CURA-5370 Small refactor for Arranger: make x and y consistent (numpy arrays start with y first in general), faster, cleanup, more unit tests, take actual build plate size in Arranger instances

This commit is contained in:
Jack Ha 2018-05-22 17:13:35 +02:00
parent 310aee07ac
commit f5bed242ed
8 changed files with 287 additions and 73 deletions

View file

@ -18,17 +18,20 @@ LocationSuggestion = namedtuple("LocationSuggestion", ["x", "y", "penalty_points
# good locations for objects that you try to put on a build place.
# Different priority schemes can be defined so it alters the behavior while using
# the same logic.
#
# Note: Make sure the scale is the same between ShapeArray objects and the Arrange instance.
class Arrange:
build_volume = None
def __init__(self, x, y, offset_x, offset_y, scale= 1.0):
self.shape = (y, x)
self._priority = numpy.zeros((x, y), dtype=numpy.int32)
self._priority_unique_values = []
self._occupied = numpy.zeros((x, y), dtype=numpy.int32)
def __init__(self, x, y, offset_x, offset_y, scale= 0.5):
self._scale = scale # convert input coordinates to arrange coordinates
self._offset_x = offset_x
self._offset_y = offset_y
world_x, world_y = int(x * self._scale), int(y * self._scale)
self._shape = (world_y, world_x)
self._priority = numpy.zeros((world_y, world_x), dtype=numpy.int32) # beware: these are indexed (y, x)
self._priority_unique_values = []
self._occupied = numpy.zeros((world_y, world_x), dtype=numpy.int32) # beware: these are indexed (y, x)
self._offset_x = int(offset_x * self._scale)
self._offset_y = int(offset_y * self._scale)
self._last_priority = 0
self._is_empty = True
@ -39,7 +42,7 @@ class Arrange:
# \param scene_root Root for finding all scene nodes
# \param fixed_nodes Scene nodes to be placed
@classmethod
def create(cls, scene_root = None, fixed_nodes = None, scale = 0.5, x = 220, y = 220):
def create(cls, scene_root = None, fixed_nodes = None, scale = 0.5, x = 350, y = 250):
arranger = Arrange(x, y, x // 2, y // 2, scale = scale)
arranger.centerFirst()
@ -61,13 +64,17 @@ class Arrange:
# If a build volume was set, add the disallowed areas
if Arrange.build_volume:
disallowed_areas = Arrange.build_volume.getDisallowedAreas()
disallowed_areas = Arrange.build_volume.getDisallowedAreasNoBrim()
for area in disallowed_areas:
points = copy.deepcopy(area._points)
shape_arr = ShapeArray.fromPolygon(points, scale = scale)
arranger.place(0, 0, shape_arr, update_empty = False)
return arranger
## This resets the optimization for finding location based on size
def resetLastPriority(self):
self._last_priority = 0
## Find placement for a node (using offset shape) and place it (using hull shape)
# return the nodes that should be placed
# \param node
@ -104,7 +111,7 @@ class Arrange:
def centerFirst(self):
# Square distance: creates a more round shape
self._priority = numpy.fromfunction(
lambda i, j: (self._offset_x - i) ** 2 + (self._offset_y - j) ** 2, self.shape, dtype=numpy.int32)
lambda j, i: (self._offset_x - i) ** 2 + (self._offset_y - j) ** 2, self._shape, dtype=numpy.int32)
self._priority_unique_values = numpy.unique(self._priority)
self._priority_unique_values.sort()
@ -112,7 +119,7 @@ class Arrange:
# This is a strategy for the arranger.
def backFirst(self):
self._priority = numpy.fromfunction(
lambda i, j: 10 * j + abs(self._offset_x - i), self.shape, dtype=numpy.int32)
lambda j, i: 10 * j + abs(self._offset_x - i), self._shape, dtype=numpy.int32)
self._priority_unique_values = numpy.unique(self._priority)
self._priority_unique_values.sort()
@ -126,9 +133,15 @@ class Arrange:
y = int(self._scale * y)
offset_x = x + self._offset_x + shape_arr.offset_x
offset_y = y + self._offset_y + shape_arr.offset_y
if offset_x < 0 or offset_y < 0:
return None # out of bounds in self._occupied
occupied_x_max = offset_x + shape_arr.arr.shape[1]
occupied_y_max = offset_y + shape_arr.arr.shape[0]
if occupied_x_max > self._occupied.shape[1] + 1 or occupied_y_max > self._occupied.shape[0] + 1:
return None # out of bounds in self._occupied
occupied_slice = self._occupied[
offset_y:offset_y + shape_arr.arr.shape[0],
offset_x:offset_x + shape_arr.arr.shape[1]]
offset_y:occupied_y_max,
offset_x:occupied_x_max]
try:
if numpy.any(occupied_slice[numpy.where(shape_arr.arr == 1)]):
return None
@ -140,7 +153,7 @@ class Arrange:
return numpy.sum(prio_slice[numpy.where(shape_arr.arr == 1)])
## Find "best" spot for ShapeArray
# Return namedtuple with properties x, y, penalty_points, priority
# Return namedtuple with properties x, y, penalty_points, priority.
# \param shape_arr ShapeArray
# \param start_prio Start with this priority value (and skip the ones before)
# \param step Slicing value, higher = more skips = faster but less accurate
@ -153,12 +166,11 @@ class Arrange:
for priority in self._priority_unique_values[start_idx::step]:
tryout_idx = numpy.where(self._priority == priority)
for idx in range(len(tryout_idx[0])):
x = tryout_idx[0][idx]
y = tryout_idx[1][idx]
projected_x = x - self._offset_x
projected_y = y - self._offset_y
x = tryout_idx[1][idx]
y = tryout_idx[0][idx]
projected_x = int((x - self._offset_x) / self._scale)
projected_y = int((y - self._offset_y) / self._scale)
# array to "world" coordinates
penalty_points = self.checkShape(projected_x, projected_y, shape_arr)
if penalty_points is not None:
return LocationSuggestion(x = projected_x, y = projected_y, penalty_points = penalty_points, priority = priority)
@ -191,8 +203,12 @@ class Arrange:
# Set priority to low (= high number), so it won't get picked at trying out.
prio_slice = self._priority[min_y:max_y, min_x:max_x]
prio_slice[numpy.where(shape_arr.arr[
min_y - offset_y:max_y - offset_y, min_x - offset_x:max_x - offset_x] == 1)] = 999
prio_slice[new_occupied] = 999
# If you want to see how the rasterized arranger build plate looks like, uncomment this code
# numpy.set_printoptions(linewidth=500, edgeitems=200)
# print(self._occupied.shape)
# print(self._occupied)
@property
def isEmpty(self):