如何将 scala 地图数组转换为 Spark df

Question

I having scala array of type Array[Map[String,String]] and i want to convert it into spark df.

input:- Array(Map("col1" -> "val1"), Map("col2" -> "val2", "col1" -> "val1"), Map("col3" -> "val3") )

expected output:-

Spark df

col1	col2	col3
val1	NA	NA
val1	val2	NA
NA	NA	val3

What is best way to do this?

Answer 1

Having:

val input = Seq(Map("col1" -> "val1"), Map("col2" -> "val2", "col1" -> "val1"), Map("col3" -> "val3") )

you can select values and keys like:

val values = input.map(_.values.toSeq)
val keys = input.flatMap(_.keys.toSeq).distinct

and then use spark implicit

import spark.implicits._
values.toDF(keys: _*)

I also THINK you can also do it this way (not sure though)

import spark.implicits._
input.toDF()