How to set an empty struct with all fields null, null in spark-CodePudding

I have this dataframe:

 ---- -------------------------------- 
|name|dates                           |
 ---- -------------------------------- 
|A   |[[1994, 12, 11], [,,]]          |
|B   |[[1994, 12, 11], [1994, 12, 15]]|
 ---- --------------------------------

with this schema:

root
 |-- name: string (nullable = true)
 |-- dates: struct (nullable = true)
 |    |-- start_date: struct (nullable = true)
 |    |    |-- year: integer (nullable = true)
 |    |    |-- month: integer (nullable = true)
 |    |    |-- day: integer (nullable = true)
 |    |-- end_date: struct (nullable = true)
 |    |    |-- year: integer (nullable = true)
 |    |    |-- month: integer (nullable = true)
 |    |    |-- day: integer (nullable = true)

I want to have this as output when all fields inside end_date are null, set end date as null

 ---- -------------------------------- 
|name|dates                           |
 ---- -------------------------------- 
|A   |[[1994, 12, 11],]               |
|B   |[[1994, 12, 11], [1994, 12, 15]]|
 ---- --------------------------------

CodePudding user response：

You can update the struct column dates by recreating a new struct from the existing attributes and use when expression to check if all end_dates attributes are null:

val df2 = df.withColumn(
  "dates",
  struct(
    col("dates.start_date"), // keep start_date
    when(
      Seq("year", "month", "day")
        .map(x => col(s"dates.end_date.$x").isNull)
        .reduce(_ and _),
      lit(null).cast("struct<year:int,month:int,day:int>")
    ).otherwise(col("dates.end_date")).alias("end_date") // set end_date to null if all attr are null
  )
)

df2.show(false)
// ---- -------------------------------- 
//|name|dates                           |
// ---- -------------------------------- 
//|A   |[[1994, 12, 11],]               |
//|B   |[[1994, 12, 11], [1994, 12, 25]]|
// ---- --------------------------------