"""
Utilities for ndcube.
"""
import numbers
import numpy as np
from astropy.units import Quantity
__all__ = ['wcs_axis_to_data_axis', 'data_axis_to_wcs_axis', 'select_order',
'convert_extra_coords_dict_to_input_format', 'get_axis_number_from_axis_name']
[docs]def data_axis_to_wcs_axis(data_axis, missing_axes):
"""
Converts a data axis number to the corresponding wcs axis number.
"""
if data_axis is None:
result = None
else:
if data_axis < 0:
data_axis = np.invert(missing_axes).sum() + data_axis
if data_axis > np.invert(missing_axes).sum() - 1 or data_axis < 0:
raise IndexError("Data axis out of range. Number data axes = {}".format(
np.invert(missing_axes).sum()))
result = len(missing_axes) - np.where(np.cumsum(
[b is False for b in missing_axes][::-1]) == data_axis + 1)[0][0] - 1
return result
[docs]def wcs_axis_to_data_axis(wcs_axis, missing_axes):
"""
Converts a wcs axis number to the corresponding data axis number.
"""
if wcs_axis is None:
result = None
else:
if wcs_axis < 0:
wcs_axis = len(missing_axes) + wcs_axis
if wcs_axis > len(missing_axes) - 1 or wcs_axis < 0:
raise IndexError("WCS axis out of range. Number WCS axes = {}".format(
len(missing_axes)))
if missing_axes[wcs_axis]:
result = None
else:
data_ordered_wcs_axis = len(missing_axes) - wcs_axis - 1
result = data_ordered_wcs_axis - sum(missing_axes[::-1][:data_ordered_wcs_axis])
return result
[docs]def select_order(axtypes):
"""
Returns indices of the correct data order axis priority given a list of WCS
CTYPEs.
For example, given ['HPLN-TAN', 'TIME', 'WAVE'] it will return
[1, 2, 0] because index 1 (time) has the lowest priority, followed by
wavelength and finally solar-x.
Parameters
----------
axtypes: str list
The list of CTYPEs to be modified.
"""
order = sorted([(0, t) if t in ['TIME', 'UTC'] else
(1, t) if t == 'WAVE' else
(2, t) if t == 'HPLT-TAN' else
(axtypes.index(t) + 3, t) for t in axtypes])
result = [axtypes.index(s) for (_, s) in order]
return result
def _format_input_extra_coords_to_extra_coords_wcs_axis(extra_coords, missing_axes,
data_shape):
extra_coords_wcs_axis = {}
coord_format_error = ("Coord must have three properties supplied, "
"name (str), axis (int), values (Quantity or array-like)."
" Input coord: {0}")
coord_0_format_error = ("1st element of extra coordinate tuple must be a "
"string giving the coordinate's name.")
coord_1_format_error = ("2nd element of extra coordinate tuple must be None "
"or an int giving the data axis "
"to which the coordinate corresponds.")
coord_len_error = ("extra coord ({0}) must have same length as data axis "
"to which it is assigned: coord length, {1} != data axis length, {2}")
for coord in extra_coords:
# Check extra coord has the right number and types of info.
if len(coord) != 3:
raise ValueError(coord_format_error.format(coord))
if not isinstance(coord[0], str):
raise ValueError(coord_0_format_error.format(coord))
if coord[1] is not None and not isinstance(coord[1], numbers.Integral):
raise ValueError(coord_1_format_error)
# Unless extra coord corresponds to a missing axis, check length
# of coord is same is data axis to which is corresponds.
if coord[1] is not None:
if not missing_axes[::-1][coord[1]]:
if len(coord[2]) != data_shape[coord[1]]:
raise ValueError(coord_len_error.format(coord[0], len(coord[2]),
data_shape[coord[1]]))
# Determine wcs axis corresponding to data axis of coord
extra_coords_wcs_axis[coord[0]] = {
"wcs axis": data_axis_to_wcs_axis(coord[1], missing_axes),
"value": coord[2]}
return extra_coords_wcs_axis
[docs]def get_axis_number_from_axis_name(axis_name, world_axis_physical_types):
"""
Returns axis number (numpy ordering) given a substring unique to a world
axis type string.
Parameters
----------
axis_name: `str`
Name or substring of name of axis as defined by NDCube.world_axis_physical_types
world_axis_physical_types: iterable of `str`
Output from NDCube.world_axis_physical_types for relevant cube,
i.e. iterable of string axis names.
Returns
-------
axis_index[0]: `int`
Axis number (numpy ordering) corresponding to axis name
"""
axis_index = [axis_name in world_axis_type for world_axis_type in world_axis_physical_types]
axis_index = np.arange(len(world_axis_physical_types))[axis_index]
if len(axis_index) != 1:
raise ValueError("User defined axis with a string that is not unique to "
"a physical axis type. {} not in any of {}".format(
axis_name, world_axis_physical_types))
return axis_index[0]
def _pixel_centers_or_edges(axis_length, edges):
"""
Returns a range of pixel_values or pixel_edges.
Parameters
----------
axis_length: `int`
The length of the axis
edges: `bool`
Boolean to signify whether pixel_edge or pixel_value requested
False stands for pixel_value, while True stands for pixel_edge
Returns
-------
`np.ndarray`
The axis_values for the given input
"""
if edges is False:
axis_values = np.arange(axis_length)
else:
axis_values = np.arange(-0.5, axis_length + 0.5)
return axis_values
def _get_dimension_for_pixel(axis_length, edges):
"""
Returns the dimensions for the given edges.
Parameters
----------
axis_length : `int`
The length of the axis
edges : `bool`
Boolean to signify whether pixel_edge or pixel_value requested
False stands for pixel_value, while True stands for pixel_edge
"""
return axis_length + 1 if edges else axis_length
def _get_extra_coord_edges(value, axis=-1):
"""Gets the pixel_edges from the pixel_values
Parameters
----------
value : `astropy.units.Quantity` or array-like
The Quantity object containing the values for a given `extra_coords`
axis : `int`
The axis about which pixel_edges needs to be calculated
Default value is -1, which is the last axis for a ndarray
"""
# Checks for corner cases
if not isinstance(value, np.ndarray):
value = np.array(value)
# Get the shape of the Quantity object
shape = value.shape
if len(shape) == 1:
shape = len(value)
if isinstance(value, Quantity):
edges = np.zeros(shape + 1) * value.unit
else:
edges = np.zeros(shape + 1)
# Calculate the pixel_edges from the given pixel_values
edges[1:-1] = value[:-1] + (value[1:] - value[:-1]) / 2
edges[0] = value[0] - (value[1] - value[0]) / 2
edges[-1] = value[-1] + (value[-1] - value[-2]) / 2
else:
# Edit the shape of the new ndarray to increase the length
# by one for a given axis
shape = list(shape)
shape[axis] += 1
shape = tuple(shape)
if isinstance(value, Quantity):
edges = np.zeros(shape) * value.unit
else:
edges = np.zeros(shape)
# Shift the axis which is point of interest to last axis
value = np.moveaxis(value, axis, -1)
edges = np.moveaxis(edges, axis, -1)
# Calculate the pixel_edges from the given pixel_values
edges[..., 1:-1] = value[..., :-1] + (value[..., 1:] - value[..., :-1]) / 2
edges[..., 0] = value[..., 0] - (value[..., 1] - value[..., 0]) / 2
edges[..., -1] = value[..., -1] + (value[..., -1] - value[..., -2]) / 2
# Revert the shape of the edges array
edges = np.moveaxis(edges, -1, axis)
return edges