[ase-users] Open trajectories from tarfile

[Peterson, Andrew]

I have a large number of trajectories contained in a tarfile, e.g.,

trajectories.tar.gz:
    subdir1/
        1.traj
        2.traj
        ...

I was expecting that I could open a trajectory without writing the
extracted file out to disk with something like this:

import tarfile
from ase import io

tf = tarfile.open('trajectories.tar.gz')
f = tf.extractfile('subdir1/2.traj')
traj = io.Trajectory(f)

Note that in the above, f is a file-like object and f.closed is False. But
if I try that, I get the below error. It appears to be traced back to
something in numpy. I know that this is a rather non-typical use-case --
but does anyone know of a workaround or a solution to this apparent bug?

  File "<stdin>", line 1, in <module>
  File "repositories/ase/ase/io/trajectory.py", line 46, in Trajectory
    return TrajectoryReader(filename)
  File "repositories/ase/ase/io/trajectory.py", line 222, in __init__
    self._open(filename)
  File "repositories/ase/ase/io/trajectory.py", line 227, in _open
    self.backend = ulmopen(filename, 'r')
  File "repositories/ase/ase/io/ulm.py", line 73, in ulmopen
    return Reader(filename, index or 0)
  File "repositories/ase/ase/io/ulm.py", line 334, in __init__
    self._offsets) = read_header(fd)
  File "repositories/ase/ase/io/ulm.py", line 312, in read_header
    version, nitems, pos0 = readints(fd, 3)
  File "repositories/ase/ase/io/ulm.py", line 101, in readints
    a = np.fromfile(fd, np.int64, n)
IOError: first argument must be an open file


Andy


-- 
Andrew Peterson
Assistant Professor
Brown University School of Engineering
Barus & Holley 247
184 Hope Street
Providence, RI 02912
(401) 863-2153
http://brown.edu/go/catalyst
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.fysik.dtu.dk/pipermail/ase-users/attachments/20161216/d0b44f4b/attachment-0001.html>