flink&kafka-connector消费 protobuf格式数据

一、背景需求

客流仿真系统是用C#写的，生成客流明细数据实时写入kafka。但是，，，

同时写的也很慢，性能达不到要求。讨论之后，因为仿真数据是在redis集群上分布生成的，现直接以收集到的对象数据封装为protobuf，经实时数据总线(接口)转入kafak，flink实时消费数据，，，那么问题是，我要解析 protobuf数据格式、拆分为明细数据、md5、去重、关联基础数据，压力在我这边了，没办法先测试一把吧！

二、protobuf模板对象生成

1、使用protobuf将模板生成java文件

首先下载一个window版本的protobuf，用的是protoc-2.5.0-win32.zip版本：
https://github.com/protocolbuffers/protobuf/releases/download/v2.5.0/protoc-2.5.0-win32.zip

其他最新的版本可参考：
https://github.com/protocolbuffers/protobuf/releases

解压后在根目录新建java文件夹

在proto新建模板

内容如下(具体根据需求定义)：

syntax = "proto2";
 
package dwd;
option java_package = "com.xxx.train.dwd";
option java_outer_classname = "BaseTrainsProto";
 
message BaseTrains{
	required string line_id = 1;
	required string section_id = 2;
	required string station_id = 3;
	required string arriving_time = 4;
	required string departure_time = 5;
	required string timetable_type = 6;
	required string trip_id = 7;
	required string ed_num = 8;
	required string pdate = 9;
	required string period_end = 10;
}

 打开cmd 进入到protoc.exe所在的目录，并执行如下命令：

protoc -I=C:\protoc-2.5.0-win32\proto --java_out=C:\protoc-2.5.0-win32 C:\protoc-2.5.0-win32\proto\BaseTrains.proto

 成功后，进入java目录即可看到java模板实例，直接将其拷贝到您的项目即可使用

注意：包路径必须和项目包路径一直！

三、flink消费kafka protobuf格式数据

1、pom.xml配置

自定义序列化器 | Apache Flink

对于 Google Protobuf 需要添加以下 Maven 依赖：

<dependency>
	<groupId>com.twitter</groupId>
	<artifactId>chill-protobuf</artifactId>
	<version>0.7.6</version>
	<!-- exclusions for dependency conversion -->
	<exclusions>
		<exclusion>
			<groupId>com.esotericsoftware.kryo</groupId>
			<artifactId>kryo</artifactId>
		</exclusion>
	</exclusions>
</dependency>
<!-- We need protobuf for chill-protobuf -->
<dependency>
	<groupId>com.google.protobuf</groupId>
	<artifactId>protobuf-java</artifactId>
	<version>3.7.0</version>
</dependency>
 
​

2、flink消费kafka数据代码

object DWDTrainAppV2ProToBuf {
 
  def main(args: Array[String]): Unit = {
 
    // 1.创建执行环境
    val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
    //设置并发度
    env.setParallelism(1)
    env.disableOperatorChaining()
    // 2. 从kafka中读取数据
    val properties = new Properties()
    properties.setProperty("bootstrap.servers", "xxxx:9092,xxxx:9092,xxxx:9092")
    val sourceTopic: String = "test_protocol"
    val consumer: FlinkKafkaConsumer[Array[Byte]] = new FlinkKafkaConsumer[Array[Byte]](sourceTopic, new ByteArrayDeserializationSchema[Array[Byte]](), properties)
    consumer.setStartFromEarliest()
    val trainMsgStream: DataStream[Array[Byte]] = env.addSource(consumer)
    //1、先获取Byte类型
    //2、传至自定义实体类BaseTrains
    val msg = trainMsgStream.map(message => new BaseTrains(message))
    msg.print()
    env.execute("start")
  }
}

类ByteArrayDeserializationSchema在下面定义

3、写入Kafka中的数据是Protobuf类型的数据，需要用二进制的Schema进行接收，可以自己实现一个类。

class ByteArrayDeserializationSchema[T] extends AbstractDeserializationSchema[Array[Byte]] {
  @throws[IOException]
  override def deserialize(message: Array[Byte]): Array[Byte] = message
}

4、BaseTrains类

public class BaseTrains implements Serializable, Protobufable {
    private String line_id;
    private String section_id;
    private String station_id;
    private String arriving_time;
    private String departure_time;
    private String timetable_type;
    private String trip_id;
    private String ed_num;
    private String pdate;
    private String period_end;
 
    public BaseTrains() {
 
    }
 
    public BaseTrains(String line_id, String section_id, String station_id, String arriving_time, String departure_time, String timetable_type, String trip_id, String ed_num, String pdate, String period_end) {
        super();
        this.line_id = line_id;
        this.section_id = section_id;
        this.station_id = station_id;
        this.arriving_time = arriving_time;
        this.departure_time = departure_time;
        this.timetable_type = timetable_type;
        this.trip_id = trip_id;
        this.ed_num = ed_num;
        this.pdate = pdate;
        this.period_end = period_end;
    }
 
    @Override
    public byte[] encode() {
        BaseTrainsProto.BaseTrains.Builder builder = BaseTrainsProto.BaseTrains.newBuilder();
        builder.setLineId(line_id);
        builder.setSectionId(section_id);
        builder.setStationId(station_id);
        builder.setArrivingTime(arriving_time);
        builder.setDepartureTime(departure_time);
        builder.setTimetableType(timetable_type);
        builder.setTripId(trip_id);
        builder.setEdNum(ed_num);
        builder.setPdate(pdate);
        builder.setPeriodEnd(period_end);
        return builder.build().toByteArray();
    }
 
    public BaseTrains(byte[] bytes) {
        try {
            //3、调用protobuf生成的模板解析二进制为文本数据
            BaseTrainsProto.BaseTrains trains = BaseTrainsProto.BaseTrains.parseFrom(bytes);
            this.line_id = trains.getLineId();
            this.section_id = trains.getSectionId();
            this.station_id = trains.getStationId();
            this.arriving_time = trains.getArrivingTime();
            this.departure_time = trains.getDepartureTime();
            this.timetable_type = trains.getTimetableType();
            this.trip_id = trains.getTripId();
            this.ed_num = trains.getEdNum();
            this.pdate = trains.getPdate();
            this.period_end = trains.getPeriodEnd();
        } catch (InvalidProtocolBufferException e) {
            e.printStackTrace();
        }
    }
 
    @Override
    public String toString() {
        return "{" + line_id + "," + section_id + "," + station_id + "," + arriving_time + "," + departure_time + "," + timetable_type + "," + trip_id + "," + ed_num + "," + pdate + "," + period_end + '}';
    }
 
    public String getLine_id() {
        return line_id;
    }
 
    public void setLine_id(String line_id) {
        this.line_id = line_id;
    }
 
    public String getSection_id() {
        return section_id;
    }
 
    public void setSection_id(String section_id) {
        this.section_id = section_id;
    }
 
    public String getStation_id() {
        return station_id;
    }
 
    public void setStation_id(String station_id) {
        this.station_id = station_id;
    }
 
    public String getArriving_time() {
        return arriving_time;
    }
 
    public void setArriving_time(String arriving_time) {
        this.arriving_time = arriving_time;
    }
 
    public String getDeparture_time() {
        return departure_time;
    }
 
    public void setDeparture_time(String departure_time) {
        this.departure_time = departure_time;
    }
 
    public String getTimetable_type() {
        return timetable_type;
    }
 
    public void setTimetable_type(String timetable_type) {
        this.timetable_type = timetable_type;
    }
 
    public String getTrip_id() {
        return trip_id;
    }
 
    public void setTrip_id(String trip_id) {
        this.trip_id = trip_id;
    }
 
    public String getEd_num() {
        return ed_num;
    }
 
    public void setEd_num(String ed_num) {
        this.ed_num = ed_num;
    }
 
    public String getPdate() {
        return pdate;
    }
 
    public void setPdate(String pdate) {
        this.pdate = pdate;
    }
 
    public String getPeriod_end() {
        return period_end;
    }
 
    public void setPeriod_end(String period_end) {
        this.period_end = period_end;
    }
}

整体过程：

1、基于提供的 BaseTrains.proto模板 调用命令protoc.exe 生成BaseTrainsProto

2、拷贝  BaseTrainsProto至项目目录

3、编写FlinkKafkaConsumer[Array[Byte]]

4、新建BaseTrains  POJO对象并调用 BaseTrainsProto解析 protouf 的二进制格式数据