BUG: HDFStore failures on timezone-aware data

[adshieh]

Code Sample, a copy-pastable example if possible

In [1]: import pandas as pd
   ...: 
   ...: def check_roundtrip(obj):
   ...:     with pd.HDFStore('test.h5', 'w') as store:
   ...:         store['obj'] = obj
   ...:         retrieved = store['obj']
   ...:         return obj.equals(retrieved)
   ...: 
   ...: s = pd.Series([], dtype='datetime64[ns, UTC]')
   ...: t = pd.Series([0], dtype='datetime64[ns, UTC]')
   ...: df = pd.DataFrame({'A': s})

In [2]: print(check_roundtrip(s))
False

In [3]: print(check_roundtrip(t))
False

In [4]: print(check_roundtrip(df))

------------------------------------------------------------------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-4-57f1e9853bf6> in <module>()
----> 1 print(check_roundtrip(df))

<ipython-input-1-1d7fb84453e9> in check_roundtrip(obj)
      4     with pd.HDFStore('test.h5', 'w') as store:
      5         store['obj'] = obj
----> 6         retrieved = store['obj']
      7         return obj.equals(retrieved)
      8 

/home/ashieh/.local/lib/python2.7/site-packages/pandas/io/pytables.pyc in __getitem__(self, key)
    481 
    482     def __getitem__(self, key):
--> 483         return self.get(key)
    484 
    485     def __setitem__(self, key, value):

/home/ashieh/.local/lib/python2.7/site-packages/pandas/io/pytables.pyc in get(self, key)
    669         if group is None:
    670             raise KeyError('No object named %s in the file' % key)
--> 671         return self._read_group(group)
    672 
    673     def select(self, key, where=None, start=None, stop=None, columns=None,

/home/ashieh/.local/lib/python2.7/site-packages/pandas/io/pytables.pyc in _read_group(self, group, **kwargs)
   1347         s = self._create_storer(group)
   1348         s.infer_axes()
-> 1349         return s.read(**kwargs)
   1350 
   1351 

/home/ashieh/.local/lib/python2.7/site-packages/pandas/io/pytables.pyc in read(self, start, stop, **kwargs)
   2902             blk_items = self.read_index('block%d_items' % i)
   2903             values = self.read_array('block%d_values' % i,
-> 2904                                      start=_start, stop=_stop)
   2905             blk = make_block(values,
   2906                              placement=items.get_indexer(blk_items))

/home/ashieh/.local/lib/python2.7/site-packages/pandas/io/pytables.pyc in read_array(self, key, start, stop)
   2464             if shape is not None:
   2465                 # length 0 axis
-> 2466                 ret = np.empty(shape, dtype=dtype)
   2467             else:
   2468                 ret = node[start:stop]

TypeError: Invalid datetime unit in metadata string "[ns, UTC]"

Problem description

HDFStore fails to save empty/non-empty Series and empty DataFrames with timezone-aware data correctly. The issue is that the timezone information is not being saved correctly. For Series, there is no timezone handling at all. For DataFrames, the timezone handling is skipped when empty.

Expected Output

The checks should pass.

Output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 2.7.12.final.0
python-bits: 64
OS: Linux
OS-release: 4.4.0-1049-aws
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
LOCALE: None.None

pandas: 0.22.0
pytest: 3.5.0
pip: 9.0.3
setuptools: 39.0.1
Cython: 0.28.1
numpy: 1.14.2
scipy: 1.0.1
pyarrow: 0.9.0
xarray: None
IPython: 5.6.0
sphinx: None
patsy: None
dateutil: 2.7.2
pytz: 2018.3
blosc: None
bottleneck: None
tables: 3.4.2
numexpr: 2.6.4
feather: None
matplotlib: 2.2.2
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: None
lxml: None
bs4: None
html5lib: 1.0.1
sqlalchemy: None
pymysql: 0.8.0
psycopg2: None
jinja2: 2.10
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: None

[jreback]

this is not supported at all for fixed stores
use format=‘table’

this is a duplicate issue

[adshieh]

Timezones seem to be well supported for non-empty DataFrames in fixed format, unless I'm missing something?

[jreback]

I had closed this a while back for non-empty: #11628

[arw2019]

Confirming this exists on 1.2 master

Output of pd.show_versions()

INSTALLED VERSIONS

commit : c8aea2c
python : 3.8.3.final.0
python-bits : 64
OS : Linux
OS-release : 5.4.0-48-generic
Version : #52-Ubuntu SMP Thu Sep 10 10:58:49 UTC 2020
machine : x86_64
processor :
byteorder : little
LC_ALL : C.UTF-8
LANG : C.UTF-8
LOCALE : en_US.UTF-8

pandas : 1.2.0.dev0+688.gc8aea2c20.dirty
numpy : 1.18.5
pytz : 2020.1
dateutil : 2.8.1
pip : 20.1.1
setuptools : 49.1.0.post20200704
Cython : 0.29.21
pytest : 5.4.3
hypothesis : 5.19.0
sphinx : 3.1.1
blosc : None
feather : None
xlsxwriter : 1.2.9
lxml.etree : 4.5.2
html5lib : 1.1
pymysql : None
psycopg2 : 2.8.5 (dt dec pq3 ext lo64)
jinja2 : 2.11.2
IPython : 7.16.1
pandas_datareader: None
bs4 : 4.9.1
bottleneck : 1.3.2
fsspec : 0.7.4
fastparquet : 0.4.0
gcsfs : 0.6.2
matplotlib : 3.2.2
numexpr : 2.7.1
odfpy : None
openpyxl : 3.0.4
pandas_gbq : None
pyarrow : 1.0.1
pytables : None
pyxlsb : None
s3fs : 0.4.2
scipy : 1.5.0
sqlalchemy : 1.3.18
tables : 3.6.1
tabulate : 0.8.7
xarray : 0.15.1
xlrd : 1.2.0
xlwt : 1.3.0
numba : 0.50.1