Skip to content

BUG: Series[dt64].__setitem__ with all-false mask incorrectly upcasting #45967

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 6 commits into from
Feb 17, 2022
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions doc/source/whatsnew/v1.5.0.rst
Original file line number Diff line number Diff line change
Expand Up @@ -326,6 +326,7 @@ Indexing
- Bug in :meth:`loc.__setitem__` treating ``range`` keys as positional instead of label-based (:issue:`45479`)
- Bug in :meth:`Series.__setitem__` when setting ``boolean`` dtype values containing ``NA`` incorrectly raising instead of casting to ``boolean`` dtype (:issue:`45462`)
- Bug in :meth:`Series.__setitem__` where setting :attr:`NA` into a numeric-dtpye :class:`Series` would incorrectly upcast to object-dtype rather than treating the value as ``np.nan`` (:issue:`44199`)
- Bug in :meth:`Series.__setitem__` with ``datetime64[ns]`` dtype, an all-``False`` boolean mask, and an incompatible value incorrectly casting to ``object`` instead of retaining ``datetime64[ns]`` dtype (:issue:`45967`)
- Bug in :meth:`Index.__getitem__` raising ``ValueError`` when indexer is from boolean dtype with ``NA`` (:issue:`45806`)
- Bug in :meth:`Series.mask` with ``inplace=True`` or setting values with a boolean mask with small integer dtypes incorrectly raising (:issue:`45750`)
- Bug in :meth:`DataFrame.mask` with ``inplace=True`` and ``ExtensionDtype`` columns incorrectly raising (:issue:`45577`)
Expand Down
8 changes: 6 additions & 2 deletions pandas/core/array_algos/putmask.py
Original file line number Diff line number Diff line change
Expand Up @@ -81,7 +81,11 @@ def putmask_without_repeat(
# TODO: this prob needs some better checking for 2D cases
nlocs = mask.sum()
if nlocs > 0 and is_list_like(new) and getattr(new, "ndim", 1) == 1:
if nlocs == len(new):
shape = np.shape(new)
# np.shape compat for if setitem_datetimelike_compat
# changed arraylike to list e.g. test_where_dt64_2d

if nlocs == shape[-1]:
# GH#30567
# If length of ``new`` is less than the length of ``values``,
# `np.putmask` would first repeat the ``new`` array and then
Expand All @@ -90,7 +94,7 @@ def putmask_without_repeat(
# to place in the masked locations of ``values``
np.place(values, mask, new)
# i.e. values[mask] = new
elif mask.shape[-1] == len(new) or len(new) == 1:
elif mask.shape[-1] == shape[-1] or shape[-1] == 1:
np.putmask(values, mask, new)
else:
raise ValueError("cannot assign mismatch length to masked array")
Expand Down
44 changes: 33 additions & 11 deletions pandas/core/internals/blocks.py
Original file line number Diff line number Diff line change
Expand Up @@ -1478,26 +1478,48 @@ def putmask(self, mask, new) -> list[Block]:
new = self._maybe_squeeze_arg(new)
mask = self._maybe_squeeze_arg(mask)

if not mask.any():
return [self]

try:
# Caller is responsible for ensuring matching lengths
values._putmask(mask, new)
except (TypeError, ValueError) as err:
_catch_deprecated_value_error(err)

if is_interval_dtype(self.dtype):
# Discussion about what we want to support in the general
# case GH#39584
blk = self.coerce_to_target_dtype(orig_new)
return blk.putmask(orig_mask, orig_new)
if self.ndim == 1 or self.shape[0] == 1:

elif isinstance(self, NDArrayBackedExtensionBlock):
# NB: not (yet) the same as
# isinstance(values, NDArrayBackedExtensionArray)
blk = self.coerce_to_target_dtype(orig_new)
return blk.putmask(orig_mask, orig_new)
if is_interval_dtype(self.dtype):
# Discussion about what we want to support in the general
# case GH#39584
blk = self.coerce_to_target_dtype(orig_new)
return blk.putmask(orig_mask, orig_new)

elif isinstance(self, NDArrayBackedExtensionBlock):
# NB: not (yet) the same as
# isinstance(values, NDArrayBackedExtensionArray)
blk = self.coerce_to_target_dtype(orig_new)
return blk.putmask(orig_mask, orig_new)

else:
raise

else:
raise
# Same pattern we use in Block.putmask
is_array = isinstance(orig_new, (np.ndarray, ExtensionArray))

res_blocks = []
nbs = self._split()
for i, nb in enumerate(nbs):
n = orig_new
if is_array:
# we have a different value per-column
n = orig_new[:, i : i + 1]

submask = orig_mask[:, i : i + 1]
rbs = nb.putmask(submask, n)
res_blocks.extend(rbs)
return res_blocks

return [self]

Expand Down
12 changes: 10 additions & 2 deletions pandas/tests/frame/indexing/test_where.py
Original file line number Diff line number Diff line change
Expand Up @@ -988,8 +988,16 @@ def _check_where_equivalences(df, mask, other, expected):
res = df.mask(~mask, other)
tm.assert_frame_equal(res, expected)

# Note: we cannot do the same with frame.mask(~mask, other, inplace=True)
# bc that goes through Block.putmask which does *not* downcast.
# Note: frame.mask(~mask, other, inplace=True) takes some more work bc
# Block.putmask does *not* downcast. The change to 'expected' here
# is specific to the cases in test_where_dt64_2d.
df = df.copy()
df.mask(~mask, other, inplace=True)
if not mask.all():
# with mask.all(), Block.putmask is a no-op, so does not downcast
expected = expected.copy()
expected["A"] = expected["A"].astype(object)
tm.assert_frame_equal(df, expected)


def test_where_dt64_2d():
Expand Down
15 changes: 15 additions & 0 deletions pandas/tests/series/indexing/test_setitem.py
Original file line number Diff line number Diff line change
Expand Up @@ -1625,3 +1625,18 @@ def test_setitem_bool_indexer_dont_broadcast_length1_values(size, mask, item, bo
expected = Series(np.arange(size, dtype=float))
expected[selection] = item
tm.assert_series_equal(ser, expected)


def test_setitem_empty_mask_dont_upcast_dt64():
dti = date_range("2016-01-01", periods=3)
ser = Series(dti)
orig = ser.copy()
mask = np.zeros(3, dtype=bool)

ser[mask] = "foo"
assert ser.dtype == dti.dtype # no-op -> dont upcast
tm.assert_series_equal(ser, orig)

ser.mask(mask, "foo", inplace=True)
assert ser.dtype == dti.dtype # no-op -> dont upcast
tm.assert_series_equal(ser, orig)