from __future__ import annotations
import itertools
import warnings
from abc import ABC, abstractmethod
from collections import defaultdict, namedtuple
from contextlib import contextmanager
from functools import cached_property
from typing import List, Optional, Tuple, Union
import magicgui as mgui
import numpy as np
from ...utils._dask_utils import configure_dask
from ...utils._magicgui import (
add_layer_to_viewer,
add_layers_to_viewer,
get_layers,
)
from ...utils.events import EmitterGroup, Event
from ...utils.events.event import WarningEmitter
from ...utils.geometry import (
find_front_back_face,
intersect_line_with_axis_aligned_bounding_box_3d,
)
from ...utils.key_bindings import KeymapProvider
from ...utils.mouse_bindings import MousemapProvider
from ...utils.naming import magic_name
from ...utils.status_messages import generate_layer_status
from ...utils.transforms import Affine, CompositeAffine, TransformChain
from ...utils.translations import trans
from .._source import current_source
from ..utils.interactivity_utils import drag_data_to_projected_distance
from ..utils.layer_utils import (
coerce_affine,
compute_multiscale_level_and_corners,
convert_to_uint8,
dims_displayed_world_to_layer,
get_extent_world,
)
from ..utils.plane import ClippingPlane, ClippingPlaneList
from ._base_constants import Blending
Extent = namedtuple('Extent', 'data world step')
def no_op(layer: Layer, event: Event) -> None:
"""
A convenient no-op event for the layer mouse binding.
This makes it easier to handle many cases by inserting this as
as place holder
Parameters
----------
layer : Layer
Current layer on which this will be bound as a callback
event : Event
event that triggered this mouse callback.
Returns
-------
None
"""
return None
[docs]@mgui.register_type(choices=get_layers, return_callback=add_layer_to_viewer)
class Layer(KeymapProvider, MousemapProvider, ABC):
"""Base layer class.
Parameters
----------
name : str
Name of the layer.
metadata : dict
Layer metadata.
scale : tuple of float
Scale factors for the layer.
translate : tuple of float
Translation values for the layer.
rotate : float, 3-tuple of float, or n-D array.
If a float convert into a 2D rotation matrix using that value as an
angle. If 3-tuple convert into a 3D rotation matrix, using a yaw,
pitch, roll convention. Otherwise assume an nD rotation. Angles are
assumed to be in degrees. They can be converted from radians with
np.degrees if needed.
shear : 1-D array or n-D array
Either a vector of upper triangular values, or an nD shear matrix with
ones along the main diagonal.
affine : n-D array or napari.utils.transforms.Affine
(N+1, N+1) affine transformation matrix in homogeneous coordinates.
The first (N, N) entries correspond to a linear transform and
the final column is a length N translation vector and a 1 or a napari
`Affine` transform object. Applied as an extra transform on top of the
provided scale, rotate, and shear values.
opacity : float
Opacity of the layer visual, between 0.0 and 1.0.
blending : str
One of a list of preset blending modes that determines how RGB and
alpha values of the layer visual get mixed. Allowed values are
{'opaque', 'translucent', 'translucent_no_depth', and 'additive'}.
visible : bool
Whether the layer visual is currently being displayed.
multiscale : bool
Whether the data is multiscale or not. Multiscale data is
represented by a list of data objects and should go from largest to
smallest.
Attributes
----------
name : str
Unique name of the layer.
opacity : float
Opacity of the layer visual, between 0.0 and 1.0.
visible : bool
Whether the layer visual is currently being displayed.
blending : Blending
Determines how RGB and alpha values get mixed.
* ``Blending.OPAQUE``
Allows for only the top layer to be visible and corresponds to
``depth_test=True``, ``cull_face=False``, ``blend=False``.
* ``Blending.TRANSLUCENT``
Allows for multiple layers to be blended with different opacity and
corresponds to ``depth_test=True``, ``cull_face=False``,
``blend=True``, ``blend_func=('src_alpha', 'one_minus_src_alpha')``.
* ``Blending.TRANSLUCENT_NO_DEPTH``
Allows for multiple layers to be blended with different opacity, but
no depth testing is performed. Corresponds to ``depth_test=False``,
``cull_face=False``, ``blend=True``,
``blend_func=('src_alpha', 'one_minus_src_alpha')``.
* ``Blending.ADDITIVE``
Allows for multiple layers to be blended together with different
colors and opacity. Useful for creating overlays. It corresponds to
``depth_test=False``, ``cull_face=False``, ``blend=True``,
``blend_func=('src_alpha', 'one')``.
scale : tuple of float
Scale factors for the layer.
translate : tuple of float
Translation values for the layer.
rotate : float, 3-tuple of float, or n-D array.
If a float convert into a 2D rotation matrix using that value as an
angle. If 3-tuple convert into a 3D rotation matrix, using a yaw,
pitch, roll convention. Otherwise assume an nD rotation. Angles are
assumed to be in degrees. They can be converted from radians with
np.degrees if needed.
shear : 1-D array or n-D array
Either a vector of upper triangular values, or an nD shear matrix with
ones along the main diagonal.
affine : n-D array or napari.utils.transforms.Affine
(N+1, N+1) affine transformation matrix in homogeneous coordinates.
The first (N, N) entries correspond to a linear transform and
the final column is a length N translation vector and a 1 or a napari
`Affine` transform object. Applied as an extra transform on top of the
provided scale, rotate, and shear values.
multiscale : bool
Whether the data is multiscale or not. Multiscale data is
represented by a list of data objects and should go from largest to
smallest.
cache : bool
Whether slices of out-of-core datasets should be cached upon retrieval.
Currently, this only applies to dask arrays.
z_index : int
Depth of the layer visual relative to other visuals in the scenecanvas.
coordinates : tuple of float
Cursor position in data coordinates.
corner_pixels : array
Coordinates of the top-left and bottom-right canvas pixels in the data
coordinates of each layer. For multiscale data the coordinates are in
the space of the currently viewed data level, not the highest resolution
level.
position : tuple
Cursor position in world coordinates.
ndim : int
Dimensionality of the layer.
thumbnail : (N, M, 4) array
Array of thumbnail data for the layer.
status : str
Displayed in status bar bottom left.
help : str
Displayed in status bar bottom right.
interactive : bool
Determine if canvas pan/zoom interactivity is enabled.
cursor : str
String identifying which cursor displayed over canvas.
cursor_size : int | None
Size of cursor if custom. None yields default size
scale_factor : float
Conversion factor from canvas coordinates to image coordinates, which
depends on the current zoom level.
source : Source
source of the layer (such as a plugin or widget)
Notes
-----
Must define the following:
* `_extent_data`: property
* `data` property (setter & getter)
May define the following:
* `_set_view_slice()`: called to set currently viewed slice
* `_basename()`: base/default name of the layer
"""
def __init__(
self,
data,
ndim,
*,
name=None,
metadata=None,
scale=None,
translate=None,
rotate=None,
shear=None,
affine=None,
opacity=1,
blending='translucent',
visible=True,
multiscale=False,
cache=True, # this should move to future "data source" object.
experimental_clipping_planes=None,
):
super().__init__()
if name is None and data is not None:
name = magic_name(data)
self._source = current_source()
self.dask_optimized_slicing = configure_dask(data, cache)
self._metadata = dict(metadata or {})
self._opacity = opacity
self._blending = Blending(blending)
self._visible = visible
self._freeze = False
self._status = 'Ready'
self._help = ''
self._cursor = 'standard'
self._cursor_size = 1
self._interactive = True
self._value = None
self.scale_factor = 1
self.multiscale = multiscale
self._experimental_clipping_planes = ClippingPlaneList()
self._ndim = ndim
self._ndisplay = 2
self._dims_order = list(range(ndim))
# Create a transform chain consisting of four transforms:
# 1. `tile2data`: An initial transform only needed to display tiles
# of an image. It maps pixels of the tile into the coordinate space
# of the full resolution data and can usually be represented by a
# scale factor and a translation. A common use case is viewing part
# of lower resolution level of a multiscale image, another is using a
# downsampled version of an image when the full image size is larger
# than the maximum allowed texture size of your graphics card.
# 2. `data2physical`: The main transform mapping data to a world-like
# physical coordinate that may also encode acquisition parameters or
# sample spacing.
# 3. `physical2world`: An extra transform applied in world-coordinates that
# typically aligns this layer with another.
# 4. `world2grid`: An additional transform mapping world-coordinates
# into a grid for looking at layers side-by-side.
if scale is None:
scale = [1] * ndim
if translate is None:
translate = [0] * ndim
self._transforms = TransformChain(
[
Affine(np.ones(ndim), np.zeros(ndim), name='tile2data'),
CompositeAffine(
scale,
translate,
rotate=rotate,
shear=shear,
ndim=ndim,
name='data2physical',
),
coerce_affine(affine, ndim=ndim, name='physical2world'),
Affine(np.ones(ndim), np.zeros(ndim), name='world2grid'),
]
)
self._dims_point = [0] * ndim
self.corner_pixels = np.zeros((2, ndim), dtype=int)
self._editable = True
self._array_like = False
self._thumbnail_shape = (32, 32, 4)
self._thumbnail = np.zeros(self._thumbnail_shape, dtype=np.uint8)
self._update_properties = True
self._name = ''
self.experimental_clipping_planes = experimental_clipping_planes
self.events = EmitterGroup(
source=self,
refresh=Event,
set_data=Event,
blending=Event,
opacity=Event,
visible=Event,
scale=Event,
translate=Event,
rotate=Event,
shear=Event,
affine=Event,
data=Event,
name=Event,
thumbnail=Event,
status=Event,
help=Event,
interactive=Event,
cursor=Event,
cursor_size=Event,
editable=Event,
loaded=Event,
_ndisplay=Event,
select=WarningEmitter(
trans._(
"'layer.events.select' is deprecated and will be removed in napari v0.4.9, use 'viewer.layers.selection.events.changed' instead, and inspect the 'added' attribute on the event.",
deferred=True,
),
type='select',
),
deselect=WarningEmitter(
trans._(
"'layer.events.deselect' is deprecated and will be removed in napari v0.4.9, use 'viewer.layers.selection.events.changed' instead, and inspect the 'removed' attribute on the event.",
deferred=True,
),
type='deselect',
),
)
self.name = name
def __str__(self):
"""Return self.name."""
return self.name
def __repr__(self):
cls = type(self)
return f"<{cls.__name__} layer {repr(self.name)} at {hex(id(self))}>"
def _mode_setter_helper(self, mode, Modeclass):
"""
Helper to manage callbacks in multiple layers
Parameters
----------
mode : Modeclass | str
New mode for the current layer.
Modeclass : Enum
Enum for the current class representing the modes it can takes,
this is usually specific on each subclass.
Returns
-------
tuple (new Mode, mode changed)
"""
mode = Modeclass(mode)
assert mode is not None
if not self.editable:
mode = Modeclass.PAN_ZOOM
if mode == self._mode:
return mode, False
if mode.value not in Modeclass.keys():
raise ValueError(
trans._(
"Mode not recognized: {mode}", deferred=True, mode=mode
)
)
old_mode = self._mode
self._mode = mode
for callback_list, mode_dict in [
(self.mouse_drag_callbacks, self._drag_modes),
(self.mouse_move_callbacks, self._move_modes),
(
self.mouse_double_click_callbacks,
getattr(
self, '_double_click_modes', defaultdict(lambda: no_op)
),
),
]:
if mode_dict[old_mode] in callback_list:
callback_list.remove(mode_dict[old_mode])
callback_list.append(mode_dict[mode])
self.cursor = self._cursor_modes[mode]
self.interactive = mode == Modeclass.PAN_ZOOM
return mode, True
@classmethod
def _basename(cls):
return f'{cls.__name__}'
@property
def name(self):
"""str: Unique name of the layer."""
return self._name
@name.setter
def name(self, name):
if name == self.name:
return
if not name:
name = self._basename()
self._name = str(name)
self.events.name()
@property
def metadata(self) -> dict:
"""Key/value map for user-stored data."""
return self._metadata
@metadata.setter
def metadata(self, value: dict) -> None:
self._metadata.clear()
self._metadata.update(value)
@property
def source(self):
return self._source
@property
def loaded(self) -> bool:
"""Return True if this layer is fully loaded in memory.
This base class says that layers are permanently in the loaded state.
Derived classes that do asynchronous loading can override this.
"""
return True
@property
def opacity(self):
"""float: Opacity value between 0.0 and 1.0."""
return self._opacity
@opacity.setter
def opacity(self, opacity):
if not 0.0 <= opacity <= 1.0:
raise ValueError(
trans._(
'opacity must be between 0.0 and 1.0; got {opacity}',
deferred=True,
opacity=opacity,
)
)
self._opacity = opacity
self._update_thumbnail()
self.events.opacity()
@property
def blending(self):
"""Blending mode: Determines how RGB and alpha values get mixed.
Blending.OPAQUE
Allows for only the top layer to be visible and corresponds to
depth_test=True, cull_face=False, blend=False.
Blending.TRANSLUCENT
Allows for multiple layers to be blended with different opacity
and corresponds to depth_test=True, cull_face=False,
blend=True, blend_func=('src_alpha', 'one_minus_src_alpha').
Blending.ADDITIVE
Allows for multiple layers to be blended together with
different colors and opacity. Useful for creating overlays. It
corresponds to depth_test=False, cull_face=False, blend=True,
blend_func=('src_alpha', 'one').
"""
return str(self._blending)
@blending.setter
def blending(self, blending):
self._blending = Blending(blending)
self.events.blending()
@property
def visible(self):
"""bool: Whether the visual is currently being displayed."""
return self._visible
@visible.setter
def visible(self, visibility):
self._visible = visibility
self.refresh()
self.events.visible()
self.editable = self._set_editable() if self.visible else False
@property
def editable(self):
"""bool: Whether the current layer data is editable from the viewer."""
return self._editable
@editable.setter
def editable(self, editable):
if self._editable == editable:
return
self._editable = editable
self._set_editable(editable=editable)
self.events.editable()
@property
def scale(self):
"""list: Anisotropy factors to scale data into world coordinates."""
return self._transforms['data2physical'].scale
@scale.setter
def scale(self, scale):
if scale is None:
scale = [1] * self.ndim
self._transforms['data2physical'].scale = np.array(scale)
self._update_dims()
self.events.scale()
@property
def translate(self):
"""list: Factors to shift the layer by in units of world coordinates."""
return self._transforms['data2physical'].translate
@translate.setter
def translate(self, translate):
self._transforms['data2physical'].translate = np.array(translate)
self._update_dims()
self.events.translate()
@property
def rotate(self):
"""array: Rotation matrix in world coordinates."""
return self._transforms['data2physical'].rotate
@rotate.setter
def rotate(self, rotate):
self._transforms['data2physical'].rotate = rotate
self._update_dims()
self.events.rotate()
@property
def shear(self):
"""array: Shear matrix in world coordinates."""
return self._transforms['data2physical'].shear
@shear.setter
def shear(self, shear):
self._transforms['data2physical'].shear = shear
self._update_dims()
self.events.shear()
@property
def affine(self):
"""napari.utils.transforms.Affine: Extra affine transform to go from physical to world coordinates."""
return self._transforms['physical2world']
@affine.setter
def affine(self, affine):
# Assignment by transform name is not supported by TransformChain and
# EventedList, so use the integer index instead. For more details, see:
# https://github.com/napari/napari/issues/3058
self._transforms[2] = coerce_affine(
affine, ndim=self.ndim, name='physical2world'
)
self._update_dims()
self.events.affine()
@property
def translate_grid(self):
warnings.warn(
trans._(
"translate_grid will become private in v0.4.14. See Layer.translate or Layer.data_to_world() instead.",
),
DeprecationWarning,
stacklevel=2,
)
return self._translate_grid
@translate_grid.setter
def translate_grid(self, translate_grid):
warnings.warn(
trans._(
"translate_grid will become private in v0.4.14. See Layer.translate or Layer.data_to_world() instead.",
),
DeprecationWarning,
stacklevel=2,
)
self._translate_grid = translate_grid
@property
def _translate_grid(self):
"""list: Factors to shift the layer by."""
return self._transforms['world2grid'].translate
@_translate_grid.setter
def _translate_grid(self, translate_grid):
if np.all(self._translate_grid == translate_grid):
return
self._transforms['world2grid'].translate = np.array(translate_grid)
self.events.translate()
@property
def _is_moving(self):
return self._private_is_moving
@_is_moving.setter
def _is_moving(self, value):
assert value in (True, False)
if value:
assert self._moving_coordinates is not None
self._private_is_moving = value
@property
def _dims_displayed(self):
"""To be removed displayed dimensions."""
# Ultimately we aim to remove all slicing information from the layer
# itself so that layers can be sliced in different ways for multiple
# canvas. See https://github.com/napari/napari/pull/1919#issuecomment-738585093
# for additional discussion.
return self._dims_order[-self._ndisplay :]
@property
def _dims_not_displayed(self):
"""To be removed not displayed dimensions."""
# Ultimately we aim to remove all slicing information from the layer
# itself so that layers can be sliced in different ways for multiple
# canvas. See https://github.com/napari/napari/pull/1919#issuecomment-738585093
# for additional discussion.
return self._dims_order[: -self._ndisplay]
@property
def _dims_displayed_order(self):
"""To be removed order of displayed dimensions."""
# Ultimately we aim to remove all slicing information from the layer
# itself so that layers can be sliced in different ways for multiple
# canvas. See https://github.com/napari/napari/pull/1919#issuecomment-738585093
# for additional discussion.
displayed = self._dims_displayed
# equivalent to: order = np.argsort(displayed)
order = sorted(range(len(displayed)), key=lambda x: displayed[x])
return tuple(order)
def _update_dims(self, event=None):
"""Update the dims model and clear the extent cache.
This function needs to be called whenever data or transform information
changes, and should be called before events get emitted.
"""
from ...components.dims import reorder_after_dim_reduction
ndim = self._get_ndim()
old_ndim = self._ndim
if old_ndim > ndim:
keep_axes = range(old_ndim - ndim, old_ndim)
self._transforms = self._transforms.set_slice(keep_axes)
self._dims_point = self._dims_point[-ndim:]
self._dims_order = list(
reorder_after_dim_reduction(self._dims_order[-ndim:])
)
elif old_ndim < ndim:
new_axes = range(ndim - old_ndim)
self._transforms = self._transforms.expand_dims(new_axes)
self._dims_point = [0] * (ndim - old_ndim) + self._dims_point
self._dims_order = list(range(ndim - old_ndim)) + [
o + ndim - old_ndim for o in self._dims_order
]
self._ndim = ndim
if 'extent' in self.__dict__:
del self.extent
self.refresh() # This call is need for invalidate cache of extent in LayerList. If you remove it pleas ad another workaround.
@property
@abstractmethod
def data(self):
# user writes own docstring
raise NotImplementedError()
@data.setter
@abstractmethod
def data(self, data):
raise NotImplementedError()
@property
@abstractmethod
def _extent_data(self) -> np.ndarray:
"""Extent of layer in data coordinates.
Returns
-------
extent_data : array, shape (2, D)
"""
raise NotImplementedError()
@property
def _extent_world(self) -> np.ndarray:
"""Range of layer in world coordinates.
Returns
-------
extent_world : array, shape (2, D)
"""
# Get full nD bounding box
return get_extent_world(
self._extent_data, self._data_to_world, self._array_like
)
@cached_property
def extent(self) -> Extent:
"""Extent of layer in data and world coordinates."""
extent_data = self._extent_data
data_to_world = self._data_to_world
extent_world = get_extent_world(
extent_data, data_to_world, self._array_like
)
return Extent(
data=extent_data,
world=extent_world,
step=abs(data_to_world.scale),
)
@property
def _slice_indices(self):
"""(D, ) array: Slice indices in data coordinates."""
if len(self._dims_not_displayed) == 0:
# all dims are displayed dimensions
return (slice(None),) * self.ndim
if self.ndim > self._ndisplay:
inv_transform = self._data_to_world.inverse
# Subspace spanned by non displayed dimensions
non_displayed_subspace = np.zeros(self.ndim)
for d in self._dims_not_displayed:
non_displayed_subspace[d] = 1
# Map subspace through inverse transform, ignoring translation
_inv_transform = Affine(
ndim=self.ndim,
linear_matrix=inv_transform.linear_matrix,
translate=None,
)
mapped_nd_subspace = _inv_transform(non_displayed_subspace)
# Look at displayed subspace
displayed_mapped_subspace = (
mapped_nd_subspace[d] for d in self._dims_displayed
)
# Check that displayed subspace is null
if any(abs(v) > 1e-8 for v in displayed_mapped_subspace):
warnings.warn(
trans._(
'Non-orthogonal slicing is being requested, but is not fully supported. Data is displayed without applying an out-of-slice rotation or shear component.',
deferred=True,
),
category=UserWarning,
)
slice_inv_transform = inv_transform.set_slice(self._dims_not_displayed)
world_pts = [self._dims_point[ax] for ax in self._dims_not_displayed]
data_pts = slice_inv_transform(world_pts)
if getattr(self, "_round_index", True):
# A round is taken to convert these values to slicing integers
data_pts = np.round(data_pts).astype(int)
indices = [slice(None)] * self.ndim
for i, ax in enumerate(self._dims_not_displayed):
indices[ax] = data_pts[i]
return tuple(indices)
@abstractmethod
def _get_ndim(self):
raise NotImplementedError()
def _set_editable(self, editable=None):
if editable is None:
self.editable = True
def _get_base_state(self):
"""Get dictionary of attributes on base layer.
Returns
-------
state : dict
Dictionary of attributes on base layer.
"""
base_dict = {
'name': self.name,
'metadata': self.metadata,
'scale': list(self.scale),
'translate': list(self.translate),
'rotate': [list(r) for r in self.rotate],
'shear': list(self.shear),
'affine': self.affine.affine_matrix,
'opacity': self.opacity,
'blending': self.blending,
'visible': self.visible,
'experimental_clipping_planes': [
plane.dict() for plane in self.experimental_clipping_planes
],
}
return base_dict
@abstractmethod
def _get_state(self):
raise NotImplementedError()
@property
def _type_string(self):
return self.__class__.__name__.lower()
def as_layer_data_tuple(self):
state = self._get_state()
state.pop('data', None)
return self.data, state, self._type_string
@property
def thumbnail(self):
"""array: Integer array of thumbnail for the layer"""
return self._thumbnail
@thumbnail.setter
def thumbnail(self, thumbnail):
if 0 in thumbnail.shape:
thumbnail = np.zeros(self._thumbnail_shape, dtype=np.uint8)
if thumbnail.dtype != np.uint8:
with warnings.catch_warnings():
warnings.simplefilter("ignore")
thumbnail = convert_to_uint8(thumbnail)
padding_needed = np.subtract(self._thumbnail_shape, thumbnail.shape)
pad_amounts = [(p // 2, (p + 1) // 2) for p in padding_needed]
thumbnail = np.pad(thumbnail, pad_amounts, mode='constant')
# blend thumbnail with opaque black background
background = np.zeros(self._thumbnail_shape, dtype=np.uint8)
background[..., 3] = 255
f_dest = thumbnail[..., 3][..., None] / 255
f_source = 1 - f_dest
thumbnail = thumbnail * f_dest + background * f_source
self._thumbnail = thumbnail.astype(np.uint8)
self.events.thumbnail()
@property
def ndim(self):
"""int: Number of dimensions in the data."""
return self._ndim
@property
def help(self):
"""str: displayed in status bar bottom right."""
return self._help
@help.setter
def help(self, help):
if help == self.help:
return
self._help = help
self.events.help(help=help)
@property
def interactive(self):
"""bool: Determine if canvas pan/zoom interactivity is enabled."""
return self._interactive
@interactive.setter
def interactive(self, interactive):
if interactive == self._interactive:
return
self._interactive = interactive
self.events.interactive(interactive=interactive)
@property
def cursor(self):
"""str: String identifying cursor displayed over canvas."""
return self._cursor
@cursor.setter
def cursor(self, cursor):
if cursor == self.cursor:
return
self._cursor = cursor
self.events.cursor(cursor=cursor)
@property
def cursor_size(self):
"""int | None: Size of cursor if custom. None yields default size."""
return self._cursor_size
@cursor_size.setter
def cursor_size(self, cursor_size):
if cursor_size == self.cursor_size:
return
self._cursor_size = cursor_size
self.events.cursor_size(cursor_size=cursor_size)
@property
def experimental_clipping_planes(self):
return self._experimental_clipping_planes
@experimental_clipping_planes.setter
def experimental_clipping_planes(
self,
value: Union[
dict,
ClippingPlane,
List[Union[ClippingPlane, dict]],
ClippingPlaneList,
],
):
self._experimental_clipping_planes.clear()
if value is None:
return
if isinstance(value, (ClippingPlane, dict)):
value = [value]
for new_plane in value:
plane = ClippingPlane()
plane.update(new_plane)
self._experimental_clipping_planes.append(plane)
def set_view_slice(self):
with self.dask_optimized_slicing():
self._set_view_slice()
@abstractmethod
def _set_view_slice(self):
raise NotImplementedError()
def _slice_dims(self, point=None, ndisplay=2, order=None):
"""Slice data with values from a global dims model.
Note this will likely be moved off the base layer soon.
Parameters
----------
point : list
Values of data to slice at in world coordinates.
ndisplay : int
Number of dimensions to be displayed.
order : list of int
Order of dimensions, where last `ndisplay` will be
rendered in canvas.
"""
if point is None:
ndim = self.ndim
else:
ndim = len(point)
if order is None:
order = list(range(ndim))
# adjust the order of the global dims based on the number of
# dimensions that a layer has - for example a global order of
# [2, 1, 0, 3] -> [0, 1] for a layer that only has two dimensions
# or -> [1, 0, 2] for a layer with three as that corresponds to
# the relative order of the last two and three dimensions
# respectively
order = self._world_to_data_dims_displayed(order, ndim_world=ndim)
if point is None:
point = [0] * ndim
nd = min(self.ndim, ndisplay)
for i in order[-nd:]:
point[i] = slice(None)
else:
point = list(point)
# If no slide data has changed, then do nothing
offset = ndim - self.ndim
if (
np.all(order == self._dims_order)
and ndisplay == self._ndisplay
and np.all(point[offset:] == self._dims_point)
):
return
self._dims_order = order
if self._ndisplay != ndisplay:
self._ndisplay = ndisplay
self.events._ndisplay()
# Update the point values
self._dims_point = point[offset:]
self._update_dims()
self._set_editable()
@abstractmethod
def _update_thumbnail(self):
raise NotImplementedError()
@abstractmethod
def _get_value(self, position):
"""Value of the data at a position in data coordinates.
Parameters
----------
position : tuple
Position in data coordinates.
Returns
-------
value : tuple
Value of the data.
"""
raise NotImplementedError()
[docs] def get_value(
self,
position,
*,
view_direction: Optional[np.ndarray] = None,
dims_displayed: Optional[List[int]] = None,
world=False,
):
"""Value of the data at a position.
If the layer is not visible, return None.
Parameters
----------
position : tuple
Position in either data or world coordinates.
view_direction : Optional[np.ndarray]
A unit vector giving the direction of the ray in nD world coordinates.
The default value is None.
dims_displayed : Optional[List[int]]
A list of the dimensions currently being displayed in the viewer.
The default value is None.
world : bool
If True the position is taken to be in world coordinates
and converted into data coordinates. False by default.
Returns
-------
value : tuple, None
Value of the data. If the layer is not visible return None.
"""
if self.visible:
if world:
ndim_world = len(position)
if dims_displayed is not None:
# convert the dims_displayed to the layer dims.This accounts
# for differences in the number of dimensions in the world
# dims versus the layer and for transpose and rolls.
dims_displayed = dims_displayed_world_to_layer(
dims_displayed,
ndim_world=ndim_world,
ndim_layer=self.ndim,
)
position = self.world_to_data(position)
if (dims_displayed is not None) and (view_direction is not None):
if len(dims_displayed) == 2 or self.ndim == 2:
value = self._get_value(position=tuple(position))
elif len(dims_displayed) == 3:
view_direction = self._world_to_data_ray(
list(view_direction)
)
start_point, end_point = self.get_ray_intersections(
position=position,
view_direction=view_direction,
dims_displayed=dims_displayed,
world=False,
)
value = self._get_value_3d(
start_point=start_point,
end_point=end_point,
dims_displayed=dims_displayed,
)
else:
value = self._get_value(position)
else:
value = None
# This should be removed as soon as possible, it is still
# used in Points and Shapes.
self._value = value
return value
def _get_value_3d(
self,
start_point: np.ndarray,
end_point: np.ndarray,
dims_displayed: List[int],
) -> Union[float, int]:
"""Get the layer data value along a ray
Parameters
----------
start_point : np.ndarray
The start position of the ray used to interrogate the data.
end_point : np.ndarray
The end position of the ray used to interrogate the data.
dims_displayed : List[int]
The indices of the dimensions currently displayed in the Viewer.
Returns
-------
value
The data value along the supplied ray.
"""
return None
[docs] def projected_distance_from_mouse_drag(
self,
start_position: np.ndarray,
end_position: np.ndarray,
view_direction: np.ndarray,
vector: np.ndarray,
dims_displayed: Union[List, np.ndarray],
):
"""Calculate the length of the projection of a line between two mouse
clicks onto a vector (or array of vectors) in data coordinates.
Parameters
----------
start_position : np.ndarray
Starting point of the drag vector in data coordinates
end_position : np.ndarray
End point of the drag vector in data coordinates
view_direction : np.ndarray
Vector defining the plane normal of the plane onto which the drag
vector is projected.
vector : np.ndarray
(3,) unit vector or (n, 3) array thereof on which to project the drag
vector from start_event to end_event. This argument is defined in data
coordinates.
dims_displayed : Union[List, np.ndarray]
(3,) list of currently displayed dimensions
Returns
-------
projected_distance : (1, ) or (n, ) np.ndarray of float
"""
start_position = self._world_to_displayed_data(
start_position, dims_displayed
)
end_position = self._world_to_displayed_data(
end_position, dims_displayed
)
view_direction = self._world_to_displayed_data_ray(
view_direction, dims_displayed
)
return drag_data_to_projected_distance(
start_position, end_position, view_direction, vector
)
@contextmanager
def block_update_properties(self):
previous = self._update_properties
self._update_properties = False
try:
yield
finally:
self._update_properties = previous
def _set_highlight(self, force=False):
"""Render layer highlights when appropriate.
Parameters
----------
force : bool
Bool that forces a redraw to occur when `True`.
"""
pass
[docs] def refresh(self, event=None):
"""Refresh all layer data based on current view slice."""
if self.visible:
self.set_view_slice()
self.events.set_data() # refresh is called in _update_dims which means that extent cache is invalidated. Then, base on this event extent cache in layerlist is invalidated.
self._update_thumbnail()
self._set_highlight(force=True)
[docs] def world_to_data(self, position):
"""Convert from world coordinates to data coordinates.
Parameters
----------
position : tuple, list, 1D array
Position in world coordinates. If longer then the
number of dimensions of the layer, the later
dimensions will be used.
Returns
-------
tuple
Position in data coordinates.
"""
if len(position) >= self.ndim:
coords = list(position[-self.ndim :])
else:
coords = [0] * (self.ndim - len(position)) + list(position)
return tuple(self._transforms[1:].simplified.inverse(coords))
[docs] def data_to_world(self, position):
"""Convert from data coordinates to world coordinates.
Parameters
----------
position : tuple, list, 1D array
Position in data coordinates. If longer then the
number of dimensions of the layer, the later
dimensions will be used.
Returns
-------
tuple
Position in world coordinates.
"""
if len(position) >= self.ndim:
coords = list(position[-self.ndim :])
else:
coords = [0] * (self.ndim - len(position)) + list(position)
return tuple(self._transforms[1:].simplified(coords))
def _world_to_displayed_data(
self, position: np.ndarray, dims_displayed: np.ndarray
) -> tuple:
"""Convert world to data coordinates for displayed dimensions only.
Parameters
----------
position : tuple, list, 1D array
Position in world coordinates. If longer then the
number of dimensions of the layer, the later
dimensions will be used.
dims_displayed : list, 1D array
Indices of displayed dimensions of the data.
Returns
-------
tuple
Position in data coordinates for the displayed dimensions only
"""
position_nd = self.world_to_data(position)
position_ndisplay = np.asarray(position_nd)[dims_displayed]
return tuple(position_ndisplay)
@property
def _data_to_world(self) -> Affine:
"""The transform from data to world coordinates.
This affine transform is composed from the affine property and the
other transform properties in the following order:
affine * (rotate * shear * scale + translate)
"""
return self._transforms[1:3].simplified
def _world_to_data_ray(self, vector) -> tuple:
"""Convert a vector defining an orientation from world coordinates to data coordinates.
For example, this would be used to convert the view ray.
Parameters
----------
vector : tuple, list, 1D array
A vector in world coordinates.
Returns
-------
tuple
Vector in data coordinates.
"""
p1 = np.asarray(self.world_to_data(vector))
p0 = np.asarray(self.world_to_data(np.zeros_like(vector)))
normalized_vector = (p1 - p0) / np.linalg.norm(p1 - p0)
return tuple(normalized_vector)
def _world_to_displayed_data_ray(
self, vector_world, dims_displayed
) -> np.ndarray:
"""Convert an orientation from world to displayed data coordinates.
For example, this would be used to convert the view ray.
Parameters
----------
vector_world : tuple, list, 1D array
A vector in world coordinates.
Returns
-------
tuple
Vector in data coordinates.
"""
vector_data_nd = np.asarray(self._world_to_data_ray(vector_world))
vector_data_ndisplay = vector_data_nd[dims_displayed]
vector_data_ndisplay /= np.linalg.norm(vector_data_ndisplay)
return vector_data_ndisplay
def _world_to_data_dims_displayed(
self, dims_displayed: List[int], ndim_world: int
) -> List[int]:
"""Convert indices of displayed dims from world to data coordinates.
This accounts for differences in dimensionality between the world
and the data coordinates. For example a world dims order of
[2, 1, 0, 3] would be [0, 1] for a layer that only has two dimensions
or [1, 0, 2] for a layer with three as that corresponds to the
relative order of the last two and three dimensions respectively
Parameters
----------
dims_displayed : List[int]
The world displayed dimensions.
ndim_world : int
The number of dimensions in the world coordinate system.
Returns
-------
dims_displayed_data : List[int]
The displayed dimensions in data coordinates.
"""
offset = ndim_world - self.ndim
order = np.array(dims_displayed)
if offset <= 0:
return list(range(-offset)) + list(order - offset)
else:
return list(order[order >= offset] - offset)
def _display_bounding_box(self, dims_displayed: np.ndarray):
"""An axis aligned (self._ndisplay, 2) bounding box around the data"""
return self._extent_data[:, dims_displayed].T
[docs] def click_plane_from_click_data(
self,
click_position: np.ndarray,
view_direction: np.ndarray,
dims_displayed: List,
) -> Tuple[np.ndarray, np.ndarray]:
"""Calculate a (point, normal) plane parallel to the canvas in data
coordinates, centered on the centre of rotation of the camera.
Parameters
----------
click_position : np.ndarray
click position in world coordinates from mouse event.
view_direction : np.ndarray
view direction in world coordinates from mouse event.
dims_displayed : List
dimensions of the data array currently in view.
Returns
-------
click_plane : Tuple[np.ndarray, np.ndarray]
tuple of (plane_position, plane_normal) in data coordinates.
"""
click_position = np.asarray(click_position)
view_direction = np.asarray(view_direction)
plane_position = self.world_to_data(click_position)[dims_displayed]
plane_normal = self._world_to_data_ray(view_direction)[dims_displayed]
return plane_position, plane_normal
[docs] def get_ray_intersections(
self,
position: List[float],
view_direction: np.ndarray,
dims_displayed: List[int],
world: bool = True,
) -> Union[Tuple[np.ndarray, np.ndarray], Tuple[None, None]]:
"""Get the start and end point for the ray extending
from a point through the data bounding box.
Parameters
----------
position
the position of the point in nD coordinates. World vs. data
is set by the world keyword argument.
view_direction : np.ndarray
a unit vector giving the direction of the ray in nD coordinates.
World vs. data is set by the world keyword argument.
dims_displayed
a list of the dimensions currently being displayed in the viewer.
world : bool
True if the provided coordinates are in world coordinates.
Default value is True.
Returns
-------
start_point : np.ndarray
The point on the axis-aligned data bounding box that the cursor click
intersects with. This is the point closest to the camera.
The point is the full nD coordinates of the layer data.
If the click does not intersect the axis-aligned data bounding box,
None is returned.
end_point : np.ndarray
The point on the axis-aligned data bounding box that the cursor click
intersects with. This is the point farthest from the camera.
The point is the full nD coordinates of the layer data.
If the click does not intersect the axis-aligned data bounding box,
None is returned.
"""
if len(dims_displayed) != 3:
return None, None
# create the bounding box in data coordinates
bounding_box = self._display_bounding_box(dims_displayed)
start_point, end_point = self._get_ray_intersections(
position=position,
view_direction=view_direction,
dims_displayed=dims_displayed,
world=world,
bounding_box=bounding_box,
)
return start_point, end_point
def _get_offset_data_position(self, position: List[float]) -> List[float]:
"""Adjust position for offset between viewer and data coordinates."""
return position
def _get_ray_intersections(
self,
position: List[float],
view_direction: np.ndarray,
dims_displayed: List[int],
world: bool = True,
bounding_box: Optional[np.ndarray] = None,
) -> Union[Tuple[np.ndarray, np.ndarray], Tuple[None, None]]:
"""Get the start and end point for the ray extending
from a point through the data bounding box.
Parameters
----------
position
the position of the point in nD coordinates. World vs. data
is set by the world keyword argument.
view_direction : np.ndarray
a unit vector giving the direction of the ray in nD coordinates.
World vs. data is set by the world keyword argument.
dims_displayed
a list of the dimensions currently being displayed in the viewer.
world : bool
True if the provided coordinates are in world coordinates.
Default value is True.
bounding_box : np.ndarray
A (2, 3) bounding box around the data currently in view
Returns
-------
start_point : np.ndarray
The point on the axis-aligned data bounding box that the cursor click
intersects with. This is the point closest to the camera.
The point is the full nD coordinates of the layer data.
If the click does not intersect the axis-aligned data bounding box,
None is returned.
end_point : np.ndarray
The point on the axis-aligned data bounding box that the cursor click
intersects with. This is the point farthest from the camera.
The point is the full nD coordinates of the layer data.
If the click does not intersect the axis-aligned data bounding box,
None is returned."""
# get the view direction and click position in data coords
# for the displayed dimensions only
if world is True:
view_dir = self._world_to_displayed_data_ray(
view_direction, dims_displayed
)
click_pos_data = self._world_to_displayed_data(
position, dims_displayed
)
else:
# adjust for any offset between viewer and data coordinates
position = self._get_offset_data_position(position)
view_dir = np.asarray(view_direction)[dims_displayed]
click_pos_data = np.asarray(position)[dims_displayed]
# Determine the front and back faces
front_face_normal, back_face_normal = find_front_back_face(
click_pos_data, bounding_box, view_dir
)
if front_face_normal is None and back_face_normal is None:
# click does not intersect the data bounding box
return None, None
# Calculate ray-bounding box face intersections
start_point_displayed_dimensions = (
intersect_line_with_axis_aligned_bounding_box_3d(
click_pos_data, view_dir, bounding_box, front_face_normal
)
)
end_point_displayed_dimensions = (
intersect_line_with_axis_aligned_bounding_box_3d(
click_pos_data, view_dir, bounding_box, back_face_normal
)
)
# add the coordinates for the axes not displayed
start_point = np.asarray(position)
start_point[dims_displayed] = start_point_displayed_dimensions
end_point = np.asarray(position)
end_point[dims_displayed] = end_point_displayed_dimensions
return start_point, end_point
@property
def _displayed_axes(self):
# assignment upfront to avoid repeated computation of properties
_dims_displayed = self._dims_displayed
_dims_displayed_order = self._dims_displayed_order
displayed_axes = [_dims_displayed[i] for i in _dims_displayed_order]
return displayed_axes
@property
def _corner_pixels_displayed(self):
displayed_axes = self._displayed_axes
corner_pixels_displayed = self.corner_pixels[:, displayed_axes]
return corner_pixels_displayed
def _update_draw(
self, scale_factor, corner_pixels_displayed, shape_threshold
):
"""Update canvas scale and corner values on draw.
For layer multiscale determining if a new resolution level or tile is
required.
Parameters
----------
scale_factor : float
Scale factor going from canvas to world coordinates.
corner_pixels_displayed : array, shape (2, 2)
Coordinates of the top-left and bottom-right canvas pixels in
world coordinates.
shape_threshold : tuple
Requested shape of field of view in data coordinates.
"""
self.scale_factor = scale_factor
displayed_axes = self._displayed_axes
# must adjust displayed_axes according to _dims_order
displayed_axes = np.asarray(
[self._dims_order[d] for d in displayed_axes]
)
# we need to compute all four corners to compute a complete,
# data-aligned bounding box, because top-left/bottom-right may not
# remain top-left and bottom-right after transformations.
all_corners = list(itertools.product(*corner_pixels_displayed.T))
# Note that we ignore the first transform which is tile2data
data_corners = (
self._transforms[1:]
.simplified.set_slice(displayed_axes)
.inverse(all_corners)
)
# find the maximal data-axis-aligned bounding box containing all four
# canvas corners
data_bbox = np.stack(
[np.min(data_corners, axis=0), np.max(data_corners, axis=0)]
)
# round and clip the bounding box values
data_bbox_int = np.stack(
[np.floor(data_bbox[0]), np.ceil(data_bbox[1])]
).astype(int)
displayed_extent = self.extent.data[:, displayed_axes]
data_bbox_clipped = np.clip(
data_bbox_int, displayed_extent[0], displayed_extent[1]
)
if self._ndisplay == 2 and self.multiscale:
level, scaled_corners = compute_multiscale_level_and_corners(
data_bbox_clipped,
shape_threshold,
self.downsample_factors[:, displayed_axes],
)
corners = np.zeros((2, self.ndim))
corners[:, displayed_axes] = scaled_corners
corners = corners.astype(int)
display_shape = tuple(
corners[1, displayed_axes] - corners[0, displayed_axes]
)
if any(s == 0 for s in display_shape):
return
if self.data_level != level or not np.all(
self.corner_pixels == corners
):
self._data_level = level
self.corner_pixels = corners
self.refresh()
else:
corners = np.zeros((2, self.ndim), dtype=int)
corners[:, displayed_axes] = data_bbox_clipped
self.corner_pixels = corners
[docs] def get_status(
self,
position: np.ndarray,
*,
view_direction: Optional[np.ndarray] = None,
dims_displayed: Optional[List[int]] = None,
world=False,
):
"""
Status message of the data at a coordinate position.
Parameters
----------
position : tuple
Position in either data or world coordinates.
view_direction : Optional[np.ndarray]
A unit vector giving the direction of the ray in nD world coordinates.
The default value is None.
dims_displayed : Optional[List[int]]
A list of the dimensions currently being displayed in the viewer.
The default value is None.
world : bool
If True the position is taken to be in world coordinates
and converted into data coordinates. False by default.
Returns
-------
msg : string
String containing a message that can be used as a status update.
"""
value = self.get_value(
position,
view_direction=view_direction,
dims_displayed=dims_displayed,
world=world,
)
return generate_layer_status(self.name, position, value)
def _get_tooltip_text(
self,
position,
*,
view_direction: Optional[np.ndarray] = None,
dims_displayed: Optional[List[int]] = None,
world: bool = False,
):
"""
tooltip message of the data at a coordinate position.
Parameters
----------
position : tuple
Position in either data or world coordinates.
view_direction : Optional[np.ndarray]
A unit vector giving the direction of the ray in nD world coordinates.
The default value is None.
dims_displayed : Optional[List[int]]
A list of the dimensions currently being displayed in the viewer.
The default value is None.
world : bool
If True the position is taken to be in world coordinates
and converted into data coordinates. False by default.
Returns
-------
msg : string
String containing a message that can be used as a tooltip.
"""
return ""
[docs] def save(self, path: str, plugin: Optional[str] = None) -> List[str]:
"""Save this layer to ``path`` with default (or specified) plugin.
Parameters
----------
path : str
A filepath, directory, or URL to open. Extensions may be used to
specify output format (provided a plugin is available for the
requested format).
plugin : str, optional
Name of the plugin to use for saving. If ``None`` then all plugins
corresponding to appropriate hook specification will be looped
through to find the first one that can save the data.
Returns
-------
list of str
File paths of any files that were written.
"""
from ...plugins.io import save_layers
return save_layers(path, [self], plugin=plugin)
def _on_selection(self, selected: bool):
# This method is a temporary workaround to the fact that the Points
# layer needs to know when its selection state changes so that it can
# update the highlight state. This, along with the events.select and
# events.deselect emitters, (and the LayerList._on_selection_event
# method) can be removed once highlighting logic has been removed from
# the layer model.
if selected:
self.events.select()
else:
self.events.deselect()
[docs] @classmethod
def create(
cls, data, meta: dict = None, layer_type: Optional[str] = None
) -> Layer:
"""Create layer from `data` of type `layer_type`.
Primarily intended for usage by reader plugin hooks and creating a
layer from an unwrapped layer data tuple.
Parameters
----------
data : Any
Data in a format that is valid for the corresponding `layer_type`.
meta : dict, optional
Dict of keyword arguments that will be passed to the corresponding
layer constructor. If any keys in `meta` are not valid for the
corresponding layer type, an exception will be raised.
layer_type : str
Type of layer to add. Must be the (case insensitive) name of a
Layer subclass. If not provided, the layer is assumed to
be "image", unless data.dtype is one of (np.int32, np.uint32,
np.int64, np.uint64), in which case it is assumed to be "labels".
Raises
------
ValueError
If ``layer_type`` is not one of the recognized layer types.
TypeError
If any keyword arguments in ``meta`` are unexpected for the
corresponding `add_*` method for this layer_type.
Examples
--------
A typical use case might be to upack a tuple of layer data with a
specified layer_type.
>>> data = (
... np.random.random((10, 2)) * 20,
... {'face_color': 'blue'},
... 'points',
... )
>>> Layer.create(*data)
"""
from ... import layers
from ..image._image_utils import guess_labels
layer_type = (layer_type or '').lower()
# assumes that big integer type arrays are likely labels.
if not layer_type:
layer_type = guess_labels(data)
if layer_type not in layers.NAMES:
raise ValueError(
trans._(
"Unrecognized layer_type: '{layer_type}'. Must be one of: {layer_names}.",
deferred=True,
layer_type=layer_type,
layer_names=layers.NAMES,
)
)
Cls = getattr(layers, layer_type.title())
try:
return Cls(data, **(meta or {}))
except Exception as exc:
if 'unexpected keyword argument' not in str(exc):
raise exc
bad_key = str(exc).split('keyword argument ')[-1]
raise TypeError(
trans._(
"_add_layer_from_data received an unexpected keyword argument ({bad_key}) for layer type {layer_type}",
deferred=True,
bad_key=bad_key,
layer_type=layer_type,
)
) from exc
mgui.register_type(type_=List[Layer], return_callback=add_layers_to_viewer)