SDK: fix string field serialization for multipart/form-data requests (#5479)

Django REST Framework ignores the Content-Type on request body parts, so
it doesn't know that they are JSON-encoded. Instead, it just tries to
decode each part as if it was an `str()`-encoded value.

Change the encoding to match the decoding. The only type this matters
for is `str`, because `json.dumps` and `str` produce different encodings
for `str` values.

Remove `none_type` from the list of encodable types since, to my
knowledge, there's no way to encode a `None` value as a
`multipart/form-data` part in a way that DRF will understand.

CHANGELOG.md

@@ -135,9 +135,14 @@ non-ascii paths while adding files from "Connected file share" (issue #4428)
 - Missing source tag in project annotations (<https://github.com/opencv/cvat/pull/5408>)
 - Creating a task with a Git repository via the SDK
   (<https://github.com/opencv/cvat/issues/4365>)
+- Queries via the low-level API using the `multipart/form-data` Content-Type with string fields
+  (<https://github.com/opencv/cvat/pull/5479>)
+
+### Security
 - `Project.import_dataset` not waiting for completion correctly
   (<https://github.com/opencv/cvat/pull/5459>)
 
+
 ## \[2.2.0] - 2022-09-12
 ### Added
 - Added ability to delete frames from a job based on (<https://github.com/openvinotoolkit/cvat/pull/4194>)

cvat-sdk/gen/templates/openapi-generator/api_client.mustache

@@ -126,8 +126,8 @@ class ApiClient(object):
         self.default_headers['User-Agent'] = value
 
     def _serialize_post_parameter(self, obj):
-        if isinstance(obj, (str, int, float, none_type, bool)):
-            return ('', json.dumps(obj), 'application/json')
+        if isinstance(obj, (str, int, float, bool)):
+            return str(obj)
         elif isinstance(obj, io.IOBase):
             return self._serialize_file(obj)
         raise ApiValueError(

tests/python/rest_api/test_tasks.py

@@ -492,6 +492,40 @@ class TestPostTaskData:
             (task, _) = api_client.tasks_api.retrieve(task_id)
             assert task.size == 4
 
+    def test_can_create_task_with_sorting_method(self):
+        task_spec = {
+            "name": f"test {self._USERNAME} to create a task with a custom sorting method",
+            "labels": [
+                {
+                    "name": "car",
+                    "color": "#ff00ff",
+                    "attributes": [],
+                }
+            ],
+        }
+
+        image_files = generate_image_files(15)
+
+        task_data = {
+            "client_files": image_files[5:] + image_files[:5],  # perturb the order
+            "image_quality": 70,
+            "sorting_method": "natural",
+        }
+
+        # Besides testing that the sorting method is applied, this also checks for
+        # regressions of <https://github.com/opencv/cvat/issues/4962>.
+        task_id = self._test_create_task(
+            self._USERNAME, task_spec, task_data, content_type="multipart/form-data"
+        )
+
+        # check that the frames were sorted again
+        with make_api_client(self._USERNAME) as api_client:
+            data_meta, _ = api_client.tasks_api.retrieve_data_meta(task_id)
+
+            # generate_image_files produces files that are already naturally sorted
+            for image_file, frame in zip(image_files, data_meta.frames):
+                assert image_file.name == frame.name
+
     def test_can_get_annotations_from_new_task_with_skeletons(self):
         spec = {
             "name": f"test admin1 to create a task with skeleton",