试试这个-

加载测试数据

val data =
      """
        |{
        |    "parentArray": [
        |        {
        |            "child 1": 0
        |        },
        |        {
        |            "child 1": 1
        |        }
        |    ]
        |}
      """.stripMargin
    val df = spark.read.option("multiLine", true)
      .json(Seq(data).toDS())
    df.show(false)
    df.printSchema()
    /**
      * +-----------+
      * |parentArray|
      * +-----------+
      * |[[0], [1]] |
      * +-----------+
      *
      * root
      * |-- parentArray: array (nullable = true)
      * |    |-- element: struct (containsNull = true)
      * |    |    |-- child 1: long (nullable = true)
      */

更改数组中的列名

val p = df.withColumn("parentArray", col("parentArray").cast("array<struct<new_col: long>>"))
    p.show(false)
    p.printSchema()

    /**
      * +-----------+
      * |parentArray|
      * +-----------+
      * |[[0], [1]] |
      * +-----------+
      *
      * root
      * |-- parentArray: array (nullable = true)
      * |    |-- element: struct (containsNull = true)
      * |    |    |-- new_col: long (nullable = true)
      */