[AstroPy] Variable Length Array Tables in PyFits/Numpy

Rudolph van der Merwe rudolph at ska.ac.za
Wed Jan 10 05:52:32 CST 2007

I'm running into trouble trying to use PyFITS 1.1b3 (built against
Numpy) to create a binary variable length array table (section 6.2.1
in the PyFITS manual).

The manual states that for such a table, the field data (attribute
'array') must be a numarray objects array. If I use the equivalent
Numpy object array, i.e.

  numpy.array([a, b], dtype=object)

PyFits aborts with an error. Here is a full listing of my sessions:

In [2]: import numpy as np

In [3]: import pyfits as pf

In [4]: a=np.array([1,2,3], dtype=int)

In [5]: b=np.array([4,5], dtype=int)

In [6]: x=np.array([a,b], dtype=object)

In [7]: c1 = pf.Column(name='var', format='PJ()', array=x)

In [8]: cols = pf.ColDefs([c1])

In [9]: tbHDU = pf.new_table(cols)
exceptions.TypeError                                 Traceback (most
recent call last)

/usr/lib/python2.4/site-packages/pyfits/NP_pyfits.py in
new_table(input, header, nrows, fill, tbtype)
   3110             elif isinstance(tmp._recformats[i], _FormatP):
   3111 #                hdu.data._convert[i] =
_makep(tmp._arrays[i][:n], hdu.data._parent.field(i)[:n],
-> 3112                 hdu.data._convert[i] =
_makep(tmp._arrays[i][:n], rec.recarray.field(hdu.data,i)[:n],
   3113             else:
   3114                 if tbtype == 'TableHDU':

/usr/lib/python2.4/site-packages/pyfits/NP_pyfits.py in _makep(input,
desp_output, dtype)
   2585     _offset = 0
-> 2586     data_output = _VLF([None]*len(input))
   2587     data_output._dtype = dtype

TypeError: an integer is required

In [10]:

Any idea how to address this?


Rudolph van der Merwe
KAT (Karoo Array Telescope) / www.kat.ac.za

More information about the AstroPy mailing list