Re: [PR] [SPARK-57561][SQL] Reject TIME inputs for unsupported datetime functions [spark]

["shrirangmhalgi (via GitHub)" <gi...@apache.org>]

shrirangmhalgi commented on code in PR #56999:
URL: https://github.com/apache/spark/pull/56999#discussion_r3523305883


##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/datetimeExpressions.scala:
##########
@@ -2631,15 +2631,37 @@ case class ParseToDate(
 
   override def children: Seq[Expression] = left +: format.toSeq
 
+  private def inputTypeForDate: AbstractDataType = {
+    TypeCollection(
+      StringTypeWithCollation(supportsTrimCollation = true),
+      DateType,
+      TimestampType,
+      TimestampNTZType)
+  }
+
   override def inputTypes: Seq[AbstractDataType] = {
     // Note: ideally this function should only take string input, but we allow more types here to
     // be backward compatible.
-    TypeCollection(
-      StringTypeWithCollation(supportsTrimCollation = true),
-        DateType,
-        TimestampType,
-        TimestampNTZType) +:
-      format.map(_ => StringTypeWithCollation(supportsTrimCollation = true)).toSeq
+    inputTypeForDate +: format.map(_ =>
+      StringTypeWithCollation(supportsTrimCollation = true)).toSeq
+  }
+
+  override def checkInputDataTypes(): TypeCheckResult = {
+    def timeInput(expression: Expression): Option[Expression] = expression match {
+      case Cast(child, _, _, _) if child.dataType.isInstanceOf[TimeType] => Some(child)

Review Comment:
   nit: With the coercion rules blocking implicit TIME -> Date/Timestamp, this `checkInputDataTypes` guard only fires for an explicit user-written `CAST(TIME'...' AS TIMESTAMP)` passed to `to_date`. Is blocking explicit casts intentional here? If a user deliberately writes `to_date(CAST(TIME'12:00:00' AS TIMESTAMP))`, they've already accepted the conversion -- rejecting it at the function level seems overly protective.
   
   If the intent is defense-in-depth for the implicit path only, consider checking `!child.resolved || child.origin.isGenerated` to distinguish user-written vs analyzer-inserted casts.



-- 
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: reviews-unsubscribe@spark.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org