Numpy recarray append_fields: unable to add numpy datetimes array
I have a recarray containing various fields and I want to add an array of datetime objects to it.
However, it looks like the function append_fields
in numpy.lib.recfunctions
won't let me add an array of objects.
Here's some sample code:
import numpy as np
import datetime
import numpy.lib.recfunctions as recfun
dtype= np.dtype([('WIND_WAVE_HGHT', '<f4'), ('WIND_WAVE_PERD', '<f4')])
obs = np.array([(0.1,10.0),(0.2,11.0),(0.3,12.0)], dtype=dtype)
dates = np.array([datetime.datetime(2001,1,1,0),
datetime.datetime(2001,1,1,0),
datetime.datetime(2001,1,1,0)])
# This doesn't work:
recfun.append_fields(obs,'obdate',dates,dtypes=np.object)
I keep getting the error TypeError: Cannot change data-type for object array.
Only seems to be a problem with np.object arrays, since I can add other ok fields. Did I miss something?
source to share
Problem
In [143]: recfun.append_fields(obs,'test',np.array([None,[],1]))
---------------------------------------------------------------------------
TypeError Traceback (most recent call last)
<ipython-input-143-5c3de23b09f7> in <module>()
----> 1 recfun.append_fields(obs,'test',np.array([None,[],1]))
/usr/local/lib/python3.5/dist-packages/numpy/lib/recfunctions.py in append_fields(base, names, data, dtypes, fill_value, usemask, asrecarray)
615 if dtypes is None:
616 data = [np.array(a, copy=False, subok=True) for a in data]
--> 617 data = [a.view([(name, a.dtype)]) for (name, a) in zip(names, data)]
618 else:
619 if not isinstance(dtypes, (tuple, list)):
/usr/local/lib/python3.5/dist-packages/numpy/lib/recfunctions.py in <listcomp>(.0)
615 if dtypes is None:
616 data = [np.array(a, copy=False, subok=True) for a in data]
--> 617 data = [a.view([(name, a.dtype)]) for (name, a) in zip(names, data)]
618 else:
619 if not isinstance(dtypes, (tuple, list)):
/usr/local/lib/python3.5/dist-packages/numpy/core/_internal.py in _view_is_safe(oldtype, newtype)
363
364 if newtype.hasobject or oldtype.hasobject:
--> 365 raise TypeError("Cannot change data-type for object array.")
366 return
367
TypeError: Cannot change data-type for object array.
So the problem is with this expression a.view([(name, a.dtype)])
. It tries to create a single structured array from a
. This works for types like int and str, but with an error object
. This failure is in the kernel view
, so this is unlikely to change.
In [148]: x=np.arange(3)
In [149]: x.view([('test', x.dtype)])
Out[149]:
array([(0,), (1,), (2,)],
dtype=[('test', '<i4')])
In [150]: x=np.array(['one','two'])
In [151]: x.view([('test', x.dtype)])
Out[151]:
array([('one',), ('two',)],
dtype=[('test', '<U3')])
In [152]: x=np.array([[1],[1,2]])
In [153]: x
Out[153]: array([[1], [1, 2]], dtype=object)
In [154]: x.view([('test', x.dtype)])
...
TypeError: Cannot change data-type for object array.
The fact that it recfunctions
requires a separate load indicates that it is somewhat of a lead, which is not used much, and is not in active development. I haven't studied the code in detail, but I suspect the fix will be kludge.
A fix
Here you can add a new field from scratch. It performs the same basic steps as append_fields
:
Define a new dtype using obs
both the new field name and dtype:
In [158]: obs.dtype.descr
Out[158]: [('WIND_WAVE_HGHT', '<f4'), ('WIND_WAVE_PERD', '<f4')]
In [159]: obs.dtype.descr+[('TEST',object)]
Out[159]: [('WIND_WAVE_HGHT', '<f4'), ('WIND_WAVE_PERD', '<f4'), ('TEST', object)]
In [160]: dt1 =np.dtype(obs.dtype.descr+[('TEST',object)])
Create an empty target array and fill it by copying the data by field name:
In [161]: newobs = np.empty(obs.shape, dtype=dt1)
In [162]: for n in obs.dtype.names:
...: newobs[n]=obs[n]
In [167]: dates
Out[167]:
array([datetime.datetime(2001, 1, 1, 0, 0),
datetime.datetime(2001, 1, 1, 0, 0),
datetime.datetime(2001, 1, 1, 0, 0)], dtype=object)
In [168]: newobs['TEST']=dates
In [169]: newobs
Out[169]:
array([( 0.1 , 10., datetime.datetime(2001, 1, 1, 0, 0)),
( 0.2 , 11., datetime.datetime(2001, 1, 1, 0, 0)),
( 0.30000001, 12., datetime.datetime(2001, 1, 1, 0, 0))],
dtype=[('WIND_WAVE_HGHT', '<f4'), ('WIND_WAVE_PERD', '<f4'), ('TEST', 'O')])
alternative datetime64
When using native numty datetimes, append works
In [179]: dates64 = dates.astype('datetime64[D]')
In [180]: recfun.append_fields(obs,'test',dates64,usemask=False)
Out[180]:
array([( 0.1 , 10., '2001-01-01'),
( 0.2 , 11., '2001-01-01'), ( 0.30000001, 12., '2001-01-01')],
dtype=[('WIND_WAVE_HGHT', '<f4'), ('WIND_WAVE_PERD', '<f4'), ('test', '<M8[D]')])
append_fields
has a few bells-n-whistles that my version doesn't - fills in values, masks arrays, repeats, etc.
array of structured dates
I could create a structured array with dates
In [197]: sdates = np.array([(i,) for i in dates],dtype=[('test',object)])
In [198]: sdates
Out[198]:
array([(datetime.datetime(2001, 1, 1, 0, 0),),
(datetime.datetime(2001, 1, 1, 0, 0),),
(datetime.datetime(2001, 1, 1, 0, 0),)],
dtype=[('test', 'O')])
There should be a function that concatenates the fields of existing arrays, but I can't find it.
Previous work
It was familiar:
https://github.com/numpy/numpy/issues/2346
TypeError when adding fields to a structured array of size ONE
source to share