to_timedelta does not recognize 'h' and 'm' units

[vfilimonov]

'h' and 'm' units are not correctly recognized in to_timedelta:

print pd.to_timedelta(1,unit='D')
print pd.to_timedelta(1,unit='h')
print pd.to_timedelta(1,unit='m')
print pd.to_timedelta(1,unit='s')
print pd.to_timedelta(1,unit='ms')
print pd.to_timedelta(1,unit='us')
print pd.to_timedelta(1,unit='ns')

results in

86400000000000 nanoseconds
1 nanoseconds
1 nanoseconds
1000000000 nanoseconds
1000000 nanoseconds
1000 nanoseconds
1 nanoseconds

version: 0.14.0

Most likely related to Issue #6423

[jreback]

these are not valid numpy units (eg have to use H and M)
that said I thought this did work

[vfilimonov]

It's the same issue for 'H' and 'M...

I picked 'h' and 'm' from to_timedelta's docstring:

unit : unit of the arg (D,h,m,s,ms,us,ns) denote the unit, which is an integer/float number

[jreback]

actually this is completely tested
what is your numpy version?

[vfilimonov]

Sorry - forgot to tell: numpy 1.8.1

[vfilimonov]

Great, @jreback, thanks a lot!

[jreback]

@vfilimonov thanks for reporting! was a bit buggy

[shoyer]

@jreback Is there still a good reason for not handling h and m in pd.to_timedelta? It looks like pd.tslib.cast_from_unit can handle them now, and everything appears to work if I add them to regex in pandas.tseries.timedeltas.

[jreback]

#7616

fixed that - it was buggy before

[shoyer]

@jreback Can we re-open this issue then? I'd like to submit a patch...

[jreback]

can u post what is not working?