如何为Scala可迭代的Spark数据集制作编码器

Question

I'm trying to create a Dataset from a RDD y

Pattern: y: RDD[(MyObj1, scala.Iterable[MyObj2])]

So I created explicitly encoder :

implicit def tuple2[A1, A2](
                              implicit e1: Encoder[A1],
                              e2: Encoder[A2]
                            ): Encoder[(A1,A2)] = Encoders.tuple[A1,A2](e1, e2) 
//Create Dataset
val z = spark.createDataset(y)(tuple2[MyObj1, Iterable[MyObj2]]) 

When I compile this code I don't have an Error but when I try to run it I get this Error :

Exception in thread "main" java.lang.UnsupportedOperationException: No Encoder found for scala.Iterable[org.bean.input.MyObj2]
- field (class: "scala.collection.Iterable", name: "_2")
- root class: "scala.Tuple2"
        at org.apache.spark.sql.catalyst.ScalaReflection$.org$apache$spark$sql$catalyst$ScalaReflection$$serializerFor(ScalaReflection.scala:625)
        at org.apache.spark.sql.catalyst.ScalaReflection$$anonfun$10.apply(ScalaReflection.scala:619)
        at org.apache.spark.sql.catalyst.ScalaReflection$$anonfun$10.apply(ScalaReflection.scala:607)
        at scala.collection.TraversableLike$$anonfun$flatMap$1.apply(TraversableLike.scala:241)
        at scala.collection.TraversableLike$$anonfun$flatMap$1.apply(TraversableLike.scala:241)
        at scala.collection.immutable.List.foreach(List.scala:381)
        at scala.collection.TraversableLike$class.flatMap(TraversableLike.scala:241)
        at scala.collection.immutable.List.flatMap(List.scala:344)
        at org.apache.spark.sql.catalyst.ScalaReflection$.org$apache$spark$sql$catalyst$ScalaReflection$$serializerFor(ScalaReflection.scala:607)
        at org.apache.spark.sql.catalyst.ScalaReflection$.serializerFor(ScalaReflection.scala:438)
        at org.apache.spark.sql.catalyst.encoders.ExpressionEncoder$.apply(ExpressionEncoder.scala:71)
        at org.apache.spark.sql.Encoders$.product(Encoders.scala:275)
        at org.apache.spark.sql.LowPrioritySQLImplicits$class.newProductEncoder(SQLImplicits.scala:233)
        at org.apache.spark.sql.SQLImplicits.newProductEncoder(SQLImplicits.scala:33)

Some explanation for my objects (MyObj1 & MyObj2)
- MyObj1 :

case class MyObj1(
                      id:String,
                      type:String
                  ) 

- MyObj2 :

trait MyObj2 {
  val o_state:Option[String]

  val n_state:Option[String]

  val ch_inf: MyObj1

  val state_updated:MyObj3
}

Any Help please

Answer 1

Spark doesn't provide Encoder for Iterables , so unless you want to use Encoder.kryo or Encoder.java this won't work.

The closest subclass of Iterable for which Spark provides Encoders is Seq , so this is probably the one you should use here. Otherwise refer to How to store custom objects in Dataset?

Answer 2

Try to change declaration to: val y: RDD[(MyObj1, Seq[MyObj2])] and it would work. I checked it for my classes:

case class Key(key: String) {}
case class Value(value: Int) {}

For:

val y: RDD[(Key, Seq[Value])] = sc.parallelize(Map(
  Key("A") -> List(Value(1), Value(2)),
  Key("B") -> List(Value(3), Value(4), Value(5))
).toSeq)

val z = sparkSession.createDataset(y)
z.show()

I got:

+---+---------------+
| _1|             _2|
+---+---------------+
|[A]|     [[1], [2]]|
|[B]|[[3], [4], [5]]|
+---+---------------+

If I change to Iterable I got exception you got.