First mypy refactorings

.pre-commit-config.yaml

@@ -21,3 +21,9 @@ repos:
     hooks:
     -   id: pyupgrade
         args: ['--py37-plus']
+
+  - repo: https://github.com/pre-commit/mirrors-mypy
+    rev: v0.971
+    hooks:
+      - id: mypy
+        files: ^darts

CHANGELOG.md

@@ -7,6 +7,7 @@ Darts is still in an early development phase, and we cannot always guarantee bac
 - Added support for retraining model(s) every `n` iteration and on custom condition in `historical_forecasts` method of `ForecastingModel` abstract class. Addressed issues [#135](https://github.com/unit8co/darts/issues/135) and [#623](https://github.com/unit8co/darts/issues/623) by [Francesco Bruzzesi](https://github.com/fbruzzesi).
 - New LayerNorm alternatives, RMSNorm and LayerNormNoBias [#1113](https://github.com/unit8co/darts/issues/1113) by [Greg DeVos](https://github.com/gdevos010).
 - Fixed type hinting for ExponentialSmoothing model [#1185](https://https://github.com/unit8co/darts/pull/1185) by [Rijk van der Meulen](https://github.com/rijkvandermeulen)
+- Enabled set-up for running mypy in pre-commit hooks and performed necessary mypy refactorings for dtw [#1202](https://https://github.com/unit8co/darts/pull/1202) by [Rijk van der Meulen](https://github.com/rijkvandermeulen)
 
 [Full Changelog](https://github.com/unit8co/darts/compare/0.21.0...master)
 

darts/dataprocessing/dtw/cost_matrix.py

@@ -93,14 +93,12 @@ def __init__(self, window: CRWindow):
         self.n = window.n
         self.m = window.m
         self.window = window
-        self.offsets = np.empty(self.n + 2, dtype=int)
         self.column_ranges = window.column_ranges
-        self.offsets[0] = 0
-        np.cumsum(window.column_lengths(), out=self.offsets[1:])
-
-        len = self.offsets[-1]
-
-        self.offsets = array.array("i", self.offsets)
+        offsets = np.empty(self.n + 2, dtype=int)
+        offsets[0] = 0
+        np.cumsum(window.column_lengths(), out=offsets[1:])
+        len = offsets[-1]
+        self.offsets = array.array("i", offsets)
         self.dense = array.array("f", repeat(np.inf, len))
 
     def fill(self, value):

darts/dataprocessing/dtw/dtw.py

@@ -1,5 +1,5 @@
 import copy
-from typing import Callable, Union
+from typing import Callable, Optional, Tuple, Union
 
 import numpy as np
 import pandas as pd
@@ -14,14 +14,15 @@
 
 logger = get_logger(__name__)
 
-SeriesValue = Union[np.ndarray, np.floating]
-DistanceFunc = Callable[[SeriesValue, SeriesValue], float]
+DistanceFunc = Union[
+    Callable[[float, float], float], Callable[[np.ndarray, np.ndarray], np.ndarray]
+]
 
 
 # CORE ALGORITHM
 def _dtw_cost_matrix(
     x: np.ndarray, y: np.ndarray, dist: DistanceFunc, window: Window
-) -> np.ndarray:
+) -> CostMatrix:
 
     dtw = CostMatrix._from_window(window)
 
@@ -120,6 +121,7 @@ def _fast_dtw(
     m = len(y)
     min_size = radius + 2
 
+    window: Union[NoWindow, CRWindow]
     if n < min_size or m < min_size or radius == -1:
         window = NoWindow()
         window.init_size(n, m)
@@ -154,7 +156,7 @@ def __init__(self, series1: TimeSeries, series2: TimeSeries, cost: CostMatrix):
         self.series2 = series2
         self.cost = cost
 
-    from ._plot import plot, plot_alignment
+    from ._plot import plot, plot_alignment  # type: ignore
 
     def path(self) -> np.ndarray:
         """
@@ -168,7 +170,7 @@ def path(self) -> np.ndarray:
 
         if hasattr(self, "_path"):
             return self._path
-        self._path = _dtw_path(self.cost)
+        self._path: np.ndarray = _dtw_path(self.cost)
         return self._path
 
     def distance(self) -> float:
@@ -191,10 +193,10 @@ def mean_distance(self) -> float:
             return self._mean_distance
 
         path = self.path()
-        self._mean_distance = self.distance() / len(path)
+        self._mean_distance: float = self.distance() / len(path)
         return self._mean_distance
 
-    def warped(self) -> (TimeSeries, TimeSeries):
+    def warped(self) -> Tuple[TimeSeries, TimeSeries]:
         """
         Warps the two time series according to the warp path returned by .path(), which minimizes
         the pair-wise distance.
@@ -254,19 +256,19 @@ def warped(self) -> (TimeSeries, TimeSeries):
         )
 
 
-def default_distance_multi(x_values: np.ndarray, y_values: np.ndarray):
+def default_distance_multi(x_values: np.ndarray, y_values: np.ndarray) -> np.ndarray:
     return np.sum(np.abs(x_values - y_values))
 
 
-def default_distance_uni(x_value: float, y_value: float):
+def default_distance_uni(x_value: float, y_value: float) -> float:
     return abs(x_value - y_value)
 
 
 def dtw(
     series1: TimeSeries,
     series2: TimeSeries,
     window: Window = NoWindow(),
-    distance: Union[DistanceFunc, None] = None,
+    distance: Optional[DistanceFunc] = None,
     multi_grid_radius: int = -1,
 ) -> DTWAlignment:
     """
@@ -338,14 +340,14 @@ def dtw(
         values_y = series2.values(copy=False)
 
     raise_if(
-        np.any(np.isnan(values_x)),
+        bool(np.any(np.isnan(values_x))),
         "Dynamic Time Warping does not support nan values. "
         "You can use the module darts.utils.missing_values to fill them, "
         "before passing them to dtw.",
         logger,
     )
     raise_if(
-        np.any(np.isnan(values_y)),
+        bool(np.any(np.isnan(values_y))),
         "Dynamic Time Warping does not support nan values. "
         "You can use the module darts.utils.missing_values to fill them,"
         "before passing it into dtw",

darts/dataprocessing/dtw/window.py

@@ -1,6 +1,7 @@
 import array
 from abc import ABC, abstractmethod
 from math import atan, tan
+from typing import Optional, Tuple
 
 import numpy as np
 
@@ -30,7 +31,7 @@ def __len__(self) -> int:
         pass
 
     @abstractmethod
-    def column_index(self, elem: (int, int)) -> int:
+    def column_index(self, elem: Tuple[int, int]) -> int:
         """
         Parameters
         ----------
@@ -96,7 +97,7 @@ class NoWindow(Window):
     def __len__(self):
         return self.n * self.m + 1  # include (0,0) element
 
-    def column_index(self, elem: (int, int)):
+    def column_index(self, elem: Tuple[int, int]):
         return elem[1] - 1
 
     def column_length(self, column: int) -> int:
@@ -106,6 +107,7 @@ def column_lengths(self) -> np.ndarray:
         result = np.empty(self.n + 1)
         result.fill(self.m)
         result[0] = 1
+        return result
 
     def __iter__(self):
         for i in range(1, self.n + 1):
@@ -117,7 +119,7 @@ def gtz(value):  # greater than zero
     return value if value > 0 else 0
 
 
-class CRWindow:
+class CRWindow(Window):
     """
     Compressed row representation window.
     Stores the range of active grid cells in each column.
@@ -128,7 +130,7 @@ class CRWindow:
     length: int
     column_ranges: array.array
 
-    def __init__(self, n: int, m: int, ranges: np.ndarray = None):
+    def __init__(self, n: int, m: int, ranges: Optional[np.ndarray] = None):
         """
         Parameters
         ----------
@@ -154,25 +156,25 @@ def __init__(self, n: int, m: int, ranges: np.ndarray = None):
             start = ranges[:, 0]
             end = ranges[:, 1]
 
-            raise_if(np.any(start < 0), "Start must be >=0")
-            raise_if(np.any(end > m), "End must be <m")
+            raise_if(bool(np.any(start < 0)), "Start must be >=0")
+            raise_if(bool(np.any(end > m)), "End must be <m")
 
             diff = np.maximum(end - start, 0)
             self.length = np.sum(diff)
 
             ranges[1:] += 1
             ranges = ranges.flatten()
+            self.column_ranges = array.array("i", ranges)
         else:
             ranges = np.zeros((n + 1) * 2, dtype=int)
             ranges[0::2] = self.m  # start
             ranges[1::2] = 0  # end
-            ranges = array.array("i", ranges)
+            ranges_array = array.array("i", ranges)
 
-            ranges[0] = 0
-            ranges[1] = 1
+            ranges_array[0] = 0
+            ranges_array[1] = 1
             self.length = 1
-
-        self.column_ranges = array.array("i", ranges)
+            self.column_ranges = array.array("i", ranges_array)
 
     def add_range(self, column: int, start: int, end: int):
         """
@@ -210,7 +212,7 @@ def add_range(self, column: int, start: int, end: int):
         self.column_ranges[start_idx] = start
         self.column_ranges[end_idx] = end
 
-    def add(self, elem: (int, int)):
+    def add(self, elem: Tuple[int, int]):
         """
         Mark grid cell as active.
 
@@ -226,7 +228,7 @@ def column_length(self, column: int) -> int:
         start, end = self.column_ranges[column]
         return gtz(end - start)
 
-    def column_index(self, elem: (int, int)) -> int:
+    def column_index(self, elem: Tuple[int, int]) -> int:
         i, j = elem
 
         start, end = self.column_ranges[i]
@@ -235,7 +237,7 @@ def column_index(self, elem: (int, int)) -> int:
         else:
             return j - start
 
-    def __contains__(self, elem: (int, int)) -> bool:
+    def __contains__(self, elem: Tuple[int, int]) -> bool:
         i, j = elem
         start, end = self.column_ranges[i]
         return start <= j < end

darts/dataprocessing/pipeline.py

@@ -78,7 +78,7 @@ def __init__(
 
         if transformers is None or len(transformers) == 0:
             logger.warning("Empty pipeline created")
-            self._transformers: Sequence[BaseDataTransformer[TimeSeries]] = []
+            self._transformers: Sequence[BaseDataTransformer] = []
         elif copy:
             self._transformers = deepcopy(transformers)
         else:
@@ -196,7 +196,7 @@ def inverse_transform(
             )
 
             for transformer in reversed(self._transformers):
-                data = transformer.inverse_transform(data)
+                data = transformer.inverse_transform(data)  # type: ignore
             return data
         else:
             for transformer in reversed(self._transformers):
@@ -237,6 +237,7 @@ def __getitem__(self, key: Union[int, slice]) -> "Pipeline":
             logger,
         )
 
+        transformers: Sequence[BaseDataTransformer]
         if isinstance(key, int):
             transformers = [self._transformers[key]]
         else:

mypy.ini

@@ -0,0 +1,71 @@
+[mypy]
+python_version = 3.9
+
+[mypy-numpy.*]
+ignore_missing_imports = True
+
+[mypy-pandas.*]
+ignore_missing_imports = True
+
+[mypy-sklearn.*]
+ignore_missing_imports = True
+
+[mypy-scipy.*]
+ignore_missing_imports = True
+
+[mypy-statsmodels.*]
+ignore_missing_imports = True
+
+[mypy-prophet.*]
+ignore_missing_imports = True
+
+[mypy-catboost.*]
+ignore_missing_imports = True
+
+[mypy-lightgbm.*]
+ignore_missing_imports = True
+
+[mypy-pmdarima.*]
+ignore_missing_imports = True
+
+[mypy-statsforecast.*]
+ignore_missing_imports = True
+
+[mypy-nfoursid.*]
+ignore_missing_imports = True
+
+[mypy-tbats.*]
+ignore_missing_imports = True
+
+[mypy-joblib.*]
+ignore_missing_imports = True
+
+[mypy-IPython.*]
+ignore_missing_imports = True
+
+[mypy-matplotlib.*]
+ignore_missing_imports = True
+
+[mypy-tqdm.*]
+ignore_missing_imports = True
+
+[mypy-darts.tests.*]
+ignore_errors = True
+
+[mypy-darts.datasets.*]
+ignore_errors = True
+
+[mypy-darts.metrics.*]
+ignore_errors = True
+
+[mypy-darts.models.*]
+ignore_errors = True
+
+[mypy-darts.utils.*]
+ignore_errors = True
+
+[mypy-darts.timeseries.*]
+ignore_errors = True
+
+[mypy-darts.dataprocessing.transformers.*]
+ignore_errors = True

requirements/dev.txt

@@ -5,3 +5,4 @@ pre-commit
 pytest-cov
 pyupgrade==2.31.0
 testfixtures
+mypy==0.971