ulysses-you commented on code in PR #37831:
URL: https://github.com/apache/spark/pull/37831#discussion_r966615335
##########
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/V1Writes.scala:
##########
@@ -102,6 +121,83 @@ object V1Writes extends Rule[LogicalPlan] with
SQLConfHelper {
}
}
+/**
+ * This rule is used to eliminate dynamic partition to static partition for v1
writes if the
+ * partition columns is foldable, so that we can avoid unnecessary sort for
dynamic partition.
+ *
+ * For example, a pure SQL:
+ * {{{
+ * INSERT INTO TABLE t1 PARTITION(p) SELECT c, 'a' as p FROM t2
+ * =>
+ * INSERT INTO TABLE t1 PARTITION(p='a') SELECT c FROM t2
+ * }}}
+ */
+object EliminateV1DynamicPartitionWrites extends Rule[LogicalPlan] {
+
+ @tailrec
+ private def queryOutput(p: LogicalPlan): Seq[NamedExpression] = p match {
+ case p: Project => p.projectList
+ case f: Filter => queryOutput(f.child)
+ case r: RepartitionOperation => queryOutput(r.child)
+ case r: RebalancePartitions => queryOutput(r.child)
+ case s: Sort => queryOutput(s.child)
+ case l: LocalLimit => queryOutput(l.child)
+ case l: GlobalLimit => queryOutput(l.child)
+ case _ => Seq.empty
+ }
+
+ private def getPartitionSpecString(part: Any): String = {
+ if (part == null) {
+ null
+ } else {
+ assert(part.isInstanceOf[UTF8String])
+
ExternalCatalogUtils.getPartitionSpecString(part.asInstanceOf[UTF8String].toString)
Review Comment:
null and empty partition are treated as null, and the final partition path
will be `__HIVE_DEFAULT_PARTITION__`
--
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]
