Skip to content
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.

Sign up for GitHub

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

TST(string dtype): Resolve replace xfails #60659

Merged
merged 3 commits into from
Jan 10, 2025
Merged

TST(string dtype): Resolve replace xfails #60659

Changes from 2 commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
f2adc92
TST(string dtype): Resolve replace xfails
rhshadrach Jan 4, 2025
d9ead32
Add test
rhshadrach Jan 4, 2025
2981e29
fixup
rhshadrach Jan 5, 2025
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
18 changes: 10 additions & 8 deletions pandas/tests/frame/methods/test_replace.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -334,7 +334,6 @@ def test_regex_replace_str_to_numeric(self, mix_abc):
return_value = res3.replace(regex=r"\s*\.\s*", value=0, inplace=True)
assert return_value is None
expec = DataFrame({"a": mix_abc["a"], "b": ["a", "b", 0, 0], "c": mix_abc["c"]})
# TODO(infer_string)
expec["c"] = expec["c"].astype(object)
tm.assert_frame_equal(res, expec)
tm.assert_frame_equal(res2, expec)
Expand Down Expand Up @@ -1469,20 +1468,23 @@ def test_regex_replace_scalar(
tm.assert_frame_equal(result, expected)

@pytest.mark.parametrize("regex", [False, True])
def test_replace_regex_dtype_frame(self, regex):
@pytest.mark.parametrize("value", [1, "1"])
def test_replace_regex_dtype_frame(self, regex, value):
# GH-48644
df1 = DataFrame({"A": ["0"], "B": ["0"]})
expected_df1 = DataFrame({"A": [1], "B": [1]}, dtype=object)
result_df1 = df1.replace(to_replace="0", value=1, regex=regex)
# When value is an integer, coerce result to object.
# When value is a string, infer the correct string dtype.
Copy link
Member

Choose a reason for hiding this comment

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

Are we sure we want to coerce to string instead of raising? The object case makes sense I'm just not as sure onn the string side if we should be implicitly casting like that

Copy link
Member Author

Choose a reason for hiding this comment

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

Not sure I understand. When infer_string=True, the input DataFrame is str dtype. Then when we go to replace "0" with value="1", certainly we want the result to still be str dtype, no?

Copy link
Member

Choose a reason for hiding this comment

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

Yea I think that makes sense, but I'm not as sure when the target value is a non-string, i.e. replace(..., value=1)

Copy link
Member Author

Choose a reason for hiding this comment

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

When the target value is a non-string, we coerce to object dtype in order to hold both integers and strings. What are you not sure about?

Copy link
Member

Choose a reason for hiding this comment

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

Sorry I was just misreading the comment - I think this is good

dtype = object if value == 1 else None

expected_df1 = DataFrame({"A": [value], "B": [value]}, dtype=dtype)
result_df1 = df1.replace(to_replace="0", value=value, regex=regex)
tm.assert_frame_equal(result_df1, expected_df1)

df2 = DataFrame({"A": ["0"], "B": ["1"]})
if regex:
# TODO(infer_string): both string columns get cast to object,
# while only needed for column A
expected_df2 = DataFrame({"A": [1], "B": ["1"]}, dtype=object)
Comment on lines -1481 to -1483
Copy link
Member Author

Choose a reason for hiding this comment

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

I think this behavior was correct - we get object dtype here because we are trying to replace string values with integer values. If we were to make the result a string dtype, then that would be introducing value-specific behavior.

expected_df2 = DataFrame({"A": [1], "B": ["1"]}, dtype=dtype)
else:
expected_df2 = DataFrame({"A": Series([1], dtype=object), "B": ["1"]})
Copy link
Member Author

Choose a reason for hiding this comment

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

This behavior looks incorrect to me, B here should also be object dtype. I think we can raise an issue (this is independent of infer_strings) if others agree.

expected_df2 = DataFrame({"A": Series([1], dtype=dtype), "B": ["1"]})
result_df2 = df2.replace(to_replace="0", value=1, regex=regex)
tm.assert_frame_equal(result_df2, expected_df2)

Expand Down
Loading