Remove previews downloading when task creating with cloud storage data (

cvat-ai#5499)

PR removes previews downloading from CS when task creating (skipped in
PR cvat-ai#5478). In addition, I had to change the test to check for the file
name existing in the message when the specified file is not found in the
bucket, because now the preview is no longer downloaded at the stage of
creating a task.

cvat/apps/engine/task.py

@@ -318,6 +318,9 @@ def _add_prefix(properties):
         content = list(map(_add_prefix, raw_content))
     else:
         sequence, content = cloud_storage_manifest.get_subset(sorted_media)
+    if not content:
+        raise ValidationError('There is no intersection of the files specified'
+                            'in the request with the contents of the bucket')
     sorted_content = (i[1] for i in sorted(zip(sequence, content)))
     manifest.create(sorted_content)
 
@@ -354,8 +357,6 @@ def _create_thread(db_task, data, isBackupRestore=False, isDatasetImport=False):
     )
 
     if is_data_in_cloud:
-        cloud_storage_instance = db_storage_to_storage_instance(db_data.cloud_storage)
-
         manifest = ImageManifestManager(db_data.get_manifest_path())
         cloud_storage_manifest = ImageManifestManager(
             os.path.join(db_data.cloud_storage.get_storage_dirname(), manifest_file),
@@ -392,13 +393,6 @@ def _create_thread(db_task, data, isBackupRestore=False, isDatasetImport=False):
         elif is_data_in_cloud:
             sorted_media = sort(media['image'], data['sorting_method'])
 
-            # download previews from cloud storage
-            data_size = len(sorted_media)
-            segment_step, *_ = _get_task_segment_data(db_task, data_size)
-            for preview_frame in range(0, data_size, segment_step):
-                preview = sorted_media[preview_frame]
-                cloud_storage_instance.download_file(preview, os.path.join(upload_dir, preview))
-
             # Define task manifest content based on cloud storage manifest content and uploaded files
             _create_task_manifest_based_on_cloud_storage_manifest(
                 sorted_media, cloud_storage_manifest_prefix,

tests/python/rest_api/test_tasks.py

@@ -425,7 +425,8 @@ def _wait_until_task_is_created(api: apis.TasksApi, task_id: int) -> models.RqSt
             sleep(1)
         raise Exception("Cannot create task")
 
-    def _test_create_task(self, username, spec, data, content_type, **kwargs):
+    @staticmethod
+    def _test_create_task(username, spec, data, content_type, **kwargs):
         with make_api_client(username) as api_client:
             (task, response) = api_client.tasks_api.create(spec, **kwargs)
             assert response.status == HTTPStatus.CREATED
@@ -435,7 +436,7 @@ def _test_create_task(self, username, spec, data, content_type, **kwargs):
             )
             assert response.status == HTTPStatus.ACCEPTED
 
-            status = self._wait_until_task_is_created(api_client.tasks_api, task.id)
+            status = TestPostTaskData._wait_until_task_is_created(api_client.tasks_api, task.id)
             assert status.state.value == "Finished"
 
         return task.id
@@ -798,16 +799,22 @@ def test_create_task_with_file_pattern(
             status = self._test_cannot_create_task(self._USERNAME, task_spec, data_spec)
             assert "No media data found" in status.message
 
+
+@pytest.mark.usefixtures("restore_db_per_function")
+@pytest.mark.usefixtures("restore_cvat_data")
+class TestWorkWithTask:
+    _USERNAME = "admin1"
+
     @pytest.mark.with_external_services
     @pytest.mark.parametrize(
         "cloud_storage_id, manifest, org",
         [(1, "manifest.jsonl", "")],  # public bucket
     )
-    def test_cannot_create_task_with_mythical_cloud_storage_data(
-        self, cloud_storage_id, manifest, org
+    def test_work_with_task_containing_non_stable_cloud_storage_files(
+        self, cloud_storage_id, manifest, org, cloud_storages, request
     ):
-        mythical_file = "mythical.jpg"
-        cloud_storage_content = [mythical_file, manifest]
+        image_name = "image_case_65_1.png"
+        cloud_storage_content = [image_name, manifest]
 
         task_spec = {
             "name": f"Task with mythical file from cloud storage {cloud_storage_id}",
@@ -821,8 +828,32 @@ def test_cannot_create_task_with_mythical_cloud_storage_data(
             "server_files": cloud_storage_content,
         }
 
-        status = self._test_cannot_create_task(self._USERNAME, task_spec, data_spec, org=org)
-        assert mythical_file in status.message
+        task_id = TestPostTaskData._test_create_task(
+            self._USERNAME, task_spec, data_spec, content_type="application/json", org=org
+        )
+
+        # save image from the "public" bucket and remove it temporary
+
+        s3_client = s3.make_client()
+        bucket_name = cloud_storages[cloud_storage_id]["resource"]
+
+        image = s3_client.download_fileobj(bucket_name, image_name)
+        s3_client.remove_file(bucket_name, image_name)
+        request.addfinalizer(
+            partial(s3_client.create_file, bucket=bucket_name, filename=image_name, data=image)
+        )
+
+        with make_api_client(self._USERNAME) as api_client:
+            try:
+                api_client.tasks_api.retrieve_data(
+                    task_id, number=0, quality="original", type="frame"
+                )
+                raise AssertionError("Frame should not exist")
+            except AssertionError:
+                raise
+            except Exception as ex:
+                assert ex.status == HTTPStatus.NOT_FOUND
+                assert image_name in ex.body
 
 
 @pytest.mark.usefixtures("restore_db_per_class")

tests/python/shared/utils/s3.py

@@ -2,6 +2,8 @@
 #
 # SPDX-License-Identifier: MIT
 
+from io import BytesIO
+
 import boto3
 from botocore.exceptions import ClientError
 
@@ -40,6 +42,11 @@ def file_exists(self, bucket: str, filename: str) -> bool:
             else:
                 raise
 
+    def download_fileobj(self, bucket: str, key: str) -> bytes:
+        with BytesIO() as data:
+            self.client.download_fileobj(Bucket=bucket, Key=key, Fileobj=data)
+            return data.getvalue()
+
 
 def make_client() -> S3Client:
     return S3Client(