arrayproxy

Array proxy base class

The proxy API is - at minimum:

  • The object has a read-only attribute shape

  • read only is_proxy attribute / property set to True

  • the object returns the data array from np.asarray(prox)

  • returns array slice from prox[<slice_spec>] where <slice_spec> is any ndarray slice specification that does not use numpy ‘advanced indexing’.

  • modifying no object outside obj will affect the result of np.asarray(obj). Specifically:

    • Changes in position (obj.tell()) of passed file-like objects will not affect the output of from np.asarray(proxy).

    • if you pass a header into the __init__, then modifying the original header will not affect the result of the array return.

See nibabel.tests.test_proxy_api for proxy API conformance checks.

ArrayLike(*args, **kwargs)

Protocol for numpy ndarray-like objects

ArrayProxy(file_like, spec, *[, mmap, ...])

Class to act as proxy for the array that can be read from a file

get_obj_dtype(obj)

Get the effective dtype of an array-like object

is_proxy(obj)

Return True if obj is an array proxy

reshape_dataobj(obj, shape)

Use obj reshape method if possible, else numpy reshape function

ArrayLike

class nibabel.arrayproxy.ArrayLike(*args, **kwargs)

Bases: Protocol

Protocol for numpy ndarray-like objects

This is more stringent than numpy.typing.ArrayLike, but guarantees access to shape, ndim and slicing.

__init__(*args, **kwargs)
property ndim: int
shape: tuple[int, ...]

ArrayProxy

class nibabel.arrayproxy.ArrayProxy(file_like, spec, *, mmap=True, order=None, keep_file_open=None)

Bases: ArrayLike

Class to act as proxy for the array that can be read from a file

The array proxy allows us to freeze the passed fileobj and header such that it returns the expected data array.

This implementation assumes a contiguous array in the file object, with one of the numpy dtypes, starting at a given file position offset with single slope and intercept scaling to produce output values.

The class __init__ requires a spec which defines how the data will be read and rescaled. The spec may be a tuple of length 2 - 5, containing the shape, storage dtype, offset, slope and intercept, or a header object with methods:

  • get_data_shape

  • get_data_dtype

  • get_data_offset

  • get_slope_inter

A header should also have a ‘copy’ method. This requirement will go away when the deprecated ‘header’ property goes away.

This implementation allows us to deal with Analyze and its variants, including Nifti1, and with the MGH format.

Other image types might need more specific classes to implement the API. See nibabel.minc1, nibabel.ecat and nibabel.parrec for examples.

Initialize array proxy instance

Parameters:
file_likeobject

File-like object or filename. If file-like object, should implement at least read and seek.

specobject or tuple

Tuple must have length 2-5, with the following values:

  1. shape: tuple - tuple of ints describing shape of data;

  2. storage_dtype: dtype specifier - dtype of array inside proxied file, or input to numpy.dtype to specify array dtype;

  3. offset: int - offset, in bytes, of data array from start of file (default: 0);

  4. slope: float - scaling factor for resulting data (default: 1.0);

  5. inter: float - intercept for rescaled data (default: 0.0).

OR

Header object implementing get_data_shape, get_data_dtype, get_data_offset, get_slope_inter

mmap{True, False, ‘c’, ‘r’}, optional, keyword only

mmap controls the use of numpy memory mapping for reading data. If False, do not try numpy memmap for data array. If one of {‘c’, ‘r’}, try numpy memmap with mode=mmap. A mmap value of True gives the same behavior as mmap='c'. If file_like cannot be memory-mapped, ignore mmap value and read array from file.

order{None, ‘F’, ‘C’}, optional, keyword only

order controls the order of the data array layout. Fortran-style, column-major order may be indicated with ‘F’, and C-style, row-major order may be indicated with ‘C’. None gives the default order, that comes from the _default_order class variable.

keep_file_open{ None, True, False }, optional, keyword only

keep_file_open controls whether a new file handle is created every time the image is accessed, or a single file handle is created and used for the lifetime of this ArrayProxy. If True, a single file handle is created and used. If False, a new file handle is created every time the image is accessed. If file_like is an open file handle, this setting has no effect. The default value (None) will result in the value of KEEP_FILE_OPEN_DEFAULT being used.

__init__(file_like, spec, *, mmap=True, order=None, keep_file_open=None)

Initialize array proxy instance

Parameters:
file_likeobject

File-like object or filename. If file-like object, should implement at least read and seek.

specobject or tuple

Tuple must have length 2-5, with the following values:

  1. shape: tuple - tuple of ints describing shape of data;

  2. storage_dtype: dtype specifier - dtype of array inside proxied file, or input to numpy.dtype to specify array dtype;

  3. offset: int - offset, in bytes, of data array from start of file (default: 0);

  4. slope: float - scaling factor for resulting data (default: 1.0);

  5. inter: float - intercept for rescaled data (default: 0.0).

OR

Header object implementing get_data_shape, get_data_dtype, get_data_offset, get_slope_inter

mmap{True, False, ‘c’, ‘r’}, optional, keyword only

mmap controls the use of numpy memory mapping for reading data. If False, do not try numpy memmap for data array. If one of {‘c’, ‘r’}, try numpy memmap with mode=mmap. A mmap value of True gives the same behavior as mmap='c'. If file_like cannot be memory-mapped, ignore mmap value and read array from file.

order{None, ‘F’, ‘C’}, optional, keyword only

order controls the order of the data array layout. Fortran-style, column-major order may be indicated with ‘F’, and C-style, row-major order may be indicated with ‘C’. None gives the default order, that comes from the _default_order class variable.

keep_file_open{ None, True, False }, optional, keyword only

keep_file_open controls whether a new file handle is created every time the image is accessed, or a single file handle is created and used for the lifetime of this ArrayProxy. If True, a single file handle is created and used. If False, a new file handle is created every time the image is accessed. If file_like is an open file handle, this setting has no effect. The default value (None) will result in the value of KEEP_FILE_OPEN_DEFAULT being used.

copy() → Self

Create a new ArrayProxy for the same file and parameters

If the proxied file is an open file handle, the new ArrayProxy will share a lock with the old one.

property dtype
get_unscaled()

Read data from file

This is an optional part of the proxy API

property inter
property is_proxy
property ndim
property offset
reshape(shape)

Return an ArrayProxy with a new shape, without modifying data

property shape
property slope

get_obj_dtype

nibabel.arrayproxy.get_obj_dtype(obj)

Get the effective dtype of an array-like object

is_proxy

nibabel.arrayproxy.is_proxy(obj)

Return True if obj is an array proxy

reshape_dataobj

nibabel.arrayproxy.reshape_dataobj(obj, shape)

Use obj reshape method if possible, else numpy reshape function