[SPARK-5789][SQL]Throw a better error message if JsonRDD.parseJson encounters unrecoverable parsing errors.

Author: Yin Huai <yhuai@databricks.com>

Closes #4582 from yhuai/jsonErrorMessage and squashes the following commits:

152dbd4 [Yin Huai] Update error message.
1466256 [Yin Huai] Throw a better error message when a JSON object in the input dataset span multiple records (lines for files or strings for an RDD of strings).

(cherry picked from commit 2e0c084528409e1c565e6945521a33c0835ebbee)
Signed-off-by: Michael Armbrust <michael@databricks.com>

1 files changed, 4 insertions, 0 deletions

diff --git a/sql/core/src/main/scala/org/apache/spark/sql/json/JsonRDD.scala b/sql/core/src/main/scala/org/apache/spark/sql/json/JsonRDD.scala
index 7dfb304021..3b8dde1823 100644
--- a/sql/core/src/main/scala/org/apache/spark/sql/json/JsonRDD.scala
+++ b/sql/core/src/main/scala/org/apache/spark/sql/json/JsonRDD.scala
@@ -306,6 +306,10 @@ private[sql] object JsonRDD extends Logging {
           val parsed = mapper.readValue(record, classOf[Object]) match {
             case map: java.util.Map[_, _] => scalafy(map).asInstanceOf[Map[String, Any]] :: Nil
             case list: java.util.List[_] => scalafy(list).asInstanceOf[Seq[Map[String, Any]]]
+            case _ =>
+              sys.error(
+                s"Failed to parse record $record. Please make sure that each line of the file " +
+                "(or each string in the RDD) is a valid JSON object or an array of JSON objects.")
           }
 
           parsed