extract time step from hdf5 directly

[PrometheusPi]

This pull request solves issue #144 by extracting the time step / iteration from hdf5 directly and not from the file name.

However, extracting the time step from hdf5 files is slower.

[ax3l]

uh, append all found iterations, not only the first one :-)

[ax3l]

cool, let's see if we can even open a groupBased file now: openPMD/openPMD-converter#4

[RemiLehe]

@PrometheusPi : Thanks, that was fast!
The automated tests indicate that there is a minor linting issue. Could you try running

pep8 --ignore=E201,E202,E122,E127,E128,E131 opmd_viewer

and fix the issues?

Otherwise, the PR is fine with me; thanks again!

[PrometheusPi]

@RemiLehe Thanks for pointing out the pep8 issues - I think I fixed them.
Now it fails here with a return value of 256.

[RemiLehe]

The automated tests return the following error:

File "/home/travis/miniconda2/envs/testing/lib/python3.6/site-packages/openPMD_viewer-0.5.2-py3.6-linux-x86_64.egg/opmd_viewer/openpmd_timeseries/utilities.py", line 50, in list_h5_files

    iters_and_names.append((int(str_iteration), full_name))

TypeError: int() argument must be a string, a bytes-like object or a number, not 'Group'

My guess is that .items() should be replaced by .keys(). Could you check?

Note: you can run the automated tests on your local computer by doing:

python setup.py install
python setup.py test

[RemiLehe]

Also, I'm not sure if [0] is needed. On the whole, I guess the line

iterations = list(f['/data'].items())[0]

should be replaced by

iterations = list(f['/data'].keys())

But again, I have not checked.

[PrometheusPi]

@RemiLehe Thanks!
Using .keys() is even simpler and faster - thanks for the input.

[RemiLehe]

Great, I'll merge. Thanks for the PR.

Of course, if you need this feature in production, you'll need to install the dev version, at least until our next release.

[PrometheusPi]

@ax3l I tested the groupBased file and it works like a charm 👍

[ax3l]

awesome, thanks! 👯‍♂️