Skip to content

BUG: DataFrame bugs in ndarray assignments #53367

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

Closed
Closed
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
2 changes: 2 additions & 0 deletions doc/source/whatsnew/v2.1.0.rst
Original file line number Diff line number Diff line change
Expand Up @@ -432,6 +432,8 @@ Groupby/resample/rolling
Reshaping
^^^^^^^^^
- Bug in :func:`crosstab` when ``dropna=False`` would not keep ``np.nan`` in the result (:issue:`10772`)
- Bug in :meth:`BlockManager.insert` allows assignments for multiple-dimension ndarray into one column (:issue:`53366`)
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

BlockManager isn't user-facing. can you write this in terms of a DataFrame or Series method?

- Bug in :meth:`DataFrame.__setitem__` allows assignments for multiple-dimension ndarray into one column (:issue:`51925`)
- Bug in :meth:`DataFrame.agg` and :meth:`Series.agg` on non-unique columns would return incorrect type when dist-like argument passed in (:issue:`51099`)
- Bug in :meth:`DataFrame.idxmin` and :meth:`DataFrame.idxmax`, where the axis dtype would be lost for empty frames (:issue:`53265`)
- Bug in :meth:`DataFrame.merge` not merging correctly when having ``MultiIndex`` with single level (:issue:`52331`)
Expand Down
18 changes: 18 additions & 0 deletions pandas/core/frame.py
Original file line number Diff line number Diff line change
Expand Up @@ -3966,6 +3966,24 @@ def __setitem__(self, key, value):
):
# Column to set is duplicated
self._setitem_array([key], value)
elif (
isinstance(value, np.ndarray) and value.ndim > 1 and self.columns.is_unique
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

does it matter what 'key' is here?

):
# TODO: a check for MultiIndex should be added
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

does the check on the next line not take care of this TODO?

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

does the check on the next line not take care of this TODO?

No, the check here is just placeholder for further improvements, and the real check should be inserted between "isinstance" line and "_set_item" line. I did not figure out what and how checks for MultiIndex should be added so I skipped this. But I think the current PR is a mitigation covering and fixing frequently used scenes.

if isinstance(self.columns, MultiIndex):
self._set_item(key, value)
return
# squeeze 2d ndarray to 1d if possible
# this keeps the backward compatibility
if np.ndim(value) == 2 and (1 in np.shape(value)):
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

can you use value.ndim and values.shape instead of the numpy methods

# if value is np.matrix, convert to np.ndarray
value = np.asarray(value).flatten()
self._set_item(key, value)
else:
# avoid assign non 1d array to column
raise ValueError(
f"Expected a 1D array, got an array with shape {value.shape}"
)
else:
# set column
self._set_item(key, value)
Expand Down
2 changes: 1 addition & 1 deletion pandas/core/internals/managers.py
Original file line number Diff line number Diff line change
Expand Up @@ -1401,7 +1401,7 @@ def insert(self, loc: int, item: Hashable, value: ArrayLike, refs=None) -> None:
# insert to the axis; this could possibly raise a TypeError
new_axis = self.items.insert(loc, item)

if value.ndim == 2:
if value.ndim >= 2:
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

why is this needed?

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

why is this needed?

A ndarray where dim >= 2 will bypass the check here, but still raise an Exception when accessing the DataFrame. For an example for this bug, see #53366

value = value.T
if len(value) > 1:
raise ValueError(
Expand Down
5 changes: 5 additions & 0 deletions pandas/tests/frame/test_constructors.py
Original file line number Diff line number Diff line change
Expand Up @@ -283,6 +283,11 @@ def test_constructor_cast_failure(self):
with pytest.raises(ValueError, match=msg):
df["test"] = np.ones((4, 2))

# this is not ok
msg = "Expected a 1D array, got an array with shape \\(4, 2, 3\\)"
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this file is for constructor tests. can you put a dedicated test in tests/frame/indexing/test_setitem.py

with pytest.raises(ValueError, match=msg):
df["test"] = np.ones((4, 2, 3))

# this is ok
df["foo2"] = np.ones((4, 2)).tolist()

Expand Down