unittests now use a mock ray implementation

tests/_mock-site-packages/ray/__init__.py

@@ -0,0 +1,44 @@
+# Copyright 2026 NNAISENSE SA
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+"""
+This is a mock ray implementation which does not create processes.
+
+This mock implementation is meant to be used as a replacement for the
+now-obsolete local mode of the actual 'ray' library.
+"""
+
+
+from . import util
+from .core import (
+    ObjectRef,
+    get,
+    init,
+    is_initialized,
+    remote,
+    shutdown,
+)
+
+_THIS_IS_MOCK = True
+
+__all__ = (
+    "init",
+    "is_initialized",
+    "ObjectRef",
+    "remote",
+    "get",
+    "util",
+    "shutdown",
+)

tests/_mock-site-packages/ray/core.py

@@ -0,0 +1,257 @@
+# Copyright 2026 NNAISENSE SA
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+from collections.abc import Callable, Mapping, Sequence
+from typing import Any
+
+try:
+    # We assume that 'cloudpickle' is available because it is a dependency of the 'ray' library.
+    # So, in this mock ray implementation, we try to use cloudpickle if possible.
+    import cloudpickle as pickle
+except ImportError:
+    import pickle
+
+
+def init(*args, **kwargs):
+    """
+    Fake representation of ray initialization procedure.
+
+    Registers that the mock ray is initialized and ready-to-use.
+
+    Args:
+        args: Ignored.
+        kwargs: Ignored.
+    """
+    init.ialized = True
+
+
+init.ialized = False
+
+
+def is_initialized() -> bool:
+    """
+    Return True if 'ray' is initialized; False otherwise.
+    """
+    return init.ialized
+
+
+def shutdown():
+    """
+    Fake representation of ray de-initialization procedure.
+
+    Registers that the mock ray is de-initialized.
+    """
+    init.ialized = False
+
+
+def _pass_through_pickle(obj: Any) -> Any:
+    tmp = pickle.dumps(obj)
+    obj = pickle.loads(tmp)
+    return obj
+
+
+def _pass_args_through_pickle(obj: Mapping | Sequence) -> Mapping | Sequence:
+    if isinstance(obj, Sequence):
+        return tuple(_pass_through_pickle(x) for x in obj)
+    elif isinstance(obj, Mapping):
+        return {k: _pass_through_pickle(v) for k, v in obj.items()}
+    else:
+        raise TypeError(f"Expected a Sequence or a Mapping, but got {type(obj)}")
+
+
+class ObjectRef:
+    """
+    Fake representation of a remote object.
+    """
+
+    def __init__(self, obj: Any):
+        """
+        `__init__(...)`: Initialize the `ObjectRef`.
+
+        Args:
+            obj: The object for which the remote representation will be made.
+        """
+        self.__object = obj
+
+    def get(self) -> Any:
+        """
+        Get the object stored by the so-called remote actor.
+
+        To mimic the transfer of the object from the remote actor to the
+        requester process, the object is serialized and then de-serialized,
+        and then returned.
+        """
+        return _pass_through_pickle(self.__object)
+
+
+class _RemoteMethod:
+    """
+    Representation of a method of a so-called remote object.
+    """
+
+    def __init__(self, obj: Any, attrname: str):
+        """
+        `__init__(...)`: Initialize the `_RemoteMethod`.
+
+        Args:
+            obj: The object that owns the method to be represented.
+            attrname: Name of the method, as string.
+        """
+        self.__object = obj
+        self.__attrname = attrname
+
+    def remote(self, *args, **kwargs) -> Any:
+        """
+        Call the method with the given positional and/or keyword arguments.
+
+        To mimic the transfer of the arguments to the remote instance,
+        the arguments are serialized and then de-serialized.
+
+        Args:
+            args: Positional arguments to be passed to the method.
+            kwargs: Keyword arguments to be passed to the method.
+
+        Returns:
+            The result, wrapped by an `ObjectRef`.
+        """
+        args = _pass_args_through_pickle(args)
+        kwargs = _pass_args_through_pickle(kwargs)
+        return ObjectRef(getattr(self.__object, self.__attrname)(*args, **kwargs))
+
+
+class _RemoteInstance:
+    """
+    Representation of an object that lives on a so-called remote actor.
+
+    Let us say that we have an object `remote_obj` that is wrapped by
+    `_RemoteInstance`. A method `f` of `remote_obj` can be called like this:
+
+    ```python
+    remote_result = remote_obj.f.remote(...)
+    ```
+
+    At this point, `remote_result` is the result of the call, wrapped by
+    `ObjectRef`. The actual result can be fetched by using the
+    function `get`.
+    """
+
+    def __init__(self, obj: Any):
+        """
+        `__init__(...)`: Initialize the `_RemoteInstance`.
+
+        Args:
+            obj: The object to be treated as if it lives on a remote actor.
+        """
+        self.__object = obj
+
+    def __getattr__(self, attrname: str) -> _RemoteMethod:
+        return _RemoteMethod(self.__object, attrname)
+
+
+class _RemoteCaller:
+    """
+    Counterpart of a function or a class that is decorated via `remote`.
+    """
+
+    def __init__(self, obj: Any):
+        """
+        `__init__(...)`: Initialize the `_RemoteCaller`.
+        """
+        self.__object = obj
+
+    def options(self, *args, **kwargs) -> "_RemoteCaller":
+        """
+        No-op method for compatibility reasons.
+        """
+        return self
+
+    def remote(self, *args, **kwargs) -> Any:
+        """
+        Call the `remote`-decorated object with the given arguments.
+
+        If the decorated object is a function, the function will be called with
+        the given arguments, its result will be wrapped via `ObjectRef`,
+        and then returned.
+
+        If the decorated object is a class, the class will be instantiated with
+        the given arguments, the instance will be wrapped via `_RemoteInstance`,
+        and then returned.
+
+        Args:
+            args: Positional arguments to pass to the `remote`-decorated object.
+            kwargs: Keyword arguments to pass to the `remote`-decorated object.
+
+        Returns:
+            The result of the call, as an `ObjectRef` or as a
+            `_RemoteInstance`.
+
+        Raises:
+            RuntimeError: ray is not initialized
+        """
+        if not is_initialized():
+            raise RuntimeError("Cannot make a remote call without initializing 'ray' first")
+        args = _pass_args_through_pickle(args)
+        kwargs = _pass_args_through_pickle(kwargs)
+        result = self.__object(*args, **kwargs)
+        if isinstance(self.__object, type):
+            result = _RemoteInstance(result)
+        else:
+            result = ObjectRef(result)
+        return result
+
+
+def remote(obj: Callable | None = None, **kwargs):
+    """
+    Wrap a function or class so that it is handled by a so-called remote actor.
+
+    If the wrapped object is a function named `func`, it has to be called like
+    this: `func.remote(*args, **kwargs)`.
+
+    If the wrapped object is a class named `cls`, it has to be instantiated
+    like this: `cls.remote(*args, **kwargs)`.
+
+    To mimic the transfer of `*args` and `**kwargs` to a remote actor,
+    these arguments are first serialized and then de-serialized.
+    """
+    if obj is None:
+        return remote
+    else:
+        return _RemoteCaller(obj)
+
+
+def get(obj: ObjectRef | Sequence[ObjectRef]) -> Any:
+    """
+    Get the value(s) of so-called remote object(s).
+
+    Args:
+        obj: An `ObjectRef`, or a sequence (e.g. list or tuple) of `ObjectRef`
+            instances.
+
+    Returns:
+        The value(s) of the so-called remote object(s) (as a list of values
+        if `obj` was given as a sequence).
+    """
+    if isinstance(obj, ObjectRef):
+        return obj.get()
+    elif isinstance(obj, Sequence):
+        result = []
+        for x in obj:
+            if isinstance(x, ObjectRef):
+                result.append(x.get())
+            else:
+                raise TypeError(f"The list contains an object of this unexpected type: {type(x)}")
+        return result
+    else:
+        raise TypeError(f"Encountered an argument of this unexpected type: {type(obj)}")

tests/_mock-site-packages/ray/util.py

@@ -0,0 +1,75 @@
+# Copyright 2026 NNAISENSE SA
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+import random
+from collections.abc import Callable, Iterator, Sequence
+from typing import Any
+
+from .core import _RemoteInstance, get
+
+
+class ActorPool:
+    """
+    A mock implementation of `ray.util.ActorPool`.
+    """
+
+    SEED = 1
+
+    def __init__(self, actors: Sequence[_RemoteInstance]):
+        """
+        Initialize the `ActorPool`.
+
+        Args:
+            actors: List of so-called remote actors.
+        """
+        self.__actors = actors
+        self.__generator = random.Random(self.SEED)
+
+    def map(self, fn: Callable[[_RemoteInstance, Any], Any], objects: Sequence[Any]) -> Iterator[Any]:
+        """
+        Pair actors with objects, and for each pair, call `fn(actor, object)`.
+
+        Args:
+            fn: The function which specifies how an object is to be used
+                with its associated actor.
+            objects: A sequence (e.g. list or tuple) of object to be processed
+                by the pool of actors.
+
+        Returns:
+            Processed counterparts of the objects, as an iterator.
+        """
+        num_actors = len(self.__actors)
+        result = []
+        for i_obj, obj in enumerate(objects):
+            i_actor = i_obj % num_actors
+            actor = self.__actors[i_actor]
+            result.append(fn(actor, obj))
+        return iter(get(result))
+
+    def map_unordered(self, fn: Callable[[_RemoteInstance, Any], Any], objects: Sequence[Any]) -> Iterator[Any]:
+        """
+        Like the `map` method, but the results are shuffled.
+
+        The reasoning behind the shuffling is to mimic the fact that
+        `map_unordered` of the actual 'ray' library returns the processed
+        results in an undeterministic order.
+
+        *Note:* for reproducibility during tests, the shuffling depends
+        on a constant random seed, which is stored by class-level constant
+        `SEED`.
+        """
+        result = list(self.map(fn, objects))
+        self.__generator.shuffle(result)
+        return iter(result)