[SPARK-27134] array_distinct function does not work correctly with columns containing array of array

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.4.0
• Fix Version/s: 2.4.1, 3.0.0
• Component/s: SQL
• Labels: correctness

Environment

Spark 2.4, scala 2.11.11

• Assignee: Dilip Biswal
• Reporter: Mike Trenaman
• Created: 12/Mar/19 10:39
• Updated: 17/Jul/19 00:44
• Resolved: 16/Mar/19 19:33

Description

The array_distinct function introduced in spark 2.4 is producing strange results when used on an array column which contains a nested array. The resulting output can still contain duplicate values, and furthermore, previously distinct values may be removed.

This is easily repeatable, e.g. with this code:

val df = Seq(
Seq(Seq(1, 2), Seq(1, 2), Seq(1, 2), Seq(3, 4), Seq(4, 5))
).toDF("Number_Combinations")

val dfWithDistinct = df.withColumn("distinct_combinations",
array_distinct(col("Number_Combinations")))

 

The initial 'df' DataFrame contains one row, where column 'Number_Combinations' contains the following values:

[[1, 2], [1, 2], [1, 2], [3, 4], [4, 5]]

 

The array_distinct function run on this column produces a new column containing the following values:

[[1, 2], [1, 2], [1, 2]]

 

As you can see, this contains three occurrences of the same value (1, 2), and furthermore, the distinct values (3, 4), (4, 5) have been removed.

 

 

Issue Links

• is caused by: SPARK-23912 High-order function: array_distinct(x) → array
• links to: GitHub Pull Request #24073