BUG: The difference in behaviour of indexing between 1.2.4 and 1.3.0

[xiki-tempula]

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• (optional) I have confirmed this bug exists on the master branch of pandas.

---

I have some code that I used to extract the index columns as data frame, which worked in pandas 1.2.4 but not in 1.3.0. I have read what's new (https://pandas.pydata.org/pandas-docs/version/1.3.0/whatsnew/v1.3.0.html). To my best knowledge, this seems to be an undocumented change. I wonder if I can get some advice on how to make the code work for both 1.2 and 1.3 and potentially 1.4.

Code Sample, a copy-pastable example

# To extract the index columns
col = df.reset_index()[df.index.names[:]]

Problem description

Say if we have a data frame.

>>> d = {'col1': [1, 2], 'col2': [3, 4]}
>>> df = pd.DataFrame(data=d)
>>> df = df.set_index('col1')
>>> df
      col2
col1      
1        3
2        4

Expected Output

For pandas == 1.2.4,

>>> df.reset_index()[df.index.names[:]]
   col1
0     1
1     2

For pandas >= 1.3.0,

>>> df.reset_index()[df.index.names[:]]
Traceback (most recent call last):
  File "<input>", line 1, in <module>
  File "~/GitHub/pandas/pandas/core/frame.py", line 3444, in __getitem__
    if com.is_bool_indexer(key):
  File "~/GitHub/pandas/pandas/core/common.py", line 147, in is_bool_indexer
    return len(key) > 0 and lib.is_bool_list(key)
TypeError: Argument 'obj' has incorrect type (expected list, got FrozenList)

[pwwang]

I think this is related to #42433

[xiki-tempula]

@pwwang Thanks. I think we are getting the same problem, but I'm also hoping that someone could give me an alternative solution that works in both 1.2 and 1.3.

[pwwang]

You can do

​df​.​reset_index​()[​list(df​.​index​.​names​[:])]

[stephan-hesselmann-by]

Ups sorry I think I linked the wrong issue

[xiki-tempula]

@pwwang I was hoping to have some more elegant solution but thanks.

[pwwang]

I am hoping so, too. But for 1.3.0, we cannot bypass this as I mentioned in my issue.