DEPR: squeeze() argument in read_csv/read_table

[lithomas1]

I'm pretty sure you can do this directly with a .squeeze() afterwards.
The code does look pretty similar
squeeze()

pandas/pandas/core/generic.py

Lines 979 to 985 in 5f648bf

	axis = range(self._AXIS_LEN) if axis is None else (self._get_axis_number(axis),)
	return self.iloc[
	tuple(
	0 if i in axis and len(a) == 1 else slice(None)
	for i, a in enumerate(self.axes)
	)
	]

vs

read_csv(squeeze=True)

pandas/pandas/io/parsers/readers.py

Lines 1063 to 1064 in 1530210

	if self.squeeze and len(df.columns) == 1:
	return df[df.columns[0]].copy()

[phofl]

This is not completly the same, see

df = pd.DataFrame({"a": [1]})

df.squeeze()
df["a"]

First returns a scalar, the second a series.

But I think it is worth discussing this nevertheless.

[jreback]

iirc there is another issue about deprecating squeeze but in any event +1 to do it (i thought we did this long ago)

[phofl]

We deprecated squeeze in groupby early last year, maybe this?

[lithomas1]

I think the equivalent is squeeze("columns")

>>> import pandas as pd
>>> a = pd.DataFrame({"a": [1]})
>>> a.squeeze("columns")
0    1
Name: a, dtype: int64

[phofl]

Good point, thx

[lithomas1]

This should also be deprecated in read_excel, and ExcelFile.parse too.
cc @rhshadrach

[rhshadrach]

+1 on deprecating squeeze for read_excel