Re: [Pytables-users] Follow-up on nested structures

[Francesc A. <fa...@ca...>]

A Dimarts 22 Agost 2006 08:10, Jan Strube va escriure:
> Oh, and yet another question:
> In my structure, I have scalar and vectorial quantities. How can I
> determine which is which on an unfilled table ?
> I am trying row.getTable().col(column_name).ndim, but that seems only to

Please, don't use row.getTable() (it is not documented and, in fact, it is=
=20
going to disappear in forthcoming PyTables 1.4 series). Instead, bind the=20
table to a name (say, 'table') and use the more effective :

table.col(column_name)

as this saves quite a few name lookups.

> work on filled tables. In my case it simply returns 1 for all columns.
>
> Is there a way to determine the dimension of the row I am currently
> filling ?

This is because attribute .ndim of NumPy returns the *number* of dimensions=
 of=20
the object, and not the shape that it is what you want. Please note that,=20
despite the fact that there is a number of different definitions for=20
the "number of dimensions", in pythonic numerical packages this is always=20
defined as len(shape).

The best way of getting the shape for columns without doing an actual read =
of=20
the data is accessing directly to the metadata of the Table object:

print table.description.baryon2.decayVector.shape

or

print table.colshapes['baryon2/decayVector']

Check the different attributes of Table in the manual (section 4.6.1)=20
as well as the Description class (section 4.8) for more accessible metadata=
=20
info.

HTH,

=2D-=20
>0,0<   Francesc Altet =C2=A0 =C2=A0 http://www.carabos.com/
V   V   C=C3=A1rabos Coop. V. =C2=A0=C2=A0Enjoy Data
 "-"