25

Avro 序列化在 Hadoop 用户中很受欢迎,但很难找到示例。

谁能帮我这个示例代码?我最感兴趣的是使用 Reflect API 读取/写入文件以及使用 Union 和 Null 注释。

public class Reflect {

    public class Packet {
        int cost;
        @Nullable TimeStamp stamp;
        public Packet(int cost, TimeStamp stamp){
            this.cost = cost;
            this.stamp = stamp;
        }
    }

    public class TimeStamp {
        int hour = 0;
        int second = 0;
        public TimeStamp(int hour, int second){
            this.hour = hour;
            this.second = second;
        }
    }

    public static void main(String[] args) throws IOException {
        TimeStamp stamp;
        Packet packet;

        stamp = new TimeStamp(12, 34);
        packet = new Packet(9, stamp);
        write(file, packet);

        packet = new Packet(8, null);
        write(file, packet);
        file.close();

        // open file to read.
        packet = read(file);
        packet = read(file);
    }
}
4

1 回答 1

40

这是上述程序的一个有效版本。

这也对文件使用压缩。

import java.io.File;
import org.apache.avro.Schema;
import org.apache.avro.file.DataFileWriter;
import org.apache.avro.file.DataFileReader;
import org.apache.avro.file.CodecFactory;
import org.apache.avro.io.DatumWriter;
import org.apache.avro.io.DatumReader;
import org.apache.avro.reflect.ReflectData;
import org.apache.avro.reflect.ReflectDatumWriter;
import org.apache.avro.reflect.ReflectDatumReader;
import org.apache.avro.reflect.Nullable;

public class Reflect {

  public static class Packet {
    int cost;
    @Nullable TimeStamp stamp;
    public Packet() {}                        // required to read
    public Packet(int cost, TimeStamp stamp){
      this.cost = cost;
      this.stamp = stamp;
    }
  }

  public static class TimeStamp {
    int hour = 0;
    int second = 0;
    public TimeStamp() {}                     // required to read
    public TimeStamp(int hour, int second){
      this.hour = hour;
      this.second = second;
    }
  }

  public static void main(String[] args) throws Exception {
    // one argument: a file name
    File file = new File(args[0]);

    // get the reflected schema for packets
    Schema schema = ReflectData.get().getSchema(Packet.class);

    // create a file of packets
    DatumWriter<Packet> writer = new ReflectDatumWriter<Packet>(Packet.class);
    DataFileWriter<Packet> out = new DataFileWriter<Packet>(writer)
      .setCodec(CodecFactory.deflateCodec(9))
      .create(schema, file);

    // write 100 packets to the file, odds with null timestamp
    for (int i = 0; i < 100; i++) {
      out.append(new Packet(i, (i%2==0) ? new TimeStamp(12, i) : null));
    }

    // close the output file
    out.close();

    // open a file of packets
    DatumReader<Packet> reader = new ReflectDatumReader<Packet>(Packet.class);
    DataFileReader<Packet> in = new DataFileReader<Packet>(file, reader);

    // read 100 packets from the file & print them as JSON
    for (Packet packet : in) {
      System.out.println(ReflectData.get().toString(packet));
    }

    // close the input file
    in.close();
  }

}
于 2012-08-08T18:54:38.603 回答