dycw-utilities 0.109.16__tar.gz → 0.109.18__tar.gz

{dycw_utilities-0.109.16 → dycw_utilities-0.109.18}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dycw-utilities
-Version: 0.109.16
+Version: 0.109.18
 Author-email: Derek Wan <d.wan@icloud.com>
 License-File: LICENSE
 Requires-Python: >=3.12

{dycw_utilities-0.109.16 → dycw_utilities-0.109.18}/pyproject.toml

@@ -92,7 +92,7 @@ dependencies = [
 name = "dycw-utilities"
 readme = "README.md"
 requires-python = ">= 3.12"
-version = "0.109.16"
+version = "0.109.18"
 
 [project.optional-dependencies]
 test = [
@@ -335,7 +335,7 @@ zzz-test-zoneinfo = [
 # bump-my-version
 [tool.bumpversion]
 allow_dirty = true
-current_version = "0.109.16"
+current_version = "0.109.18"
 
 [[tool.bumpversion.files]]
 filename = "src/utilities/__init__.py"

{dycw_utilities-0.109.16 → dycw_utilities-0.109.18}/src/tests/test_hypothesis.py

@@ -29,7 +29,7 @@ from hypothesis.strategies import (
 from luigi import Task
 from numpy import inf, int64, isfinite, isinf, isnan, ravel, rint
 from pathvalidate import validate_filepath
-from pytest import raises
+from pytest import mark, raises
 from sqlalchemy import Column, Integer, MetaData, Table, insert, select
 from sqlalchemy.ext.asyncio import AsyncEngine
 
@@ -1033,6 +1033,7 @@ class TestTempPaths:
         assert len(set(path.iterdir())) == 0
 
     @given(path=temp_paths(), contents=sets(text_ascii(min_size=1), max_size=10))
+    @mark.flaky
     def test_writing_files(self, *, path: Path, contents: AbstractSet[str]) -> None:
         assert len(set(path.iterdir())) == 0
         as_set = set(maybe_yield_lower_case(contents))

{dycw_utilities-0.109.16 → dycw_utilities-0.109.18}/src/tests/test_numpy.py

@@ -19,6 +19,7 @@ from numpy import (
     ndarray,
     ones,
     pi,
+    where,
     zeros,
     zeros_like,
 )
@@ -27,7 +28,7 @@ from numpy.random import Generator
 from numpy.testing import assert_equal
 from pytest import mark, param, raises
 
-from utilities.hypothesis import float_arrays
+from utilities.hypothesis import float_arrays, pairs
 from utilities.numpy import (
     DEFAULT_RNG,
     AsIntError,
@@ -36,11 +37,16 @@ from utilities.numpy import (
     NDArrayF,
     NDArrayI,
     ShiftError,
+    SigmoidError,
+    _BoxCarLocationsError,
+    _BoxCarLowerBoundSlopeError,
+    _BoxCarUpperBoundSlopeError,
+    adjust_frequencies,
     array_indexer,
     as_int,
+    boxcar,
     discretize,
     fillna,
-    filter_frequencies,
     flatn0,
     get_frequency_spectrum,
     has_dtype,
@@ -94,12 +100,39 @@ from utilities.numpy import (
     minimum,
     shift,
     shift_bool,
+    sigmoid,
 )
 
 if TYPE_CHECKING:
     from collections.abc import Sequence
 
 
+class TestAdjustFrequencies:
+    def test_filter(self) -> None:
+        n = 1000
+        x = linspace(0, 2 * pi, n)
+        noise = DEFAULT_RNG.normal(scale=0.25, size=n)
+        y = x + noise
+        result = adjust_frequencies(y, filters=lambda f: np.abs(f) <= 0.02)
+        assert result.shape == (n,)
+        amplitudes = fft(result)
+        freqs = fftfreq(n)
+        assert np.allclose(amplitudes[np.abs(freqs) > 0.02], 0.0)
+
+    def test_weight(self) -> None:
+        n = 1000
+        x = linspace(0, 2 * pi, n)
+        noise = DEFAULT_RNG.normal(scale=0.25, size=n)
+        y = x + noise
+        result = adjust_frequencies(
+            y, weights=lambda f: where(np.abs(f) <= 0.02, 1.0, 0.0)
+        )
+        assert result.shape == (n,)
+        amplitudes = fft(result)
+        freqs = fftfreq(n)
+        assert np.allclose(amplitudes[np.abs(freqs) > 0.02], 0.0)
+
+
 class TestArrayIndexer:
     @mark.parametrize(
         ("i", "ndim", "expected"),
@@ -252,6 +285,56 @@ class TestDiscretize:
         assert_equal(result, expected)
 
 
+class TestBoxCar:
+    @given(
+        locs=pairs(floats(-10.0, 10.0), sorted=True),
+        slope_low=floats(0.1, 10.0),
+        slope_high=floats(0.1, 10.0),
+    )
+    def test_main(
+        self, *, locs: tuple[float, float], slope_low: float, slope_high: float
+    ) -> None:
+        loc_low, loc_high = locs
+        n = 1000
+        x = linspace(0, 2 * pi, n)
+        y = boxcar(
+            x,
+            loc_low=loc_low,
+            slope_low=slope_low,
+            loc_high=loc_high,
+            slope_high=slope_high,
+        )
+        assert y.shape == (n,)
+        assert is_between(y, 0.0, 1.0).all()
+
+    def test_error_locations(self) -> None:
+        n = 1000
+        x = linspace(0, 2 * pi, n)
+        with raises(
+            _BoxCarLocationsError,
+            match="Location parameters must be consistent; got 1.0 and -1.0",
+        ):
+            _ = boxcar(x, loc_low=1.0, loc_high=-1.0)
+
+    def test_error_lower_bound_slope(self) -> None:
+        n = 1000
+        x = linspace(0, 2 * pi, n)
+        with raises(
+            _BoxCarLowerBoundSlopeError,
+            match="Lower-bound slope parameter must be positive; got 0.0",
+        ):
+            _ = boxcar(x, slope_low=0.0)
+
+    def test_error_upper_bound_slope(self) -> None:
+        n = 1000
+        x = linspace(0, 2 * pi, n)
+        with raises(
+            _BoxCarUpperBoundSlopeError,
+            match="Upper-bound slope parameter must be positive; got 0.0",
+        ):
+            _ = boxcar(x, slope_high=0.0)
+
+
 class TestFillNa:
     @mark.parametrize(
         ("init", "value", "expected_v"),
@@ -274,19 +357,6 @@ class TestFillNa:
         assert_equal(result, expected)
 
 
-class TestFilterFrequencies:
-    def test_main(self) -> None:
-        n = 1000
-        x = linspace(0, 2 * pi, n)
-        noise = DEFAULT_RNG.normal(scale=0.25, size=n)
-        y = x + noise
-        result = filter_frequencies(y, lambda f: np.abs(f) <= 0.02)
-        assert result.shape == (n,)
-        fft_vals = fft(result)
-        freqs = fftfreq(n)
-        assert np.allclose(fft_vals[np.abs(freqs) > 0.02], 0.0)
-
-
 class TestFlatN0:
     @given(data=data(), n=integers(1, 10))
     def test_main(self, *, data: DataObject, n: int) -> None:
@@ -313,10 +383,10 @@ class TestGetFrequencySpectrum:
         x = linspace(0, 2 * pi, n)
         noise = DEFAULT_RNG.normal(scale=0.25, size=n)
         y = x + noise
-        y2 = filter_frequencies(y, lambda f: np.abs(f) <= 0.02)
+        y2 = adjust_frequencies(y, filters=lambda f: np.abs(f) <= 0.02)
         result = get_frequency_spectrum(y2)
         assert result.shape == (n, 2)
-        assert np.allclose(result[result[:, 0] > 0.02, 1], 0.0)
+        assert np.allclose(result[np.abs(result[:, 0]) > 0.02, 1], 0.0)
 
 
 class TestHasDtype:
@@ -1078,3 +1148,19 @@ class TestShiftBool:
             [fill_value if e is None else e for e in expected_v], dtype=bool
         )
         assert_equal(result, expected)
+
+
+class TestSigmoid:
+    @given(loc=floats(-10.0, 10.0), slope=floats(-10.0, -0.1) | floats(0.1, 10.0))
+    def test_main(self, *, loc: float, slope: float) -> None:
+        n = 1000
+        x = linspace(0, 2 * pi, n)
+        y = sigmoid(x, loc=loc, slope=slope)
+        assert y.shape == (n,)
+        assert is_between(y, 0.0, 1.0).all()
+
+    def test_error(self) -> None:
+        n = 1000
+        x = linspace(0, 2 * pi, n)
+        with raises(SigmoidError, match="Slope must be non-zero"):
+            _ = sigmoid(x, slope=0.0)

{dycw_utilities-0.109.16 → dycw_utilities-0.109.18}/src/tests/test_orjson.py

@@ -23,7 +23,7 @@ from hypothesis.strategies import (
 )
 from orjson import JSONDecodeError
 from polars import Object, String, UInt64
-from pytest import approx, raises
+from pytest import approx, mark, raises
 
 from tests.conftest import SKIPIF_CI_AND_WINDOWS
 from tests.test_operator import (
@@ -204,6 +204,7 @@ class TestGetLogRecords:
         min_log_file_line_num=integers() | none(),
         max_log_file_line_num=integers() | none(),
     )
+    @mark.flaky
     def test_filter(
         self,
         *,

{dycw_utilities-0.109.16 → dycw_utilities-0.109.18}/src/tests/test_polars.py

@@ -12,6 +12,7 @@ from typing import TYPE_CHECKING, Any, ClassVar, Literal, cast
 from uuid import UUID, uuid4
 
 import hypothesis.strategies
+import numpy as np
 import polars as pl
 from hypothesis import given
 from hypothesis.strategies import (
@@ -26,6 +27,7 @@ from hypothesis.strategies import (
     sampled_from,
     timezones,
 )
+from numpy import allclose, linspace, pi
 from polars import (
     Boolean,
     DataFrame,
@@ -60,6 +62,7 @@ from utilities.hypothesis import (
     zoned_datetimes,
 )
 from utilities.math import number_of_decimals
+from utilities.numpy import DEFAULT_RNG
 from utilities.pathlib import PWD
 from utilities.polars import (
     AppendDataClassError,
@@ -104,6 +107,7 @@ from utilities.polars import (
     _InsertBetweenMissingColumnsError,
     _InsertBetweenNonConsecutiveError,
     _yield_struct_series_element_remove_nulls,
+    adjust_frequencies,
     append_dataclass,
     are_frames_equal,
     ceil_datetime,
@@ -121,6 +125,7 @@ from utilities.polars import (
     finite_ewm_mean,
     floor_datetime,
     get_data_type_or_series_time_zone,
+    get_frequency_spectrum,
     get_series_number_of_decimals,
     insert_after,
     insert_before,
@@ -162,6 +167,16 @@ if TYPE_CHECKING:
     from utilities.types import MaybeType, StrMapping, WeekDay
 
 
+class TestAdjustFrequencies:
+    def test_main(self) -> None:
+        n = 1000
+        x = linspace(0, 2 * pi, n)
+        noise = DEFAULT_RNG.normal(scale=0.25, size=n)
+        y = Series(values=x + noise)
+        result = adjust_frequencies(y, filters=lambda f: np.abs(f) <= 0.02)
+        assert isinstance(result, Series)
+
+
 class TestAppendDataClass:
     @given(
         data=fixed_dictionaries({
@@ -1213,6 +1228,20 @@ class TestGetDataTypeOrSeriesTimeZone:
             _ = get_data_type_or_series_time_zone(Datetime)
 
 
+class TestGetFrequencySpectrum:
+    def test_main(self) -> None:
+        n = 1000
+        x = linspace(0, 2 * pi, n)
+        noise = DEFAULT_RNG.normal(scale=0.25, size=n)
+        y = Series(x + noise)
+        y2 = adjust_frequencies(y, filters=lambda f: np.abs(f) <= 0.02)
+        result = get_frequency_spectrum(y2)
+        check_polars_dataframe(
+            result, height=n, schema_list={"frequency": Float64, "amplitude": Float64}
+        )
+        assert allclose(result.filter(col("frequency").abs() > 0.02)["amplitude"], 0.0)
+
+
 class TestGetSeriesNumberOfDecimals:
     @given(data=data(), n=hypothesis.strategies.integers(1, 10), nullable=booleans())
     def test_main(self, *, data: DataObject, n: int, nullable: bool) -> None:

{dycw_utilities-0.109.16 → dycw_utilities-0.109.18}/src/utilities/__init__.py

@@ -1,3 +1,3 @@
 from __future__ import annotations
 
-__version__ = "0.109.16"
+__version__ = "0.109.18"

{dycw_utilities-0.109.16 → dycw_utilities-0.109.18}/src/utilities/numpy.py

@@ -14,12 +14,12 @@ from numpy import (
     digitize,
     dtype,
     errstate,
+    exp,
     flatnonzero,
-    float64,
     floating,
     full_like,
     inf,
-    int64,
+    integer,
     isclose,
     isfinite,
     isinf,
@@ -38,11 +38,13 @@ from numpy.linalg import det, eig
 from numpy.random import default_rng
 from numpy.typing import NDArray
 
-from utilities.iterables import is_iterable_not_str
+from utilities.iterables import always_iterable, is_iterable_not_str
 
 if TYPE_CHECKING:
     from collections.abc import Callable, Iterable
 
+    from utilities.types import MaybeIterable
+
 
 ##
 
@@ -88,8 +90,9 @@ timedelta64as = dtype("timedelta64[as]")
 
 NDArrayA = NDArray[Any]
 NDArrayB = NDArray[bool_]
-NDArrayF = NDArray[float64]
-NDArrayI = NDArray[int64]
+NDArrayC128 = NDArray[complex128]
+NDArrayF = NDArray[floating[Any]]
+NDArrayI = NDArray[integer[Any]]
 NDArrayO = NDArray[object_]
 
 
@@ -133,6 +136,65 @@ class AsIntError(Exception): ...
 ##
 
 
+def boxcar(
+    array: NDArrayF,
+    /,
+    *,
+    loc_low: float = -1.0,
+    slope_low: float = 1.0,
+    loc_high: float = 1.0,
+    slope_high: float = 1.0,
+    rtol: float | None = None,
+    atol: float | None = None,
+) -> NDArrayF:
+    """Construct a boxcar function."""
+    if not is_at_most(loc_low, loc_high, rtol=rtol, atol=atol):
+        raise _BoxCarLocationsError(low=loc_low, high=loc_high)
+    if not is_positive(slope_low, rtol=rtol, atol=atol):
+        raise _BoxCarLowerBoundSlopeError(slope=slope_low)
+    if not is_positive(slope_high, rtol=rtol, atol=atol):
+        raise _BoxCarUpperBoundSlopeError(slope=slope_high)
+    return (
+        sigmoid(array, loc=loc_low, slope=slope_low)
+        + sigmoid(array, loc=loc_high, slope=-slope_high)
+    ) / 2
+
+
+@dataclass(kw_only=True, slots=True)
+class BoxCarError(Exception): ...
+
+
+@dataclass(kw_only=True, slots=True)
+class _BoxCarLocationsError(BoxCarError):
+    low: float
+    high: float
+
+    @override
+    def __str__(self) -> str:
+        return f"Location parameters must be consistent; got {self.low} and {self.high}"
+
+
+@dataclass(kw_only=True, slots=True)
+class _BoxCarLowerBoundSlopeError(BoxCarError):
+    slope: float
+
+    @override
+    def __str__(self) -> str:
+        return f"Lower-bound slope parameter must be positive; got {self.slope}"
+
+
+@dataclass(kw_only=True, slots=True)
+class _BoxCarUpperBoundSlopeError(BoxCarError):
+    slope: float
+
+    @override
+    def __str__(self) -> str:
+        return f"Upper-bound slope parameter must be positive; got {self.slope}"
+
+
+##
+
+
 def discretize(x: NDArrayF, bins: int | Iterable[float], /) -> NDArrayF:
     """Discretize an array of floats.
 
@@ -164,28 +226,43 @@ def fillna(array: NDArrayF, /, *, value: float = 0.0) -> NDArrayF:
 ##
 
 
-def filter_frequencies(
+def adjust_frequencies(
     array: NDArrayF,
     /,
-    *filters: Callable[[NDArray[floating[Any]]], NDArrayB],
+    *,
+    filters: MaybeIterable[Callable[[NDArrayF], NDArrayB]] | None = None,
+    weights: MaybeIterable[Callable[[NDArrayF], NDArrayF]] | None = None,
     d: int = 1,
 ) -> NDArrayF:
-    """Filter an array by the frequencies of its FFT."""
+    """Adjust an array via its FFT frequencies."""
     (n,) = array.shape
-    fft_vals = fft(array)
+    amplitudes = fft(array)
     freqs = fftfreq(n, d=d)
-    reduced = reduce(partial(_filter_frequencies_one, freqs=freqs), filters, fft_vals)
-    return ifft(reduced).real
+    if filters is not None:
+        amplitudes = reduce(
+            partial(_adjust_frequencies_filter_one, freqs=freqs),
+            always_iterable(filters),
+            amplitudes,
+        )
+    if weights is not None:
+        amplitudes = reduce(
+            partial(_adjust_frequencies_weight_one, freqs=freqs),
+            always_iterable(weights),
+            amplitudes,
+        )
+    return ifft(amplitudes).real
+
+
+def _adjust_frequencies_filter_one(
+    acc: NDArrayC128, el: Callable[[NDArrayF], NDArrayB], /, *, freqs: NDArrayF
+) -> NDArrayC128:
+    return where(el(freqs), acc, 0.0)
 
 
-def _filter_frequencies_one(
-    acc: NDArray[complex128],
-    el: Callable[[NDArray[floating[Any]]], NDArrayB],
-    /,
-    *,
-    freqs: NDArray[floating[Any]],
-) -> NDArray[complex128]:
-    return where(el(freqs), acc, 0.0)
+def _adjust_frequencies_weight_one(
+    acc: NDArrayC128, el: Callable[[NDArrayF], NDArrayF], /, *, freqs: NDArrayF
+) -> NDArrayC128:
+    return acc * el(freqs)
 
 
 ##
@@ -224,12 +301,12 @@ class FlatN0MultipleError(FlatN0Error):
 ##
 
 
-def get_frequency_spectrum(array: NDArrayF, /, *, d: int = 1) -> NDArray[floating[Any]]:
+def get_frequency_spectrum(array: NDArrayF, /, *, d: int = 1) -> NDArrayF:
     """Get the frequency spectrum."""
     (n,) = array.shape
-    fft_vals = fft(array)
+    amplitudes = fft(array)
     freqs = fftfreq(n, d=d)
-    amplitudes = np.abs(fft_vals)
+    amplitudes = np.abs(amplitudes)
     data = np.hstack([freqs.reshape(-1, 1), amplitudes.reshape(-1, 1)])
     return data[argsort(data[:, 0])]
 
@@ -847,6 +924,31 @@ def shift_bool(
 ##
 
 
+def sigmoid(
+    array: NDArrayF,
+    /,
+    *,
+    loc: float = 0.0,
+    slope: float = 1.0,
+    rtol: float | None = None,
+    atol: float | None = None,
+) -> NDArrayF:
+    """Construct a sigmoid function."""
+    if is_zero(slope, rtol=rtol, atol=atol):
+        raise SigmoidError
+    return 1 / (1 + exp(-slope * (array - loc)))
+
+
+@dataclass(kw_only=True, slots=True)
+class SigmoidError(Exception):
+    @override
+    def __str__(self) -> str:
+        return "Slope must be non-zero"
+
+
+##
+
+
 def _is_close(
     x: Any,
     y: Any,
@@ -869,6 +971,7 @@ def _is_close(
 __all__ = [
     "DEFAULT_RNG",
     "AsIntError",
+    "BoxCarError",
     "FlatN0EmptyError",
     "FlatN0Error",
     "FlatN0MultipleError",
@@ -878,8 +981,11 @@ __all__ = [
     "NDArrayI",
     "NDArrayO",
     "ShiftError",
+    "SigmoidError",
+    "adjust_frequencies",
     "array_indexer",
     "as_int",
+    "boxcar",
     "datetime64D",
     "datetime64M",
     "datetime64W",
@@ -895,7 +1001,6 @@ __all__ = [
     "datetime64us",
     "discretize",
     "fillna",
-    "filter_frequencies",
     "flatn0",
     "get_frequency_spectrum",
     "has_dtype",
@@ -948,4 +1053,5 @@ __all__ = [
     "maximum",
     "minimum",
     "shift_bool",
+    "sigmoid",
 ]

{dycw_utilities-0.109.16 → dycw_utilities-0.109.18}/src/utilities/polars.py

@@ -119,6 +119,7 @@ if TYPE_CHECKING:
         TimeUnit,  # pyright: ignore[reportPrivateImportUsage]
     )
 
+    from utilities.numpy import NDArrayB, NDArrayF
     from utilities.types import (
         Dataclass,
         MaybeIterable,
@@ -141,6 +142,27 @@ _FINITE_EWM_MIN_WEIGHT = 0.9999
 ##
 
 
+def adjust_frequencies(
+    series: Series,
+    /,
+    *,
+    filters: MaybeIterable[Callable[[NDArrayF], NDArrayB]] | None = None,
+    weights: MaybeIterable[Callable[[NDArrayF], NDArrayF]] | None = None,
+    d: int = 1,
+) -> Series:
+    """Adjust a Series via its FFT frequencies."""
+    import utilities.numpy
+
+    array = series.to_numpy()
+    adjusted = utilities.numpy.adjust_frequencies(
+        array, filters=filters, weights=weights, d=d
+    )
+    return Series(name=series.name, values=adjusted, dtype=Float64)
+
+
+##
+
+
 def append_dataclass(df: DataFrame, obj: Dataclass, /) -> DataFrame:
     """Append a dataclass object to a DataFrame."""
     non_null_fields = {k: v for k, v in asdict(obj).items() if v is not None}
@@ -1027,6 +1049,20 @@ class _GetDataTypeOrSeriesTimeZoneNotZonedError(GetDataTypeOrSeriesTimeZoneError
 ##
 
 
+def get_frequency_spectrum(series: Series, /, *, d: int = 1) -> DataFrame:
+    """Get the frequency spectrum."""
+    import utilities.numpy
+
+    array = series.to_numpy()
+    spectrum = utilities.numpy.get_frequency_spectrum(array, d=d)
+    return DataFrame(
+        data=spectrum, schema={"frequency": Float64, "amplitude": Float64}, orient="row"
+    )
+
+
+##
+
+
 @overload
 def get_series_number_of_decimals(
     series: Series, /, *, nullable: Literal[True]
@@ -1736,6 +1772,7 @@ __all__ = [
     "SetFirstRowAsColumnsError",
     "StructFromDataClassError",
     "YieldStructSeriesElementsError",
+    "adjust_frequencies",
     "append_dataclass",
     "are_frames_equal",
     "ceil_datetime",
@@ -1753,6 +1790,7 @@ __all__ = [
     "finite_ewm_mean",
     "floor_datetime",
     "get_data_type_or_series_time_zone",
+    "get_frequency_spectrum",
     "get_series_number_of_decimals",
     "insert_after",
     "insert_before",