Merge pull request #315 from MachineLearningLifeScience/feature/strict-typing

Adds strict typing using pyright, downgrades pytdc

Author: miguelgondu

.pre-commit-config.yaml

@@ -1,6 +1,6 @@
 repos:
 -   repo: https://github.com/pre-commit/pre-commit-hooks
-    rev: v3.2.0
+    rev: v5.0.0
     hooks:
     -   id: trailing-whitespace
         exclude: '.*\.pdb$'
@@ -20,4 +20,12 @@ repos:
   hooks:
     # Run the linter.
     - id: ruff
-      args: [ --fix ]
+      args: [ --fix ]
+- repo: local
+  hooks:
+    - id: pyright
+      name: pyright
+      entry: pyright
+      language: system
+      require_serial: true
+      types: [python]

pyproject.toml

@@ -53,7 +53,7 @@ ehrlichholo = [
     "pytorch-holo",
 ]
 tdc = [
-    "pytdc",
+    "pytdc==1.1.14",
 ]
 dockstring = [
     "dockstring"
@@ -65,7 +65,7 @@ rosetta_energy = [
     "biopython",
     "pyrosetta-installer",
 ]
-dev = ["black", "tox", "pytest", "bump-my-version"]
+dev = ["black", "tox", "pytest", "bump-my-version", "pre-commit", "pyright"]
 docs = ["sphinx", "furo"]
 
 [project.urls]
@@ -83,9 +83,15 @@ markers = [
     "poli__rosetta_energy: marks tests that run in poli__rosetta_energy",
     "poli__ehrlich_holo: marks tests that run in poli__ehrlich_holo environment",
     "poli__dms: marks tests that run in poli__dms environment",
+    "isolation: marks tests that require isolation of the black box function",
     "unmarked: All other tests, which usually run in the base environment",
 ]
 
+[tool.pyright]
+include = ["src/poli"]
+exclude = ["src/poli/core/util/proteins/rasp/inner_rasp", "src/poli/objective_repository/gfp_cbas", "examples", "src/poli/tests"]
+reportIncompatibleMethodOverride = "none"
+
 [tool.isort]
 profile = "black"
 
@@ -148,4 +154,7 @@ replace = 'version: {new_version}'
 [dependency-groups]
 dev = [
     "pre-commit>=4.2.0",
+    "pyright>=1.1.403",
+    "pytest>=8.4.0",
+    "ruff>=0.12.3",
 ]

src/poli/benchmarks/guacamol.py

@@ -125,7 +125,7 @@ def _initialize_problem(self, index: int) -> Problem:
         problem_factory = self.problem_factories[index]
 
         problem = problem_factory.create(
-            string_representation=self.string_representation,
+            string_representation=self.string_representation,  # type: ignore
             seed=self.seed,
             batch_size=self.batch_size,
             parallelize=self.parallelize,

src/poli/benchmarks/pmo.py

@@ -75,7 +75,7 @@ def __init__(
         batch_size: Union[int, None] = None,
         parallelize: bool = False,
         num_workers: Union[int, None] = None,
-        evaluation_budget: int = None,
+        evaluation_budget: int | None = None,
     ) -> None:
         super().__init__(
             string_representation=string_representation,

src/poli/benchmarks/toy_continuous_functions_benchmark.py

@@ -15,13 +15,14 @@
     https://www.sfu.ca/~ssurjano/optimization.html.
 """
 
-from typing import List, Union
+from typing import Sequence, Union, cast
 
 from poli.core.abstract_benchmark import AbstractBenchmark
 from poli.core.problem import Problem
 from poli.objective_repository import ToyContinuousProblemFactory
 from poli.objective_repository.toy_continuous_problem.toy_continuous_problem import (
     POSSIBLE_FUNCTIONS,
+    POSSIBLE_FUNCTIONS_TYPE,
     SIX_DIMENSIONAL_PROBLEMS,
     TWO_DIMENSIONAL_PROBLEMS,
 )
@@ -49,12 +50,12 @@ def __init__(
         self,
         n_dimensions: int = 2,
         embed_in: Union[int, None] = None,
-        dimensions_to_embed_in: Union[List[int], None] = None,
+        dimensions_to_embed_in: Union[list[int], None] = None,
         seed: Union[int, None] = None,
         batch_size: Union[int, None] = None,
         parallelize: bool = False,
         num_workers: Union[int, None] = None,
-        evaluation_budget: Union[int, List[int]] = None,
+        evaluation_budget: int | None = None,
     ) -> None:
         super().__init__(
             seed=seed,
@@ -66,7 +67,7 @@ def __init__(
         self.n_dimensions = n_dimensions
         self.embed_in = embed_in
         self.dimensions_to_embed_in = dimensions_to_embed_in
-        self.function_names = list(
+        self.function_names: Sequence[POSSIBLE_FUNCTIONS_TYPE] = list(  # type: ignore
             (
                 set(POSSIBLE_FUNCTIONS)
                 - set(TWO_DIMENSIONAL_PROBLEMS)
@@ -78,7 +79,9 @@ def __init__(
         )
 
     def _initialize_problem(self, index: int) -> Problem:
-        problem_factory: ToyContinuousProblemFactory = self.problem_factories[index]
+        problem_factory: ToyContinuousProblemFactory = cast(
+            ToyContinuousProblemFactory, self.problem_factories[index]
+        )
 
         problem = problem_factory.create(
             function_name=self.function_names[index],
@@ -121,7 +124,7 @@ def __init__(
         batch_size: Union[int, None] = None,
         parallelize: bool = False,
         num_workers: Union[int, None] = None,
-        evaluation_budget: Union[int, List[int]] = None,
+        evaluation_budget: int | None = None,
     ) -> None:
         super().__init__(
             seed,
@@ -134,7 +137,9 @@ def __init__(
         self.problem_factories = [ToyContinuousProblemFactory()] * len(self.embed_in)
 
     def _initialize_problem(self, index: int) -> Problem:
-        problem_factory: ToyContinuousProblemFactory = self.problem_factories[index]
+        problem_factory: ToyContinuousProblemFactory = cast(
+            ToyContinuousProblemFactory, self.problem_factories[index]
+        )
 
         problem = problem_factory.create(
             function_name="branin_2d",
@@ -174,7 +179,7 @@ def __init__(
         batch_size: Union[int, None] = None,
         parallelize: bool = False,
         num_workers: Union[int, None] = None,
-        evaluation_budget: Union[int, List[int]] = None,
+        evaluation_budget: int | None = None,
     ) -> None:
         super().__init__(
             seed,
@@ -187,7 +192,9 @@ def __init__(
         self.problem_factories = [ToyContinuousProblemFactory()] * len(self.embed_in)
 
     def _initialize_problem(self, index: int) -> Problem:
-        problem_factory: ToyContinuousProblemFactory = self.problem_factories[index]
+        problem_factory: ToyContinuousProblemFactory = cast(
+            ToyContinuousProblemFactory, self.problem_factories[index]
+        )
 
         if index == 0:
             problem = problem_factory.create(

src/poli/core/abstract_benchmark.py

@@ -1,13 +1,13 @@
 from __future__ import annotations
 
-from typing import List, Union
+from typing import Union
 
 from poli.core.abstract_problem_factory import AbstractProblemFactory
 from poli.core.problem import Problem
 
 
 class AbstractBenchmark:
-    problem_factories: List[AbstractProblemFactory]
+    problem_factories: list[AbstractProblemFactory]
     index: int = 0
 
     def __init__(
@@ -16,7 +16,7 @@ def __init__(
         batch_size: Union[int, None] = None,
         parallelize: bool = False,
         num_workers: Union[int, None] = None,
-        evaluation_budget: int = None,
+        evaluation_budget: int | None = None,
     ) -> None:
         self.seed = seed
         self.batch_size = batch_size
@@ -46,7 +46,7 @@ def info(self) -> str:
         raise NotImplementedError
 
     @property
-    def problem_names(self) -> List[str]:
+    def problem_names(self) -> list[str]:
         return [
             problem_factory.__module__.replace(
                 "poli.objective_repository.", ""

src/poli/core/abstract_black_box.py

@@ -5,9 +5,11 @@
 from __future__ import annotations
 
 from multiprocessing import Pool, cpu_count
+from typing import cast
 from warnings import warn
 
 import numpy as np
+from numpy.typing import NDArray
 
 from poli.core.black_box_information import BlackBoxInformation
 from poli.core.exceptions import BudgetExhaustedException
@@ -22,17 +24,17 @@ class AbstractBlackBox:
 
     Parameters
     ----------
-    batch_size : int, optional
+    batch_size : int | None, optional
         The batch size for evaluating the black box function. Default is None.
     parallelize : bool, optional
         Flag indicating whether to evaluate the black box function in parallel.
         Default is False.
-    num_workers : int, optional
+    num_workers : int | None, optional
         The number of workers to use for parallel evaluation. Default is None,
         which uses half of the available CPU cores.
-    evaluation_budget : int, optional
+    evaluation_budget : int | None, optional
         The maximum number of evaluations allowed for the black box function.
-        Default is None).
+        Default is None, which means an infinite budget.
 
     Attributes
     ----------
@@ -44,7 +46,7 @@ class AbstractBlackBox:
         Flag indicating whether to evaluate the black box function in parallel.
     num_workers : int
         The number of workers to use for parallel evaluation.
-    batch_size : int or None
+    batch_size : int | None
         The batch size for evaluating the black box function.
 
     Methods
@@ -84,13 +86,13 @@ def __init__(
 
         Parameters
         ----------
-        batch_size : int, optional
+        batch_size : int | None, optional
             The batch size for parallel execution, by default None.
         parallelize : bool, optional
             Flag indicating whether to parallelize the execution, by default False.
-        num_workers : int, optional
+        num_workers : int | None, optional
             The number of workers for parallel execution, by default we use half the available CPUs.
-        evaluation_budget : int, optional
+        evaluation_budget : int | None, optional
             The maximum number of evaluations allowed for the black box function, by default it is None, which means no limit.
         """
         self.observer = None
@@ -145,13 +147,13 @@ def set_observer(self, observer: AbstractObserver):
             )
         self.observer = observer
 
-    def set_observer_info(self, observer_info: object):
+    def set_observer_info(self, observer_info: dict[str, object] | None):
         """
         Set the observer information after initialization.
 
         Parameters
         ----------
-        observer_info : object
+        observer_info : dict[str, object]
             The information given by the observer after initialization.
         """
         self.observer_info = observer_info
@@ -160,7 +162,7 @@ def reset_evaluation_budget(self):
         """Resets the evaluation budget by setting the number of evaluations made to 0."""
         self.num_evaluations = 0
 
-    def __call__(self, x: np.array, context=None):
+    def __call__(self, x: NDArray[np.str_], context=None):
         """Calls the black box function.
 
         The purpose of this function is to enforce that inputs are equal across
@@ -340,7 +342,7 @@ def terminate(self) -> None:
         Terminate the black box optimization problem.
         """
         if hasattr(self, "inner_function"):
-            self.inner_function.terminate()
+            self.inner_function.terminate()  # type: ignore
         # if self.observer is not None:
         #     # NOTE: terminating a problem should gracefully end the observer process -> write the last state.
         #     self.observer.finish()
@@ -387,13 +389,13 @@ def __init__(self, f: AbstractBlackBox):
             batch_size=f.batch_size,
             parallelize=f.parallelize,
             num_workers=f.num_workers,
-            evaluation_budget=f.evaluation_budget,
+            evaluation_budget=cast(int | None, f.evaluation_budget),
         )
 
-    def __call__(self, x, context=None):
+    def __call__(self, x: NDArray[np.str_], context=None):
         return -self.f.__call__(x, context)
 
-    def _black_box(self, x, context=None):
+    def _black_box(self, x: NDArray[np.str_], context=None):
         return self.f._black_box(x, context)
 
     def __str__(self) -> str:

src/poli/core/abstract_problem_factory.py

@@ -30,11 +30,11 @@ class AbstractProblemFactory(metaclass=MetaProblemFactory):
 
     def create(
         self,
-        seed: int = None,
-        batch_size: int = None,
+        seed: int | None = None,
+        batch_size: int | None = None,
         parallelize: bool = False,
-        num_workers: int = None,
-        evaluation_budget: int = None,
+        num_workers: int | None = None,
+        evaluation_budget: int | None = None,
         force_isolation: bool = False,
     ) -> Problem:
         """

src/poli/core/benchmark_information.py

@@ -10,7 +10,7 @@ def __init__(
         fixed_length: bool,
         deterministic: bool,
         alphabet: list,
-        log_transform_recommended: bool = None,
+        log_transform_recommended: bool | None = None,
         discrete: bool = True,
         fidelity: Union[Literal["high", "low"], None] = None,
         padding_token: str = "",
@@ -111,7 +111,7 @@ def get_alphabet(self) -> list:
         """
         return self.alphabet
 
-    def log_transform_recommended(self) -> bool:
+    def is_log_transform_recommended(self) -> bool | None:
         """
         Returns whether the black-box recommends log-transforming the targets.