[SPARK-4959] [SQL] Attributes are case sensitive when using a select query from a projection

Author: Cheng Hao <hao.cheng@intel.com>

Closes #3796 from chenghao-intel/spark_4959 and squashes the following commits:

3ec08f8 [Cheng Hao] Replace the attribute in comparing its exprId other than itself

2 files changed, 17 insertions, 5 deletions

diff --git a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/Optimizer.scala b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/Optimizer.scala
index 806c1394eb..0f2eae6400 100644
--- a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/Optimizer.scala
+++ b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/Optimizer.scala
@@ -142,16 +142,16 @@ object ColumnPruning extends Rule[LogicalPlan] {
     case Project(projectList1, Project(projectList2, child)) =>
       // Create a map of Aliases to their values from the child projection.
       // e.g., 'SELECT ... FROM (SELECT a + b AS c, d ...)' produces Map(c -> Alias(a + b, c)).
-      val aliasMap = projectList2.collect {
-        case a @ Alias(e, _) => (a.toAttribute: Expression, a)
-      }.toMap
+      val aliasMap = AttributeMap(projectList2.collect {
+        case a @ Alias(e, _) => (a.toAttribute, a)
+      })
 
       // Substitute any attributes that are produced by the child projection, so that we safely
       // eliminate it.
       // e.g., 'SELECT c + 1 FROM (SELECT a + b AS C ...' produces 'SELECT a + b + 1 ...'
       // TODO: Fix TransformBase to avoid the cast below.
       val substitutedProjection = projectList1.map(_.transform {
-        case a if aliasMap.contains(a) => aliasMap(a)
+        case a: Attribute if aliasMap.contains(a) => aliasMap(a)
       }).asInstanceOf[Seq[NamedExpression]]
 
       Project(substitutedProjection, child)
diff --git a/sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveTableScanSuite.scala b/sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveTableScanSuite.scala
index a0ace91060..16f77a438e 100644
--- a/sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveTableScanSuite.scala
+++ b/sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveTableScanSuite.scala
@@ -18,7 +18,8 @@
 package org.apache.spark.sql.hive.execution
 
 import org.apache.spark.sql.hive.test.TestHive
-import org.apache.spark.sql.{Row, SchemaRDD}
+import org.apache.spark.sql.hive.test.TestHive._
+import org.apache.spark.sql.Row
 
 import org.apache.spark.util.Utils
 
@@ -76,4 +77,15 @@ class HiveTableScanSuite extends HiveComparisonTest {
       === Array(Row(java.sql.Timestamp.valueOf("2014-12-11 00:00:00")),Row(null)))
     TestHive.sql("DROP TABLE timestamp_query_null")
   }
+
+  test("Spark-4959 Attributes are case sensitive when using a select query from a projection") {
+    sql("create table spark_4959 (col1 string)")
+    sql("""insert into table spark_4959 select "hi" from src limit 1""")
+    table("spark_4959").select(
+      'col1.as('CaseSensitiveColName),
+      'col1.as('CaseSensitiveColName2)).registerTempTable("spark_4959_2")
+
+    assert(sql("select CaseSensitiveColName from spark_4959_2").first() === Row("hi"))
+    assert(sql("select casesensitivecolname from spark_4959_2").first() === Row("hi"))
+  }
 }