[SPARK-18137][SQL] Fix RewriteDistinctAggregates UnresolvedException when a UDAF has a foldable TypeCheck

## What changes were proposed in this pull request?

In RewriteDistinctAggregates rewrite funtion,after the UDAF's childs are mapped to AttributeRefference, If the UDAF(such as ApproximatePercentile) has a foldable TypeCheck for the input, It will failed because the AttributeRefference is not foldable,then the UDAF is not resolved, and then nullify on the unresolved object will throw a Exception.

In this PR, only map Unfoldable child to AttributeRefference, this can avoid the UDAF's foldable TypeCheck. and then only Expand Unfoldable child, there is no need to Expand a static value(foldable value).

**Before sql result**

> select percentile_approxy(key,0.99999),count(distinct key),sume(distinc key) from src limit 1
> org.apache.spark.sql.catalyst.analysis.UnresolvedException: Invalid call to dataType on unresolved object, tree: 'percentile_approx(CAST(src.`key` AS DOUBLE), CAST(0.99999BD AS DOUBLE), 10000)
> at org.apache.spark.sql.catalyst.analysis.UnresolvedAttribute.dataType(unresolved.scala:92)
>     at org.apache.spark.sql.catalyst.optimizer.RewriteDistinctAggregates$.org$apache$spark$sql$catalyst$optimizer$RewriteDistinctAggregates$$nullify(RewriteDistinctAggregates.scala:261)

**After sql result**

> select percentile_approxy(key,0.99999),count(distinct key),sume(distinc key) from src limit 1
> [498.0,309,79136]
## How was this patch tested?

Add a test case in HiveUDFSuit.

Author: root <root@iZbp1gsnrlfzjxh82cz80vZ.(none)>

Closes #15668 from windpiger/RewriteDistinctUDAFUnresolveExcep.

1 files changed, 35 insertions, 0 deletions

diff --git a/sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveUDFSuite.scala b/sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveUDFSuite.scala
index f690035c84..48adc833f4 100644
--- a/sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveUDFSuite.scala
+++ b/sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveUDFSuite.scala
@@ -150,6 +150,41 @@ class HiveUDFSuite extends QueryTest with TestHiveSingleton with SQLTestUtils {
   }
 
   test("Generic UDAF aggregates") {
+
+    checkAnswer(sql(
+     """
+       |SELECT percentile_approx(2, 0.99999),
+       |       sum(distinct 1),
+       |       count(distinct 1,2,3,4) FROM src LIMIT 1
+     """.stripMargin), sql("SELECT 2, 1, 1 FROM src LIMIT 1").collect().toSeq)
+
+    checkAnswer(sql(
+      """
+        |SELECT ceiling(percentile_approx(distinct key, 0.99999)),
+        |       count(distinct key),
+        |       sum(distinct key),
+        |       count(distinct 1),
+        |       sum(distinct 1),
+        |       sum(1) FROM src LIMIT 1
+      """.stripMargin),
+      sql(
+        """
+          |SELECT max(key),
+          |       count(distinct key),
+          |       sum(distinct key),
+          |       1, 1, sum(1) FROM src LIMIT 1
+        """.stripMargin).collect().toSeq)
+
+    checkAnswer(sql(
+      """
+        |SELECT ceiling(percentile_approx(distinct key, 0.9 + 0.09999)),
+        |       count(distinct key), sum(distinct key),
+        |       count(distinct 1), sum(distinct 1),
+        |       sum(1) FROM src LIMIT 1
+      """.stripMargin),
+      sql("SELECT max(key), count(distinct key), sum(distinct key), 1, 1, sum(1) FROM src LIMIT 1")
+        .collect().toSeq)
+
     checkAnswer(sql("SELECT ceiling(percentile_approx(key, 0.99999D)) FROM src LIMIT 1"),
       sql("SELECT max(key) FROM src LIMIT 1").collect().toSeq)