vladimirg-db commented on code in PR #50192:
URL: https://github.com/apache/spark/pull/50192#discussion_r1986999174
##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/Analyzer.scala:
##########
@@ -4082,3 +4084,31 @@ object RemoveTempResolvedColumn extends
Rule[LogicalPlan] {
}
}
}
+
+object ReassignAliasNamesWithCollations extends Rule[LogicalPlan] {
+ override def apply(plan: LogicalPlan): LogicalPlan = {
+ plan.resolveExpressionsWithPruning(_.containsPattern(ALIAS)) {
+ case a: Alias if
+ a.resolved &&
+ a.metadata.contains(AUTO_GENERATED_ALIAS) &&
+ hasNonDefaultCollationInTheSubtree(a.child) =>
+ val newName = toPrettySQL(a.child)
+ if (newName != a.name) {
+ a.withName(newName)
+ } else {
+ a
+ }
+ }
+ }
+
+ private def hasNonDefaultCollationInTheSubtree(rootExpression: Expression) =
{
+ rootExpression.exists {
+ case expression =>
+ val dataType = try { Some(expression.dataType) } catch { case ex:
Throwable => None }
Review Comment:
Some expressions don't support calling `dataType` on them. Any idea how to
implement that cleanly?
##########
sql/core/src/test/resources/sql-tests/analyzer-results/collations.sql.out:
##########
@@ -821,15 +821,15 @@ InsertIntoHadoopFsRelationCommand file:[not included in
comparison]/{warehouse_d
-- !query
select concat_ws(' ', utf8_lcase, utf8_lcase) from t5
-- !query analysis
-Project [concat_ws( , utf8_lcase#x, utf8_lcase#x) AS concat_ws( , utf8_lcase,
utf8_lcase)#x]
+Project [concat_ws( , utf8_lcase#x, utf8_lcase#x) AS concat_ws(' ' collate
UTF8_LCASE, utf8_lcase, utf8_lcase)#x]
+- SubqueryAlias spark_catalog.default.t5
+- Relation spark_catalog.default.t5[s#x,utf8_binary#x,utf8_lcase#x] parquet
-- !query
select concat_ws(' ', utf8_binary, utf8_lcase) from t5
-- !query analysis
-Project [concat_ws( , cast(utf8_binary#x as string collate null),
cast(utf8_lcase#x as string collate null)) AS concat_ws( , utf8_binary,
utf8_lcase)#x]
+Project [concat_ws( , cast(utf8_binary#x as string collate null),
cast(utf8_lcase#x as string collate null)) AS concat_ws(' ' collate null,
utf8_binary, utf8_lcase)#x]
Review Comment:
I wonder why we get `collate null` here...
##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/Analyzer.scala:
##########
@@ -451,6 +451,8 @@ class Analyzer(override val catalogManager: CatalogManager)
extends RuleExecutor
RewriteMergeIntoTable),
Batch("Subquery", Once,
UpdateOuterReferences),
+ Batch("ReassignAliasNamesWithCollations", Once,
Review Comment:
@cloud-fan @stefankandic @mihailom-db please comment what are your opinions
on that approach?
--
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]
