high-level SDK does not support uploading annotations from the cloud

[burak43]

Actions before raising this issue

• I searched the existing issues and did not find anything similar.
• I read/searched the docs

Steps to Reproduce

from cvat_sdk import make_client
from cvat_sdk.core.proxies.tasks import ResourceType

with make_client(host="host", credentials=('user', 'password')) as client:
    task_spec = {
        "name": "dummy task",
        "project_id": 1,
        "source_storage": {'location': 'cloud_storage', 'cloud_storage_id': 1},
        "target_storage": {'location': 'cloud_storage', 'cloud_storage_id': 1},
    }
    data_params = {
        "cloud_storage_id": 1,
        "image_quality": 70,
        "sorting_method": "natural",
        "frame_step": 1,
    }

    task = client.tasks.create_from_data(
        spec=task_spec,
        resource_type=ResourceType.SHARE,
        resources=['path/to/s3/key/dir/'],
        data_params=data_params
    )

    # Upload annotations (since filename is s3 path, filename.stat() will fail.)
    task.import_annotations(
        format_name="COCO 1.0",
        filename='path/to/s3/key/dir/coco.json'
    )

when importing annotaions from cloud storage, filename.stat() will fail as filename is not a local file. Passing annotation_path and annotation_format to create_from_data will also fail. Then, what's the purpose of having source_storage in the task_spec if one cannot import resources from the cloud using high-level SDK?

Expected Behavior

Import annotations from the specified cloud storage.

Possible Solution

Provide a modified Uploader class that does not depent on a stream for the annotation file.

Context

Open a task using data and annotations both are available on the attached cloud storage. Task with data (images) can be opened but annotations cannot be uploaded from the cloud storage.

Environment

Python 3.11.9
VERSION = "2.25.0"

[Sunnykumar926]

I thought that when we try to import annotations from cloud storage, we get an error due to filename.stat(), as it checks for a local path while we provide 'path/to/s3/key/dir/coco.json'. To import the annotations, we first download the annotation file from cloud storage to local and then import it. please tell my i am correct or not