Replace Black with Ruff formatter

.devcontainer/devcontainer.json

@@ -10,16 +10,15 @@
                     "source.organizeImports": true
                 },
                 "[python]": {
-                    "editor.defaultFormatter": "ms-python.black-formatter"
+                    "editor.defaultFormatter": "charliermarsh.ruff"
                 },
                 "editor.rulers": [
                     80
                 ]
             },
             "extensions": [
                 "charliermarsh.ruff",
-                "ms-python.python",
-                "ms-python.black-formatter"
+                "ms-python.python"
             ]
         }
     },

CONTRIBUTING.md

@@ -109,8 +109,7 @@ section of the README.
 
 ## Code style
 
-Keras uses [Black](https://black.readthedocs.io/en/stable/) and
-[Ruff](https://docs.astral.sh/ruff/) to format the code. Please refer to
+Keras uses [Ruff](https://docs.astral.sh/ruff/) to format the code. Please refer to
 [requirements-common.txt](https://github.com/keras-team/keras/blob/master/requirements-common.txt)
 for the required versions. Run the following command **at the root directory of
 the repo** to format your code.

examples/demo_custom_layer_backend_agnostic.py

@@ -47,9 +47,7 @@ def __init__(self, rate, name=None):
 
     def call(self, inputs):
         # Use `keras.random` for random ops.
-        return keras.random.dropout(
-            inputs, self.rate, seed=self.seed_generator
-        )
+        return keras.random.dropout(inputs, self.rate, seed=self.seed_generator)
 
 
 class MyModel(Model):

examples/demo_jax_distributed.py

@@ -27,9 +27,9 @@
 
 BATCH_SIZE = 192
 
-(x_train, train_labels), (
-    x_eval,
-    eval_labels,
+(
+    (x_train, train_labels),
+    (x_eval, eval_labels),
 ) = keras.datasets.mnist.load_data()
 x_train = np.expand_dims(x_train, axis=-1).astype(
     np.float32

guides/making_new_layers_and_models_via_subclassing.py

@@ -643,7 +643,7 @@ def __init__(
         intermediate_dim=64,
         latent_dim=32,
         name="autoencoder",
-        **kwargs
+        **kwargs,
     ):
         super().__init__(name=name, **kwargs)
         self.original_dim = original_dim

integration_tests/dataset_tests/boston_housing_test.py

@@ -3,7 +3,6 @@
 
 
 class BostonHousingTest(testing.TestCase):
-
     def test_load_data(self):
         (x_train, y_train), (x_test, y_test) = boston_housing.load_data()
         self.assertEqual(x_train.shape[1], 13)

integration_tests/dataset_tests/california_housing_test.py

@@ -3,7 +3,6 @@
 
 
 class CaliforniaHousingTest(testing.TestCase):
-
     def test_load_data_large(self):
         (x_train, y_train), (x_test, y_test) = california_housing.load_data(
             version="large"

integration_tests/model_visualization_test.py

@@ -41,7 +41,6 @@ def get_edge_dict(dot):
 
 
 class ModelVisualizationTest(testing.TestCase):
-
     def test_plot_sequential_model(self):
         model = keras.Sequential(
             [

keras/src/backend/common/masking_test.py

@@ -6,7 +6,6 @@
 
 
 class MaskingTest(testing.TestCase):
-
     def test_mask_on_eager_tensor(self):
         x = ops.zeros((2, 3))
         self.assertIsNone(get_keras_mask(x))
@@ -25,7 +24,6 @@ def test_mask_on_eager_tensor(self):
         self.assertIsNone(get_keras_mask(x))
 
     def test_mask_on_tracer_tensor(self):
-
         def fn(x):
             self.assertIsNone(get_keras_mask(x))
 

keras/src/backend/common/symbolic_scope_test.py

@@ -8,7 +8,6 @@
 
 class TestSymbolicScope(testing.TestCase):
     def test_basic_flow(self):
-
         # Define a function that behaves differently according to
         # `in_symbolic_scope`.
         def compute_loss(y, y_pred):

keras/src/backend/jax/core.py

@@ -153,7 +153,6 @@ def convert_keras_tensor_to_jax(x, fill_value=None):
             return x
 
         def wrapped_fn(*args, **kwargs):
-
             # Turn inputs that are sparse to BCOO tensors
             def to_bcoo_if_sparse(x, maybe_symbolic_x):
                 if (

keras/src/backend/jax/optimizer.py

@@ -13,7 +13,6 @@
 
 
 class JaxOptimizer(base_optimizer.BaseOptimizer):
-
     def _backend_apply_gradients(self, grads, trainable_variables):
         if self.gradient_accumulation_steps:
             is_update_step = (

keras/src/backend/jax/trainer.py

@@ -354,10 +354,9 @@ def fit(
             # Create the validation data using the training data. Only supported
             # for TF/numpy/jax arrays.
             (
-                x,
-                y,
-                sample_weight,
-            ), validation_data = array_slicing.train_validation_split(
+                (x, y, sample_weight),
+                validation_data,
+            ) = array_slicing.train_validation_split(
                 (x, y, sample_weight), validation_split=validation_split
             )
 

keras/src/backend/tensorflow/core.py

@@ -499,7 +499,6 @@ def _base_case():
             )
 
         def _recursive_case():
-
             odd_elems = _scan(reduced_elems)
 
             def _even_length_case():

keras/src/backend/tensorflow/optimizer.py

@@ -18,7 +18,6 @@
 
 
 class TFOptimizer(KerasAutoTrackable, base_optimizer.BaseOptimizer):
-
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
         self._distribution_strategy = tf.distribute.get_strategy()

keras/src/backend/tensorflow/saved_model_test.py

@@ -150,22 +150,18 @@ def call(self, inputs):
         named_product(struct_type=["tuple", "array", "dict"])
     )
     def test_model_with_input_structure(self, struct_type):
-
         class TupleModel(models.Model):
-
             def call(self, inputs):
                 x, y = inputs
                 return x + ops.mean(y, axis=1)
 
         class ArrayModel(models.Model):
-
             def call(self, inputs):
                 x = inputs[0]
                 y = inputs[1]
                 return x + ops.mean(y, axis=1)
 
         class DictModel(models.Model):
-
             def call(self, inputs):
                 x = inputs["x"]
                 y = inputs["y"]

keras/src/backend/tensorflow/trainer.py

@@ -99,7 +99,6 @@ def predict_step(self, data):
         return y_pred
 
     def _make_function(self, step_function):
-
         @tf.autograph.experimental.do_not_convert
         def one_step_on_data(data):
             """Runs a single training step on a batch of data."""
@@ -271,10 +270,9 @@ def fit(
             # Create the validation data using the training data. Only supported
             # for TF/numpy/jax arrays.
             (
-                x,
-                y,
-                sample_weight,
-            ), validation_data = array_slicing.train_validation_split(
+                (x, y, sample_weight),
+                validation_data,
+            ) = array_slicing.train_validation_split(
                 (x, y, sample_weight), validation_split=validation_split
             )
 

keras/src/backend/torch/trainer.py

@@ -195,10 +195,9 @@ def fit(
             # for TF/numpy/jax arrays.
             # TODO: Support torch tensors for validation data.
             (
-                x,
-                y,
-                sample_weight,
-            ), validation_data = array_slicing.train_validation_split(
+                (x, y, sample_weight),
+                validation_data,
+            ) = array_slicing.train_validation_split(
                 (x, y, sample_weight), validation_split=validation_split
             )
 

keras/src/datasets/reuters.py

@@ -124,8 +124,9 @@ def load_data(
         xs = [[w for w in x if skip_top <= w < num_words] for x in xs]
 
     idx = int(len(xs) * (1 - test_split))
-    x_train, y_train = np.array(xs[:idx], dtype="object"), np.array(
-        labels[:idx]
+    x_train, y_train = (
+        np.array(xs[:idx], dtype="object"),
+        np.array(labels[:idx]),
     )
     x_test, y_test = np.array(xs[idx:], dtype="object"), np.array(labels[idx:])
 

keras/src/export/export_lib.py

@@ -166,7 +166,6 @@ def track(self, resource):
             # Variables in the lists below are actually part of the trackables
             # that get saved, because the lists are created in __init__.
             if backend.backend() == "jax":
-
                 trainable_variables = tree.flatten(resource.trainable_variables)
                 non_trainable_variables = tree.flatten(
                     resource.non_trainable_variables
@@ -328,7 +327,6 @@ def serving_fn(x):
                     fn, input_signature=input_signature, autograph=False
                 )
             else:  # JAX backend
-
                 # 1. Create a stateless wrapper for `fn`
                 # 2. jax2tf the stateless wrapper
                 # 3. Create a stateful function that binds the variables with

keras/src/export/export_lib_test.py

@@ -74,7 +74,6 @@ def test_standard_model_export(self, model_type):
         named_product(model_type=["sequential", "functional", "subclass"])
     )
     def test_model_with_rng_export(self, model_type):
-
         class RandomLayer(layers.Layer):
             def __init__(self):
                 super().__init__()
@@ -104,7 +103,6 @@ def call(self, inputs):
         named_product(model_type=["sequential", "functional", "subclass"])
     )
     def test_model_with_non_trainable_state_export(self, model_type):
-
         class StateLayer(layers.Layer):
             def __init__(self):
                 super().__init__()
@@ -151,22 +149,18 @@ def test_model_with_tf_data_layer(self, model_type):
         named_product(struct_type=["tuple", "array", "dict"])
     )
     def test_model_with_input_structure(self, struct_type):
-
         class TupleModel(models.Model):
-
             def call(self, inputs):
                 x, y = inputs
                 return ops.add(x, y)
 
         class ArrayModel(models.Model):
-
             def call(self, inputs):
                 x = inputs[0]
                 y = inputs[1]
                 return ops.add(x, y)
 
         class DictModel(models.Model):
-
             def call(self, inputs):
                 x = inputs["x"]
                 y = inputs["y"]
@@ -214,7 +208,6 @@ def call(self, inputs):
         export_lib.export_model(revived_model, self.get_temp_dir())
 
     def test_model_with_multiple_inputs(self):
-
         class TwoInputsModel(models.Model):
             def call(self, x, y):
                 return x + y
@@ -302,7 +295,6 @@ def test_low_level_model_export_with_alias(self):
         named_product(model_type=["sequential", "functional", "subclass"])
     )
     def test_low_level_model_export_with_dynamic_dims(self, model_type):
-
         class ReductionLayer(layers.Layer):
             def call(self, inputs):
                 return ops.max(inputs, axis=1)
@@ -382,7 +374,6 @@ def test_low_level_model_export_with_jax2tf_kwargs(self):
         reason="This test is only for the JAX backend.",
     )
     def test_low_level_model_export_with_jax2tf_polymorphic_shapes(self):
-
         class SquareLayer(layers.Layer):
             def call(self, inputs):
                 return ops.matmul(inputs, inputs)

keras/src/layers/activations/prelu.py

@@ -37,7 +37,7 @@ def __init__(
         alpha_regularizer=None,
         alpha_constraint=None,
         shared_axes=None,
-        **kwargs
+        **kwargs,
     ):
         super().__init__(**kwargs)
         self.supports_masking = True

keras/src/layers/convolutional/conv1d.py

@@ -110,7 +110,7 @@ def __init__(
         activity_regularizer=None,
         kernel_constraint=None,
         bias_constraint=None,
-        **kwargs
+        **kwargs,
     ):
         super().__init__(
             rank=1,
@@ -130,7 +130,7 @@ def __init__(
             activity_regularizer=activity_regularizer,
             kernel_constraint=kernel_constraint,
             bias_constraint=bias_constraint,
-            **kwargs
+            **kwargs,
         )
 
     def _compute_causal_padding(self):

keras/src/layers/convolutional/conv1d_transpose.py

@@ -108,7 +108,7 @@ def __init__(
         activity_regularizer=None,
         kernel_constraint=None,
         bias_constraint=None,
-        **kwargs
+        **kwargs,
     ):
         super().__init__(
             rank=1,
@@ -127,5 +127,5 @@ def __init__(
             activity_regularizer=activity_regularizer,
             kernel_constraint=kernel_constraint,
             bias_constraint=bias_constraint,
-            **kwargs
+            **kwargs,
         )

keras/src/layers/convolutional/conv2d.py

@@ -104,7 +104,7 @@ def __init__(
         activity_regularizer=None,
         kernel_constraint=None,
         bias_constraint=None,
-        **kwargs
+        **kwargs,
     ):
         super().__init__(
             rank=2,
@@ -124,5 +124,5 @@ def __init__(
             activity_regularizer=activity_regularizer,
             kernel_constraint=kernel_constraint,
             bias_constraint=bias_constraint,
-            **kwargs
+            **kwargs,
         )

keras/src/layers/convolutional/conv2d_transpose.py

@@ -110,7 +110,7 @@ def __init__(
         activity_regularizer=None,
         kernel_constraint=None,
         bias_constraint=None,
-        **kwargs
+        **kwargs,
     ):
         super().__init__(
             rank=2,
@@ -129,5 +129,5 @@ def __init__(
             activity_regularizer=activity_regularizer,
             kernel_constraint=kernel_constraint,
             bias_constraint=bias_constraint,
-            **kwargs
+            **kwargs,
         )

keras/src/layers/convolutional/conv3d.py

@@ -110,7 +110,7 @@ def __init__(
         activity_regularizer=None,
         kernel_constraint=None,
         bias_constraint=None,
-        **kwargs
+        **kwargs,
     ):
         super().__init__(
             rank=3,
@@ -130,5 +130,5 @@ def __init__(
             activity_regularizer=activity_regularizer,
             kernel_constraint=kernel_constraint,
             bias_constraint=bias_constraint,
-            **kwargs
+            **kwargs,
         )