From 07a1788ee04597700f101183e67d237f9a866c46 Mon Sep 17 00:00:00 2001
From: gatorsmile <gatorsmile@gmail.com>
Date: Mon, 30 Jan 2017 18:38:14 -0800
Subject: [PATCH] [SPARK-19406][SQL] Fix function to_json to respect
 user-provided options

### What changes were proposed in this pull request?
Currently, the function `to_json` allows users to provide options for generating JSON. However, it does not pass it to `JacksonGenerator`. Thus, it ignores the user-provided options. This PR is to fix it. Below is an example.

```Scala
val df = Seq(Tuple1(Tuple1(java.sql.Timestamp.valueOf("2015-08-26 18:00:00.0")))).toDF("a")
val options = Map("timestampFormat" -> "dd/MM/yyyy HH:mm")
df.select(to_json($"a", options)).show(false)
```
The current output is like
```
+--------------------------------------+
|structtojson(a)                       |
+--------------------------------------+
|{"_1":"2015-08-26T18:00:00.000-07:00"}|
+--------------------------------------+
```

After the fix, the output is like
```
+-------------------------+
|structtojson(a)          |
+-------------------------+
|{"_1":"26/08/2015 18:00"}|
+-------------------------+
```
### How was this patch tested?
Added test cases for both `from_json` and `to_json`

Author: gatorsmile <gatorsmile@gmail.com>

Closes #16745 from gatorsmile/toJson.

(cherry picked from commit f9156d2956a8e751720bf63071c504a3e86f267d)
Signed-off-by: gatorsmile <gatorsmile@gmail.com>
---
 .../expressions/jsonExpressions.scala         |  5 ++++-
 .../apache/spark/sql/JsonFunctionsSuite.scala | 21 ++++++++++++++++++-
 2 files changed, 24 insertions(+), 2 deletions(-)

diff --git a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/jsonExpressions.scala b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/jsonExpressions.scala
index 667ff649d1..92d0888fc6 100644
--- a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/jsonExpressions.scala
+++ b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/jsonExpressions.scala
@@ -517,7 +517,10 @@ case class StructToJson(options: Map[String, String], child: Expression)
 
   @transient
   lazy val gen =
-    new JacksonGenerator(child.dataType.asInstanceOf[StructType], writer)
+    new JacksonGenerator(
+      child.dataType.asInstanceOf[StructType],
+      writer,
+      new JSONOptions(options))
 
   override def dataType: DataType = StringType
 
diff --git a/sql/core/src/test/scala/org/apache/spark/sql/JsonFunctionsSuite.scala b/sql/core/src/test/scala/org/apache/spark/sql/JsonFunctionsSuite.scala
index 890cc5b560..9c39b3c7f0 100644
--- a/sql/core/src/test/scala/org/apache/spark/sql/JsonFunctionsSuite.scala
+++ b/sql/core/src/test/scala/org/apache/spark/sql/JsonFunctionsSuite.scala
@@ -19,7 +19,7 @@ package org.apache.spark.sql
 
 import org.apache.spark.sql.functions.{from_json, struct, to_json}
 import org.apache.spark.sql.test.SharedSQLContext
-import org.apache.spark.sql.types.{CalendarIntervalType, IntegerType, StructType}
+import org.apache.spark.sql.types.{CalendarIntervalType, IntegerType, StructType, TimestampType}
 
 class JsonFunctionsSuite extends QueryTest with SharedSQLContext {
   import testImplicits._
@@ -105,6 +105,16 @@ class JsonFunctionsSuite extends QueryTest with SharedSQLContext {
       Row(Row(1)) :: Nil)
   }
 
+  test("from_json with option") {
+    val df = Seq("""{"time": "26/08/2015 18:00"}""").toDS()
+    val schema = new StructType().add("time", TimestampType)
+    val options = Map("timestampFormat" -> "dd/MM/yyyy HH:mm")
+
+    checkAnswer(
+      df.select(from_json($"value", schema, options)),
+      Row(Row(java.sql.Timestamp.valueOf("2015-08-26 18:00:00.0"))))
+  }
+
   test("from_json missing columns") {
     val df = Seq("""{"a": 1}""").toDS()
     val schema = new StructType().add("b", IntegerType)
@@ -131,6 +141,15 @@ class JsonFunctionsSuite extends QueryTest with SharedSQLContext {
       Row("""{"_1":1}""") :: Nil)
   }
 
+  test("to_json with option") {
+    val df = Seq(Tuple1(Tuple1(java.sql.Timestamp.valueOf("2015-08-26 18:00:00.0")))).toDF("a")
+    val options = Map("timestampFormat" -> "dd/MM/yyyy HH:mm")
+
+    checkAnswer(
+      df.select(to_json($"a", options)),
+      Row("""{"_1":"26/08/2015 18:00"}""") :: Nil)
+  }
+
   test("to_json unsupported type") {
     val df = Seq(Tuple1(Tuple1("interval -3 month 7 hours"))).toDF("a")
       .select(struct($"a._1".cast(CalendarIntervalType).as("a")).as("c"))
-- 
GitLab