BUG: joining empty series with dtype: datetime64[ns, UTC]

[kcajf]

Code Sample

>>> s1 = pd.Series(pd.to_datetime([], utc=True))
>>> s2 = pd.Series([1,2,3])
>>> pd.concat([s1, s2], axis=1)

IndexError: cannot do a non-empty take from an empty axes.


>>> s1 = pd.Series(pd.to_datetime([], utc=False))
>>> s2 = pd.Series([1,2,3])
>>> pd.concat([s1, s2], axis=1)

Empty DataFrame
Columns: [0, 1]
Index: []


>>> s1 = pd.Series(pd.to_datetime([], utc=True))
>>> s2 = pd.Series([])
>>> pd.concat([s1, s2], axis=1)

Empty DataFrame
Columns: [0, 1]
Index: []


>>> df1 = pd.DataFrame(columns=['a', 'b'])
>>> df2 = pd.DataFrame(np.random.random((2, 2)), columns=['c', 'd'])
>>> df1['a'] = pd.to_datetime(df1['b'], utc=True) 
>>> pd.concat([df1, df2], axis=1)

IndexError: cannot do a non-empty take from an empty axes.


>>> df1.join(df2, how='outer')

IndexError: cannot do a non-empty take from an empty axes.


>>> df1['a'] = pd.to_datetime(df1['b'], utc=False) 
>>> pd.concat([df1, df2], axis=1)

    a    b         c         d
0 NaT  NaN  0.777252  0.657679
1 NaT  NaN  0.274332  0.981532

Problem description

When trying to concatenate multiple series (or dataframes) along axis 1, if one of them is empty and has a UTC datetime column, the concatenation will fail with IndexError. This applies to joins as well. If you set the datetime column to be non-utc (i.e. tz-naive), it works as expected. If you concatenate 2 empty objects, one of which has a UTC datetime column, it works as expected.

Output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.6.2.final.0
python-bits: 64
OS: Linux
OS-release: 4.12.9-300.fc26.x86_64
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_GB.utf8
LOCALE: en_GB.UTF-8

pandas: 0.21.0
pytest: 3.0.7
pip: 9.0.1
setuptools: 36.6.0
Cython: None
numpy: 1.13.3
scipy: 0.19.1
pyarrow: 0.7.1
xarray: 0.9.6
IPython: 6.2.1
sphinx: None
patsy: 0.4.1
dateutil: 2.6.0
pytz: 2017.2
blosc: None
bottleneck: None
tables: 3.4.2
numexpr: 2.6.4
feather: 0.4.0
matplotlib: 2.0.2
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: None
lxml: None
bs4: None
html5lib: 1.0b10
sqlalchemy: None
pymysql: None
psycopg2: None
jinja2: 2.9.6
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: 0.4.0

[jreback]

this is relateed to #12396

should be straightforward to fix, if you can do a PR!