[Enhancement]: Add internal cache clean up

openml/datasets/__init__.py

@@ -7,6 +7,7 @@
     check_datasets_active,
     create_dataset,
     delete_dataset,
+    delete_dataset_cache,
     edit_dataset,
     fork_dataset,
     get_dataset,
@@ -22,6 +23,7 @@
     "create_dataset",
     "get_dataset",
     "get_datasets",
+    "delete_dataset_cache",
     "list_datasets",
     "OpenMLDataset",
     "OpenMLDataFeature",

openml/datasets/functions.py

@@ -1461,3 +1461,21 @@ def delete_dataset(dataset_id: int) -> bool:
         True if the deletion was successful. False otherwise.
     """
     return openml.utils._delete_entity("data", dataset_id)
+
+
+def delete_dataset_cache(dataset_id: int | str) -> None:
+    """Delete the cache directory for a specific dataset.
+
+    Parameters
+    ----------
+    dataset_id : int | str
+        Dataset ID (integer) or dataset name (string).
+    """
+    if isinstance(dataset_id, str):
+        try:
+            dataset_id = int(dataset_id)
+        except ValueError:
+            dataset_id = _name_to_id(str(dataset_id))
+
+    did_cache_dir = _get_cache_dir_for_id(DATASETS_CACHE_DIR_NAME, dataset_id)
+    _remove_cache_dir_for_id(DATASETS_CACHE_DIR_NAME, did_cache_dir)

openml/flows/__init__.py

@@ -4,6 +4,7 @@
 from .functions import (
     assert_flows_equal,
     delete_flow,
+    delete_flow_cache,
     flow_exists,
     get_flow,
     get_flow_id,
@@ -13,6 +14,7 @@
 __all__ = [
     "OpenMLFlow",
     "get_flow",
+    "delete_flow_cache",
     "list_flows",
     "get_flow_id",
     "flow_exists",

openml/flows/functions.py

@@ -552,3 +552,21 @@ def delete_flow(flow_id: int) -> bool:
         True if the deletion was successful. False otherwise.
     """
     return openml.utils._delete_entity("flow", flow_id)
+
+
+def delete_flow_cache(flow_id: int) -> None:
+    """Delete the cache directory for a specific flow.
+
+    Parameters
+    ----------
+    flow_id : int
+        Flow ID (integer).
+    """
+    if isinstance(flow_id, str):
+        try:
+            flow_id = int(flow_id)
+        except ValueError:
+            raise ValueError("Flow ID should be an integer") from None
+
+    fid_cache_dir = openml.utils._create_cache_directory_for_id(FLOWS_CACHE_DIR_NAME, flow_id)
+    openml.utils._remove_cache_dir_for_id(FLOWS_CACHE_DIR_NAME, fid_cache_dir)

openml/runs/__init__.py

@@ -2,6 +2,7 @@
 
 from .functions import (
     delete_run,
+    delete_run_cache,
     get_run,
     get_run_trace,
     get_runs,
@@ -22,6 +23,7 @@
     "run_model_on_task",
     "run_flow_on_task",
     "get_run",
+    "delete_run_cache",
     "list_runs",
     "get_runs",
     "get_run_trace",

openml/runs/functions.py

@@ -1301,3 +1301,21 @@ def delete_run(run_id: int) -> bool:
         True if the deletion was successful. False otherwise.
     """
     return openml.utils._delete_entity("run", run_id)
+
+
+def delete_run_cache(run_id: int) -> None:
+    """Delete the cache directory for a specific run.
+
+    Parameters
+    ----------
+    run_id : int
+        Run ID (integer).
+    """
+    if isinstance(run_id, str):
+        try:
+            run_id = int(run_id)
+        except ValueError:
+            raise ValueError("Run ID should be an integer") from None
+
+    run_cache_dir = openml.utils._create_cache_directory_for_id(RUNS_CACHE_DIR_NAME, run_id)
+    openml.utils._remove_cache_dir_for_id(RUNS_CACHE_DIR_NAME, run_cache_dir)

openml/tasks/__init__.py

@@ -3,6 +3,7 @@
 from .functions import (
     create_task,
     delete_task,
+    delete_task_cache,
     get_task,
     get_tasks,
     list_tasks,
@@ -28,6 +29,7 @@
     "create_task",
     "get_task",
     "get_tasks",
+    "delete_task_cache",
     "list_tasks",
     "OpenMLSplit",
     "TaskType",

openml/tasks/functions.py

@@ -612,3 +612,21 @@ def delete_task(task_id: int) -> bool:
         True if the deletion was successful. False otherwise.
     """
     return openml.utils._delete_entity("task", task_id)
+
+
+def delete_task_cache(task_id: int) -> None:
+    """Delete the cache directory for a specific task.
+
+    Parameters
+    ----------
+    task_id : int
+        Task ID (integer).
+    """
+    if isinstance(task_id, str):
+        try:
+            task_id = int(task_id)
+        except ValueError:
+            raise ValueError("Task ID should be an integer") from None
+
+    tid_cache_dir = openml.utils._create_cache_directory_for_id(TASKS_CACHE_DIR_NAME, task_id)
+    openml.utils._remove_cache_dir_for_id(TASKS_CACHE_DIR_NAME, tid_cache_dir)