fix: add dangling repo cleanup to hf repos (#891)

* feat: add clean_hf script to all.sh

scripts/all.sh

@@ -13,3 +13,4 @@ set +a
 ./scripts/doctest.sh
 ./scripts/notebook_runner.sh
 ./scripts/test.sh
+python "$(dirname "$0")/clean_hf.py"

scripts/clean_hf.py

@@ -0,0 +1,23 @@
+import os
+import warnings
+
+from dotenv import load_dotenv
+from huggingface_hub import HfApi  # type: ignore
+
+
+def clean_up_dangling_hf_repos(hugging_face_token: str) -> None:
+    api = HfApi(token=hugging_face_token)
+    datasets = list(
+        api.list_datasets(author="Aleph-Alpha", dataset_name="IL-temp-tests")
+    )
+    if len(datasets) > 0:
+        warnings.warn("dangling hf datasets found, attempting to delete")
+    for dataset in datasets:
+        api.delete_repo(dataset.id, repo_type="dataset", missing_ok=True)
+
+
+if __name__ == "__main__":
+    load_dotenv()
+    token = os.getenv("HUGGING_FACE_TOKEN")
+    assert isinstance(token, str)
+    clean_up_dangling_hf_repos(token)

tests/evaluation/infrastructure/test_hugging_face_repository.py

@@ -1,5 +1,3 @@
-from uuid import uuid4
-
 import huggingface_hub  # type: ignore
 
 from intelligence_layer.evaluation.infrastructure.hugging_face_repository import (
@@ -8,15 +6,15 @@
 
 
 def test_hugging_face_repository_can_create_and_delete_a_repository(
-    hugging_face_token: str,
+    hugging_face_token: str, hugging_face_test_repository_id: str
 ) -> None:
-    repository_id = f"Aleph-Alpha/test-{uuid4()}"
+    repository_id = hugging_face_test_repository_id + "unused-suffix"
 
     assert not huggingface_hub.repo_exists(
         repo_id=repository_id,
         token=hugging_face_token,
         repo_type="dataset",
-    ), f"This is very unlikely but it seems that the repository with the ID {repository_id} already exists."
+    ), f"The repository with the ID {repository_id} already exists. Try to run the clean_hf script."
 
     created_repository = HuggingFaceRepository(
         repository_id=repository_id,