dtenedor commented on a change in pull request #35982:
URL: https://github.com/apache/spark/pull/35982#discussion_r836864523
##########
File path:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/util/ResolveDefaultColumns.scala
##########
@@ -18,14 +18,227 @@
package org.apache.spark.sql.catalyst.util
import org.apache.spark.sql.AnalysisException
-import org.apache.spark.sql.catalyst.analysis.Analyzer
+import org.apache.spark.sql.catalyst.TableIdentifier
+import org.apache.spark.sql.catalyst.analysis.{Analyzer, NoSuchTableException,
UnresolvedAttribute, UnresolvedInlineTable, UnresolvedRelation}
+import org.apache.spark.sql.catalyst.catalog.{SessionCatalog,
UnresolvedCatalogRelation}
import org.apache.spark.sql.catalyst.expressions._
import org.apache.spark.sql.catalyst.optimizer.ConstantFolding
import org.apache.spark.sql.catalyst.parser.{CatalystSqlParser, ParseException}
import org.apache.spark.sql.catalyst.plans.logical._
+import org.apache.spark.sql.catalyst.rules.Rule
+import org.apache.spark.sql.catalyst.util.ResolveDefaultColumns._
import org.apache.spark.sql.internal.SQLConf
import org.apache.spark.sql.types._
+/**
+ * This is a rule to process DEFAULT columns in statements such as
CREATE/REPLACE TABLE.
+ *
+ * Background: CREATE TABLE and ALTER TABLE invocations support setting column
default values for
+ * later operations. Following INSERT, and INSERT MERGE commands may then
reference the value
+ * using the DEFAULT keyword as needed.
+ *
+ * Example:
+ * CREATE TABLE T(a INT DEFAULT 4, b INT NOT NULL DEFAULT 5);
+ * INSERT INTO T VALUES (1, 2);
+ * INSERT INTO T VALUES (1, DEFAULT);
+ * INSERT INTO T VALUES (DEFAULT, 6);
+ * SELECT * FROM T;
+ * (1, 2)
+ * (1, 5)
+ * (4, 6)
+ *
+ * @param analyzer analyzer to use for processing DEFAULT values stored as
text.
+ * @param catalog the catalog to use for looking up the schema of INSERT INTO
table objects.
+ * @param insert the enclosing INSERT statement for which this rule is
processing the query, if any.
+ */
+case class ResolveDefaultColumns(
+ analyzer: Analyzer,
+ catalog: SessionCatalog,
+ insert: Option[InsertIntoStatement] = None) extends Rule[LogicalPlan] {
+ override def apply(plan: LogicalPlan): LogicalPlan =
plan.resolveOperatorsWithPruning(
+ (_ => SQLConf.get.enableDefaultColumns), ruleId) {
+ case i@InsertIntoStatement(_, _, _, _, _, _)
+ if i.query.collectFirst { case u: UnresolvedInlineTable => u }.isDefined
=>
+ // Create a helper instance of this same rule with the `insert` argument
populated as `i`.
+ // Then recursively apply it on the `query` of `i` to transform the
result. This recursive
+ // application lets the below case matching against
`UnresolvedInlineTable` to trigger
+ // anywhere that operator may appear in the descendants of `i.query`.
+ val helper = ResolveDefaultColumns(analyzer, catalog, Some(i))
+ val newQuery = helper.apply(i.query)
+ i.copy(query = newQuery)
+
+ case table: UnresolvedInlineTable
+ if insert.isDefined &&
+ table.rows.nonEmpty && table.rows.forall(_.size == table.rows(0).size)
=>
+ val expanded: UnresolvedInlineTable =
addMissingDefaultColumnValues(table).getOrElse(table)
+ replaceExplicitDefaultColumnValues(analyzer, expanded).getOrElse(table)
+
+ case i@InsertIntoStatement(_, _, _, project: Project, _, _)
+ if !project.resolved =>
+ val helper = ResolveDefaultColumns(analyzer, catalog, Some(i))
+ val replaced: Option[Project] =
helper.replaceExplicitDefaultColumnValues(analyzer, project)
+ if (replaced.isDefined) i.copy(query = replaced.get) else i
+
+ case i@InsertIntoStatement(_, _, _, project: Project, _, _)
+ if project.resolved =>
+ val helper = ResolveDefaultColumns(analyzer, catalog, Some(i))
+ val expanded: Project =
helper.addMissingDefaultColumnValues(project).getOrElse(project)
Review comment:
I removed this check. It turns out this was a hold-out from the previous
version of the rule logic. We do not need to keep this anymore, since the
"check analysis" step that errors-out in response to the unresolved DEFAULT
column happens later in the sequence. The code is simpler now.
--
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]
