BUG: fix pandas-dev#30353 invalid end

Author: hasB4K

pandas/core/arrays/timedeltas.py

@@ -38,6 +38,7 @@
 from pandas.core import nanops
 from pandas.core.algorithms import checked_add_with_arr
 from pandas.core.arrays import datetimelike as dtl
+from pandas.core.arrays._ranges import _generate_range_overflow_safe
 import pandas.core.common as com
 from pandas.core.construction import extract_array
 
@@ -1061,14 +1062,15 @@ def _generate_regular_range(start, end, periods, offset):
     stride = offset.nanos
     if periods is None:
         b = Timedelta(start).value
-        e = Timedelta(end).value
-        e += stride - e % stride
+        # cannot just use e = Timestamp(end) + 1 because arange breaks when
+        # stride is too large, see GH 10887 & GH 30353
+        e = b + (Timedelta(end).value - b) // stride * stride + stride // 2 + 1
     elif start is not None:
         b = Timedelta(start).value
-        e = b + periods * stride
+        e = _generate_range_overflow_safe(b, periods, stride, side="start")
     elif end is not None:
         e = Timedelta(end).value + stride
-        b = e - periods * stride
+        b = _generate_range_overflow_safe(e, periods, stride, side="end")
     else:
         raise ValueError(
             "at least 'start' or 'end' should be specified if a 'period' is given."

pandas/tests/resample/test_base.py

@@ -10,7 +10,7 @@
 from pandas.core.groupby.grouper import Grouper
 from pandas.core.indexes.datetimes import date_range
 from pandas.core.indexes.period import PeriodIndex, period_range
-from pandas.core.indexes.timedeltas import TimedeltaIndex, timedelta_range
+from pandas.core.indexes.timedeltas import timedelta_range
 from pandas.core.resample import _asfreq_compat
 
 # a fixture value can be overridden by the test parameter value. Note that the
@@ -182,7 +182,6 @@ def test_resample_size_empty_dataframe(freq, empty_frame_dti):
 @pytest.mark.parametrize("index", tm.all_timeseries_index_generator(0))
 @pytest.mark.parametrize("dtype", [np.float, np.int, np.object, "datetime64[ns]"])
 def test_resample_empty_dtypes(index, dtype, resample_method):
-
     # Empty series were sometimes causing a segfault (for the functions
     # with Cython bounds-checking disabled) or an IndexError.  We just run
     # them to ensure they no longer do.  (GH #10228)
@@ -215,13 +214,7 @@ def test_resample_loffset_arg_type(frame, create_index, arg):
     if isinstance(arg, list):
         expected.columns = pd.MultiIndex.from_tuples([("value", "mean")])
 
-    # GH 13022, 7687 - TODO: fix resample w/ TimedeltaIndex
-    if isinstance(expected.index, TimedeltaIndex):
-        msg = "DataFrame are different"
-        with pytest.raises(AssertionError, match=msg):
-            tm.assert_frame_equal(result_agg, expected)
-    else:
-        tm.assert_frame_equal(result_agg, expected)
+    tm.assert_frame_equal(result_agg, expected)
 
 
 @all_ts

pandas/tests/resample/test_timedelta.py

@@ -1,6 +1,7 @@
 from datetime import timedelta
 
 import numpy as np
+import pytest
 
 import pandas as pd
 from pandas import DataFrame, Series
@@ -114,14 +115,33 @@ def test_resample_timedelta_values():
     # check that timedelta dtype is preserved when NaT values are
     # introduced by the resampling
 
-    times = timedelta_range("1 day", "4 day", freq="4D")
+    times = timedelta_range("1 day", "6 day", freq="4D")
     df = DataFrame({"time": times}, index=times)
 
-    times2 = timedelta_range("1 day", "4 day", freq="2D")
+    times2 = timedelta_range("1 day", "6 day", freq="2D")
     exp = Series(times2, index=times2, name="time")
     exp.iloc[1] = pd.NaT
 
     res = df.resample("2D").first()["time"]
     tm.assert_series_equal(res, exp)
     res = df["time"].resample("2D").first()
     tm.assert_series_equal(res, exp)
+
+
+@pytest.mark.parametrize(
+    "freq, resample_freq, start, periods, expected_resample_end",
+    [("10S", "3H", "8H", 5040, "20H")],
+)
+def test_resample_timedelta_end_already_included_in_bins(
+    freq, resample_freq, start, periods, expected_resample_end,
+):
+    # GH 30353
+    # check that the timedelta bins does not contains an extra bin
+    idx = pd.timedelta_range(start=start, freq=freq, periods=periods)
+    s = pd.Series(np.arange(periods), index=idx)
+    result = s.resample(resample_freq).min()
+    expected_index = pd.timedelta_range(
+        freq=resample_freq, start=start, end=expected_resample_end
+    )
+    tm.assert_index_equal(result.index, expected_index)
+    assert not np.isnan(result[-1])