Fix for #1171 utf-16 issue in smartserializer

Author: ezimuel

src/Elasticsearch/Serializers/SmartSerializer.php

@@ -20,6 +20,7 @@
 
 use Elasticsearch\Common\Exceptions;
 use Elasticsearch\Common\Exceptions\Serializer\JsonErrorException;
+use JsonException;
 
 if (!defined('JSON_INVALID_UTF8_SUBSTITUTE')) {
     //PHP < 7.2 Define it as 0 so it does nothing
@@ -84,9 +85,12 @@ private function decode(?string $data): array
             try {
                 $result = json_decode($data, true, 512, JSON_THROW_ON_ERROR);
                 return $result;
-            } catch (\JsonException $e) {
-                $result = $result ?? [];
-                throw new JsonErrorException($e->getCode(), $data, $result);
+            } catch (JsonException $e) {
+                switch ($e->getCode()) {
+                    case JSON_ERROR_UTF16:
+                        return $this->decode(str_replace('\\', '\\\\', $data));
+                }
+                throw new JsonErrorException($e->getCode(), $data, $result ?? []);
             }
         }
 

tests/Elasticsearch/Tests/Serializers/SmartSerializerTest.php

@@ -45,4 +45,18 @@ public function testThrowJsonErrorException()
 
         $result = $this->serializer->deserialize('{ "foo" : bar" }', []);
     }
+
+    /**
+     * Single unpaired UTF-16 surrogate in unicode escape
+     * 
+     * @requires PHP 7.3
+     * @see https://github.com/elastic/elasticsearch-php/issues/1171
+     */
+    public function testSingleUnpairedUTF16SurrogateInUnicodeEscape()
+    {
+        $json = '{ "data": "ud83d\ude4f" }';
+
+        $result = $this->serializer->deserialize($json, []);
+        $this->assertEquals($result['data'], 'ud83d\ude4f');
+    }
 }