You are viewing a plain text version of this content. The canonical link for it is here.
Posted to reviews@spark.apache.org by GitBox <gi...@apache.org> on 2020/07/29 07:35:36 UTC
[GitHub] [spark] viirya commented on a change in pull request #29261: [SPARK-32459][SQL] Support WrappedArray as customCollectionCls in MapObjects
viirya commented on a change in pull request #29261:
URL: https://github.com/apache/spark/pull/29261#discussion_r461245294
##########
File path: sql/core/src/test/scala/org/apache/spark/sql/UDFSuite.scala
##########
@@ -787,4 +789,13 @@ class UDFSuite extends QueryTest with SharedSparkSession {
Seq(Array(1)).toDF("a").createOrReplaceTempView("t")
checkAnswer(sql("SELECT key(a) AS k FROM t GROUP BY key(a)"), Row(1) :: Nil)
}
+
+ test("SPARK-32459: WrappedArray") {
Review comment:
nit: WrappedArray -> UDF should not fail on WrappedArray.
##########
File path: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/objects/objects.scala
##########
@@ -912,6 +916,24 @@ case class MapObjects private(
val (initCollection, addElement, getResult): (String, String => String, String) =
customCollectionCls match {
+ case Some(cls) if classOf[WrappedArray[_]].isAssignableFrom(cls) =>
+ // Scala WrappedArray
+ val getBuilder = s"${cls.getName}$$.MODULE$$.newBuilder()"
+ val builder = ctx.freshName("collectionBuilder")
+ val toWrappedArray = (arrayBuffer: ArrayBuffer[_]) =>
+ WrappedArray.make(arrayBuffer.toArray)
Review comment:
+1
##########
File path: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/objects/objects.scala
##########
@@ -912,6 +915,20 @@ case class MapObjects private(
val (initCollection, addElement, getResult): (String, String => String, String) =
customCollectionCls match {
+ case Some(cls) if classOf[WrappedArray[_]].isAssignableFrom(cls) =>
+ // Scala WrappedArray
+ val getBuilder = s"${cls.getName}$$.MODULE$$.newBuilder()"
+ val builder = ctx.freshName("collectionBuilder")
+ (
+ s"""
+ ${classOf[Builder[_, _]].getName} $builder = $getBuilder;
+ $builder.sizeHint($dataLength);
+ """,
+ (genValue: String) => s"$builder.$$plus$$eq($genValue);",
+ s"(${cls.getName}) ${classOf[WrappedArray[_]].getName}$$." +
+ s"MODULE$$.make(((${classOf[ArrayBuffer[_]].getName})$builder" +
Review comment:
`$builder.result()` is a `ArrayBuffer`?
##########
File path: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/objects/objects.scala
##########
@@ -912,6 +915,20 @@ case class MapObjects private(
val (initCollection, addElement, getResult): (String, String => String, String) =
customCollectionCls match {
+ case Some(cls) if classOf[WrappedArray[_]].isAssignableFrom(cls) =>
+ // Scala WrappedArray
+ val getBuilder = s"${cls.getName}$$.MODULE$$.newBuilder()"
+ val builder = ctx.freshName("collectionBuilder")
+ (
+ s"""
+ ${classOf[Builder[_, _]].getName} $builder = $getBuilder;
+ $builder.sizeHint($dataLength);
+ """,
+ (genValue: String) => s"$builder.$$plus$$eq($genValue);",
+ s"(${cls.getName}) ${classOf[WrappedArray[_]].getName}$$." +
+ s"MODULE$$.make(((${classOf[ArrayBuffer[_]].getName})$builder" +
Review comment:
Isn't `WrappedArray` `HasNewBuilder[T, WrappedArray[T]]`, so `result` should return a `WrappedArray`?
##########
File path: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/objects/objects.scala
##########
@@ -912,6 +915,20 @@ case class MapObjects private(
val (initCollection, addElement, getResult): (String, String => String, String) =
customCollectionCls match {
+ case Some(cls) if classOf[WrappedArray[_]].isAssignableFrom(cls) =>
+ // Scala WrappedArray
+ val getBuilder = s"${cls.getName}$$.MODULE$$.newBuilder()"
+ val builder = ctx.freshName("collectionBuilder")
+ (
+ s"""
+ ${classOf[Builder[_, _]].getName} $builder = $getBuilder;
+ $builder.sizeHint($dataLength);
+ """,
+ (genValue: String) => s"$builder.$$plus$$eq($genValue);",
+ s"(${cls.getName}) ${classOf[WrappedArray[_]].getName}$$." +
+ s"MODULE$$.make(((${classOf[ArrayBuffer[_]].getName})$builder" +
Review comment:
https://www.scala-lang.org/api/2.12.8/scala/collection/mutable/WrappedArray.html and https://www.scala-lang.org/api/2.12.8/scala/collection/generic/HasNewBuilder.html
##########
File path: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/objects/objects.scala
##########
@@ -912,6 +915,20 @@ case class MapObjects private(
val (initCollection, addElement, getResult): (String, String => String, String) =
customCollectionCls match {
+ case Some(cls) if classOf[WrappedArray[_]].isAssignableFrom(cls) =>
+ // Scala WrappedArray
+ val getBuilder = s"${cls.getName}$$.MODULE$$.newBuilder()"
+ val builder = ctx.freshName("collectionBuilder")
+ (
+ s"""
+ ${classOf[Builder[_, _]].getName} $builder = $getBuilder;
+ $builder.sizeHint($dataLength);
+ """,
+ (genValue: String) => s"$builder.$$plus$$eq($genValue);",
+ s"(${cls.getName}) ${classOf[WrappedArray[_]].getName}$$." +
+ s"MODULE$$.make(((${classOf[ArrayBuffer[_]].getName})$builder" +
Review comment:
Oh, I see. Object `WrappedArray`'s `newBuilder` is `def newBuilder[A]: Builder[A, IndexedSeq[A]]`.
##########
File path: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/objects/objects.scala
##########
@@ -912,6 +915,20 @@ case class MapObjects private(
val (initCollection, addElement, getResult): (String, String => String, String) =
customCollectionCls match {
+ case Some(cls) if classOf[WrappedArray[_]].isAssignableFrom(cls) =>
+ // Scala WrappedArray
+ val getBuilder = s"${cls.getName}$$.MODULE$$.newBuilder()"
+ val builder = ctx.freshName("collectionBuilder")
+ (
+ s"""
+ ${classOf[Builder[_, _]].getName} $builder = $getBuilder;
+ $builder.sizeHint($dataLength);
+ """,
+ (genValue: String) => s"$builder.$$plus$$eq($genValue);",
+ s"(${cls.getName}) ${classOf[WrappedArray[_]].getName}$$." +
+ s"MODULE$$.make(((${classOf[ArrayBuffer[_]].getName})$builder" +
Review comment:
Then I think `IndexedSeq` is safer?
----------------------------------------------------------------
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:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org