Neeraj Bhadani NeerajBhadani

## arr_array_distinct.scala
val arr_distinct_df = df.withColumn("result", array_distinct($"array_col2"))

arr_distinct_df.show()

## arr_array_repeat.scala
val arr_repeat_df = df.withColumn("result", array_repeat($"array_col2", 2))

arr_repeat_df.show(truncate = false)

## arr_array_sort.scala
val arr_sort_df = df.withColumn("result", array_sort($"array_col2"))

arr_sort_df.show()

## arr_array_union.scala
val arr_union_df = full_df
                    .withColumn("result", array_union($"array_col1", $"array_col2"))
                    .drop("col1")

arr_union_df.show(truncate=false)

## arr_arrays_overlap.scala
val arr_overlap_df = full_df
                      .withColumn("result", arrays_overlap($"array_col1", $"array_col2"))

arr_overlap_df.show()

## arr_arrays_zip.scala
// remove element "2" from array column "array_col2"
val temp_df = full_df
               .withColumn("new_array_col", array_remove($"array_col2",2))

// zip column "array_col1" with newly created column "new_array_col"
val arr_zip_df = temp_df
                  .withColumn("result", arrays_zip($"array_col1", $"new_array_col"))
                  .select("array_col1", "new_array_col", "result")

arr_zip_df.show(truncate=false)

## arr_zip_schema.scala
arr_zip_df.select("result").printSchema()

## arr_concat.scala
val arr_cat_df = full_df.withColumn("result", concat($"array_col1", $"array_col2"))

arr_cat_df.show(truncate=false)

## arr_element_at.scala
val arr_element_at_df = df.withColumn("result", element_at($"array_col2", 1))

arr_element_at_df.show()

## arr_flatten.scala
// Generate the nested array using the function "array_repeat".
val arr_repeat_df = df.withColumn("repeat", array_repeat($"array_col2", 2))

// flatten the nested array.
val arr_flat_df = arr_repeat_df
                   .withColumn("result", flatten($"repeat")).select("repeat", "result")
                   .select("repeat", "result")

arr_flat_df.show(truncate=false)
	val arr_distinct_df = df.withColumn("result", array_distinct($"array_col2"))

	arr_distinct_df.show()
	val arr_repeat_df = df.withColumn("result", array_repeat($"array_col2", 2))

	arr_repeat_df.show(truncate = false)
	val arr_sort_df = df.withColumn("result", array_sort($"array_col2"))

	arr_sort_df.show()
	val arr_union_df = full_df
	.withColumn("result", array_union($"array_col1", $"array_col2"))
	.drop("col1")

	arr_union_df.show(truncate=false)
	val arr_overlap_df = full_df
	.withColumn("result", arrays_overlap($"array_col1", $"array_col2"))

	arr_overlap_df.show()
	// remove element "2" from array column "array_col2"
	val temp_df = full_df
	.withColumn("new_array_col", array_remove($"array_col2",2))

	// zip column "array_col1" with newly created column "new_array_col"
	val arr_zip_df = temp_df
	.withColumn("result", arrays_zip($"array_col1", $"new_array_col"))
	.select("array_col1", "new_array_col", "result")

	arr_zip_df.show(truncate=false)
	val arr_cat_df = full_df.withColumn("result", concat($"array_col1", $"array_col2"))

	arr_cat_df.show(truncate=false)
	val arr_element_at_df = df.withColumn("result", element_at($"array_col2", 1))

	arr_element_at_df.show()
	// Generate the nested array using the function "array_repeat".
	val arr_repeat_df = df.withColumn("repeat", array_repeat($"array_col2", 2))

	// flatten the nested array.
	val arr_flat_df = arr_repeat_df
	.withColumn("result", flatten($"repeat")).select("repeat", "result")
	.select("repeat", "result")

	arr_flat_df.show(truncate=false)