Datetime as coordinaets does not convert back to datetime (returns int)

[feefladder]

What happened:
datetime was in np.datetime64 formet. When converted t datetime.datetime format it returned an int
What you expected to happen:
`to get a datetime returned
Minimal Complete Verifiable Example:

# Put your MCVE code here
import xarray as xr
import numpy as np
import datetime
date_frame = xr.DataArray(dims='time',coords={'time':pd.date_range('2000-01-01',periods=365)},data=np.zeros(365))
print('pandas date range (datetime): ',pd.date_range('2000-01-01',periods=365)[0])
print('dataframe datetime converted to datetime (int): ',date_frame.coords['time'].data[0].astype(datetime.datetime))
print("normal numpy datetime64 converted to datetime (datetime): ",np.datetime64(datetime.datetime(2000,1,1)).astype(datetime.datetime))

output:

pandas date range (datetime):  2000-01-01 00:00:00
dataframe datetime converted to datetime (int):  946684800000000000
normal numpy datetime64 converted to datetime (datetime):  2000-01-01 00:00:00

if converted to int, it also gives different lengths of int :
date_frame:
946684800000000000
946684800000000
normal datetime64^
Anything else we need to know?:

it is also mentioned in this SO thread appears to be a problem in the datetime64....

numpy version 1.20.0
pandas version 1.2.1

Environment:

Output of xr.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.7.9 | packaged by conda-forge | (default, Dec 9 2020, 21:08:20)
[GCC 9.3.0]
python-bits: 64
OS: Linux
OS-release: 5.4.0-59-generic
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
LOCALE: en_US.UTF-8
libhdf5: None
libnetcdf: None

xarray: 0.16.2
pandas: 1.2.1
numpy: 1.20.0
scipy: None
netCDF4: None
pydap: None
h5netcdf: None
h5py: None
Nio: None
zarr: 2.6.1
cftime: None
nc_time_axis: None
PseudoNetCDF: None
rasterio: None
cfgrib: None
iris: None
bottleneck: None
dask: 2021.01.1
distributed: 2021.01.1
matplotlib: None
cartopy: None
seaborn: None
numbagg: None
pint: None
setuptools: 49.6.0.post20210108
pip: 21.0.1
conda: None
pytest: None
IPython: 7.20.0
sphinx: None

[max-sixty]

Thanks for clear example. As per @keewis 's label, this does look like an upstream issue.

[keewis]

yes, I believe this to be a numpy issue:

date_frame.coords['time'].data[0]

returns a numpy.datetime64[ns] object, which, casted to datetime.datetime using astype becomes a int. See numpy/numpy#18363 for the upstream issue.

[feefladder]

Yes, I created that issue :), should this then be closed?

[keewis]

Possibly, if this doesn't require any action from us. @spencerkclark, what do you think?

[spencerkclark]

Yes I think unfortunately there is not much we can do here, since as @keewis noted, data_frame.coords["time"].data is a NumPy array.

This issue in general is a little tricky since xarray and pandas use nanosecond-precision datetime values, but datetime.datetime objects only have microsecond precision. See also discussion in numpy/numpy#8546, numpy/numpy#12550, and #3256.

[max-sixty]

Closing as dupe of #3256