Github user kiszk commented on a diff in the pull request:
https://github.com/apache/spark/pull/21045#discussion_r181489957
--- Diff:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/collectionOperations.scala
---
@@ -87,6 +87,62 @@ case class MapKeys(child: Expression)
override def prettyName: String = "map_keys"
}
+@ExpressionDescription(
+ usage = """_FUNC_(a1, a2) - Returns a merged array matching N-th element
of first
+ array with the N-th element of second.""",
+ examples = """
+ Examples
+ > SELECT _FUNC_(array(1, 2, 3), array(2, 3, 4));
+ [[1, 2], [2, 3], [3, 4]]
+ """,
+ since = "2.4.0")
+case class Zip(left: Expression, right: Expression)
+ extends BinaryExpression with ExpectsInputTypes {
+
+ override def inputTypes: Seq[AbstractDataType] = Seq(ArrayType,
ArrayType)
+
+ override def dataType: DataType =
ArrayType(left.dataType.asInstanceOf[ArrayType].elementType)
+
+ override def doGenCode(ctx: CodegenContext, ev: ExprCode): ExprCode = {
+ nullSafeCodeGen(ctx, ev, (arr1, arr2) => {
+ val i = ctx.freshName("i")
+ s"""
+ for (int $i = 0; $i < $arr1.numElements(); $i ++) {
+ if ($arr1.isNullAt($i)) {
+ ${ev.isNull} = true;
+ } else {
+ ${ev.value}[$i] = ($arr1[$i], $arr2[$i]);
--- End diff --
I might be wrong. We have the similar discussion
[here](https://github.com/apache/spark/pull/21061#discussion_r181399858).
If we need to generate Java code, how about writing runtime routine in
scala and calling the routine from Java?
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
