dtenedor commented on code in PR #40710:
URL: https://github.com/apache/spark/pull/40710#discussion_r1161011837
##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/ResolveDefaultColumns.scala:
##########
@@ -91,6 +90,25 @@ case class ResolveDefaultColumns(catalog: SessionCatalog)
extends Rule[LogicalPl
}
}
+ /**
+ * Checks if a logical plan is an INSERT INTO command where the inserted
data comes from a SELECT
+ * list, with possible other unary operators like sorting and/or alias(es)
in between.
+ */
+ private def insertsFromProject(i: InsertIntoStatement): Option[Project] = {
+ var node = i.query
+ def matches(node: LogicalPlan): Boolean = node match {
+ case _: GlobalLimit | _: LocalLimit | _: Offset | _: SubqueryAlias | _:
Sort => true
+ case _ => false
+ }
+ while (matches(node)) {
+ node = node.children.head
Review Comment:
Good question, I added a test for this:
```
insert into t2 (i, s) select default, default from t1 inner join t1 using
(i, s);
> true, 42L, "abc",
> true, 42L, "abc"
```
We want the `default` resolution to cover the topmost `select` list before
the join, but not any below, per the specification. For example, this should
not work:
```
insert into t2 (i, s) select i, s from (
select default as i, default as s from t1)
inner join t1
using (i, s);
```
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
