A Monday 01 June 2009 21:09:58 escriguéreu:
> On Jun 1, 2009, at 12:00 PM, Francesc Alted wrote:
> > A Monday 01 June 2009 19:25:17 Robert Ferrell escrigué:
> >> I'm trying to write a table of numpy records, but failing.
> >>
> >> I have a numpy dtype
> >>
> >> dtp = np.dtype([('x',np.float64), ('y', np.float64)])
> >>
> >> I was hoping I could make columns with this:
> >>
> >> class DoesNotWork(tables.IsDescription):
> >> XY = tables.Col.from_atom(tables.Atom.from_dtype(dtype=dtp))
> >>
> >> but PyTables doesn't like that. The complaint is "compound data
> >> types
> >> are not supported: dtype([('x', '<f8'), ('y', '<f8')])".
> >>
> >> I can make a compound data type manually just fine:
> >>
> >> class ThisWorks(tables.IsDescription):
> >> x = tables.Float64Col()
> >> y = tables.Float64Col()
> >>
> >> Is it possible to make a table description directly from my dtype?
> >> If
> >> not, is there some other way to create the description directly from
> >> the dtype?
> >
> > Mmh, the `createTable()` constructor does accept a record array as
> > input.
> > Would that be enough for you?
>
> I'm not sure. What I actually have is a list of (label, array)
> tuples, where label is a string and array has dtype like above. So
> what I really want is something like:
>
> class MyTable(tables.IsDescription):
> label = tables.StringCol(16)
> XY = tables.Col.from_atom(tables.Atom.from_dtype(dtype=dtp))
>
> That would define a row, and then I'd just loop over my list and
> insert the rows.
>
> for l in myList:
> row['label'] = l[0]
> row['XY'] = l[1]
> row.append()
>
> Am I close?
Ah, OK. You can pass a dictionary instead. See this example:
import numpy as np
import tables as tb
descr = {'label':tb.StringCol(16)} # Initial table description
dtp = np.dtype('f8,f4') # NumPy type to append to above description
# Create a dictionary with the fields in PyTables format
dtp_ = dict([(f, tb.Col.from_dtype(t[0])) for f, t in dtp.fields.items()])
descr.update(dtp_) # Update the description dictionary
f = tb.openFile("test.h5", 'w')
t = f.createTable(f.root, 'table', descr) # Create the table
print `t`
f.close()
and the output:
/table (Table(0,)) ''
description := {
"f0": Float64Col(shape=(), dflt=0.0, pos=0),
"f1": Float32Col(shape=(), dflt=0.0, pos=1),
"label": StringCol(itemsize=16, shape=(), dflt='', pos=2)}
byteorder := 'little'
chunkshape := (292,)
Is that what you are after?
--
Francesc Alted
------------------------------------------------------------------------------
OpenSolaris 2009.06 is a cutting edge operating system for enterprises
looking to deploy the next generation of Solaris that includes the latest
innovations from Sun and the OpenSource community. Download a copy and
enjoy capabilities such as Networking, Storage and Virtualization.
Go to: http://p.sf.net/sfu/opensolaris-get
_______________________________________________
Pytables-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/pytables-users
