HIVE-29277: Invalid index should return null for get_json_object

Author: wecharyu

ql/src/java/org/apache/hadoop/hive/ql/udf/UDFJson.java

@@ -239,11 +239,12 @@ private Object extract(Object json, String path, boolean skipMapProc) {
       indexListCache.put(path, indexList);
     }
 
-    if (indexList.size() > 0) {
-      json = extract_json_withindex(json, indexList);
+    if (indexList.isEmpty()) {
+      // Return null if index is invalid.
+      return path.contains("[") ? null : json;
     }
 
-    return json;
+    return extract_json_withindex(json, indexList);
   }
 
   private transient AddingList jsonList = new AddingList();

ql/src/test/org/apache/hadoop/hive/ql/udf/TestUDFJson.java

@@ -105,6 +105,9 @@ public void testRootArray() throws HiveException {
     runTest("[1,2,3]", "$0", null, udf);
     runTest("[1,2,3]", "0", null, udf);
     runTest("[1,2,3]", "$.", null, udf);
+    runTest("[1,2,3]", "$[1 ]", null, udf);
+    runTest("[1,2,3]", "$[ 1]", null, udf);
+    runTest("[1,2,3]", "$[invalid]", null, udf);
 
     runTest("[1,2,3]", "$", "[1,2,3]", udf);
     runTest("{\"a\":4}", "$", "{\"a\":4}", udf);

ql/src/test/queries/clientpositive/udf_get_json_object.q

@@ -64,4 +64,6 @@ get_json_object('[{"k1":[{"k11":[1,2,3]}]}]', '$[0].k1[0].k11[1]');
 SELECT
 get_json_object('[1,2,3]', '[2]'),
 get_json_object('[1,2,3]', '$0'),
+get_json_object('[1,2,3]', '$[0 ]'),
+get_json_object('[1,2,3]', '$[ 0]'),
 get_json_object('[1,2,3]', '$[3]');

ql/src/test/results/clientpositive/llap/udf_get_json_object.q.out

@@ -227,15 +227,19 @@ POSTHOOK: Input: _dummy_database@_dummy_table
 PREHOOK: query: SELECT
 get_json_object('[1,2,3]', '[2]'),
 get_json_object('[1,2,3]', '$0'),
+get_json_object('[1,2,3]', '$[0 ]'),
+get_json_object('[1,2,3]', '$[ 0]'),
 get_json_object('[1,2,3]', '$[3]')
 PREHOOK: type: QUERY
 PREHOOK: Input: _dummy_database@_dummy_table
 #### A masked pattern was here ####
 POSTHOOK: query: SELECT
 get_json_object('[1,2,3]', '[2]'),
 get_json_object('[1,2,3]', '$0'),
+get_json_object('[1,2,3]', '$[0 ]'),
+get_json_object('[1,2,3]', '$[ 0]'),
 get_json_object('[1,2,3]', '$[3]')
 POSTHOOK: type: QUERY
 POSTHOOK: Input: _dummy_database@_dummy_table
 #### A masked pattern was here ####
-NULL	NULL	NULL
+NULL	NULL	NULL	NULL	NULL