refactor(framework) Correct numpy client fit - return data type - error message

src/py/flwr/client/numpy_client.py

@@ -14,10 +14,11 @@
 # ==============================================================================
 """Flower client app."""
 
-
 from abc import ABC
 from typing import Callable
 
+import numpy as np
+
 from flwr.client.client import Client
 from flwr.common import (
     Config,
@@ -259,7 +260,10 @@ def _fit(self: Client, ins: FitIns) -> FitRes:
     results = self.numpy_client.fit(parameters, ins.config)  # type: ignore
     if not (
         len(results) == 3
-        and isinstance(results[0], list)
+        and isinstance(results[0], list)  # Check if it's a list
+        and all(
+            isinstance(p, np.ndarray) for p in results[0]
+        )  # Check elements are np.ndarray
         and isinstance(results[1], int)
         and isinstance(results[2], dict)
     ):

src/py/flwr/client/numpy_client_test.py

@@ -14,6 +14,7 @@
 # ==============================================================================
 """Flower NumPyClient tests."""
 
+import numpy as np
 
 from flwr.common import Config, NDArrays, Properties, Scalar
 
@@ -40,8 +41,15 @@ def get_parameters(self, config: Config) -> NDArrays:
     def fit(
         self, parameters: NDArrays, config: dict[str, Scalar]
     ) -> tuple[NDArrays, int, dict[str, Scalar]]:
-        """Simulate training by returning empty weights, 0 samples, empty metrics."""
-        return [], 0, {}
+        """Simulate training by returning updated weights, 0 samples, and metrics."""
+        # Simulate updated parameters as NumPy arrays
+        updated_parameters = [np.array([0.1, 0.2]), np.array([0.3, 0.4])]
+
+        # Simulate training metrics
+        metrics = {"accuracy": 0.95}
+
+        # Return updated parameters, number of examples, and metrics
+        return updated_parameters, 0, metrics
 
     def evaluate(
         self, parameters: NDArrays, config: dict[str, Scalar]
@@ -156,3 +164,75 @@ def test_has_evaluate_false() -> None:
 
     # Assert
     assert actual == expected
+
+
+def test_fit_return_type() -> None:
+    """Test that fit returns the correct type."""
+    # Prepare
+    client = OverridingClient()
+
+    # Execute
+    parameters, num_examples, metrics = client.fit(
+        parameters=[np.array([0.1, 0.2])], config={"epochs": 5}
+    )
+
+    # Assert
+    # Check if parameters is a list and all elements are np.ndarray
+    assert isinstance(parameters, list)
+    assert all(isinstance(p, np.ndarray) for p in parameters)
+
+    # Check other return types
+    assert isinstance(num_examples, int)
+    assert isinstance(metrics, dict)
+    assert all(
+        isinstance(k, str) and isinstance(v, (bool, bytes, float, int, str))
+        for k, v in metrics.items()
+    )
+
+
+def test_evaluate_return_type() -> None:
+    """Test that evaluate returns the correct type."""
+    # Prepare
+    client = OverridingClient()
+
+    # Execute
+    loss, num_examples, metrics = client.evaluate(
+        parameters=[np.array([0.1, 0.2])], config={"batch_size": 32}
+    )
+
+    # Assert
+    assert isinstance(loss, float)
+    assert isinstance(num_examples, int)
+    assert isinstance(metrics, dict)
+    assert all(isinstance(k, str) for k in metrics)  # Fix: Removed `.keys()`
+    assert all(isinstance(v, (bool, bytes, float, int, str)) for v in metrics.values())
+
+
+def test_get_parameters_return_type() -> None:
+    """Test that get_parameters returns the correct type."""
+    # Prepare
+    client = OverridingClient()
+
+    # Execute
+    parameters = client.get_parameters(config={})
+
+    # Assert
+    # Check if parameters is a list and all elements are np.ndarray
+    assert isinstance(parameters, list)
+    assert all(isinstance(p, np.ndarray) for p in parameters)
+
+
+def test_get_properties_return_type() -> None:
+    """Test that get_properties returns the correct type."""
+    # Prepare
+    client = OverridingClient()
+
+    # Execute
+    properties = client.get_properties(config={})
+
+    # Assert
+    assert isinstance(properties, dict)  # Properties is a dict[str, Scalar]
+    assert all(isinstance(k, str) for k in properties)
+    assert all(
+        isinstance(v, (bool, bytes, float, int, str)) for v in properties.values()
+    )