Extend server tests for dataset manager. #3192
Conversation
Commented out tests passed Ok Uncommented tests failed if launched in single test session, seems that tasks created via POST request to url /api/v1/tasks have the same id (1)
…zuevx/cvat into mz/server_test_dataset_manager
…zuevx/cvat into mz/server_test_dataset_manager
…zuevx/cvat into mz/server_test_dataset_manager
…o mz/server_test_dataset_manager
…zuevx/cvat into ms/server_test_dataset_manager # Conflicts: # kubernetes-templates/03_database_deployment.yml # kubernetes-templates/03_redis_deployment.yml # kubernetes-templates/04_cvat_backend_deployment.yml # kubernetes-templates/04_cvat_frontend_deployment.yml # kubernetes-templates/04_database_service.yml # kubernetes-templates/04_redis_service.yml # kubernetes-templates/05_cvat_backend_service.yml # kubernetes-templates/05_cvat_frontend_service.yml # kubernetes-templates/05_cvat_proxy_deployment.yml # kubernetes-templates/05_cvat_proxy_service.yml # kubernetes-templates/README.md
| content = BytesIO(b"".join(response.streaming_content)) | ||
| with open(file_zip_name, "wb") as f: | ||
| f.write(content.getvalue()) |
There was a problem hiding this comment.
| content = BytesIO(b"".join(response.streaming_content)) | |
| with open(file_zip_name, "wb") as f: | |
| f.write(content.getvalue()) | |
| content = b"".join(response.streaming_content) | |
| with open(file_zip_name, "wb") as f: | |
| f.write(content) |
There was a problem hiding this comment.
It doesn't work that way.
| task_ann_data = task_ann.data | ||
| self.assertEqual(len(task_ann_data["shapes"]), len(task_ann_prev_data["shapes"])) | ||
|
|
||
| def test_api_v1_unit_test_on_normalize_shape_function(self): |
There was a problem hiding this comment.
What does this test test? Also, I see repeated cases with different comments. Also, this test does not use rest api, maybe its better place in test_annotations.py?
There was a problem hiding this comment.
Ok, what about other problems with this test? Currently, it only checks for not throwing an exception. It has 2 repeated test cases.
There was a problem hiding this comment.
Repeated tests fixed and check length of returned values.
…o ms/server_test_dataset_manager # Conflicts: # cvat/apps/dataset_manager/tests/test_rest_api_formats.py
| dump_format_name = dump_format.DISPLAY_NAME | ||
| with self.subTest(): | ||
| if dump_format_name in [ | ||
| "MOT 1.1", # issue #2925 |
There was a problem hiding this comment.
@MashaSS , I see that some of these problems were fixed but tests are disabled. Could you please re-check?
There was a problem hiding this comment.
MOT 1.1 and MOTS PNG 1.0 cannot be checked by datumaro. I removed old comments about this issue.
| from cvat.apps.dataset_manager.bindings import CvatTaskDataExtractor, TaskData | ||
| from cvat.apps.dataset_manager.task import TaskAnnotation | ||
| from cvat.apps.dataset_manager.annotation import TrackManager |
There was a problem hiding this comment.
@MashaSS , the import is unused for now. It leads to a warning from pylint. Could you please fix?
| "name": "camera_id", | ||
| "mutable": false, | ||
| "input_type": "number", | ||
| "values": ["1", "2", "3", "4", "5", "6"] |
There was a problem hiding this comment.
@MashaSS , could you please help me to recall? Why do we have a list of values for "text" and "number"?
There was a problem hiding this comment.
Values for "number" define range of possible values (min, max, step). Values for "text" are used in requests but don't define any value restrictions.
There was a problem hiding this comment.
Values for text should be text. Do you agree?
There was a problem hiding this comment.
Yes, could you please clarify what test values should be changed? Attributes color and center in icdar_segmentation task?
| def test_api_v1_unit_test_on_normalize_shape_function(self): | ||
| # 3 points | ||
| norm_shape = TrackManager.normalize_shape({ |
There was a problem hiding this comment.
@MashaSS , @yasakova-anastasia , why do we test "normalize_shape" function?
There was a problem hiding this comment.
It's the part of code related to dataset manager. It was decided to place it in test_annotation.py because test doesn't use rest api.
There was a problem hiding this comment.
@MashaSS , the function should be removed in the code. It is dead code. No need to cover it by tests. Just remove.
There was a problem hiding this comment.
Removed dead code function and test for this function.
| url = self._generate_url_upload_tasks_annotations(task_id, upload_format_name) | ||
|
|
||
| with open(file_zip_name, 'rb') as binary_file: | ||
| response = self._put_request_with_data(url, {"annotation_file": binary_file}, user) |
There was a problem hiding this comment.
@MashaSS , I miss real checks in the test. Basically you create a task, create some annotations, dump the file. The same for upload is done. But the test doesn't check that uploaded and dumped information corresponds to each over.
@yasakova-anastasia , need to design tests which are going to check real data. Not just HTTP codes.
Extend server tests for pull request #3004.