NumPy (NPY)#
This reader supports reading and writing of numpy arrays in the .npy
format,
which is a binary file format for storing numpy arrays. This reader supports reading
large npy
files lazily and in a distributed fashion.
Note
This format doesn’t support storing metadata like scale, units, etc.
API functions#
- rsciio.numpy.file_reader(filename, lazy=False, chunks='auto', navigation_axes=None, **kwargs)#
Read data from npy files.
- Parameters:
- filename
str
,pathlib.Path
Filename of the file to read or corresponding pathlib.Path.
- lazybool, default=False
Whether to open the file lazily or not. The file will stay open until closed in
compute()
or closed manually.get_file_handle()
can be used to access the file handler and close it manually.- chunks
tuple
ofint
orNone
, default=None Define the chunking used for saving the dataset. If
None
, calculates chunks for the signal, with preferably at least one chunk per signal space.- navigation_axes
list
, optional List of axes that should be treated as navigation axes. If not provided, all axes will be treated as signal axes.
- **kwargs
dict
, optional Pass keyword arguments to the
numpy.load()
, when lazy is False, otherwise torsciio.utils.distributed.memmap_distributed()
.
- filename
- Returns:
list
ofdict
List of dictionaries containing the following fields:
‘data’ – multidimensional
numpy.ndarray
ordask.array.Array
‘axes’ – list of dictionaries describing the axes containing the fields ‘name’, ‘units’, ‘index_in_array’, and either ‘size’, ‘offset’, and ‘scale’ or a numpy array ‘axis’ containing the full axes vector
‘metadata’ – dictionary containing the parsed metadata
‘original_metadata’ – dictionary containing the full metadata tree from the input file
When the file contains several datasets, each dataset will be loaded as separate dictionary.
Notes
This plugin does not support metadata like scale, units, etc.
Examples
To load a numpy file with lazy loading and specified chunks:
>>> from rsciio.numpy import file_reader >>> d = file_reader('data.npy', lazy=True, chunks=("auto", "auto", 250, 250), navigation_axes=[0, 1]) >>> d[0]['data'] dask.array<array, shape=(1000, 1000, 500, 500), dtype=float64, chunksize=(100, 100, 250, 250)>
- rsciio.numpy.file_writer(filename, signal, **kwargs)#
Write data to npy files.
- Parameters:
- filename
str
,pathlib.Path
Filename of the file to write to or corresponding pathlib.Path.
- signal
dict
Dictionary containing the signal object. Should contain the following fields:
‘data’ – multidimensional numpy array
‘axes’ – list of dictionaries describing the axes containing the fields ‘name’, ‘units’, ‘index_in_array’, and either ‘size’, ‘offset’, and ‘scale’ or a numpy array ‘axis’ containing the full axes vector
‘metadata’ – dictionary containing the metadata tree
- **kwargs
dict
, optional Additional keyword arguments passed to
numpy.save()
.
- filename
Notes
This plugin does not support metadata like scale, units, etc.