cloud-fan commented on a change in pull request #31440:
URL: https://github.com/apache/spark/pull/31440#discussion_r570797933
##########
File path:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/Analyzer.scala
##########
@@ -964,11 +964,37 @@ class Analyzer(override val catalogManager:
CatalogManager)
* columns are not accidentally selected by *.
*/
object AddMetadataColumns extends Rule[LogicalPlan] {
+ import
org.apache.spark.sql.execution.datasources.v2.DataSourceV2Implicits._
+
+ private def hasMetadataCol(plan: LogicalPlan): Boolean = {
+ plan.expressions.exists(_.find {
+ case a: Attribute => a.isMetadataCol
+ case _ => false
+ }.isDefined)
+ }
+
+ private def addMetadataCol(plan: LogicalPlan): LogicalPlan = plan match {
+ case r: DataSourceV2Relation => r.withMetadataColumns()
+ case _ => plan.withNewChildren(plan.children.map(addMetadataCol))
+ }
+
def apply(plan: LogicalPlan): LogicalPlan = plan resolveOperatorsUp {
- case node if node.resolved && node.children.nonEmpty &&
node.missingInput.nonEmpty =>
- node resolveOperatorsUp {
- case rel: DataSourceV2Relation =>
- rel.withMetadataColumns()
+ case node if node.children.nonEmpty && node.resolved &&
hasMetadataCol(node) =>
+ val inputAttrs = AttributeSet(node.children.flatMap(_.output))
+ val metaCols = node.expressions.flatMap(_.collect {
+ case a: Attribute if a.isMetadataCol && !inputAttrs.contains(a) => a
+ })
+ if (metaCols.isEmpty) {
+ node
+ } else {
+ val newNode = addMetadataCol(node)
Review comment:
Yea, it's good enough because:
1. We guarantee that the outer plan will project out extra columns, so the
final output schema won't change
2. Column pruning will work and eventually the data source doesn't need to
produce un-referenced columns.
----------------------------------------------------------------
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:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
