fix: resample method for TimeSeriesBoolean (#349)

If a period between two timesteps in the resampled boolean variable corresponds to
several periods in the original boolean variable, the resampling method should
return False if any of the respecitve periods in the original boolean variable are False.

src/libecalc/common/time_utils.py

@@ -2,7 +2,7 @@
 
 import enum
 from dataclasses import dataclass
-from datetime import date, datetime
+from datetime import date, datetime, timedelta
 from typing import Any, Dict, List, Optional, Tuple, Union
 
 import numpy as np
@@ -211,7 +211,10 @@ def create_time_steps(
         the requested frequency
 
     """
-    date_range = pd.date_range(start=start, end=end, freq=frequency.value)
+    # If the start date or end date is part of the date_range made by the frequency, the returned date range will
+    # always include the start and end date (no matter what the include_start_date and include_end_date booleans are).
+    # To avoid this add one day to start and subtract one day from end.
+    date_range = pd.date_range(start=start + timedelta(days=1), end=end - timedelta(days=1), freq=frequency.value)
 
     time_steps = [clear_time(time_step) for time_step in date_range]
     if include_start_date:

src/libecalc/common/utils/rates.py

@@ -429,8 +429,9 @@ def resample(self, freq: Frequency, include_start_date: bool = True, include_end
 class TimeSeriesBoolean(TimeSeries[bool]):
     def resample(self, freq: Frequency, include_start_date: bool = True, include_end_date: bool = True) -> Self:
         """
-        Resample using forward-fill This means that a value is assumed to be the same until the next observation,
-        e.g. covering the whole period interval.
+        If a period between two time steps in the return time vector contains more than one time step in the
+        original vector, check if any of the relevant values in the time original time vector is False. Then the
+        resampled value for that time step will be False.
 
         Args:
             freq: The frequency the time series should be resampled to
@@ -441,16 +442,26 @@ def resample(self, freq: Frequency, include_start_date: bool = True, include_end
         if freq is Frequency.NONE:
             return self.copy()
 
-        ds = pd.Series(index=self.timesteps, data=self.values)
-
+        # Always make new time series WITH end date, but remove it later is not needed
         new_timeseries = resample_time_steps(
-            self.timesteps, frequency=freq, include_start_date=include_start_date, include_end_date=include_end_date
+            self.timesteps, frequency=freq, include_start_date=include_start_date, include_end_date=True
         )
-        ds_resampled = ds.reindex(new_timeseries).ffill()
+        resampled = []
+
+        # Iterate over all pairs of subsequent dates in the new time vector
+        for start_period, end_period in zip(new_timeseries[:-1], new_timeseries[1:]):
+            start_index = self.timesteps.index(max([date for date in self.timesteps if date <= start_period]))
+            end_index = self.timesteps.index(max([date for date in self.timesteps if date < end_period]))
+            resampled.append(all(self.values[start_index : end_index + 1]))
+
+        if include_end_date:
+            resampled.append(self.values[-1])
+        else:
+            new_timeseries.pop()
 
         return TimeSeriesBoolean(
             timesteps=new_timeseries,
-            values=[bool(x) for x in ds_resampled.values.tolist()],
+            values=resampled,
             unit=self.unit,
         )
 

src/tests/libecalc/common/utils/test_rates.py

@@ -72,13 +72,16 @@ class TestBooleanTimeSeries:
     @pytest.fixture
     def boolean_series(self):
         return TimeSeriesBoolean(
-            values=[False, True, True, False, True],
+            values=[False, True, True, False, True, True, False, True],
             timesteps=[
                 datetime(2019, 7, 1),
                 datetime(2020, 1, 1),
                 datetime(2020, 7, 1),
                 datetime(2021, 1, 1),
                 datetime(2021, 7, 1),
+                datetime(2022, 1, 1),
+                datetime(2022, 7, 1),
+                datetime(2023, 1, 1),
             ],
             unit=Unit.NONE,
         )
@@ -92,36 +95,43 @@ def two_first_timesteps(self, boolean_series):
     def test_resample_boolean(self, boolean_series):
         # resample including start and end date
         yearly_values = boolean_series.resample(freq=Frequency.YEAR)
-        assert yearly_values.values == [False, True, False, True]
+        assert yearly_values.values == [False, True, False, False, True]
         assert yearly_values.timesteps == [
             datetime(2019, 7, 1),
             datetime(2020, 1, 1),
             datetime(2021, 1, 1),
-            datetime(2021, 7, 1),
+            datetime(2022, 1, 1),
+            datetime(2023, 1, 1),
         ]
 
         # resample including start and without end date
         yearly_values = boolean_series.resample(freq=Frequency.YEAR, include_end_date=False)
-        assert yearly_values.values == [False, True, False]
+        assert yearly_values.values == [False, True, False, False]
         assert yearly_values.timesteps == [
             datetime(2019, 7, 1),
             datetime(2020, 1, 1),
             datetime(2021, 1, 1),
+            datetime(2022, 1, 1),
         ]
 
         # resample without start and including end date
         yearly_values = boolean_series.resample(freq=Frequency.YEAR, include_start_date=False)
-        assert yearly_values.values == [True, False, True]
+        assert yearly_values.values == [True, False, False, True]
         assert yearly_values.timesteps == [
             datetime(2020, 1, 1),
             datetime(2021, 1, 1),
-            datetime(2021, 7, 1),
+            datetime(2022, 1, 1),
+            datetime(2023, 1, 1),
         ]
 
         # resample without start and end date
         yearly_values = boolean_series.resample(freq=Frequency.YEAR, include_start_date=False, include_end_date=False)
-        assert yearly_values.values == [True, False]
-        assert yearly_values.timesteps == [datetime(2020, 1, 1), datetime(2021, 1, 1)]
+        assert yearly_values.values == [True, False, False]
+        assert yearly_values.timesteps == [
+            datetime(2020, 1, 1),
+            datetime(2021, 1, 1),
+            datetime(2022, 1, 1),
+        ]
 
     def test_indexing(self, boolean_series):
         first_timestep = TimeSeriesBoolean(