fix

apache · Sep 21, 2018 · edc261c · edc261c
1 parent 3030b82
commit edc261c
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 3 deletions.
diff --git a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/Analyzer.scala b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/Analyzer.scala
@@ -554,8 +554,11 @@ class Analyzer(
           Cast(value, pivotColumn.dataType, Some(conf.sessionLocalTimeZone)).eval(EmptyRow)
         }
         // Group-by expressions coming from SQL are implicit and need to be deduced.
-        val groupByExprs = groupByExprsOpt.getOrElse(
-          (child.outputSet -- aggregates.flatMap(_.references) -- pivotColumn.references).toSeq)
+        val groupByExprs = groupByExprsOpt.getOrElse {
+          val pivotColAndAggRefs =
+            (pivotColumn.references ++ aggregates.flatMap(_.references)).toSet
+          child.output.filterNot(pivotColAndAggRefs.contains(_))
+        }
         val singleAgg = aggregates.size == 1
         def outputName(value: Expression, aggregate: Expression): String = {
           val stringValue = value match {

diff --git a/sql/core/src/test/resources/sql-tests/inputs/pivot.sql b/sql/core/src/test/resources/sql-tests/inputs/pivot.sql
@@ -287,3 +287,13 @@ PIVOT (
   sum(earnings)
   FOR (course, m) IN (('dotNET', map('1', 1)), ('Java', map('2', 2)))
 );
+
+-- grouping columns output in the same order as input
+SELECT * FROM (
+  SELECT course, earnings, "a" as a, "z" as z, "b" as b, "y" as y, "c" as c, "x" as x, "d" as d, "w" as w
+  FROM courseSales
+)
+PIVOT (
+  sum(earnings)
+  FOR course IN ('dotNET', 'Java')
+);
diff --git a/sql/core/src/test/resources/sql-tests/results/pivot.sql.out b/sql/core/src/test/resources/sql-tests/results/pivot.sql.out
@@ -1,5 +1,5 @@
 -- Automatically generated by SQLQueryTestSuite
--- Number of queries: 31
+-- Number of queries: 32
 
 
 -- !query 0
@@ -476,3 +476,18 @@ struct<>
 -- !query 30 output
 org.apache.spark.sql.AnalysisException
 Invalid pivot column 'named_struct(course, course#x, m, m#x)'. Pivot columns must be comparable.;
+
+
+-- !query 31
+SELECT * FROM (
+  SELECT course, earnings, "a" as a, "z" as z, "b" as b, "y" as y, "c" as c, "x" as x, "d" as d, "w" as w
+  FROM courseSales
+)
+PIVOT (
+  sum(earnings)
+  FOR course IN ('dotNET', 'Java')
+)
+-- !query 31 schema
+struct<a:string,z:string,b:string,y:string,c:string,x:string,d:string,w:string,dotNET:bigint,Java:bigint>
+-- !query 31 output
+a	z	b	y	c	x	d	w	63000	50000