。大厂使用Flink尚自合理，毕竟数据量大且需要整套的集群管理调度监控功能。但一般的中小公司业务简单体量小，何必要费时费力地搭建一整套Flink集群。有很多简单的流处理业务场景使用Kafka Streams绰绰有余了，更何况Kafka Streams在设计上特别是在消息精确处理语义上一点都不必Flink差，只是它的定位不同罢了。如果你对Kafka Streams感兴趣，不妨关注一下这个系列。不过有点令人不爽的是，该教程使用Confluent Kafka作为开发环境进行演示，而且大量用到了Avro进行消息的序列化/反序列化。Confluent Kafka默认提供Schema Regstry、avro-console-producer和avro-console-consumer工具可以很方便地把消息转换成Avro格式并进行测试，但是如果你使用原生的Kafka（也就是社区版的Apache Kafka），这些功能要自行开发，非常地不方便。

鉴于此，我打算对照着这系列教程中的例子，使用Apache Kafka重新实现一遍。虽然有些科普，不过我觉得这是个很好的学习过程。需要注意的是，我不是使用Avro，而是使用Google的Protocol Buffer（下称protobuf）来进行演示。

 演示功能说明

第一篇要演示的功能很简单，就是流处理的map功能：map函数或算子将一个消息流中的每条事件进行转换，变更成另一个格式或另一条新的事件。今天输入消息是表示电影的消息，格式如下：

{"id": 294, "title": "Die Hard::1988", "genre": "action"}

我们使用Kafka Streams实时地将每条消息中的title字段分开，将里面的发行年份字段提取出来，变成下面这样：

{"id":294,"title":"Die Hard","release_year":1988,"genre":"action"}

 初始化项目

第一步是创建项目文件夹。在执行这步前，你需要安装并配置好Java环境和Gradle环境。Gradle是用于构建Java工程用的，下载地址是：https://gradle.org/。然后执行下列命令去创建项目：

mkdir movie-streams/
cd movie-streams/

 配置项目

在movie-streams目录下，创建build.gradle文件——该文件是Gradle的项目配置文件，类似于Maven的pom.xml。该文件内容如下：

 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.MovieStreamApp'
        )
    }
}

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

 其中比较关键的是：首先我们要引入Gradle的shadow插件，用于fat jar打包；其次，我们指定了Gradle的protobuf插件用于帮助我们将*.proto文件自动编译成Java类。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movie-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：

mkdir configuration

同时创建配置文件dev.properties，内容如下：

application.id=movie-transformer

bootstrap.servers=localhost:9092

input.topic.name=raw-movies
input.topic.partitions=1
input.topic.replication.factor=1

output.topic.name=movies
output.topic.partitions=1
output.topic.replication.factor=1

 该文件设置了我们要连接的Kafka集群信息以及输入topic和输出topic详情。

 创建消息Schema

下一步是创建输入消息和输出消息的schema。在movie-streams下执行命令创建保存schema的文件夹：

mkdir -p src/main/proto 

 然后分别创建raw-movie.proto和parsed-movie.proto文件，内容分别是：

syntax = "proto3";

package huxihx.kafkastreams.proto;

message RawMovie {
    uint64 id = 1;
    string title = 2;
    string genre = 3;
}

syntax = "proto3";

package huxihx.kafkastreams.proto;

message Movie {
    uint64 id = 1;
    string title = 2;
    uint32 release_year = 3;
    string genre = 4;
}

文件内容是标准的protobuf语法，定义了电影事件的id、title、release_year和genre信息。保存这两个文件，在movie-streams下之后运行gradlew命令将它们自动编译成Java类：

./gradlew build

此时，你应该可以在src/main/java/huxihx/kafkastreams/proto下看到生成的两个Java类：ParsedMovie和RawMovieOuterClass。

 创建Serdes

这一步我们为RawMovie和Movie消息创建各自的Serdes。所谓的Serdes是serializer和deserializer的合称。Kafka Streams程序在读取消息时需要用到Serdes中的deserializer将消息字节序列转换成对应的Java对象实例，而生产消息时则会用到Serdes的serializer将Java对象实例转换成字节序列。由于我们使用protobuf框架进行序列化和反序列化，因此我们需要创建支持protobuf的Serdes。

在movie-streams下执行:

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

然后在新创建的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();
    }
}

然后创建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);
        }
    }
}

最后创建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;
    }
}

 开发主流程

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

package huxihx.kafkastreams;

import huxihx.kafkastreams.proto.ParsedMovie;
import huxihx.kafkastreams.proto.RawMovieOuterClass;
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.KeyValue;
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.Produced;

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 MovieStreamApp {

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

        MovieStreamApp app = new MovieStreamApp();
        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);
    }


    /**
     * 构建Streams拓扑对象实例
     *
     * @param envProps
     * @return
     */
    private Topology buildTopology(Properties envProps) {
        final StreamsBuilder builder = new StreamsBuilder();
        final String inputTopic = envProps.getProperty("input.topic.name");
        final String outputTopic = envProps.getProperty("output.topic.name");

        builder.stream(inputTopic, Consumed.with(Serdes.String(), rawMovieProtobufSerdes()))
                .map((key, rawMovie) -> new KeyValue<>(rawMovie.getId(), parseRawMovie(rawMovie)))
                .to(outputTopic, Produced.with(Serdes.Long(), movieProtobufSerdes()));

        return builder.build();
    }

    /**
     * 为Kafka Streams程序构建所需的Properties实例
     *
     * @param envProps
     * @return
     */
    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;
    }

    /**
     * 预创建输入/输出topic，如果topic已存在则忽略
     *
     * @param envProps
     * @throws Exception
     */
    private void preCreateTopics(Properties envProps) throws Exception {
        Map<String, Object> config = new HashMap<>();
        config.put(AdminClientConfig.BOOTSTRAP_SERVERS_CONFIG, envProps.getProperty("bootstrap.servers"));
        String inputTopic = envProps.getProperty("input.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(inputTopic))
                topics.add(new NewTopic(
                        envProps.getProperty("input.topic.name"),
                        Integer.parseInt(envProps.getProperty("input.topic.partitions")),
                        Short.parseShort(envProps.getProperty("input.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();
        }
    }

    /**
     * 加载configuration下的配置文件
     *
     * @param fileName
     * @return
     * @throws IOException
     */
    private Properties loadEnvProperties(String fileName) throws IOException {
        Properties envProps = new Properties();
        try (FileInputStream input = new FileInputStream(fileName)) {
            envProps.load(input);
        }
        return envProps;
    }

    /**
     * 构建输出topic所需的Serdes
     *
     * @return
     */
    private static ProtobufSerdes<ParsedMovie.Movie> movieProtobufSerdes() {
        return new ProtobufSerdes<>(ParsedMovie.Movie.parser());
    }

    /**
     * 构建输入topic所需的Serdes
     *
     * @return
     */
    private static ProtobufSerdes<RawMovieOuterClass.RawMovie> rawMovieProtobufSerdes() {
        return new ProtobufSerdes<>(RawMovieOuterClass.RawMovie.parser());
    }

    /**
     * 执行map逻辑提取release_year字段
     *
     * @param rawMovie
     * @return
     */
    private static ParsedMovie.Movie parseRawMovie(RawMovieOuterClass.RawMovie rawMovie) {
        String[] titleParts = rawMovie.getTitle().split("::");
        String title = titleParts[0];
        int releaseYear = Integer.parseInt(titleParts[1]);
        return ParsedMovie.Movie.newBuilder()
                .setId(rawMovie.getId())
                .setTitle(title)
                .setReleaseYear(releaseYear)
                .setGenre(rawMovie.getGenre())
                .build();
    }
}

 编写测试Producer和Consumer

在src/main/java/huxihx/kafkastreams/tests/TestProducer.java和TestConsumer.java，内容分别如下：

package huxihx.kafkastreams.tests;

import huxihx.kafkastreams.proto.RawMovieOuterClass;
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.ProducerRecord;

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

public class TestProducer {

    // 测试输入事件
    private static final List<RawMovieOuterClass.RawMovie> TEST_RAW_MOVIES = Arrays.asList(
            RawMovieOuterClass.RawMovie.newBuilder()
                    .setId(294).setTitle("Die Hard::1988").setGenre("action").build(),

            RawMovieOuterClass.RawMovie.newBuilder()
                    .setId(354).setTitle("Tree of Life::2011").setGenre("drama").build(),

            RawMovieOuterClass.RawMovie.newBuilder()
                    .setId(782).setTitle("A Walk in the Clouds::1995").setGenre("romance").build(),

            RawMovieOuterClass.RawMovie.newBuilder()
                    .setId(128).setTitle("The Big Lebowski::1998").setGenre("comedy").build());

    public static void main(String[] args) {
        Properties props = new Properties();
        props.put("bootstrap.servers", "localhost:9092");
        props.put("acks", "all");
        props.put("key.serializer", "org.apache.kafka.common.serialization.StringSerializer");
        props.put("value.serializer", new ProtobufSerializer<RawMovieOuterClass.RawMovie>().getClass());

        try (final Producer<String, RawMovieOuterClass.RawMovie> producer = new KafkaProducer<>(props)) {
            TEST_RAW_MOVIES.stream()
                    .map(rawMovie -> new ProducerRecord<String, RawMovieOuterClass.RawMovie>("raw-movies", rawMovie))
                    .forEach(producer::send);
        }
    }
}

package huxihx.kafkastreams.tests;

import com.google.protobuf.Parser;
import huxihx.kafkastreams.proto.ParsedMovie;
import huxihx.kafkastreams.serdes.ProtobufDeserializer;
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.LongDeserializer;

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) {
        // 为输出事件构造protobuf deserializer
        Deserializer<ParsedMovie.Movie> deserializer = new ProtobufDeserializer<>();
        Map<String, Parser<ParsedMovie.Movie>> config = new HashMap<>();
        config.put("parser", ParsedMovie.Movie.parser());
        deserializer.configure(config, false);

        Properties props = new Properties();
        props.put("bootstrap.servers", "localhost:9092");
        props.put("group.id", "test-group");
        props.put("enable.auto.commit", "true");
        props.put("auto.commit.interval.ms", "1000");
        props.put("auto.offset.reset", "earliest");
        KafkaConsumer<Long, ParsedMovie.Movie> consumer = new KafkaConsumer<>(props, new LongDeserializer(), deserializer);
        consumer.subscribe(Arrays.asList("movies"));
        while (true) {
            ConsumerRecords<Long, ParsedMovie.Movie> records = consumer.poll(Duration.ofSeconds(1));
            for (ConsumerRecord<Long, ParsedMovie.Movie> record : records)
                System.out.printf("offset = %d, key = %s, value = %s%n", record.offset(), record.key(), record.value());
        }
    }
}

 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

然后启动TestProducer发送测试事件：

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

最后启动TestConsumer验证Kafka Streams提取出了每条输入事件的release_year字段：

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

.......

offset = 0, key = 294, value = id: 294
title: "Die Hard"
release_year: 1988
genre: "action"

offset = 1, key = 354, value = id: 354
title: "Tree of Life"
release_year: 2011
genre: "drama"

offset = 2, key = 782, value = id: 782
title: "A Walk in the Clouds"
release_year: 1995
genre: "romance"

offset = 3, key = 128, value = id: 128
title: "The Big Lebowski"
release_year: 1998
genre: "comedy"

 总结

okay，第一篇的演示至此结束。总体上来看，Kafka Streams的转换操作算子map还是非常好用的。它能够实时地为每条入站消息执行你指定的逻辑。下一篇中我将演示另一个经典的transformation操作：filter。