[SPARK-14372][SQL] Dataset.randomSplit() needs a Java version

## What changes were proposed in this pull request?

1.Added method randomSplitAsList() in Dataset for java
for https://issues.apache.org/jira/browse/SPARK-14372

## How was this patch tested?

TestSuite

Author: Rekha Joshi <rekhajoshm@gmail.com>
Author: Joshi <rekhajoshm@gmail.com>

Closes #12184 from rekhajoshm/SPARK-14372.

1 files changed, 10 insertions, 0 deletions

diff --git a/sql/core/src/test/java/test/org/apache/spark/sql/JavaDatasetSuite.java b/sql/core/src/test/java/test/org/apache/spark/sql/JavaDatasetSuite.java
index f26c57b301..5abd62cbc2 100644
--- a/sql/core/src/test/java/test/org/apache/spark/sql/JavaDatasetSuite.java
+++ b/sql/core/src/test/java/test/org/apache/spark/sql/JavaDatasetSuite.java
@@ -454,6 +454,16 @@ public class JavaDatasetSuite implements Serializable {
     Assert.assertEquals(data, ds.collectAsList());
   }
 
+  @Test
+  public void testRandomSplit() {
+    List<String> data = Arrays.asList("hello", "world", "from", "spark");
+    Dataset<String> ds = context.createDataset(data, Encoders.STRING());
+    double[] arraySplit = {1, 2, 3};
+
+    List<Dataset<String>> randomSplit =  ds.randomSplitAsList(arraySplit, 1);
+    Assert.assertEquals("wrong number of splits", randomSplit.size(), 3);
+  }
+
   /**
    * For testing error messages when creating an encoder on a private class. This is done
    * here since we cannot create truly private classes in Scala.