How can I parse stream of avro byteString in akka stream

Question

I am having Avro files in the s3 bucket and trying to stream and parse into a case class. I have the schema to parse but don't know how to proceed with it.

I use s3.download to download and stream the file from the s3 bucket and then convert it to utf8string.

Please assist, how can I parse with the schema we have considering the input stream I am getting.

Answer 1

I am gonna answer this question based on what you asked to use the schema to (de)serialize messages using Avro.

I have the schema to parse but don't know how to proceed with it.

and assume that you already download messages from s3.buckets. Then I am gonna use my example that I persist messages on PostgreSQL, just for the sake of a working example. But you can assume your s3.bucket connection.

I am using the com.sksamuel.avro4s lib to create my Avro serializer. Here is the necessary libs to put in the build.sbt:

  val akkaVersion = "2.6.10"
  "com.typesafe.akka" %% "akka-actor" % akkaVersion,
  "com.typesafe.akka" %% "akka-persistence" % akkaVersion,
  "com.sksamuel.avro4s" %% "avro4s-core" % "4.0.4",
  "org.xerial.snappy" % "snappy-java" % "1.1.8.2",
  "org.postgresql" % "postgresql" % "42.2.2",
  "com.github.dnvriend" %% "akka-persistence-jdbc" % "3.4.0",

Then you create your serializer, which in my case is the MyFirstAvroSerializer, extending akka.serialization.Serializer. It has a schema that in my case is the case class CompanyRegistry. Basically, you have to implement the methods identifier that has to have a unique ID, toBinary and fromBinary to convert the messages, and the includeManifest which is false because I don't need a manifest.

import akka.serialization.Serializer
import com.sksamuel.avro4s.{AvroInputStream, AvroOutputStream, AvroSchema}
import com.typesafe.config.ConfigFactory

import java.io.{ByteArrayInputStream, ByteArrayOutputStream}

case class BankAccount(iban: String, bankCode: String, amount: Double, currency: String)
case class CompanyRegistry(name: String, accounts: Seq[BankAccount], activityCode: String, marketCap: Double)

class MyFirstAvroSerializer extends Serializer {
  val schema = AvroSchema[CompanyRegistry]
  override def identifier: Int = 454874
  override def toBinary(o: AnyRef): Array[Byte] = o match {
    case c: CompanyRegistry =>
      val baos = new ByteArrayOutputStream()
      val avroOutputStream = AvroOutputStream.binary[CompanyRegistry].to(baos).build() // schema
      avroOutputStream.write(c)
      avroOutputStream.flush()
      avroOutputStream.close()
      baos.toByteArray
    case _ => throw new IllegalArgumentException(s"we only support CompanyRegistry for Avro")
  }
  override def fromBinary(bytes: Array[Byte], manifest: Option[Class[_]]): AnyRef = {
    val avroInputStream = AvroInputStream.binary[CompanyRegistry].from(new ByteArrayInputStream(bytes)).build(schema)
    val companyRegistryIterator: Iterator[CompanyRegistry] = avroInputStream.iterator
    val companyRegistry = companyRegistryIterator.next()
    avroInputStream.close()
    companyRegistry
  }
  override def includeManifest: Boolean = false
}

Then you have to configure your project to call this serializer when exchanging akka messages among actors. Configure it on the application.conf by adding a specific config. In my case is the avroSerializable. You set the MyFirstAvroSerializer under the serializers scope and the case classes under the serialization-bindings scope. I also configured for Akka-remote but you can ignore it.

avroSerializable {
  akka {
    actor {
      provider = remote
      #allow-java-serialization = off
      serializers {
        java = "akka.serialization.JavaSerializer"
        avro = "org.github.felipegutierrez.explore.akka.classic.remote.serialization.MyFirstAvroSerializer"
      }
      serialization-bindings {
        "org.github.felipegutierrez.explore.akka.classic.remote.serialization.CompanyRegistry" = avro
        "java.io.Serializable" = java
      }
    }
    remote {
      artery {
        enabled = on
        transport = aeron-udp
        canonical.hostname = "localhost"
      }
    }
  }
}

Like I said in the beginning, I am using PostgreSQL. But in your case, it will be the s3 bucket storage configuration. I am gonna leave here just for the sake of completeness and because I call this configuration when I am creating the actor system.

postgresStore {
  akka.persistence.journal.plugin = "jdbc-journal"
  akka.persistence.snapshot-store.plugin = "jdbc-snapshot-store"
  akka.actor.allow-java-serialization = on
  # create JDBC configuration to Akka persistence
  akka-persistence-jdbc {
    shared-databases {
      slick {
        profile = "slick.jdbc.PostgresProfile$"
        db {
          numThreads = 10
          driver = "org.postgresql.Driver"
          url = "jdbc:postgresql://localhost:5432/rtjvm"
          user = "docker"
          password = "docker"
        }
      }
    }
  }
  # dbinding the JDBC plugins with the configureation created above
  jdbc-journal {
    use-shared-db = "slick"
  }
  jdbc-snapshot-store {
    use-shared-db = "slick"
  }
}

Now it is time to create the actor system and a simple actor SimplePersistentActor and send a message over the wire. SimplePersistentActor is just a very simple actor that accepts the messages that I am sending, nothing very fancy.

object AvroSerialization_Persistence {
  def main(args: Array[String]): Unit = {
    val config = ConfigFactory.load().getConfig("postgresStore")
      .withFallback(ConfigFactory.load("avroSerializable"))
    val system = ActorSystem("postgresStoreSystem", config)
    val simplePersistentActor = system.actorOf(SimplePersistentActor.props("avro-actor"), "personAvroActor")
    val companyRegistryMsg = CompanyRegistry(
      "Google",
      Seq(
        BankAccount("US-1234", "google-bank", 4.3, "gazillion dollars"),
        BankAccount("GB-4321", "google-bank", 0.5, "trillion pounds")
      ),
      "ads",
      523895
    )
    simplePersistentActor ! companyRegistryMsg
  }
}