Function parse_assignment_indices

dask/array/slicing.py:1236–1425

Reformat the indices for assignment. The aim of this is to convert the indices to a standardised form so that it is easier to ascertain which chunks are touched by the indices. This function is intended to be called by `setitem_array`. A slice object that is decreasing (i.e. w

(indices, shape)

def parse_assignment_indices(indices, shape):
    """Reformat the indices for assignment.

    The aim of this is to convert the indices to a standardised form
    so that it is easier to ascertain which chunks are touched by the
    indices.

    This function is intended to be called by `setitem_array`.

    A slice object that is decreasing (i.e. with a negative step), is
    recast as an increasing slice (i.e. with a positive step. For
    example ``slice(7,3,-1)`` would be cast as ``slice(4,8,1)``. This
    is to facilitate finding which blocks are touched by the
    index. The dimensions for which this has occurred are returned by
    the function.

    Parameters
    ----------
    indices : numpy-style indices
        Indices to array defining the elements to be assigned.
    shape : sequence of `int`
        The shape of the array.

    Returns
    -------
    parsed_indices : `list`
        The reformatted indices that are equivalent to the input
        indices.
    implied_shape : `list`
        The shape implied by the parsed indices. For instance, indices
        of ``(slice(0,2), 5, [4,1,-1])`` will have implied shape
        ``[2,3]``.
    reverse : `list`
        The positions of the dimensions whose indices in the
        parsed_indices output are reversed slices.
    implied_shape_positions: `list`
        The positions of the dimensions whose indices contribute to
        the implied_shape. For instance, indices of ``(slice(0,2), 5,
        [4,1,-1])`` will have implied_shape ``[2,3]`` and
        implied_shape_positions ``[0,2]``.

    Examples
    --------
    >>> parse_assignment_indices((slice(1, -1),), (8,))
    ([slice(1, 7, 1)], [6], [], [0])

    >>> parse_assignment_indices(([1, 2, 6, 5],), (8,))
    ([array([1, 2, 6, 5])], [4], [], [0])

    >>> parse_assignment_indices((3, slice(-1, 2, -1)), (7, 8))
    ([3, slice(3, 8, 1)], [5], [1], [1])

    >>> parse_assignment_indices((slice(-1, 2, -1), 3, [1, 2]), (7, 8, 9))
    ([slice(3, 7, 1), 3, array([1, 2])], [4, 2], [0], [0, 2])

    >>> parse_assignment_indices((slice(0, 5), slice(3, None, 2)), (5, 4))
    ([slice(0, 5, 1), slice(3, 4, 2)], [5, 1], [], [0, 1])