PySpark - Sum values inside a struct

Question

I have a schema like this :

root
 |-- DataColumn1: struct (nullable = true)
 |    |-- colA: double (nullable = true)
 |    |-- colB: struct (nullable = true)
 |    |    |-- fieldA: double (nullable = true)
 |    |    |-- fieldB: double (nullable = true)
 |    |    |-- fieldC: double (nullable = true)
 |    |-- colC: long (nullable = true)
 |    |-- colD: string (nullable = true)
 |-- DataColumn2: string (nullable = true)
 |-- DataColumn3: string (nullable = true)

My goal is to create a new column say 'DataColumn4' which is the sum of all the fields 'fieldA', 'fieldB' and 'fieldC' (fieldA + fieldB + fieldC) inside the struct 'colB' which is inside 'DataColumn1'.

There could be N number of fields inside 'colB' so how do I sum them all without accessing the fields one by one through DataColumn1.colB.fieldA, DataColumn1.colB.fieldB and so on?

Example data:

DataColumn1                 DataColumn2             DataColumn3
(1, (1, 2, 3), 4, 5)           XXX                      YYY
(1, (2, 3, 3), 8, 9)           XYZ                      XYX

My expected result must have a new column that is a summation of the nested fields

DataColumn1             DataColumn2             DataColumn3.   DataColumn4
(1, (1, 2, 3), 4, 5)       XXX                      YYY.         6 (since 1+2+3 = 6)
(1, (2, 3, 3), 8, 9)       XYZ                      XYX          8 (since 2+3+3 = 8)

How do I write a code for this in PySpark preferably without a PandasUDF?

tip - you can access the fields inside a struct with `struct_column.struct_field`. e.g. - `DataColumn1.colB.fieldA` — samkart, Aug 27 '22 at 10:51
You're right, thanks. Any idea how to sum those fields? That's where I'm stuck — AnonymousMe, Aug 27 '22 at 10:52
do you want fielda+fieldb+fieldc, or sum(fielda), sum(fieldb)... — samkart, Aug 27 '22 at 10:54
then access those fields as stated in my tip, and create new column with their addition — samkart, Aug 27 '22 at 11:03
Is there a way to do it without accessing them one by one? There could be 'N' fields, is there a way to sum all the fields of 'colB' by aggregation? — AnonymousMe, Aug 27 '22 at 11:07
why not add these intricate details in the question? members would understand your problem statement better. also, please add a sample data ([mcve]) — samkart, Aug 27 '22 at 11:10

score 2 · Accepted Answer · answered Aug 27 '22 at 11:34

you can access the fields inside a struct with struct_column.struct_field. e.g. - DataColumn1.colB.fieldA. And, you can select all struct fields by using DataColumn1.colB.*.

Here's an example of how you could do the summation. Given the following data.

+--------------------+
|        c1_c2c3c4_c5|
+--------------------+
|{1, {1, 2, 3}, 4, 5}|
|{1, {2, 3, 3}, 8, 9}|
+--------------------+

root
 |-- c1_c2c3c4_c5: struct (nullable = false)
 |    |-- c1: long (nullable = true)
 |    |-- c2c3c4: struct (nullable = false)
 |    |    |-- c2: long (nullable = true)
 |    |    |-- c3: long (nullable = true)
 |    |    |-- c4: long (nullable = true)
 |    |-- c5: long (nullable = true)
 |    |-- c6: long (nullable = true)

To get the sum of those fields, we'll need the fields which can be extracted using a select.

data_sdf.select('c1_c2c3c4_c5.c2c3c4.*').columns
# ['c2', 'c3', 'c4']

Actual summation code

# use reduce to add all struct fields
struct_field_sum = reduce(lambda x, y: x + y, 
                          [func.col('c1_c2c3c4_c5.c2c3c4.'+k) 
                           for k in data_sdf.select('c1_c2c3c4_c5.c2c3c4.*').columns]
                          )
# Column<'((c1_c2c3c4_c5.c2c3c4.c2 + c1_c2c3c4_c5.c2c3c4.c3) + c1_c2c3c4_c5.c2c3c4.c4)'>

data_sdf. \
    withColumn('reqd_sum', struct_field_sum). \
    show()

# +--------------------+--------+
# |        c1_c2c3c4_c5|reqd_sum|
# +--------------------+--------+
# |{1, {1, 2, 3}, 4, 5}|       6|
# |{1, {2, 3, 3}, 8, 9}|       8|
# +--------------------+--------+

score 1 · Answer 2 · answered Aug 27 '22 at 11:57

To achieve this you would have to parse the child fields within your StructType and finally reduce them or you can also sum [native python] by projecting them as a list

Example

structureData = [
    ("James","Smith","36636","M",(3100,200)),
    ("Michael","Rose","40288","M",(4300,200)),
    ("Robert","Williams","42114","M",(1400,300)),
    ("Maria","Jones","39192","F",(5500,300)),
    ("Jen","Brown","39156","F",(5000,600))
  ]
structureSchema = StructType([
         StructField('firstname', StringType(), True),
         StructField('lastname', StringType(), True),
         StructField('id', StringType(), True),
         StructField('gender', StringType(), True),
         StructField('salary', StructType([
             StructField('component1', IntegerType(), True),
             StructField('component2', IntegerType(), True)
             ]))
         ])

sparkDF = sql.createDataFrame(data=structureData,schema=structureSchema)
sparkDF.printSchema()
sparkDF.show(truncate=False)


sparkDF = sparkDF.withColumn('total_salary',reduce(lambda a, b: a + b,
                                        [F.col(f'salary.{c}') for c in sparkDF.schema['salary'].dataType.names ]
                                    ) 
                )\
                .withColumn('total_salary_2'
                            ,sum(sparkDF[f'salary.{c}'] for c in sparkDF.schema['salary'].dataType.names)
                          ) 


sparkDF.show()

+---------+--------+-----+------+-----------+------------+--------------+
|firstname|lastname|   id|gender|     salary|total_salary|total_salary_2|
+---------+--------+-----+------+-----------+------------+--------------+
|    James|   Smith|36636|     M|{3100, 200}|        3300|          3300|
|  Michael|    Rose|40288|     M|{4300, 200}|        4500|          4500|
|   Robert|Williams|42114|     M|{1400, 300}|        1700|          1700|
|    Maria|   Jones|39192|     F|{5500, 300}|        5800|          5800|
|      Jen|   Brown|39156|     F|{5000, 600}|        5600|          5600|
+---------+--------+-----+------+-----------+------------+--------------+

References -

Whats is the correct way to sum different dataframe columns in a list in pyspark?

PySpark - Sum values inside a struct

2 Answers2

Example