In this blog, I will discuss a problem that I faced recently. One thing to keep in mind is that this problem is specifically related to Spark version 2.0.0. Other than this version, this problem does not occur.

Problem: Spark code was reading a CSV file. This particular CSV file had one timestamp column that could have null values, as well. So when Spark tried to read the CSV, it was throwing an error whenever it got null values for the timestamp field. I needed the solution that could handle null timestamp fields.

You can find the code snippet below:

import org.apache.spark.sql.SparkSession
 
object CsvReader extends App {
 
val sparkSession = SparkSession.builder()
 .master("local")
 .appName("POC")
 .getOrCreate()
 
val df = sparkSession.read
 .format("com.databricks.spark.csv")
 .option("header", "true")
 .option("inferSchema", "true")
 .load("test.csv")
 
df.printSchema()
 df.show()
}

You can see easily that the above code is inferring the schema while reading the CSV file.

Solution: To solve the above problem, we need to follow the below approach:

1. Provide a custom schema where the timestamp field must be read as a string type.
2. Cast the timestamp field explicitly.

By using the above approach, we can solve the null timestamp field issue. But there is one thing to notice that we must know first: the field for the timestamp in CSV and the schema for the whole CSV file. Only then will we be able to cast that field from the string to the timestamp explicitly and while maintaining the original schema for the file.

In my case, I am taking below CSV file test.csv:

The schema for the CSV file is as:

• ID: String.

• PHONE: Integer.

• BIRTH_DT: Timestamp.

The solution code must be as follows:

import org.apache.spark.sql.SparkSession
import org.apache.spark.sql.functions.{col, unix_timestamp}
import org.apache.spark.sql.types._

object CsvReader extends App {

  val sparkSession = SparkSession.builder()
    .master("local")
    .appName("POC")
    .getOrCreate()

  val schema = StructType(List(
    StructField("ID", StringType),
    StructField("PHONE", IntegerType),
    StructField("BIRTH_DT", StringType)
  ))

  val df = sparkSession.read
    .format("com.databricks.spark.csv")
    .schema(schema)
    .option("header", "true")
    .load("test.csv")

  val columnName = "BIRTH_DT"
  val updatedDF = df.withColumn(columnName, unix_timestamp(col(columnName), "yyyy-MM-dd HH:mm:ss").cast("timestamp"))

  updatedDF.printSchema()
  updatedDF.show()
}

That’s it. I hope this blog is helpful to you!