在没有本地迭代器的情况下,如何返回spark rdd分区值?

我正在学习Spark及其与RDD分区分布相关的并行性。我有一台4 CPU的机器,因此我有4个并行单元。要返回分区索引“0”的成员,在不强制RDD使用本地化器的情况下,我找不到返回此分区的方法。

我习惯于保持简洁。是否有更简洁的方法按分区筛选RDD?以下两种方法是可行的,但似乎很笨拙。

scala> val data = 1 to 20
data: scala.collection.immutable.Range.Inclusive = Range(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20)

scala> val distData = sc.parallelize(data)
distData: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[75] at parallelize at <console>:26

scala> distData.mapPartitionsWithIndex{
   (index,it) => {
      it.toList.map(x => if (index == 0) (x)).iterator
   }
}.toLocalIterator.toList.filterNot(
   _.isInstanceOf[Unit]
)
res107: List[AnyVal] = List(1, 2, 3, 4, 5)

scala> distData.mapPartitionsWithIndex{
   (index,it) => {
      it.toList.map(x => if (index == 0) (x)).iterator
   }
}.toLocalIterator.toList.filter(
   _ match{
      case x: Unit => false
      case x => true
   }
)
res108: List[AnyVal] = List(1, 2, 3, 4, 5)