correctly check analysis for scalar sub-queries

Author: cloud-fan

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/CheckAnalysis.scala

@@ -130,6 +130,8 @@ trait CheckAnalysis extends PredicateHelper {
             }
 
           case s @ ScalarSubquery(query, conditions, _) =>
+            checkAnalysis(query)
+
             // If no correlation, the output must be exactly one column
             if (conditions.isEmpty && query.output.size != 1) {
               failAnalysis(
@@ -179,7 +181,6 @@ trait CheckAnalysis extends PredicateHelper {
                 case fail => failAnalysis(s"Correlated scalar subqueries must be Aggregated: $fail")
               }
             }
-            checkAnalysis(query)
             s
 
           case s: SubqueryExpression =>

sql/core/src/test/scala/org/apache/spark/sql/SubquerySuite.scala

@@ -72,7 +72,7 @@ class SubquerySuite extends QueryTest with SharedSQLContext {
     }
   }
 
-  test("rdd deserialization does not crash [SPARK-15791]") {
+  test("SPARK-15791: rdd deserialization does not crash") {
     sql("select (select 1 as b) as b").rdd.count()
   }
 
@@ -867,4 +867,12 @@ class SubquerySuite extends QueryTest with SharedSQLContext {
       sql("select * from l, r where l.a = r.c + 1 AND (exists (select * from r) OR l.a = r.c)"),
       Row(3, 3.0, 2, 3.0) :: Row(3, 3.0, 2, 3.0) :: Nil)
   }
+
+  test("SPARK-20688: correctly check analysis for scalar sub-queries") {
+    withTempView("t") {
+      Seq(1 -> "a").toDF("i", "j").createTempView("t")
+      val e = intercept[AnalysisException](sql("SELECT (SELECT count(*) FROM t WHERE a = 1)"))
+      assert(e.message.contains("cannot resolve '`a`' given input columns: [i, j]"))
+    }
+  }
 }