maropu commented on a change in pull request #27965:
[SPARK-29774][SQL][followup] support string literal as the second argument of
date_add/date_sub functions
URL: https://github.com/apache/spark/pull/27965#discussion_r395524006
##########
File path:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/TypeCoercion.scala
##########
@@ -1043,6 +1044,22 @@ object TypeCoercion {
}
}
}
+
+ /**
+ * A special rule to support string literal as the second argument of
date_add/date_sub functions,
+ * to keep backward compatibility.
+ * TODO: revisit the type coercion rules for string.
+ */
+ object StringLiteralCoercion extends TypeCoercionRule {
+ override protected def coerceTypes(plan: LogicalPlan): LogicalPlan = plan
resolveExpressions {
+ // Skip nodes who's children have not been resolved yet.
+ case e if !e.childrenResolved => e
+ case DateAdd(l, r) if r.dataType == StringType && r.foldable =>
+ DateAdd(l, AnsiCast(r, IntegerType))
Review comment:
We need to depend on `AnsiCast` here, even though we can know cast errors in
an analysis phase? IMHO an analysis exception with an explicit error message is
better than the current number format exception:
https://github.com/apache/spark/pull/27965/files#diff-79dd276be45ede6f34e24ad7005b0a7cR279
----------------------------------------------------------------
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:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
