beliefer commented on a change in pull request #29800:
URL: https://github.com/apache/spark/pull/29800#discussion_r503179138
##########
File path:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/windowExpressions.scala
##########
@@ -347,18 +341,51 @@ abstract class OffsetWindowFunction
val default: Expression
/**
- * (Foldable) expression that contains the number of rows between the
current row and the row
- * where the input expression is evaluated.
+ * (Foldable) expression that contains the number of rows between the
starting row (current row
+ * if isRelative=true, or the first row of the window frame otherwise) and
the row where the
+ * input expression is evaluated.
*/
- val offset: Expression
+ val offsetExpr: Expression
/**
- * Direction of the number of rows between the current row and the row where
the input expression
- * is evaluated.
+ * Direction of the number of rows between the current row or the first row
of the entire
+ * window partition and the row where the input expression is evaluated.
*/
val direction: SortDirection
- override def children: Seq[Expression] = Seq(input, offset, default)
+ /**
+ * An optional specification that indicates the offset window function
should skip null values in
+ * the determination of which row to use.
+ */
+ val ignoreNulls: Boolean
+
+ /**
+ * Whether the offset is starts with the current row. If `isRelative` is
true, `offset` means
+ * the offset is start with the current row. otherwise, the offset is starts
with the first
+ * row of the entire window frame.
+ */
+ val isRelative: Boolean = true
+
+ lazy val boundary = direction match {
+ case Ascending => offsetExpr
+ case Descending => UnaryMinus(offsetExpr) match {
+ case e: Expression if e.foldable => Literal.create(e.eval(EmptyRow),
e.dataType)
+ case o => o
+ }
+ }
+
+ lazy val fakeFrame = SpecifiedWindowFrame(RowFrame, boundary, boundary)
Review comment:
Just tell every one that this frame don't play the real role of the
frame, just a fake.
##########
File path:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/windowExpressions.scala
##########
@@ -347,18 +341,51 @@ abstract class OffsetWindowFunction
val default: Expression
/**
- * (Foldable) expression that contains the number of rows between the
current row and the row
- * where the input expression is evaluated.
+ * (Foldable) expression that contains the number of rows between the
starting row (current row
+ * if isRelative=true, or the first row of the window frame otherwise) and
the row where the
+ * input expression is evaluated.
*/
- val offset: Expression
+ val offsetExpr: Expression
/**
- * Direction of the number of rows between the current row and the row where
the input expression
- * is evaluated.
+ * Direction of the number of rows between the current row or the first row
of the entire
+ * window partition and the row where the input expression is evaluated.
*/
val direction: SortDirection
- override def children: Seq[Expression] = Seq(input, offset, default)
+ /**
+ * An optional specification that indicates the offset window function
should skip null values in
+ * the determination of which row to use.
+ */
+ val ignoreNulls: Boolean
+
+ /**
+ * Whether the offset is starts with the current row. If `isRelative` is
true, `offset` means
+ * the offset is start with the current row. otherwise, the offset is starts
with the first
+ * row of the entire window frame.
+ */
+ val isRelative: Boolean = true
+
+ lazy val boundary = direction match {
Review comment:
I got it.
----------------------------------------------------------------
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]
