[WIP] Server tests. Dataset manager. #2943
Conversation
| else: | ||
| upload_format_name = dump_format_name | ||
|
|
||
| url = f"/api/v1/tasks/{task_id}/annotations?format={upload_format_name}" |
There was a problem hiding this comment.
I suggest introducing a function for this code block.
There was a problem hiding this comment.
Done. Created new function.
| task_ann.init_from_db() | ||
| task_data = TaskData(task_ann.ir_data, Task.objects.get(pk=task_id)) | ||
| extractor = CvatTaskDataExtractor(task_data) | ||
| data_from_task_after_upload = datumaro.components.project.Dataset.from_extractors(extractor) |
There was a problem hiding this comment.
It's better to import Dataset and use it directly instead of having datumaro.components.project.. Currently, it resides in datumaro.components.dataset
There was a problem hiding this comment.
Done. Fixed import.
| dump_format_name = dump_format.DISPLAY_NAME | ||
|
|
||
| # TODO skip failed formats | ||
| if dump_format_name in [ |
There was a problem hiding this comment.
changed points values
Points in segmentation formats (CamVid, Segmentation mask, Mots) can't be stored exactly, because they need to be recognized after converting into masks.
There was a problem hiding this comment.
OK. These formats are not checked.
| ]: | ||
| continue | ||
|
|
||
| print("dump_format_name:", dump_format_name) |
There was a problem hiding this comment.
Consider using self.subTest() instead. You can find examples below in the file.
There was a problem hiding this comment.
Done. Added it actions.
| "Segmentation mask 1.1", # changed points values | ||
| "WiderFace 1.0", # issue #XXX | ||
| ]: | ||
| continue |
There was a problem hiding this comment.
Consider using self.skip() instead.
There was a problem hiding this comment.
Done. Added it actions.
| # upload annotations | ||
| url_upload = f"/api/v1/tasks/{task_id}/annotations?format={upload_format_name}" | ||
| with open(file_zip_name_after_change, 'rb') as binary_file: | ||
| with self.assertRaises(ValueError) as context: |
There was a problem hiding this comment.
Consider using assertRaisesRegex
There was a problem hiding this comment.
Done. Added it actions.
| task_data = TaskData(task_ann.ir_data, Task.objects.get(pk=task_id)) | ||
| extractor = CvatTaskDataExtractor(task_data) | ||
| data_from_task_after_upload = datumaro.components.project.Dataset.from_extractors(extractor) | ||
| compare_datasets(self, data_from_task_before_upload, data_from_task_after_upload) |
There was a problem hiding this comment.
Consider using require_images=True along with save_images=True in line 648 for dataset export to compare image data. This can affect the kind of exported format in several formats.
There was a problem hiding this comment.
Done. Added include_images actions.
| task = self._create_task(tasks["main"], images) | ||
|
|
||
| # create annotations | ||
| if dump_format_name in ["MOT 1.1", "MOTS PNG 1.0", "PASCAL VOC 1.1", "Segmentation mask 1.1", "TFRecord 1.0", "YOLO 1.1", "ImageNet 1.0", "WiderFace 1.0", "VGGFace2 1.0"]: |
There was a problem hiding this comment.
Strive to keep line length under 80 characters.
| "action": "download", | ||
| } | ||
| file_zip_name = osp.join(osp.dirname(__file__), 'assets', f'{test_name}_{dump_format_name}.zip') | ||
| self._download_file(url, data, self.admin, file_zip_name) |
There was a problem hiding this comment.
Try to split and simplify tests as much as possible. For example, I think, we could split format tests and API tests.
There was a problem hiding this comment.
Moved tests to other file.
| compare_datasets(self, data_datumaro["dataset"], data_datumaro["annotations"]) | ||
|
|
||
| @skip("Fail") | ||
| def test_api_v1_tasks_annotations_dump_and_upload_with_datumaro_many_jobs(self): |
There was a problem hiding this comment.
Please clarify, what is the problem here? A task with multiple jobs is a regular case.
|
We have another PR for these tests: #3004 |
Server tests. Dataset manager.
Motivation and context
How has this been tested?
Checklist
developbranchcvat-core, cvat-data and cvat-ui)
License
Feel free to contact the maintainers if that's a concern.