Generate null values for nullable fields (#303)

holdenk · Oct 17, 2019 · 79eef40 · 79eef40
1 parent 4b071e6
commit 79eef40
Show file tree

Hide file tree

Showing 2 changed files with 62 additions and 26 deletions.
diff --git a/core/src/main/1.3/scala/com/holdenkarau/spark/testing/DataframeGenerator.scala b/core/src/main/1.3/scala/com/holdenkarau/spark/testing/DataframeGenerator.scala
@@ -95,20 +95,20 @@ object DataframeGenerator {
       List[Gen[Any]] = {
     val generatorMap = userGenerators.map(
       generator => (generator.columnName -> generator)).toMap
-    (0 until fields.length).toList.map(index => {
-      if (generatorMap.contains(fields(index).name)) {
-        generatorMap.get(fields(index).name) match {
+    fields.toList.map { field =>
+    if (generatorMap.contains(field.name)) {
+        generatorMap.get(field.name) match {
           case Some(gen: Column) => gen.gen
-          case Some(list: ColumnList) => getGenerator(fields(index).dataType, list.gen)
+          case Some(list: ColumnList) => getGenerator(field.dataType, list.gen, nullable = field.nullable)
         }
       }
-      else getGenerator(fields(index).dataType)
-    })
+      else getGenerator(field.dataType, nullable = field.nullable)
+    }
   }
 
   private def getGenerator(
-    dataType: DataType, generators: Seq[ColumnGenerator] = Seq()): Gen[Any] = {
-    dataType match {
+    dataType: DataType, generators: Seq[ColumnGenerator] = Seq(), nullable: Boolean = false): Gen[Any] = {
+    val nonNullGen = dataType match {
       case ByteType => Arbitrary.arbitrary[Byte]
       case ShortType => Arbitrary.arbitrary[Short]
       case IntegerType => Arbitrary.arbitrary[Int]
@@ -121,12 +121,12 @@ object DataframeGenerator {
       case TimestampType => Arbitrary.arbLong.arbitrary.map(new Timestamp(_))
       case DateType => Arbitrary.arbLong.arbitrary.map(new Date(_))
       case arr: ArrayType => {
-        val elementGenerator = getGenerator(arr.elementType)
+        val elementGenerator = getGenerator(arr.elementType, nullable = arr.containsNull)
         Gen.listOf(elementGenerator)
       }
       case map: MapType => {
         val keyGenerator = getGenerator(map.keyType)
-        val valueGenerator = getGenerator(map.valueType)
+        val valueGenerator = getGenerator(map.valueType, nullable = map.valueContainsNull)
         val keyValueGenerator: Gen[(Any, Any)] = for {
           key <- keyGenerator
           value <- valueGenerator
@@ -139,6 +139,10 @@ object DataframeGenerator {
       case _ => throw new UnsupportedOperationException(
         s"Type: $dataType not supported")
     }
+    if (nullable)
+      Gen.oneOf(nonNullGen, Gen.const(null))
+    else
+      nonNullGen
   }
 
 }

diff --git a/core/src/test/1.3/scala/com/holdenkarau/spark/testing/SampleScalaCheckTest.scala b/core/src/test/1.3/scala/com/holdenkarau/spark/testing/SampleScalaCheckTest.scala
@@ -140,7 +140,7 @@ class SampleScalaCheckTest extends FunSuite
 
   test("assert rows' types like schema type") {
     val schema = StructType(
-      List(StructField("name", StringType), StructField("age", IntegerType)))
+      List(StructField("name", StringType, nullable = false), StructField("age", IntegerType, nullable = false)))
     val rowGen: Gen[Row] = DataframeGenerator.getRowGenerator(schema)
     val property =
       forAll(rowGen) {
@@ -316,24 +316,24 @@ class SampleScalaCheckTest extends FunSuite
     check(property)
   }
 
+  val fields = StructField("byteType", ByteType) ::
+    StructField("shortType", ShortType) ::
+    StructField("intType", IntegerType) ::
+    StructField("longType", LongType) ::
+    StructField("doubleType", DoubleType) ::
+    StructField("stringType", StringType) ::
+    StructField("binaryType", BinaryType) ::
+    StructField("booleanType", BooleanType) ::
+    StructField("timestampType", TimestampType) ::
+    StructField("dateType", DateType) ::
+    StructField("arrayType", ArrayType(TimestampType)) ::
+    StructField("mapType",
+      MapType(LongType, TimestampType, valueContainsNull = true)) ::
+    StructField("structType",
+      StructType(StructField("timestampType", TimestampType) :: Nil)) :: Nil
   test("second dataframe's evaluation has the same values as first") {
     implicit val generatorDrivenConfig =
       PropertyCheckConfig(minSize = 1, maxSize = 1)
-    val fields = StructField("byteType", ByteType) ::
-      StructField("shortType", ShortType) ::
-      StructField("intType", IntegerType) ::
-      StructField("longType", LongType) ::
-      StructField("doubleType", DoubleType) ::
-      StructField("stringType", StringType) ::
-      StructField("binaryType", BinaryType) ::
-      StructField("booleanType", BooleanType) ::
-      StructField("timestampType", TimestampType) ::
-      StructField("dateType", DateType) ::
-      StructField("arrayType", ArrayType(TimestampType)) ::
-      StructField("mapType",
-        MapType(LongType, TimestampType, valueContainsNull = true)) ::
-      StructField("structType",
-        StructType(StructField("timestampType", TimestampType) :: Nil)) :: Nil
 
     val sqlContext = new SQLContext(sc)
     val dataframeGen =
@@ -352,6 +352,38 @@ class SampleScalaCheckTest extends FunSuite
 
     check(property)
   }
+  test("nullable fields contain null values as well") {
+    implicit val generatorDrivenConfig =
+      PropertyCheckConfig(minSize = 1, maxSize = 1)
+    val nullableFields = fields.map(f => f.copy(nullable = true, name = s"${f.name}Nullable"))
+    val sqlContext = new SQLContext(sc)
+    val allFields = fields ::: nullableFields
+    val dataframeGen =
+      DataframeGenerator.arbitraryDataFrame(sqlContext, StructType(allFields))
+
+    val property =
+      forAll(Gen.resize(100, dataframeGen.arbitrary)) {
+        dataframe => {
+          allFields.forall { f =>
+            val colValues = dataframe.select(f.name).collect().map(_.get(0))
+            if (f.nullable)
+              colValues.contains(null) ||
+                // Unfortunately, dataframeGen.arbitrary sometimes generates DataFrames where all
+                // rows have exactly identical values.
+                // In that case, even generating many rows doesn't help to get some nulls...
+                // To work around this we check if we generated at least some distinct values.
+                colValues.distinct.size < 4 ||
+                // This is needed for Array-valued fields where .distinct returns all values, even when
+                // they're identical.
+                colValues.size == colValues.distinct.size
+            else
+              !colValues.contains(null)
+          }
+        }
+      }
+
+    check(property)
+  }
 
   private def filterOne(rdd: RDD[String]): RDD[Int] = {
     rdd.filter(_.length > 2).map(_.length)