我们正在尝试将protobuf与Akka结合使用,并通过protobuf序列化所有Akka消息。对于Scala,我们有一个名为ScalaPB的库,可帮助我们生成该类,其中包括诸如parseFrom
,toByteArray
等方法,用于对数据进行序列化或反序列化。但是,当我们尝试运行该程序时,出现以下异常:
akka.actor.dungeon.SerializationCheckFailedException: Failed to serialize and deserialize message of type com.knoldus.akkaserialization.msg.example.Bang$ for testing. To avoid this error, either disable 'akka.actor.serialize-messages', mark the message with 'akka.actor.NoSerializationVerificationNeeded', or configure serialization to support this message
application.conf文件包含以下配置:
akka {
actor {
allow-java-serialization = off
serialize-messages = on
serializers {
proto = "akka.remote.serialization.ProtobufSerializer"
}
serialization-bindings {
"com.knoldus.akkaserialization.msg.example.Bang" = proto
"com.knoldus.akkaserialization.msg.example.Message" = proto
}
}
}
这些类com.knoldus.akkaserialization.msg.example.Bang
和com.knoldus.akkaserialization.msg.example.Message
通过ScalaPB生成,并包含所有require方法。
akka.remote.serialization.ProtobufSerializer
的源代码定义
This Serializer serializes `akka.protobuf.Message` and `com.google.protobuf.Message` It is using reflection to find the `parseFrom` and `toByteArray` methods to avoid dependency to `com.google.protobuf`
因此,我们希望这会自动读取我们的案例类Bang
和Message
并执行序列化,但不幸的是遇到了序列化异常。
您能帮助找出ScalaPB和ProtoBuff的确切问题吗?
答案 0 :(得分:1)
您正在使用的序列化程序旨在与Java protobufs一起使用,而不与ScalaPB一起使用。您需要包括自己的序列化器。这是您可以使用的一个:
package com.example.protoser
import java.util.concurrent.atomic.AtomicReference
import akka.actor.ExtendedActorSystem
import akka.serialization.BaseSerializer
import scalapb.GeneratedMessageCompanion
class ScalaPbSerializer(val system: ExtendedActorSystem) extends BaseSerializer {
private val classToCompanionMapRef = new AtomicReference[Map[Class[_], GeneratedMessageCompanion[_]]](Map.empty)
override def toBinary(o: AnyRef): Array[Byte] = o match {
case e: scalapb.GeneratedMessage => e.toByteArray
case _ => throw new IllegalArgumentException("Need a subclass of scalapb.GeneratedMessage")
}
override def includeManifest: Boolean = true
override def fromBinary(bytes: Array[Byte], manifest: Option[Class[_]]): AnyRef = {
manifest match {
case Some(clazz) =>
@scala.annotation.tailrec
def messageCompanion(companion: GeneratedMessageCompanion[_] = null): GeneratedMessageCompanion[_] = {
val classToCompanion = classToCompanionMapRef.get()
classToCompanion.get(clazz) match {
case Some(cachedCompanion) => cachedCompanion
case None =>
val uncachedCompanion =
if (companion eq null) Class.forName(clazz.getName + "$", true, clazz.getClassLoader)
.getField("MODULE$").get().asInstanceOf[GeneratedMessageCompanion[_]]
else companion
if (classToCompanionMapRef.compareAndSet(classToCompanion, classToCompanion.updated(clazz, uncachedCompanion)))
uncachedCompanion
else
messageCompanion(uncachedCompanion)
}
}
messageCompanion().parseFrom(bytes).asInstanceOf[AnyRef]
case _ => throw new IllegalArgumentException("Need a ScalaPB companion class to be able to deserialize.")
}
}
}
配置应如下所示:
akka {
actor {
serializers {
scalapb = "com.example.protoser.ScalaPbSerializer"
}
serialization-bindings {
"scalapb.GeneratedMessage" = scalapb
}
serialization-identifiers {
"com.example.protoser.ScalaPbSerializer" = 10000
}
}
}
以上是旧代码的适配器,因此欢迎修改和建议!