chore: avoid the double evaluation of entityId in ClusterSharding

cluster-sharding-typed/src/main/scala/org/apache/pekko/cluster/sharding/typed/internal/ClusterShardingImpl.scala

@@ -20,6 +20,7 @@ import java.util.concurrent.CompletionStage
 import java.util.concurrent.ConcurrentHashMap
 
 import scala.concurrent.Future
+import scala.runtime.AbstractPartialFunction
 
 import org.apache.pekko
 import pekko.actor.ActorRefProvider
@@ -173,10 +174,16 @@ import pekko.util.JavaDurationConverters._
       allocationStrategy: Option[ShardAllocationStrategy]): ActorRef[E] = {
 
     val extractorAdapter = new ExtractorAdapter(extractor)
-    val extractEntityId: ShardRegion.ExtractEntityId = {
-      // TODO is it possible to avoid the double evaluation of entityId
-      case message if extractorAdapter.entityId(message) != null =>
-        (extractorAdapter.entityId(message), extractorAdapter.unwrapMessage(message))
+    // !!!important is only applicable if you know that isDefinedAt(x) is always called before apply(x) (with the same x)
+    val extractEntityId: ShardRegion.ExtractEntityId = new AbstractPartialFunction[Any, (String, Any)] {
+      var cache: String = _
+
+      override def isDefinedAt(msg: Any): Boolean = {
+        cache = extractorAdapter.entityId(msg)
+        cache != null
+      }
+
+      override def apply(x: Any): (DataCenter, Any) = (cache, extractorAdapter.unwrapMessage(x))
     }
     val extractShardId: ShardRegion.ExtractShardId = { message =>
       extractorAdapter.entityId(message) match {

cluster-sharding/src/main/scala/org/apache/pekko/cluster/sharding/ClusterSharding.scala

@@ -19,6 +19,7 @@ import java.util.concurrent.ConcurrentHashMap
 
 import scala.collection.immutable
 import scala.concurrent.Await
+import scala.runtime.AbstractPartialFunction
 import scala.util.control.NonFatal
 
 import org.apache.pekko
@@ -429,15 +430,26 @@ class ClusterSharding(system: ExtendedActorSystem) extends Extension {
       typeName,
       _ => entityProps,
       settings,
-      extractEntityId = {
-        case msg if messageExtractor.entityId(msg) ne null =>
-          (messageExtractor.entityId(msg), messageExtractor.entityMessage(msg))
-      },
+      extractEntityId = extractEntityIdFromExtractor(messageExtractor),
       extractShardId = msg => messageExtractor.shardId(msg),
       allocationStrategy = allocationStrategy,
       handOffStopMessage = handOffStopMessage)
   }
 
+  // !!!important is only applicable if you know that isDefinedAt(x) is always called before apply(x) (with the same x)
+  private def extractEntityIdFromExtractor(
+      messageExtractor: ShardRegion.MessageExtractor): ShardRegion.ExtractEntityId =
+    new AbstractPartialFunction[Any, (String, Any)] {
+      var cache: String = _
+
+      override def isDefinedAt(msg: Any): Boolean = {
+        cache = messageExtractor.entityId(msg)
+        cache != null
+      }
+
+      override def apply(x: Any): (DataCenter, Any) = (cache, messageExtractor.entityMessage(x))
+    }
+
   /**
    * Java/Scala API: Register a named entity type by defining the [[pekko.actor.Props]] of the entity actor
    * and functions to extract entity and shard identifier from messages. The [[ShardRegion]] actor
@@ -612,11 +624,12 @@ class ClusterSharding(system: ExtendedActorSystem) extends Extension {
       dataCenter: Optional[String],
       messageExtractor: ShardRegion.MessageExtractor): ActorRef = {
 
-    startProxy(typeName, Option(role.orElse(null)), Option(dataCenter.orElse(null)),
-      extractEntityId = {
-        case msg if messageExtractor.entityId(msg) ne null =>
-          (messageExtractor.entityId(msg), messageExtractor.entityMessage(msg))
-      }, extractShardId = msg => messageExtractor.shardId(msg))
+    startProxy(
+      typeName,
+      Option(role.orElse(null)),
+      Option(dataCenter.orElse(null)),
+      extractEntityId = extractEntityIdFromExtractor(messageExtractor),
+      msg => messageExtractor.shardId(msg))
 
   }