How to convert array of array (string type) to struct - Spark/Scala?

To create an array of structs given an array of arrays of strings, you can use struct function to build a struct given a list of columns combined with element_at function to extract column element at a specific index of an array.

To solve your specific problem, as you correctly stated you need to do two things:

• First, transform your string to an array of arrays of strings
• Then, use this array of arrays of strings to build your struct

In Spark 3.0 and greater

Using Spark 3.0, we can perform all those steps using spark built-in functions.

For the first step, I would do as follows:

• first remove [[ and ]] from family_name string using regexp_replace function
• then, create first array level by splitting this string using split function
• then, create second array level by splitting each element of previous array using transform and split functions

And for the second step, use struct function to build a struct, picking element in arrays using element_at function.

Thus, complete code using Spark 3.0 and greater would be as follows, with data as input dataframe:

import org.apache.spark.sql.functions.{col, element_at, regexp_replace, split, struct, transform}

val result = data
  .withColumn(
    "family_name", 
    transform( 
      split( // first level split
        regexp_replace(col("family_name"), "\\[\\[|]]", ""), // remove [[ and ]]
        "],\\["
      ), 
      x => split(x, ",") // split for each element in first level array
    )
  )
  .withColumn("family_name", transform(col("family_name"), x => struct(
    element_at(x, 1).as("f_name"), // index starts at 1
    element_at(x, 2).as("l_name"),
    element_at(x, 3).as("status"),
    element_at(x, -1).as("ph_no"), // get last element of array
  )))

In Spark 2.X

Using Spark 2.X, we have to rely on an user-defined function. First, we need to define a case class that represent our struct:

case class FamilyName(
  f_name: String, 
  l_name: String, 
  status: String, 
  ph_no: String
)

Then, we define our user-defined function and apply it to our input dataframe:

import org.apache.spark.sql.functions.{col, udf}

val extract_array = udf((familyName: String) => familyName
  .replaceAll("\\[\\[|]]", "")
  .split("],\\[")
  .map(familyName => {
    val explodedFamilyName = familyName.split(",", -1)
    FamilyName(
      f_name = explodedFamilyName(0),
      l_name = explodedFamilyName(1),
      status = explodedFamilyName(2),
      ph_no = explodedFamilyName(explodedFamilyName.length - 1)
    )
  })
)

val result = data.withColumn("family_name", extract_array(col("family_name")))

Result

If you have the following data dataframe:

+---------------------------------------------------------------+---+
|family_name                                                    |id |
+---------------------------------------------------------------+---+
|[[John, Doe, Married, 999-999-9999],[Jane, Doe, Married,Wife,]]|id1|
|[[Tom, Riddle, Single, 888-888-8888]]                          |id2|
+---------------------------------------------------------------+---+

You get the following result dataframe:

+-----------------------------------------------------------------+---+
|family_name                                                      |id |
+-----------------------------------------------------------------+---+
|[{John,  Doe,  Married,  999-999-9999}, {Jane,  Doe,  Married, }]|id1|
|[{Tom,  Riddle,  Single,  888-888-8888}]                         |id2|
+-----------------------------------------------------------------+---+

having the following schema:

root
 |-- family_name: array (nullable = true)
 |    |-- element: struct (containsNull = false)
 |    |    |-- f_name: string (nullable = true)
 |    |    |-- l_name: string (nullable = true)
 |    |    |-- status: string (nullable = true)
 |    |    |-- ph_no: string (nullable = true)
 |-- id: string (nullable = true)