Merge branch 'main' into fix/race-condition-test-split

Author: Alm0stSurely

.github/workflows/test.yml

@@ -114,9 +114,9 @@ jobs:
         fi
 
         if [ "${{ matrix.sklearn-only }}" = "true" ]; then
-          marks="sklearn and not production and not uses_test_server"
+          marks="sklearn and not production_server and not test_server"
         else
-          marks="not production and not uses_test_server"
+          marks="not production_server and not test_server"
         fi
 
         pytest -n 4 --durations=20 --dist load -sv $codecov -o log_cli=true -m "$marks"
@@ -131,9 +131,9 @@ jobs:
         fi
 
         if [ "${{ matrix.sklearn-only }}" = "true" ]; then
-          marks="sklearn and production and not uses_test_server"
+          marks="sklearn and production_server and not test_server"
         else
-          marks="production and not uses_test_server"
+          marks="production_server and not test_server"
         fi
 
         pytest -n 4 --durations=20 --dist load -sv $codecov -o log_cli=true -m "$marks"
@@ -143,7 +143,7 @@ jobs:
       env:
         OPENML_TEST_SERVER_ADMIN_KEY: ${{ secrets.OPENML_TEST_SERVER_ADMIN_KEY }}
       run: |  # we need a separate step because of the bash-specific if-statement in the previous one.
-        pytest -n 4 --durations=20 --dist load -sv --reruns 5 --reruns-delay 1 -m "not uses_test_server"
+        pytest -n 4 --durations=20 --dist load -sv --reruns 5 --reruns-delay 1 -m "not test_server"
 
     - name: Check for files left behind by test
       if: matrix.os != 'windows-latest' && always()

docs/developer_setup.md

@@ -170,8 +170,8 @@ The OpenML Python SDK utilizes `pytest` markers to categorize tests based on dep
 | Marker            | Description                                                                 |
 |-------------------|-----------------------------------------------------------------------------|
 | `sklearn`          | Tests requiring `scikit-learn`. Skipped if the library is missing.          |
-| `production`      | Tests that interact with the live OpenML server (real API calls).         |
-| `uses_test_server`  | Tests requiring the OpenML test server environment.                       |
+| `production_server`| Tests that interact with the live OpenML server (real API calls).         |
+| `test_server`     | Tests requiring the OpenML test server environment.                       |
 
 ### Execution Examples
 
@@ -190,7 +190,7 @@ pytest -m sklearn
 Exclude production tests (local only):
 
 ```bash
-pytest -m "not production"
+pytest -m "not production_server"
 ```
 
 ### Admin Privilege Tests

openml/_api_calls.py

@@ -22,8 +22,8 @@
 from . import config
 from .__version__ import __version__
 from .exceptions import (
+    OpenMLAuthenticationError,
     OpenMLHashException,
-    OpenMLNotAuthorizedError,
     OpenMLServerError,
     OpenMLServerException,
     OpenMLServerNoResult,
@@ -515,11 +515,7 @@ def __parse_server_exception(
         400,  # run/42 delete
         460,  # task/42 delete
     ]:
-        msg = (
-            f"The API call {url} requires authentication via an API key.\nPlease configure "
-            "OpenML-Python to use your API as described in this example:"
-            "\nhttps://openml.github.io/openml-python/latest/examples/Basics/introduction_tutorial/#authentication"
-        )
-        return OpenMLNotAuthorizedError(message=msg)
+        msg = f"The API call {url} requires authentication via an API key."
+        return OpenMLAuthenticationError(message=msg)
 
     return OpenMLServerException(code=code, message=full_message, url=url)

openml/cli.py

@@ -102,15 +102,15 @@ def check_apikey(apikey: str) -> str:
 
 def configure_server(value: str) -> None:
     def check_server(server: str) -> str:
-        is_shorthand = server in ["test", "production"]
+        is_shorthand = server in ["test", "production_server"]
         if is_shorthand or looks_like_url(server):
             return ""
-        return "Must be 'test', 'production' or a url."
+        return "Must be 'test', 'production_server' or a url."
 
     def replace_shorthand(server: str) -> str:
         if server == "test":
-            return "https://test.openml.org/api/v1/xml"
-        if server == "production":
+            return f"{config.TEST_SERVER_URL}/api/v1/xml"
+        if server == "production_server":
             return "https://www.openml.org/api/v1/xml"
         return server
 
@@ -119,7 +119,7 @@ def replace_shorthand(server: str) -> str:
         value=value,
         check_with_message=check_server,
         intro_message="Specify which server you wish to connect to.",
-        input_message="Specify a url or use 'test' or 'production' as a shorthand: ",
+        input_message="Specify a url or use 'test' or 'production_server' as a shorthand: ",
         sanitize=replace_shorthand,
     )
 

openml/config.py

@@ -28,6 +28,8 @@
 OPENML_TEST_SERVER_ADMIN_KEY_ENV_VAR = "OPENML_TEST_SERVER_ADMIN_KEY"
 _TEST_SERVER_NORMAL_USER_KEY = "normaluser"
 
+TEST_SERVER_URL = "https://test.openml.org"
+
 
 class _Config(TypedDict):
     apikey: str
@@ -214,7 +216,7 @@ class ConfigurationForExamples:
     _last_used_server = None
     _last_used_key = None
     _start_last_called = False
-    _test_server = "https://test.openml.org/api/v1/xml"
+    _test_server = f"{TEST_SERVER_URL}/api/v1/xml"
     _test_apikey = _TEST_SERVER_NORMAL_USER_KEY
 
     @classmethod
@@ -470,7 +472,8 @@ def get_cache_directory() -> str:
 
     """
     url_suffix = urlparse(server).netloc
-    reversed_url_suffix = os.sep.join(url_suffix.split(".")[::-1])  # noqa: PTH118
+    url_parts = url_suffix.replace(":", "_").split(".")[::-1]
+    reversed_url_suffix = os.sep.join(url_parts)  # noqa: PTH118
     return os.path.join(_root_cache_directory, reversed_url_suffix)  # noqa: PTH118
 
 

openml/exceptions.py

@@ -63,5 +63,28 @@ class OpenMLNotAuthorizedError(OpenMLServerError):
     """Indicates an authenticated user is not authorized to execute the requested action."""
 
 
+class OpenMLAuthenticationError(OpenMLServerError):
+    """Exception raised when API authentication fails.
+
+    This typically occurs when:
+    - No API key is configured
+    - The API key is invalid or expired
+    - The API key format is incorrect
+
+    This is different from authorization (OpenMLNotAuthorizedError), which occurs
+    when a valid API key lacks permissions for the requested operation.
+    """
+
+    def __init__(self, message: str):
+        help_text = (
+            "\n\nTo fix this:\n"
+            "1. Get your API key from https://www.openml.org/\n"
+            "   (you'll need to register for a free account if you don't have one)\n"
+            "2. Configure your API key by following the authentication guide:\n"
+            "   https://openml.github.io/openml-python/latest/examples/Basics/introduction_tutorial/#authentication"
+        )
+        super().__init__(message + help_text)
+
+
 class ObjectNotPublishedError(PyOpenMLError):
     """Indicates an object has not been published yet."""

openml/tasks/functions.py

@@ -415,9 +415,10 @@ def get_task(
     if not isinstance(task_id, int):
         raise TypeError(f"Task id should be integer, is {type(task_id)}")
 
-    cache_key_dir = openml.utils._create_cache_directory_for_id(TASKS_CACHE_DIR_NAME, task_id)
-    tid_cache_dir = cache_key_dir / str(task_id)
-    tid_cache_dir_existed = tid_cache_dir.exists()
+    task_cache_directory = openml.utils._create_cache_directory_for_id(
+        TASKS_CACHE_DIR_NAME, task_id
+    )
+    task_cache_directory_existed = task_cache_directory.exists()
     try:
         task = _get_task_description(task_id)
         dataset = get_dataset(task.dataset_id, **get_dataset_kwargs)
@@ -431,8 +432,8 @@ def get_task(
         if download_splits and isinstance(task, OpenMLSupervisedTask):
             task.download_split()
     except Exception as e:
-        if not tid_cache_dir_existed:
-            openml.utils._remove_cache_dir_for_id(TASKS_CACHE_DIR_NAME, tid_cache_dir)
+        if not task_cache_directory_existed:
+            openml.utils._remove_cache_dir_for_id(TASKS_CACHE_DIR_NAME, task_cache_directory)
         raise e
 
     return task

openml/testing.py

@@ -47,7 +47,7 @@ class TestBase(unittest.TestCase):
         "user": [],
     }
     flow_name_tracker: ClassVar[list[str]] = []
-    test_server = "https://test.openml.org/api/v1/xml"
+    test_server = f"{openml.config.TEST_SERVER_URL}/api/v1/xml"
     admin_key = os.environ.get(openml.config.OPENML_TEST_SERVER_ADMIN_KEY_ENV_VAR)
     user_key = openml.config._TEST_SERVER_NORMAL_USER_KEY
 

pyproject.toml

@@ -133,10 +133,10 @@ filterwarnings=[
     "ignore:the matrix subclass:PendingDeprecationWarning"
 ]
 markers = [
-  "server: anything that connects to a server",
   "upload: anything that uploads to a server",
-  "production: any interaction with the production server",
+  "production_server: any interaction with the production server",
   "cache: anything that interacts with the (test) cache",
+  "test_server: tests that require the OpenML test server",
 ]
 
 # https://github.com/charliermarsh/ruff

tests/conftest.py

@@ -272,12 +272,12 @@ def as_robot() -> Iterator[None]:
 
 @pytest.fixture(autouse=True)
 def with_server(request):
-    if "production" in request.keywords:
+    if "production_server" in request.keywords:
         openml.config.server = "https://www.openml.org/api/v1/xml"
         openml.config.apikey = None
         yield
         return
-    openml.config.server = "https://test.openml.org/api/v1/xml"
+    openml.config.server = f"{openml.config.TEST_SERVER_URL}/api/v1/xml"
     openml.config.apikey = TestBase.user_key
     yield