Merge two spark sql columns of type Array[string] into a new Array[string] column

Question

I have two columns in a Spark SQL DataFrame with each entry in either column as an array of strings.

val  ngramDataFrame = Seq(
  (Seq(\"curiou         


        

Answer 1

In Spark 2.4 or later you can use concat (if you want to keep duplicates):

ngramDataFrame.withColumn(
  "full_array", concat($"filtered_words", $"ngrams_array")
).show

+--------------------+---------------+--------------------+
|      filtered_words|   ngrams_array|          full_array|
+--------------------+---------------+--------------------+
|[curious, bought,...|[iwa, was, asj]|[curious, bought,...|
+--------------------+---------------+--------------------+

or array_union (if you want to drop duplicates):

ngramDataFrame.withColumn(
  "full_array",
   array_union($"filtered_words", $"ngrams_array")
)

These can be also composed from the other higher order functions, for example

ngramDataFrame.withColumn(
   "full_array",
   flatten(array($"filtered_words", $"ngrams_array"))
)

with duplicates, and

ngramDataFrame.withColumn(
   "full_array",
   array_distinct(flatten(array($"filtered_words", $"ngrams_array")))
)

without.

On a side note, you shouldn't use WrappedArray when working with ArrayType columns. Instead you should expect the guaranteed interface, which is Seq. So the udf should use function with following signature:

(Seq[String], Seq[String]) => Seq[String]

Please refer to SQL Programming Guide for details.