Handling time in string format

[SarahAlidoost]

In ewatercycle, the start/end time is given as a string in the ISO format (which uses the Grogerian calendar), and in UTC time. A date string ISO format is represented differently by different libraries:

• datetime.datetime

from datetime import datetime, timezone
datetime.fromisoformat('2000-01-01T00:00:00+00:00')
#out: datetime.datetime(2000, 1, 1, 0, 0, tzinfo=datetime.timezone.utc)
datetime(2000,1, 1, tzinfo=timezone.utc).isoformat()
#out: '2000-01-01T00:00:00+00:00'

• numpy.datetime64

import numpy as np
np.datetime64('2000-01-01T00:00:00.0')
#out: numpy.datetime64('2000-01-01T00:00:00.000')
np.datetime_as_string(np.array(np.datetime64('2000-01-01T00:00:00.0')), timezone='UTC')
#out: '2000-01-01T00:00:00.000Z'

• pandas.date_range

import pandas as pd
pd.date_range(start='1/1/2000', periods=1, tz='UTC')
#out: DatetimeIndex(['2000-01-01 00:00:00+00:00'], dtype='datetime64[ns, UTC]', freq='D')

• cftime.datetime

import cftime
cftime.datetime(2000,1, 1)
#out: cftime.datetime(2000, 1, 1, 0, 0, 0, 0)
cftime.datetime(2000,1, 1).isoformat()
#out: '2000-01-01T00:00:00'

• xarray with np.datetime64

import numpy as np
import xarray as xr
da = xr.DataArray(
        [0.1],
        coords=[[np.datetime64('2000-01-01')]],
        dims=['time'],
        )
da.coords['time'][0]
#out: array('2000-01-01T00:00:00.000000000', dtype='datetime64[ns]')
np.datetime_as_string(da.coords['time'][0], timezone='UTC')
#out: '2000-01-01T00:00:00.000000000Z'

• xarray with datetime

from datetime import datetime, timezone
import xarray as xr
da = xr.DataArray(
        [0.1],
        coords=[[datetime(2000, 1, 1, tzinfo=timezone.utc)]],
        dims=['time'],
        )
da.coords['time'][0]
#out: array('2000-01-01T00:00:00.000000000', dtype='datetime64[ns]')
np.datetime_as_string(da.coords['time'][0], timezone='UTC')
#out: '2000-01-01T00:00:00.000000000Z'

[SarahAlidoost]

I suggest using the numpy iso format i.e. '2000-01-01T00:00:00Z' and library dateutil to parse it as:

from dateutil.parser import parse
parse('2000-01-01T00:00:00Z')
#out: datetime.datetime(2000, 1, 1, 0, 0, tzinfo=tzutc())

and to convert the datetime object back to the iso format:

parse('2000-01-01T00:00:00Z').strftime("%Y-%m-%dT%H:%M:%SZ")
#out: '2000-01-01T00:00:00Z'

[Peter9192]

Nice overview! So the internal representation of dateutil is 'normal' datetime.datetime objects?

[SarahAlidoost]

Nice overview! So the internal representation of dateutil is 'normal' datetime.datetime objects?

It seems so. It is mentioned in the documentation that "the dateutil module provides powerful extensions to the standard datetime module".

[Peter9192]

Addressed in #137