def _check(self, data):
dataset = self._dataset / (data + '.pd.xz')
sha256_file = self._dataset / 'sha256.yml'
# fetch data from source if not in local
if not dataset.exists():
raise RuntimeError(
"data {0} not exist, please run <preset.sync('{0}')> to download from the repository"
.format(data),
'See also: https://xenonpy.readthedocs.io/en/latest/tutorials/1-dataset.html'
)
# check sha256 value
sha256_file.touch() # make sure sha256_file file exist.
sha256 = self._yaml.load(sha256_file)
if sha256 is None:
sha256 = {}
if data not in sha256:
sha256_ = get_sha256(str(dataset))
sha256[data] = sha256_
self._yaml.dump(sha256, sha256_file)
else:
sha256_ = sha256[data]
if sha256_ != config(data):
warn(
"local version {0} is different from the latest version {1}."
"you can use <Preset.sync('{0}', to='{1}')> to fix it.".format(
data, config('db_version')), RuntimeWarning)
def __init__(self):
self._dataset = Path(__cfg_root__) / 'dataset'
self._ext_data = config('ext_data')
super().__init__(str(self._dataset),
config('userdata'),
*self._ext_data,
backend='pandas',
prefix=('dataset', ))
yaml = YAML(typ='safe')
yaml.indent(mapping=2, sequence=4, offset=2)
self._yaml = yaml
def test_config_1():
tmp = config('name')
assert tmp == 'xenonpy'
with pytest.raises(RuntimeError):
config('no_exist')
tmp = config(new_key='test')
assert tmp is None
tmp = config('new_key')
assert tmp == 'test'
tmp = config('github_username', other_key='other')
assert tmp == 'yoshida-lab'
tmp = config('other_key')
assert tmp == 'other'
def build(self, *keys, save_to=None, **kwargs):
# build materials project dataset
def mp_builder(api_key, mp_ids):
# print('Will fetch %s inorganic compounds from Materials Project' % len(mp_ids))
# split requests into fixed number groups
# eg: grouper('ABCDEFG', 3, 'x') --> ABC DEF Gxx
def grouper(iterable, n, fillvalue=None):
"""Collect data into fixed-length chunks or blocks"""
args = [iter(iterable)] * max(n, 1)
return zip_longest(fillvalue=fillvalue, *args)
# the following props will be fetched
mp_props = [
'band_gap', 'density', 'volume', 'material_id',
'pretty_formula', 'elements', 'efermi', 'e_above_hull',
'formation_energy_per_atom', 'final_energy_per_atom',
'unit_cell_formula', 'structure'
]
entries = []
mpid_groups = [g for g in grouper(mp_ids, len(mp_ids) // 10)]
with MPRester(api_key) as mpr:
for group in tqdm(mpid_groups):
mpid_list = [id for id in filter(None, group)]
chunk = mpr.query({"material_id": {
"$in": mpid_list
}}, mp_props)
entries.extend(chunk)
df = pd.DataFrame(entries,
index=[e['material_id'] for e in entries])
df = df.drop('material_id', axis=1)
df = df.rename(columns={'unit_cell_formula': 'composition'})
df = df.reindex(columns=sorted(df.columns))
return df
for key in keys:
if key is 'mp_samples':
if 'api_key' not in kwargs:
raise RuntimeError(
'api key of materials projects database is needed')
if 'mp_ids' in kwargs:
ids = kwargs['mp_ids']
if isinstance(ids, (list, tuple)):
mp_ids = ids
elif isinstance(ids, str):
mp_ids = [
s.decode('utf-8') for s in np.loadtxt(ids, 'S20')
]
else:
raise ValueError(
'parameter `mp_ids` can only be a str to specific the ids file path'
'or a list-like object contain the ids')
else:
ids = Path(__file__).absolute().parent / 'mp_ids.txt'
mp_ids = [
s.decode('utf-8') for s in np.loadtxt(str(ids), 'S20')
]
data = mp_builder(kwargs['api_key'], mp_ids)
if not save_to:
save_to = Path(config('userdata')) / 'mp_samples.pd.xz'
save_to = save_to.expanduser().absolute()
data.to_pickle(save_to)
self._make_index(prefix=['dataset'])
return
raise ValueError('no available key(s) in %s, these can only be %s' %
(keys, self.__builder__))
