单元测试火花数据帧转换链接

Question

我是 scala spark 生态系统的新手，想知道对链式数据帧进行单元测试的最佳方法是什么transformation.So这是我想测试的方法的代码示例

def writeToParquet(spark: SparkSession, dataFrame: DataFrame, col1: DataType1, col2:DataType2): Unit {
    dataFrame
        .withColumn("date", some_columnar_date_logic)
        .withColumn("hour", some_more_functional_logic)
        .... //couple more transformation logic
        .write
        .mode(SaveMode.Append)
        .partitionBy("col1", "col2", "col3")
        .parquet("some hdfs/s3/url")        
}

问题是 parquet 是 Unit return 类型，这使得测试变得困难。转换本质上是不可变的，这使得模拟和间谍有点困难

，这一事实进一步放大了这个问题

为了创建数据框，我在 csv 中转储了测试数据集

Answer 1

请找到数据框架单元测试的简单示例。你可以把它分成两部分。第一的。测试转换，你可以做简单的shell脚本来测试写入的文件

import com.holdenkarau.spark.testing._
import org.apache.spark.sql.{DataFrame, Row}
import org.apache.spark.sql.functions.lit
import org.apache.spark.sql.types.{IntegerType, StringType, StructField, StructType}
import org.scalatest.{FunSuite, Matchers}

class SomeDFTest extends FunSuite with Matchers with DataFrameSuiteBase    {
 import spark.implicits._

  test("Testing Input customer data date transformation") {


    val inputSchema = List(
      StructField("number", IntegerType, false),
      StructField("word", StringType, false)
    )
    val expectedSchema = List(
      StructField("number", IntegerType, false),
      StructField("word", StringType, false),
      StructField("dummyColumn", StringType, false)

    )
    val inputData = Seq(
      Row(8, "bat"),
      Row(64, "mouse"),
      Row(-27, "horse")
    )

    val expectedData = Seq(
      Row (8, "bat","test"),
      Row(64, "mouse","test"),
      Row(-27, "horse","test")
    )

    val inputDF = spark.createDataFrame(
      spark.sparkContext.parallelize(inputData),
      StructType(inputSchema)
    )

    val expectedDF = spark.createDataFrame(
      spark.sparkContext.parallelize(expectedData),
      StructType(expectedSchema)
    )


    val actual = transformSomeDf(inputDF)

    assertDataFrameEquals(actual, expectedDF) // equal



  }

  def transformSomeDf(df:DataFrame):DataFrame={
    df.withColumn("dummyColumn",lit("test"))
  }
}

Sbt.build 配置

name := "SparkTest"

version := "0.1"

scalaVersion := "2.11.8"

val sparkVersion = "2.3.0"

libraryDependencies ++= Seq(
  "org.apache.spark" %% "spark-core" % sparkVersion,
  "org.apache.spark" %% "spark-sql" % sparkVersion,
  "org.apache.spark" %% "spark-hive" % sparkVersion % "provided",
"com.holdenkarau" %% "spark-testing-base" % "2.4.0_0.11.0" % Test

)

Answer 2

我在测试数据框时想到的第一件事就是将转换和 IOs

分开

所以对于以上场景我们可以把上面的链分成三部分

class Coordinator {
    def transformAndWrite(dataframe: Dataframe): Unit = {
transformedDf = dataFrame
        .withColumn("date", some_columnar_date_logic)
        .withColumn("hour", some_more_functional_logic)
        .... //couple more transformation logic

partitionedDfWriter = transformedDf.write
        .mode(SaveMode.Append)
        .partitionBy("col1", "col2", "col3")

和

partitionedDfWriter.parquet("some hdfs/s3/url")
}

现在我们可以将它们移动到三个独立的 classes，

DFTransformer , DFPartitioner 和 DataFrameParquetWriter extends ResourceWriter

所以代码会变成这样

class DFTransformer {
    def transform(dataframe:DataFrame): Dataframe = {
        return dataFrame
        .withColumn("date", some_columnar_date_logic)
        .withColumn("hour", some_more_functional_logic)
        .... //couple more transformation logic

}

class DfPartitioner {
    def partition(dataframe: DataFrame): DataFrameWriter = {
        return dataframe.write
        .mode(SaveMode.Append)
        .partitionBy("col1", "col2", "col3")
    }
}

和

class DataFrameParquetWriter extends ResourceWriter {
    overide def write(partitionedDfWriter: DataFrameWriter) = {
       partitionedDfWriter.parquet("some hdfs/s3/url") 

    }

class Coordinator(dfTransformer:DfTransformer, dfPartitioner: DFPartitioner, resourceWriter: ResourceWriter) {
    val transformedDf = dfTransformer.transform(dataframe)
    val partitionedDfWriter = dfPartitioner.partition(transformedDf)
    resourceWriter.write(partitionedDfWriter)
}

上面的优点是，当你必须测试你的 Coordinator class 时，你可以很容易地使用 Mockito 来模拟你的依赖关系。
现在测试DFTransformer也很容易，您可以传递存根数据帧并断言返回的数据帧。（使用 spark-testing-base）。我们还可以测试转换返回的列。我们还可以测试count

单元测试火花数据帧转换链接

Unit Testing spark dataframes transformation chaining

unit-testing

scala

apache-spark

parquet

apache-spark-sql