cloud-fan commented on a change in pull request #25204: [SPARK-28441][SQL][Python] Fix error when PythonUDF is used in correlated scalar subquery URL: https://github.com/apache/spark/pull/25204#discussion_r305670236
########## File path: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/subquery.scala ########## @@ -318,17 +318,30 @@ object RewriteCorrelatedScalarSubquery extends Rule[LogicalPlan] { /** * Statically evaluate an expression containing zero or more placeholders, given a set - * of bindings for placeholder values. + * of bindings for placeholder values, if the expression is evaluable. If it is not, + * bind statically evaluated expression results to an expression. */ - private def evalExpr(expr: Expression, bindings: Map[ExprId, Option[Any]]) : Option[Any] = { + private def bindingExpr( + expr: Expression, + bindings: Map[ExprId, Option[Expression]]): Option[Expression] = { val rewrittenExpr = expr transform { case r: AttributeReference => bindings(r.exprId) match { - case Some(v) => Literal.create(v, r.dataType) + case Some(v) => v case None => Literal.default(NullType) } } - Option(rewrittenExpr.eval()) + if (rewrittenExpr.find(_.isInstanceOf[PythonUDF]).isDefined) { + // SPARK-28441: `PythonUDF` can't be statically evaluated. Review comment: We can't call `Expression.eval(null)` if it's not foldable, otherwise exception may be thrown: 1. `AttributeReference.eval(null)` fails with NPE 2. `Nondeterministic.eval(null)` fails because it needs to be initialized first Whatever hack we use, I'd expect it makes the expression foldable. ---------------------------------------------------------------- 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: us...@infra.apache.org With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org