将具有 JSON 个对象数组的 Spark 数据框列转换为多行

Question

我有一个流式JSON数据，其结构可以用下面的案例class来描述

case class Hello(A: String, B: Array[Map[String, String]])

相同的示例数据如下

|  A    | B                                        |
|-------|------------------------------------------|
|  ABC  |  [{C:1, D:1}, {C:2, D:4}]                | 
|  XYZ  |  [{C:3, D :6}, {C:9, D:11}, {C:5, D:12}] |

我想将其转换为

|   A   |  C  |  D   |
|-------|-----|------|
|  ABC  |  1  |  1   |
|  ABC  |  2  |  4   |
|  XYZ  |  3  |  6   |
|  XYZ  |  9  |  11  |
|  XYZ  |  5  |  12  |

我们将不胜感激。

Answer 1

不确定这是否是最佳方法，但分两步即可完成。撇开你的案例 class 不谈，以下是：

import org.apache.spark.sql.functions._
//case class ComponentPlacement(A: String, B: Array[Map[String, String]])
val df = Seq (
              ("ABC", List(Map("C" -> "1",  "D" -> "2"))),
              ("XYZ", List(Map("C" -> "11", "D" -> "22")))
             ).toDF("A", "B")

val df2 = df.select($"A", explode($"B")).toDF("A", "Bn")

val df3 = df2.select($"A", explode($"Bn")).toDF("A", "B", "C")

val df4 = df3.select($"A", $"B", $"C").groupBy("A").pivot("B").agg(first($"C"))

returns:

+---+---+---+
|  A|  C|  D|
+---+---+---+
|XYZ| 11| 22|
|ABC|  1|  2|
+---+---+---+

Answer 2

随着问题的演变，我将原始答案留在那里，这解决了最后一个问题。

Important point, the input mentioned as follows is now catered for:

val df0 = Seq (
            ("ABC", List(Map("C" -> "1", "D" -> "2"), Map("C" -> "3", "D" -> "4"))),
            ("XYZ", List(Map("C" -> "44", "D" -> "55"), Map("C" -> "188", "D" -> "199"), Map("C" -> "88", "D" -> "99")))
              )
             .toDF("A", "B")

Can also be done like this, but then the script needs to be modified for this, although trivial:

val df0 = Seq (
           ("ABC", List(Map("C" -> "1",  "D" -> "2"))), 
           ("ABC", List(Map("C" -> "44", "D" -> "55"))),
           ("XYZ", List(Map("C" -> "11", "D" -> "22")))
              )
            .toDF("A", "B")

Following on from requested format then:

val df1 = df0.select($"A", explode($"B")).toDF("A", "Bn")

val df2 = df1.withColumn("SeqNum", monotonically_increasing_id()).toDF("A", "Bn", "SeqNum") 

val df3 = df2.select($"A", explode($"Bn"), $"SeqNum").toDF("A", "B", "C", "SeqNum")

val df4 = df3.withColumn("dummy", concat( $"SeqNum", lit("||"), $"A"))

val df5 = df4.select($"dummy", $"B", $"C").groupBy("dummy").pivot("B").agg(first($"C")) 

val df6 = df5.withColumn("A", substring_index(col("dummy"), "||", -1)).drop("dummy")

df6.show(false)

returns:

+---+---+---+
|C  |D  |A  |
+---+---+---+
|3  |4  |ABC|
|1  |2  |ABC|
|88 |99 |XYZ|
|188|199|XYZ|
|44 |55 |XYZ|
+---+---+---+

您可以对列重新排序。

将具有 JSON 个对象数组的 Spark 数据框列转换为多行

Convert an Spark dataframe columns with an array of JSON objects to multiple rows

apache-spark

spark-streaming

apache-spark-sql