BUG: fix a timezone bug between origin and index on df.resample

Author: hasB4K

pandas/core/generic.py

@@ -7821,8 +7821,8 @@ def resample(
             For a MultiIndex, level (name or number) to use for
             resampling. `level` must be datetime-like.
         origin : {'epoch', 'start', 'start_day'}, Timestamp or str, default 'start_day'
-            The timestamp on which to adjust the grouping. It must be timezone aware if
-            the index of the resampled data is.
+            The timestamp on which to adjust the grouping. The timezone of origin
+            must match the timezone of the index.
             If a timestamp is not used, these values are also supported:
 
             - 'epoch': `origin` is 1970-01-01

pandas/core/groupby/grouper.py

@@ -84,8 +84,8 @@ class Grouper:
             See: :class:`DataFrame.resample`
 
     origin : {'epoch', 'start', 'start_day'}, Timestamp or str, default 'start_day'
-        The timestamp on which to adjust the grouping. It must be timezone aware if
-        the index of the resampled data is.
+        The timestamp on which to adjust the grouping. The timezone of origin must
+        match the timezone of the index.
         If a timestamp is not used, these values are also supported:
 
         - 'epoch': `origin` is 1970-01-01

pandas/core/resample.py

@@ -1666,10 +1666,11 @@ def _get_timestamp_range_edges(
         The dateoffset to which the Timestamps will be adjusted.
     closed : {'right', 'left'}, default None
         Which side of bin interval is closed.
-    origin : {'epoch', 'start', 'start_day'}, Timestamp or str, default 'start_day'
-        The timestamp on which to adjust the grouping. It must be timezone aware if
-        the index of the resampled data is.
+    origin : {'epoch', 'start', 'start_day'} or Timestamp, default 'start_day'
+        The timestamp on which to adjust the grouping. The timezone of origin must
+        match the timezone of the index.
         If a timestamp is not used, these values are also supported:
+
         - 'epoch': `origin` is 1970-01-01
         - 'start': `origin` is the first value of the timeseries
         - 'start_day': `origin` is the first day at midnight of the timeseries
@@ -1680,38 +1681,34 @@ def _get_timestamp_range_edges(
     -------
     A tuple of length 2, containing the adjusted pd.Timestamp objects.
     """
-    if isinstance(freq, Tick):
-        if origin not in {"epoch", "start", "start_day"}:
-            is_idx_tz_aware = first.tz is not None or last.tz is not None
-            if origin.tz is None and is_idx_tz_aware:
-                raise ValueError("The origin must have the same timezone as the index.")
+    index_tz = first.tz
+    if isinstance(origin, Timestamp) and (origin.tz is None) != (index_tz is None):
+        raise ValueError("The origin must have the same timezone as the index.")
 
+    if isinstance(freq, Tick):
         if isinstance(freq, Day):
             # _adjust_dates_anchored assumes 'D' means 24H, but first/last
             # might contain a DST transition (23H, 24H, or 25H).
             # So "pretend" the dates are naive when adjusting the endpoints
-            tz = first.tz
             first = first.tz_localize(None)
             last = last.tz_localize(None)
 
         first, last = _adjust_dates_anchored(
             first, last, freq, closed=closed, origin=origin, offset=offset,
         )
         if isinstance(freq, Day):
-            first = first.tz_localize(tz)
-            last = last.tz_localize(tz)
-        return first, last
-
+            first = first.tz_localize(index_tz)
+            last = last.tz_localize(index_tz)
     else:
         first = first.normalize()
         last = last.normalize()
 
-    if closed == "left":
-        first = Timestamp(freq.rollback(first))
-    else:
-        first = Timestamp(first - freq)
+        if closed == "left":
+            first = Timestamp(freq.rollback(first))
+        else:
+            first = Timestamp(first - freq)
 
-    last = Timestamp(last + freq)
+        last = Timestamp(last + freq)
 
     return first, last
 
@@ -1733,14 +1730,15 @@ def _get_period_range_edges(
         The freq to which the Periods will be adjusted.
     closed : {'right', 'left'}, default None
         Which side of bin interval is closed.
-    origin : {'epoch', 'start', 'start_day'}, Timestamp or str, default 'start_day'
-        The timestamp on which to adjust the grouping. It must be timezone aware if
-        the index of the resampled data is.
+    origin : {'epoch', 'start', 'start_day'}, Timestamp, default 'start_day'
+        The timestamp on which to adjust the grouping. The timezone of origin must
+        match the timezone of the index.
 
         If a timestamp is not used, these values are also supported:
-        - If 'epoch': `origin` is 1970-01-01
-        - If 'start': then `origin` is the first value of the timeseries
-        - If 'start_day', then `origin` is the first day at midnight of the timeseries
+
+        - 'epoch': `origin` is 1970-01-01
+        - 'start': `origin` is the first value of the timeseries
+        - 'start_day': `origin` is the first day at midnight of the timeseries
     offset : pd.Timedelta, default is None
         An offset timedelta added to the origin.
 

pandas/tests/resample/test_datetime_index.py

@@ -814,6 +814,12 @@ def test_resample_origin_with_tz():
     with pytest.raises(ValueError, match=msg):
         ts.resample("5min", origin="12/31/1999 23:57:00").mean()
 
+    # if the series is not tz aware, origin should not be tz aware
+    rng = date_range("2000-01-01 00:00:00", "2000-01-01 02:00", freq="s")
+    ts = Series(np.random.randn(len(rng)), index=rng)
+    with pytest.raises(ValueError, match=msg):
+        ts.resample("5min", origin="12/31/1999 23:57:00+03:00").mean()
+
 
 def test_resample_daily_anchored():
     rng = date_range("1/1/2000 0:00:00", periods=10000, freq="T")