list of day-light savings days

[gmohandas]

Would it be useful to add a feature that returns a list of days within a given time interval where the clock changes due to daylight savings? Personally, I've had occasion to find such days/dates as part of my work. I do not know of a one-shot way with which to do so in pandas at the moment. I'd be happy to contribute this functionality as an enhancement if there is some interest.

[gfyoung]

cc @mroeschke

Not aware of a one-shot way either. As for your proposal, I'm uncertain of the degree of use that such functionality would get vs. the maintenance efforts involved.

If you do have an implementation, feel free to share it here though.

[gmohandas]

@gfyoung I was wondering of possibly including a function like the one given below in pandas.core.indexes.datetimes. Not including all the necessary tests and accompaniments of course.

def get_dls_days(startdate, enddate):
   # array of all Sundays within the specified time interval.
    dran = pd.date_range(start=startdate, end=enddate, freq='W-SUN')

    # keep only the last Sunday falling in March and October
    dls_days = dran[(dran.month == 3) & (dran.day >= 25) |
                    (dran.month == 10) & (dran.day >= 25)]

    if not dls_days.size:
        print('No daylight saving days in the specified time interval.')
    else:
        return dls_days

To be honest, this was also my idea of making a first contribution to the pandas codebase (or to any major project for that matter). If it isn't worth the effort, never mind, but if it might be a nice extra feature, then I'd welcome any thoughts and suggestions.

[jreback]

see #22560; this method was rejected as not generally that useful for user facing operations as tz's are complicated. In any event it is pretty easy to generically do this (your method about is not very generic).

In [45]: idx = date_range('2018-01-01', '2018-12-31', freq='H', tz='US/Eastern')                                                                                                                                                                                 

In [46]: dst = idx.map(lambda x: x.dst())                                                                                                                                                                                                                      

In [47]: idx[1:][dst[1:]!=dst[:-1]]                                                                                                                                                                                                                            
Out[47]: DatetimeIndex(['2018-03-11 03:00:00-04:00', '2018-11-04 01:00:00-05:00'], dtype='datetime64[ns, US/Eastern]', freq=None)