fix: Fix double JSON encoding when saving eval set results

The `model_dump_json()` method already returns a JSON string, so wrapping it in `json.dumps()` was causing double encoding

Close #3993

Co-authored-by: George Weale <gweale@google.com>
PiperOrigin-RevId: 852455534

tests/unittests/evaluation/test_gcs_eval_set_results_manager.py

@@ -12,6 +12,8 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+import json
+
 from google.adk.errors.not_found_error import NotFoundError
 from google.adk.evaluation._eval_set_results_manager_utils import _sanitize_eval_set_result_name
 from google.adk.evaluation._eval_set_results_manager_utils import create_eval_set_result
@@ -165,6 +167,33 @@ class TestGcsEvalSetResultsManager:
     )
     assert retrieved_eval_set_result == eval_set_result
 
+  def test_get_eval_set_result_double_encoded_legacy(
+      self, gcs_eval_set_results_manager, mocker
+  ):
+    mocker.patch("time.time", return_value=12345678)
+    app_name = "test_app"
+    eval_set_id = "test_eval_set"
+    eval_case_results = _get_test_eval_case_results()
+    eval_set_result = create_eval_set_result(
+        app_name, eval_set_id, eval_case_results
+    )
+
+    blob_name = gcs_eval_set_results_manager._get_eval_set_result_blob_name(
+        app_name, eval_set_result.eval_set_result_id
+    )
+    blob = gcs_eval_set_results_manager.bucket.blob(blob_name)
+    double_encoded_json = json.dumps(eval_set_result.model_dump_json())
+    blob.upload_from_string(
+        double_encoded_json, content_type="application/json"
+    )
+
+    retrieved_eval_set_result = (
+        gcs_eval_set_results_manager.get_eval_set_result(
+            app_name, eval_set_result.eval_set_result_id
+        )
+    )
+    assert retrieved_eval_set_result == eval_set_result
+
   def test_list_eval_set_results(self, gcs_eval_set_results_manager, mocker):
     mocker.patch("time.time", return_value=123)
     app_name = "test_app"

tests/unittests/evaluation/test_local_eval_set_results_manager.py

@@ -85,11 +85,12 @@ class TestLocalEvalSetResultsManager:
     )
     assert os.path.exists(expected_file_path)
     with open(expected_file_path, "r") as f:
-      actual_eval_set_result_json = json.load(f)
+      actual_eval_set_result_data = json.load(f)
 
-    # need to convert eval_set_result to json
-    expected_eval_set_result_json = self.eval_set_result.model_dump_json()
-    assert expected_eval_set_result_json == actual_eval_set_result_json
+    # Verify the file contains a proper JSON object (not double-encoded)
+    # Use mode='json' to serialize enums to their values for comparison
+    expected_eval_set_result_data = self.eval_set_result.model_dump(mode="json")
+    assert expected_eval_set_result_data == actual_eval_set_result_data
 
   def test_get_eval_set_result(self, mocker):
     mock_time = mocker.patch("time.time")
@@ -102,6 +103,24 @@ class TestLocalEvalSetResultsManager:
     )
     assert retrieved_result == self.eval_set_result
 
+  def test_get_eval_set_result_double_encoded_legacy(self):
+    eval_history_dir = os.path.join(
+        self.agents_dir, self.app_name, _ADK_EVAL_HISTORY_DIR
+    )
+    os.makedirs(eval_history_dir, exist_ok=True)
+    eval_set_result_file_path = os.path.join(
+        eval_history_dir,
+        self.eval_set_result_name + _EVAL_SET_RESULT_FILE_EXTENSION,
+    )
+    double_encoded_json = json.dumps(self.eval_set_result.model_dump_json())
+    with open(eval_set_result_file_path, "w", encoding="utf-8") as f:
+      f.write(double_encoded_json)
+
+    retrieved_result = self.manager.get_eval_set_result(
+        self.app_name, self.eval_set_result_name
+    )
+    assert retrieved_result == self.eval_set_result
+
   def test_get_eval_set_result_not_found(self, mocker):
     mock_time = mocker.patch("time.time")
     mock_time.return_value = self.timestamp