Closes #3: add tests for PeriodAxis

Author: xiaotongchang-creator

src/backtest_lib/market/__init__.py

@@ -4,7 +4,7 @@
 from abc import ABC, abstractmethod
 from collections.abc import Iterable, Iterator, Mapping, Sequence
 from dataclasses import replace
-from enum import Enum, auto
+from enum import Enum, StrEnum, auto
 from typing import (
     TYPE_CHECKING,
     Any,
@@ -32,6 +32,12 @@
     from backtest_lib.universe.vector_mapping import VectorMapping
 
 
+class Closed(StrEnum):
+    LEFT = "left"
+    RIGHT = "right"
+    BOTH = "both"
+
+
 def get_pastview_from_mapping(backend: str) -> type[PastView]:
     if backend == "polars":
         from backtest_lib.market.polars_impl import PolarsPastView
@@ -121,6 +127,8 @@ def between(
         self,
         start: Index | str,
         end: Index | str,
+        *,
+        closed: Closed | str = Closed.LEFT,
     ) -> Self:
         """PLACEHOLDER"""
         ...  # will not clone data, must be contiguous, performs a binary search

src/backtest_lib/market/polars_impl/_axis.py

@@ -7,6 +7,8 @@
 import polars as pl
 from numpy.typing import NDArray
 
+from backtest_lib.market import Closed
+
 
 @dataclass(frozen=True)
 class SecurityAxis:
@@ -48,11 +50,11 @@ def take(self, idxs: Sequence[int] | NDArray[np.int64]) -> PeriodAxis:
         new_labels = tuple(self.labels[i] for i in idxs)
         return PeriodAxis(
             dt64=self.dt64[idxs],
-            labels=tuple(new_labels),
+            labels=new_labels,
             pos={lbl: i for i, lbl in enumerate(new_labels)},
         )
 
-    def _slice(self, key: slice) -> PeriodAxis:
+    def slice(self, key: slice) -> PeriodAxis:
         """
         Creates a new PeriodAxis from a slice
         of the current PeriodAxis.
@@ -94,32 +96,10 @@ def bounds_between(
         start: np.datetime64,
         end: np.datetime64,
         *,
-        closed: str = "both",
+        closed: str | Closed = Closed.BOTH,
     ) -> tuple[int, int]:
-        inc_start = closed in ("both", "left")
-        inc_end = closed in ("both", "right")
+        inc_start = closed in (Closed.BOTH, Closed.LEFT)
+        inc_end = closed in (Closed.BOTH, Closed.RIGHT)
         left, _ = self.bounds_after(start, inclusive=inc_start)
         _, right = self.bounds_before(end, inclusive=inc_end)
         return left, right
-
-    def after(
-        self, start: np.datetime64, *, inclusive: bool = True
-    ) -> NDArray[np.int64]:
-        left, right = self.bounds_after(start, inclusive=inclusive)
-        return np.arange(left, right, dtype=np.int64)
-
-    def before(
-        self, end: np.datetime64, *, inclusive: bool = False
-    ) -> NDArray[np.int64]:
-        left, right = self.bounds_before(end, inclusive=inclusive)
-        return np.arange(left, right, dtype=np.int64)
-
-    def between(
-        self,
-        start: np.datetime64,
-        end: np.datetime64,
-        *,
-        closed: str = "left",
-    ) -> NDArray[np.int64]:
-        left, right = self.bounds_between(start, end, closed=closed)
-        return np.arange(left, right, dtype=np.int64)

src/backtest_lib/market/polars_impl/_past_view.py

@@ -16,7 +16,7 @@
 import numpy as np
 import polars as pl
 
-from backtest_lib.market import ByPeriod, BySecurity, PastView
+from backtest_lib.market import ByPeriod, BySecurity, Closed, PastView
 from backtest_lib.market.plotting import (
     ByPeriodPlotAccessor,
     BySecurityPlotAccessor,
@@ -589,7 +589,7 @@ def between(
         start: np.datetime64 | str,
         end: np.datetime64 | str,
         *,
-        closed: str = "left",
+        closed: Closed | str = Closed.LEFT,
     ) -> PolarsPastView:
         left, right = self._period_axis.bounds_between(
             to_npdt64(start), to_npdt64(end), closed=closed

src/backtest_lib/market/polars_impl/_timeseries.py

@@ -89,7 +89,7 @@ def __getitem__(self, key: int | slice) -> T | Self:
             # return self._scalar_type(self._vec[key])
         else:
             return PolarsTimeseries[T](
-                self._vec[key], self._axis._slice(key), self._name, self._scalar_type
+                self._vec[key], self._axis.slice(key), self._name, self._scalar_type
             )
 
     def before(self, end: np.datetime64 | str, *, inclusive=False) -> Self:

tests/market/polars_impl/test_axis.py

@@ -1,4 +1,11 @@
-from backtest_lib.market.polars_impl._axis import SecurityAxis
+import datetime
+
+import numpy as np
+import polars as pl
+import pytest
+from polars.exceptions import InvalidOperationError
+
+from backtest_lib.market.polars_impl._axis import PeriodAxis, SecurityAxis
 
 
 def test_static_constructor_security_axis():
@@ -17,3 +24,77 @@ def test_returning_length_of_names_security_axis():
 def test_ability_to_handle_empty_names_security_axis():
     security_axis = SecurityAxis.from_names([])
     assert len(security_axis.names) == 0
+
+
+def test_casting_incompatible_data_to_date_should_throw_error_period_axis():
+    s_string = pl.Series("string", ["a", "b", "c"])
+    with pytest.raises(InvalidOperationError) as e_info:
+        PeriodAxis.from_series(s_string)
+    assert "conversion from `str` to `datetime[μs]` failed in column 'string'" in str(
+        e_info.value
+    )
+
+
+def test_constructor_casting_required_period_axis():
+    series_date = pl.Series(
+        "dates",
+        [
+            datetime.date(2023, 1, 1),
+            datetime.date(2023, 1, 2),
+            datetime.date(2023, 1, 3),
+        ],
+    )
+    period_axis = PeriodAxis.from_series(series_date)
+    assert period_axis.labels == ("2023-01-01", "2023-01-02", "2023-01-03")
+    assert period_axis.pos == {"2023-01-01": 0, "2023-01-02": 1, "2023-01-03": 2}
+    expected_dt64_array = np.array(
+        ["2023-01-01", "2023-01-02", "2023-01-03"], dtype="datetime64[us]"
+    )
+    np.testing.assert_array_equal(period_axis.dt64, expected_dt64_array)
+
+
+def test_len_period_axis():
+    series_date = pl.Series(
+        "dates",
+        [
+            datetime.date(2023, 1, 1),
+            datetime.date(2023, 1, 2),
+            datetime.date(2023, 1, 3),
+        ],
+    )
+    period_axis = PeriodAxis.from_series(series_date)
+    assert len(period_axis) == 3
+
+
+def test_slicing_incontiguous_sequence():
+    series_date = pl.Series(
+        "dates",
+        [
+            datetime.date(2023, 1, 1),
+            datetime.date(2023, 1, 2),
+            datetime.date(2023, 1, 3),
+        ],
+    )
+    period_axis = PeriodAxis.from_series(series_date)
+    sliced_period_axis = period_axis.slice(slice(None, None, 2))
+    assert sliced_period_axis.labels == ("2023-01-01", "2023-01-03")
+    assert sliced_period_axis.pos == {"2023-01-01": 0, "2023-01-03": 1}
+    expected_dt64_array = np.array(["2023-01-01", "2023-01-03"], dtype="datetime64[us]")
+    np.testing.assert_array_equal(sliced_period_axis.dt64, expected_dt64_array)
+
+
+def test_slicing_contiguous_sequence():
+    series_date = pl.Series(
+        "dates",
+        [
+            datetime.date(2023, 1, 1),
+            datetime.date(2023, 1, 2),
+            datetime.date(2023, 1, 3),
+        ],
+    )
+    period_axis = PeriodAxis.from_series(series_date)
+    sliced_period_axis = period_axis.slice(slice(1, 3))
+    assert sliced_period_axis.labels == ("2023-01-02", "2023-01-03")
+    assert sliced_period_axis.pos == {"2023-01-02": 0, "2023-01-03": 1}
+    expected_dt64_array = np.array(["2023-01-02", "2023-01-03"], dtype="datetime64[us]")
+    np.testing.assert_array_equal(sliced_period_axis.dt64, expected_dt64_array)