chore: adjusting types and tests

keras_tuner/__init__.py

@@ -1,17 +1,18 @@
 """Keras Tuner Version."""
 
-from keras_tuner import oracles, tuners
-from keras_tuner.engine.hypermodel import HyperModel
-from keras_tuner.engine.hyperparameters import HyperParameter, HyperParameters
+# from keras_tuner import oracles, tuners
+# from keras_tuner.engine.hypermodel import HyperModel
+# from keras_tuner.engine.hyperparameters import HyperParameter, HyperParameters
 from keras_tuner.engine.objective import Objective
-from keras_tuner.engine.oracle import Oracle, synchronized
-from keras_tuner.engine.tuner import Tuner
-from keras_tuner.tuners import (
-    BayesianOptimization,
-    GridSearch,
-    Hyperband,
-    RandomSearch,
-    SklearnTuner,
-)
+
+# from keras_tuner.engine.oracle import Oracle, synchronized
+# from keras_tuner.engine.tuner import Tuner
+# from keras_tuner.tuners import (
+#     BayesianOptimization,
+#     GridSearch,
+#     Hyperband,
+#     RandomSearch,
+#     SklearnTuner,
+# )
 
 __version__ = "2.0.0"

keras_tuner/distribute/oracle_client.py

@@ -14,13 +14,16 @@
 """OracleClient class."""
 
 import os
+from typing import TYPE_CHECKING
 
 import grpc
 
 from keras_tuner import protos
 from keras_tuner.engine.hyperparameters import HyperParameters
-from keras_tuner.engine.oracle import Oracle
-from keras_tuner.engine.trial import Trial
+
+if TYPE_CHECKING:
+    from keras_tuner.engine.oracle import Oracle
+    from keras_tuner.engine.trial import Trial
 
 # The timeout is so high to prevent a rare race condition from happening.
 # We need clients to wait till chief oracle server starts. This normally takes

keras_tuner/engine/base_tuner.py

@@ -18,18 +18,23 @@
 import traceback
 import warnings
 from pathlib import Path
-from typing import cast
+from typing import TYPE_CHECKING, Any, cast
 
 from keras_tuner import backend, errors, utils
 from keras_tuner import config as config_module
 from keras_tuner.distribute import utils as dist_utils
-from keras_tuner.engine import hypermodel as hm_module
 from keras_tuner.engine import stateful, tuner_utils
-from keras_tuner.engine.hypermodel import HyperModel
-from keras_tuner.engine.oracle import Oracle
-from keras_tuner.engine.trial import Trial, TrialStatus
+from keras_tuner.engine.hypermodel import HyperModel, get_hypermodel
+from keras_tuner.engine.trial import TrialStatus
 from keras_tuner.types import _Verbose
 
+if TYPE_CHECKING:
+    from keras_tuner.engine.oracle import Oracle
+    from keras_tuner.engine.trial import Trial
+else:
+    Oracle = Any
+    Trial = Any
+
 
 class BaseTuner(stateful.Stateful):
     """Tuner base class.
@@ -107,7 +112,7 @@ def __init__(
             raise ValueError(msg)
 
         self.oracle = oracle
-        self.hypermodel = hm_module.get_hypermodel(hypermodel)
+        self.hypermodel = get_hypermodel(hypermodel)
 
         # Ops and metadata
         self.directory = directory

keras_tuner/engine/hyperparameters/hyperparameter.py → keras_tuner/engine/hyperparameters/HyperParameter.py

@@ -35,7 +35,7 @@ class HyperParameter:
 
     """
 
-    def __init__(self, name, default=None, conditions=None):
+    def __init__(self, name: str, default=None, conditions=None):
         self.name = name
         self._default = default
 

keras_tuner/engine/hyperparameters/hyperparameters.py → keras_tuner/engine/hyperparameters/HyperParameters.py

@@ -20,7 +20,7 @@
 from keras_tuner import protos
 from keras_tuner.engine import conditions as conditions_mod
 from keras_tuner.engine.hyperparameters import hp_types
-from keras_tuner.engine.hyperparameters import hyperparameter as hp_module
+from keras_tuner.engine.hyperparameters.HyperParameter import HyperParameter
 
 
 class HyperParameters:
@@ -38,7 +38,7 @@ class HyperParameters:
 
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         # Current name scopes.
         self._name_scopes = []
         # Current `Condition`s, managed by `conditional_scope`.
@@ -78,7 +78,7 @@ def name_scope(self, name):
 
     @contextlib.contextmanager
     def conditional_scope(self, parent_name, parent_values):
-        """Opens a scope to create conditional HyperParameters.
+        """Open a scope to create conditional HyperParameters.
 
         All `HyperParameter`s created under this scope will only be active when
         the parent `HyperParameter` specified by `parent_name` is equal to one
@@ -127,9 +127,8 @@ def build(self, hp):
         """
         parent_name = self._get_name(parent_name)  # Add name_scopes.
         if not self._exists(parent_name):
-            raise ValueError(
-                f"`HyperParameter` named: {parent_name} not defined."
-            )
+            msg = f"`HyperParameter` named: {parent_name} not defined."
+            raise ValueError(msg)
 
         condition = conditions_mod.Parent(parent_name, parent_values)
         self._conditions.append(condition)
@@ -145,7 +144,7 @@ def build(self, hp):
             self._conditions.pop()
 
     def is_active(self, hyperparameter):
-        """Checks if a hyperparameter is currently active for a `Trial`.
+        """Check if a hyperparameter is currently active for a `Trial`.
 
         A hyperparameter is considered active if and only if all its parent
         conditions are active, and not affected by whether the hyperparameter
@@ -163,7 +162,7 @@ def is_active(self, hyperparameter):
 
         """
         hp = hyperparameter
-        if isinstance(hp, hp_module.HyperParameter):
+        if isinstance(hp, HyperParameter):
             return self._conditions_are_active(hp.conditions)
         hp_name = str(hp)
         return any(
@@ -175,7 +174,7 @@ def _conditions_are_active(self, conditions):
         return all(condition.is_active(self.values) for condition in conditions)
 
     def _exists(self, name, conditions=None):
-        """Checks for a hyperparameter with the same name and conditions."""
+        """Check for a hyperparameter with the same name and conditions."""
         if conditions is None:
             conditions = self._conditions
 
@@ -187,7 +186,7 @@ def _exists(self, name, conditions=None):
         return False
 
     def _retrieve(self, hp):
-        """Gets or creates a hyperparameter.
+        """Get or creates a hyperparameter.
 
         Args:
             hp: A `HyperParameter` instance.
@@ -204,7 +203,7 @@ def _retrieve(self, hp):
         return self._register(hp)
 
     def _register(self, hyperparameter, overwrite=False):
-        """Registers a hyperparameter in this container.
+        """Register a hyperparameter in this container.
 
         Args:
             hp: A `HyperParameter` instance.
@@ -502,8 +501,8 @@ def Boolean(
             )
             return self._retrieve(hp)
 
-    def Fixed(self, name, value, parent_name=None, parent_values=None):
-        """Fixed, untunable value.
+    def Fixed(self, name: str, value, parent_name=None, parent_values=None):
+        """Untunable value.
 
         Args:
             name: A string. the name of parameter. Must be unique for each
@@ -659,7 +658,8 @@ def to_proto(self):
             elif isinstance(hp, hp_types.Boolean):
                 boolean_space.append(hp.to_proto())
             else:
-                raise ValueError(f"Unrecognized HP type: {hp}")
+                msg = f"Unrecognized HP type: {hp}"
+                raise TypeError(msg)
 
         values = {}
         for name, value in self.values.items():
@@ -705,7 +705,7 @@ def _get_name(self, name, name_scopes=None):
             else str(name)
         )
 
-    def _validate_name(self, name):
+    def _validate_name(self, name: str) -> None:
         for condition in self._conditions:
             if condition.name == name:
                 raise ValueError(

keras_tuner/engine/hyperparameters/__init__.py

@@ -14,18 +14,18 @@
 
 from keras import utils
 
-from keras_tuner.engine.hyperparameters import hp_types
-
-# these below are needed apparently
+# we re-export to make them available further up the chain.
 from keras_tuner.engine.hyperparameters.hp_types import (
     Boolean,
     Choice,
     Fixed,
     Float,
     Int,
 )
-from keras_tuner.engine.hyperparameters.hyperparameter import HyperParameter
-from keras_tuner.engine.hyperparameters.hyperparameters import HyperParameters
+
+from . import hp_types
+from .HyperParameter import HyperParameter
+from .HyperParameters import HyperParameters
 
 OBJECTS = [*hp_types.OBJECTS, HyperParameter, HyperParameters]
 

keras_tuner/engine/hyperparameters/hp_types/__init__.py

@@ -14,11 +14,13 @@
 
 from keras import utils
 
-from keras_tuner.engine.hyperparameters.hp_types.boolean_hp import Boolean
-from keras_tuner.engine.hyperparameters.hp_types.choice_hp import Choice
-from keras_tuner.engine.hyperparameters.hp_types.fixed_hp import Fixed
-from keras_tuner.engine.hyperparameters.hp_types.float_hp import Float
-from keras_tuner.engine.hyperparameters.hp_types.int_hp import Int
+# we export the modules' main functions so that they can be imported from
+# hp_types instead.
+from .boolean_hp import Boolean
+from .choice_hp import Choice
+from .fixed_hp import Fixed
+from .float_hp import Float
+from .int_hp import Int
 
 OBJECTS = (
     Fixed,

keras_tuner/engine/hyperparameters/hp_types/boolean_hp.py

@@ -15,10 +15,10 @@
 
 from keras_tuner import protos
 from keras_tuner.engine import conditions as conditions_mod
-from keras_tuner.engine.hyperparameters import hyperparameter
+from keras_tuner.engine.hyperparameters.HyperParameter import HyperParameter
 
 
-class Boolean(hyperparameter.HyperParameter):
+class Boolean(HyperParameter):
     """Choice between True and False.
 
     Args:

keras_tuner/engine/hyperparameters/hp_types/boolean_hp_test.py

@@ -14,22 +14,22 @@
 
 import pytest
 
-from keras_tuner.engine import hyperparameters as hp_module
+from . import Boolean
 
 
 def test_boolean():
     # Test default default
-    boolean = hp_module.Boolean("bool")
+    boolean = Boolean("bool")
     assert boolean.default is False
     # Test default setting
-    boolean = hp_module.Boolean("bool", default=True)
+    boolean = Boolean("bool", default=True)
     assert boolean.default is True
     # Wrong default type
     with pytest.raises(ValueError, match="must be a Python boolean"):
-        hp_module.Boolean("bool", default=None)
+        Boolean("bool", default=None)
     # Test serialization
-    boolean = hp_module.Boolean("bool", default=True)
-    boolean = hp_module.Boolean.from_config(boolean.get_config())
+    boolean = Boolean("bool", default=True)
+    boolean = Boolean.from_config(boolean.get_config())
     assert boolean.default is True
     assert boolean.name == "bool"
 
@@ -43,8 +43,8 @@ def test_boolean():
 
 
 def test_boolean_repr():
-    assert repr(hp_module.Boolean("bool")) == repr(hp_module.Boolean("bool"))
+    assert repr(Boolean("bool")) == repr(Boolean("bool"))
 
 
 def test_boolean_values_property():
-    assert list(hp_module.Boolean("bool").values) == [True, False]
+    assert list(Boolean("bool").values) == [True, False]

keras_tuner/engine/hyperparameters/hp_types/choice_hp.py

@@ -15,10 +15,11 @@
 
 from keras_tuner import protos
 from keras_tuner.engine import conditions as conditions_mod
-from keras_tuner.engine.hyperparameters import hp_utils, hyperparameter
+from keras_tuner.engine.hyperparameters import hp_utils
+from keras_tuner.engine.hyperparameters.HyperParameter import HyperParameter
 
 
-class Choice(hyperparameter.HyperParameter):
+class Choice(HyperParameter):
     """Choice of one value among a predefined set of possible values.
 
     Args:
@@ -39,15 +40,17 @@ class Choice(hyperparameter.HyperParameter):
     def __init__(self, name, values, ordered=None, default=None, **kwargs):
         super().__init__(name=name, default=default, **kwargs)
         if not values:
-            raise ValueError("`values` must be provided for `Choice`.")
+            msg = "`values` must be provided for `Choice`."
+            raise ValueError(msg)
 
         # Type checking.
         types = {type(v) for v in values}
         if len(types) > 1:
-            raise TypeError(
+            msg = (
                 "A `Choice` can contain only one type of value, "
                 f"found values: {values!s} with types {types}."
             )
+            raise TypeError(msg)
 
         # Standardize on str, int, float, bool.
         if isinstance(values[0], str):
@@ -67,26 +70,25 @@ def __init__(self, name, values, ordered=None, default=None, **kwargs):
         self._values = values
 
         if default is not None and default not in values:
-            raise ValueError(
+            msg = (
                 "The default value should be one of the choices. "
                 f"You passed: values={values}, default={default}"
             )
+            raise ValueError(msg)
         self._default = default
 
         # Get or infer ordered.
         self.ordered = ordered
         is_numeric = isinstance(values[0], (int | float))
         if self.ordered and not is_numeric:
-            raise ValueError("`ordered` must be `False` for non-numeric types.")
+            msg = "`ordered` must be `False` for non-numeric types."
+            raise ValueError(msg)
         if self.ordered is None:
             self.ordered = is_numeric
 
     def __repr__(self):
-        return (
-            f"Choice(name: '{self.name}', "
-            + f"values: {self._values}, "
-            + f"ordered: {self.ordered}, default: {self.default})"
-        )
+        return f"""Choice(name: '{self.name}', values: {self._values},
+        ordered: {self.ordered}, default: {self.default})"""
 
     @property
     def values(self):