[SPARK-4693] [SQL] PruningPredicates may be wrong if predicates contains an empty AttributeSet() references

Author: YanTangZhai

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/AttributeSet.scala

@@ -112,4 +112,6 @@ class AttributeSet private (val baseSet: Set[AttributeEquals])
   override def toSeq: Seq[Attribute] = baseSet.map(_.a).toArray.toSeq
 
   override def toString = "{" + baseSet.map(_.a).mkString(", ") + "}"
+
+	def isEmpty: Boolean = baseSet.isEmpty
 }

sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveStrategies.scala

@@ -194,10 +194,9 @@ private[hive] trait HiveStrategies {
         // Filter out all predicates that only deal with partition keys, these are given to the
         // hive table scan operator to be used for partition pruning.
         val partitionKeyIds = AttributeSet(relation.partitionKeys)
-        val (pruningPredicates, otherPredicates) = predicates.partition { x =>
-          x.references.baseSet != null &&
-          !x.references.baseSet.isEmpty &&
-          x.references.subsetOf(partitionKeyIds)
+        val (pruningPredicates, otherPredicates) = predicates.partition { predicate =>
+          !predicate.references.isEmpty &&
+          predicate.references.subsetOf(partitionKeyIds)
         }
 
         pruneFilterProject(