How to create a udf in PySpark which returns an array of strings?

Question

I have a udf which returns a list of strings. this should not be too hard. I pass in the datatype when executing the udf since it returns an array of strings: ArrayType(StringType).

Now, somehow this is not working:

the dataframe i'm operating on is df_subsets_concat and looks like this:

df_subsets_concat.show(3,False)

+----------------------+
|col1                  |
+----------------------+
|oculunt               |
|predistposed          |
|incredulous           |
+----------------------+
only showing top 3 rows

and the code is

from pyspark.sql.types import ArrayType, FloatType, StringType

my_udf = lambda domain: ['s','n']
label_udf = udf(my_udf, ArrayType(StringType))
df_subsets_concat_with_md = df_subsets_concat.withColumn('subset', label_udf(df_subsets_concat.col1))

and the result is

/usr/lib/spark/python/pyspark/sql/types.py in __init__(self, elementType, containsNull)
    288         False
    289         """
--> 290         assert isinstance(elementType, DataType), "elementType should be DataType"
    291         self.elementType = elementType
    292         self.containsNull = containsNull

AssertionError: elementType should be DataType

It is my understanding that this was the correct way to do this. Here are some resources: pySpark Data Frames "assert isinstance(dataType, DataType), "dataType should be DataType" How to return a "Tuple type" in a UDF in PySpark?

But neither of these have helped me resolve why this is not working. i am using pyspark 1.6.1.

How to create a udf in pyspark which returns an array of strings?

Answer 1

You need to initialize a StringType instance:

label_udf = udf(my_udf, ArrayType(StringType()))
#                                           ^^ 
df.withColumn('subset', label_udf(df.col1)).show()
+------------+------+
|        col1|subset|
+------------+------+
|     oculunt|[s, n]|
|predistposed|[s, n]|
| incredulous|[s, n]|
+------------+------+