columns comparison with Enum column name triggers TypeError

[s-wakaba]

Code Sample, a copy-pastable example if possible

from enum import Enum
import pandas as pd

Cols = Enum('Cols', 'col1 col2')

q1 = pd.DataFrame({Cols.col1: [1, 2, 3]})
q2 = pd.DataFrame({Cols.col1: [1, 2, 3]})

print((q1[Cols.col1] == q2[Cols.col1]).all())

Problem description

On pandas 0.23.x, comparison of columns in DataFrames which have enum.Enum column names triggers off a TypeError exception. it looks that pandas calls user defined Enum class as a function.

Traceback (most recent call last):
  File "./pandas_enum.py", line 9, in <module>
    print((q1[Cols.col1] == q2[Cols.col1]).all())
  File "/.../lib/python3.6/site-packages/pandas/core/ops.py", line 1217, in wrapper
    name=res_name).rename(res_name)
  File "/.../lib/python3.6/site-packages/pandas/core/series.py", line 3318, in rename
    return super(Series, self).rename(index=index, **kwargs)
  File "/.../lib/python3.6/site-packages/pandas/core/generic.py", line 973, in rename
    level=level)
  File "/.../lib/python3.6/site-packages/pandas/core/internals.py", line 3340, in rename_axis
    obj.set_axis(axis, _transform_index(self.axes[axis], mapper, level))
  File "/.../lib/python3.6/site-packages/pandas/core/internals.py", line 5298, in _transform_index
    items = [func(x) for x in index]
  File "/.../lib/python3.6/site-packages/pandas/core/internals.py", line 5298, in <listcomp>
    items = [func(x) for x in index]
TypeError: 'Cols' object is not callable

Expected Output

just print True. Pandas 0.22.0 does the expected behavior. Pandas 0.23.0 and later has this problem.

Output of pd.show_versions()

>>> pd.show_versions()

INSTALLED VERSIONS
------------------
commit: None
python: 3.6.1.final.0
python-bits: 64
OS: Linux
OS-release: 3.10.0-327.4.5.el7.x86_64
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
LOCALE: en_US.UTF-8

pandas: 0.23.4
pytest: None
pip: 18.0
setuptools: 40.2.0
Cython: None
numpy: 1.15.1
scipy: None
pyarrow: None
xarray: None
IPython: None
sphinx: None
patsy: None
dateutil: 2.7.3
pytz: 2018.5
blosc: None
bottleneck: None
tables: None
numexpr: None
feather: None
matplotlib: None
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: None
lxml: None
bs4: None
html5lib: None
sqlalchemy: None
pymysql: None
psycopg2: None
jinja2: None
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: None

[WillAyd]

Hmm is there any reason why you wouldn't just use the .name of your Enum? Would seem like a much safer approach not sure we want to (or rather can) support indexing by arbitrary objects

[s-wakaba]

My motivation is the same with using Symbol() objects as property keys in Javascript (ES6). It can be an explicit key for both coding and debugging and can also avoid name conflict.

[jkfm]

I have the same problem, and would also prefer not to use the .name-property to ensure specificity of the key.

Or rather: If the reason for this error is an unintended effect of a change in 0.22 to 0.23, what would speak against restoring the original behavior?

[hayd]

Same issue: https://stackoverflow.com/q/55822555/1240268

[sdementen]

I get a similar issue but I am not indexing at all the dataframe or the series, just comparing it

from uuid import uuid4

import pandas as pd

col = uuid4()

sr = pd.Series([4, 5, 6], name=col)
(sr == sr)
# raise TypeError: 'UUID' object is not callable

df = pd.DataFrame({col: [1, 2, 3]})
(df == df)
# raise TypeError: 'UUID' object is not callable

[mroeschke]

Looks to work on master. Could use a test

In [23]: from enum import Enum
    ...: import pandas as pd
    ...:
    ...: Cols = Enum('Cols', 'col1 col2')
    ...:
    ...: q1 = pd.DataFrame({Cols.col1: [1, 2, 3]})
    ...: q2 = pd.DataFrame({Cols.col1: [1, 2, 3]})
    ...:
    ...: print((q1[Cols.col1] == q2[Cols.col1]).all())
True

[AnjaTRPES]

Hey! I am a first-time contributor and would like to help on this. My experience with tests is pretty non-existant, so as far as I understand it one should convert the code-example to a function, and then add this to the appropriate pandas/test suite (maybe to pandas/test/dtypes)?