mda.Universe(topology, array_like) should load into a MemoryReader

[kain88-de]

Expected behaviour

Using a u = mda.Universe(PSF, np.random.uniform(size=(...))) should just work like with any other reader.

Actual behaviour

I get a error that the fileformat can't be determined

[richardjgowers]

If you do format='memory' this will work. This needs a hack into get_reader_for which makes it return the MemoryReader if the "filename" is actually a np array

[orbeckst]

@richardjgowers : Using format="memory" does not work (on develop pre-363) because we haven't special-cased it in util.guess_format(filename) – it is assumed to be an array for the CHAIN reader:

u0 = mda.Universe(PSF, np.random.uniform(size=(u0.atoms.n_atoms, 100, 3)), format="memory")

fails with

...

/Volumes/Data/oliver/Biop/Library/python/mdanalysis/package/MDAnalysis/lib/util.pyc in guess_format(filename)
    855         # and list of filenames is handled by ChainReader
    856         format = (format_from_filename_extension(filename)
--> 857                   if not iterable(filename) else 'CHAIN')
    858
    859     return format.upper()

...

TypeError: Cannot determine file format for file '0.749369681291'.
           You can set the format explicitly with 'Universe(..., format=FORMAT)'.

Explicit MemoryReader works:

>>> import MDAnalysis as mda; from MDAnalysisTests.datafiles import PSF
>>> from MDAnalysis.coordinates.memory import MemoryReader

>>> u0 = mda.Universe(PSF, np.random.uniform(size=(u0.atoms.n_atoms, 100, 3)), format=MemoryReader)
>>> print(u0.trajectory)
<MemoryReader with 100 frames of 3341 atoms>

[orbeckst]

@richardjgowers – does format="memory" work in #363 because you do the fancy autoregistering of readers?

[richardjgowers]

Nah this won't be any different in 363 afaik. I thought format= would override everything, but it seems that CHAIN has higher precedence.

[orbeckst]

Ok, then we should fix this to be consistent.

Oliver Beckstein
email: orbeckst@gmail.com

Am Nov 1, 2016 um 4:13 schrieb Richard Gowers notifications@github.com:

Nah this won't be any different in 363 afaik. I thought format= would override everything, but it seems that CHAIN has higher precedence.

—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub, or mute the thread.