Invoking Pyspark script from Scala Spark Code

user_5412315 0
Viewed 372

I have a Scala Spark application and want to invoke pySpark/python (pyspark_script.py) for further processing.

There are multiple resources to use Java/Scala code in Python but I am looking for scala->Pyspark

I explored Jython for Scala/Java to include Python code as follows:

PythonInterpreter.initialize(System.getProperties, properties, sysArgs)
val pi = new PythonInterpreter()
pi.execfile("path/to/pyscript/mypysparkscript.py")

I see error that says: "ImportError: No module named pyspark"

Is there any way on how Scala spark can talk to PYSpark with same sparkContext/session?

pyspark apache-spark jython scala
1 Answers

You can run shell commands in scala using process object.

// Spark codes goes here .....
// Call pyspark code 
import sys.process._
"python3 /path/to/python/file.py.!!

To use same session add below line to python file.

spark = SparkSession.builder.getOrCreate()

You can use getActiveSession() method also.

NOTE: Make sure you installed pyspark module. You can do that by using pip3 install pyspark command.

user_87733090
Related