cloud-fan commented on a change in pull request #29522:
URL: https://github.com/apache/spark/pull/29522#discussion_r475480962
##########
File path:
sql/core/src/test/scala/org/apache/spark/sql/ColumnExpressionSuite.scala
##########
@@ -1452,4 +1452,89 @@ class ColumnExpressionSuite extends QueryTest with
SharedSparkSession {
.select($"struct_col".withField("a.c", lit(3)))
}.getMessage should include("Ambiguous reference to fields")
}
+
+ test("SPARK-32641: extracting field from non-null struct column after
withField should return " +
+ "field value") {
+ // extract newly added field
+ checkAnswerAndSchema(
+ structLevel1.withColumn("a", $"a".withField("d", lit(4)).getField("d")),
+ Row(4) :: Nil,
+ StructType(Seq(StructField("a", IntegerType, nullable = false))))
+
+ // extract newly replaced field
+ checkAnswerAndSchema(
+ structLevel1.withColumn("a", $"a".withField("a", lit(4)).getField("a")),
+ Row(4) :: Nil,
+ StructType(Seq(StructField("a", IntegerType, nullable = false))))
+
+ // add new field, extract another field from original struct
+ checkAnswerAndSchema(
+ structLevel1.withColumn("a", $"a".withField("d", lit(4)).getField("c")),
+ Row(3):: Nil,
+ StructType(Seq(StructField("a", IntegerType, nullable = false))))
+
+ // replace field, extract another field from original struct
+ checkAnswerAndSchema(
+ structLevel1.withColumn("a", $"a".withField("a", lit(4)).getField("c")),
+ Row(3):: Nil,
+ StructType(Seq(StructField("a", IntegerType, nullable = false))))
+ }
+
+ test("SPARK-32641: extracting field from null struct column after withField
should return " +
+ "null if the original struct was null") {
+ // extract newly added field
+ checkAnswerAndSchema(
+ nullStructLevel1.withColumn("a", $"a".withField("d",
lit(4)).getField("d")),
+ Row(null) :: Nil,
+ StructType(Seq(StructField("a", IntegerType, nullable = true))))
+
+ // extract newly replaced field
+ checkAnswerAndSchema(
+ nullStructLevel1.withColumn("a", $"a".withField("a",
lit(4)).getField("a")),
+ Row(null):: Nil,
+ StructType(Seq(StructField("a", IntegerType, nullable = true))))
+
+ // add new field, extract another field from original struct
+ checkAnswerAndSchema(
+ nullStructLevel1.withColumn("a", $"a".withField("d",
lit(4)).getField("c")),
+ Row(null):: Nil,
+ StructType(Seq(StructField("a", IntegerType, nullable = true))))
+
+ // replace field, extract another field from original struct
+ checkAnswerAndSchema(
+ nullStructLevel1.withColumn("a", $"a".withField("a",
lit(4)).getField("c")),
+ Row(null):: Nil,
+ StructType(Seq(StructField("a", IntegerType, nullable = true))))
+ }
+
+ test("SPARK-32641: extracting field from nullable struct column which
contains both null and " +
+ "non-null values after withField should return null if the original struct
was null") {
+ val df = structLevel1.union(nullStructLevel1).cache()
Review comment:
is the cache necessary?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
