Adding a test to outline resolution and support of the upper limits o…

…f file uris in cloud storage

label_studio/tests/tasks/test_presign_storage_data.py

@@ -1,4 +1,5 @@
 import pytest
+import base64
 from unittest.mock import patch
 from django.urls import reverse
 from rest_framework import status
@@ -147,3 +148,63 @@ def mock_task_get(*args, **kwargs):
             response.url
             == "https://presigned-url.com/czM6Ly9oeXBlcnRleHQtYnVja2V0L2ZpbGUgd2l0aCAvc3BhY2VzIGFuZCcgLyAnIC8gcXVvdGVzLmpwZw=="
         )
+
+    def test_successful_request_with_long_fileuri(
+        self, view, task, project, user, monkeypatch
+    ):
+        task.resolve_storage_uri.return_value = dict(
+            url="https://presigned-url.com/fileuri",
+            presign_ttl=3600,
+        )
+        project.has_permission.return_value = True
+        task.project = project
+
+        def mock_task_get(*args, **kwargs):
+            if kwargs["pk"] == 1:
+                return task
+            else:
+                raise Task.DoesNotExist
+
+        obj = MagicMock()
+        obj.get = mock_task_get
+        monkeypatch.setattr("tasks.models.Task.objects", obj)
+
+        # This is a long fileuri that will be hashed
+        # The total length of the fileuri can not be more than 1024 characters
+        # The length of the fileuri below is 1024 characters including the extension
+        longest_allowable_cloud_storage_path = "is/a/long/path/that/needs/to/be/1024/characters/long/so/that/it/gets/hashedis/a/long/path/that/needs/to/be/1024/characters/long/so/that/it/gets/hashedis/a/long/path/that/needs/to/be/1024/characters/long/so/that/it/gets/hashedis/a/long/path/that/needs/to/be/1024/characters/long/so/that/it/gets/hashedis/a/long/path/that/needs/to/be/1024/characters/long/so/that/it/gets/hashedis/a/long/path/that/needs/to/be/1024/characters/long/so/that/it/gets/hashedis/a/long/path/that/needs/to/be/1024/characters/long/so/that/it/gets/hashedis/a/long/path/that/needs/to/be/1024/characters/long/so/that/it/gets/hashedis/a/long/path/that/needs/to/be/1024/characters/long/so/that/it/gets/hashedis/a/long/path/that/needs/to/be/1024/characters/long/so/that/it/gets/hashedis/a/long/path/that/needs/to/be/1024/characters/long/so/that/it/gets/hashedis/a/long/path/that/needs/to/be/1024/characters/long/so/that/it/gets/hashedis/a/long/path/that/needs/to/be/1024/characters/long/so/that/it/gets/hashedis/long/path/that/needs/to/be/1024/characters.png"
+        longest_uri = f"azure-blob://{longest_allowable_cloud_storage_path}"
+
+        base64_encoded_uri = base64.urlsafe_b64encode(longest_uri.encode()).decode()
+
+        # Determining the absolute upper bounds which could be possible, and ensuring it resolves and is supported
+        longest_allowable_url_length = 2000  # This is the maximum length of a url in most browsers, and is the absolute upper bound
+        largest_allowable_task_key = 9223372036854775807
+        longest_presign_path = f"/tasks/{largest_allowable_task_key}/presign/?fileuri="
+        scheme_length = len("https://")
+        longest_presign_path_length = len(longest_presign_path)
+        longest_allowable_fileuri_hash_length = len(base64_encoded_uri)
+        remaining_url_origin_length = (
+            longest_allowable_url_length
+            - scheme_length
+            + longest_presign_path_length
+            + longest_allowable_fileuri_hash_length
+        )
+
+        # The user domain should be the shortest part of the url, but factoring lengthy subdomains with nested levels in staging and dev environments this is a safe allowance
+        assert remaining_url_origin_length >= 512
+
+        # Check this resolves correctly on the server
+        request = APIRequestFactory().get(
+            reverse("data_import:storage-data-presign", kwargs={"task_id": 1})
+            + f"?fileuri={base64_encoded_uri}"
+        )
+
+        request.user = user
+        force_authenticate(request, user)
+
+        response = view(request, task_id=1)
+
+        # And that the response is correct
+        assert response.status_code == status.HTTP_303_SEE_OTHER
+        assert response.url == "https://presigned-url.com/fileuri"