def test_splitext2(convert_path):
filepath0 = '/foo/bar/baz.txt'
expected = ('/foo/bar', 'baz', '.txt')
filepath = convert_path(filepath0)
actual = splitext2(filepath)
assert actual == expected
filepath0 = 'baz.txt'
expected = ('', 'baz', '.txt')
filepath = convert_path(filepath0)
actual = splitext2(filepath)
assert actual == expected
def _read_tabular_h5(filepath):
_, fn, ext = splitext2(filepath)
_check_ext(ext, '.h5')
with h5py.File(filepath, 'r') as hf:
dataset = hf[fn]
data = dataset[:]
return data
def _write_tabular_h5(obj, filepath):
_, fn, ext = splitext2(filepath)
_check_ext(ext, '.h5')
if isinstance(obj, np.ndarray):
with h5py.File(filepath, 'w') as hf:
hf.create_dataset(fn, data=obj)
elif isinstance(obj, pd.core.frame.NDFrame):
obj.to_hdf(filepath, key=fn)
else:
raise NotImplementedError
def _write_tabular_pickle(obj, filepath):
_, fn, ext = splitext2(filepath)
_check_ext(ext, '.pkl')
if isinstance(obj, np.ndarray):
with open(filepath, 'wb') as f:
pickle.dump(obj, f)
elif isinstance(obj, pd.core.frame.NDFrame):
obj.to_pickle(filepath)
else:
raise NotImplementedError
def read_tabular(filepath: Pathy):
"""Read tabular object in HDF5 or pickle format
Args:
filepath: path to read to; must end in '.h5' or '.pkl'
"""
_, fn, ext = splitext2(filepath)
if ext == '.h5':
return _read_tabular_h5(filepath)
elif ext == '.pkl':
return _read_tabular_pickle(filepath)
else:
raise NotImplementedError
def write_tabular(obj: Union[np.ndarray, pd.DataFrame], filepath: Pathy):
"""Write tabular object in HDF5 or pickle format
Args:
obj: tabular object to write
filepath: path to write to; must end in '.h5' or '.pkl'
"""
_, fn, ext = splitext2(filepath)
if ext == '.h5':
_write_tabular_h5(obj, filepath)
elif ext == '.pkl':
_write_tabular_pickle(obj, filepath)
else:
raise NotImplementedError
def write_tabular(obj, filepath):
"""Write tabular object in HDF5 or pickle format
Args:
obj (array or DataFrame): tabular object to write
filepath (path-like): path to write to; must end in '.h5' or '.pkl'
"""
_, fn, ext = splitext2(filepath)
if ext == '.h5':
_write_tabular_h5(obj, filepath)
elif ext == '.pkl':
_write_tabular_pickle(obj, filepath)
else:
raise NotImplementedError
def _read_tabular_pickle(filepath):
_, fn, ext = splitext2(filepath)
_check_ext(ext, '.pkl')
with open(filepath, 'rb') as f:
return pickle.load(f)