Comparing two arrays and getting the difference in PySpark

Question

I have two array fields in a data frame.

I have a requirement to compare these two arrays and get the difference as an array(new column) in the same data frame.

Expected output is:

Column B is a subset of column A. Also the words is going to be in the same order in both arrays.

Can any one please help me to get a solution for this?

Answer 1

Since Spark 2.4.0, this can be solved easily using array_except. Taking the example

from pyspark.sql import functions as F

#example df
df=sqlContext.createDataFrame(pd.DataFrame(data=[[["hello", "world"], 
["world"]],[["sample", "overflow", "text"], ["sample", "text"]]], columns=["A", "B"]))


df=df.withColumn('difference', F.array_except('A', 'B'))

for more similar operations on arrays, I suggest this blogpost https://www.waitingforcode.com/apache-spark-sql/apache-spark-2.4.0-features-array-higher-order-functions/read

Answer 2

You can use a user-defined function. My example dataframe differs a bit from yours, but the code should work fine:

import pandas as pd
from pyspark.sql.types import *

#example df
df=sqlContext.createDataFrame(pd.DataFrame(data=[[["hello", "world"], 
["world"]],[["sample", "overflow", "text"], ["sample", "text"]]], columns=["A", "B"]))

# define udf
differencer=udf(lambda x,y: list(set(x)-set(y)), ArrayType(StringType()))
df=df.withColumn('difference', differencer('A', 'B'))

EDIT:

This does not work if there are duplicates as set retains only uniques. So you can amend the udf as follows:

differencer=udf(lambda x,y: [elt for elt in x if elt not in y] ), ArrayType(StringType()))