def init():
# inserting a constraint
dt = exact_delaunay.Triangulation()
pnts = [[0, 0], [5, 0], [10, 0], [5, 5], [3, 0], [6, 2], [12, 4]]
for pnt in pnts:
dt.add_node(x=pnt)
return dt
def test_basic1():
plot = False
dt = exact_delaunay.Triangulation()
pnts = [[0, 0], [5, 0], [10, 0], [5, 5]]
dt.add_node(x=pnts[0]) # This tests insert into empty
dt.add_node(x=pnts[1]) # adjacent_vertex
dt.add_node(x=pnts[2]) # adjacent_vertex
dt.add_node(x=pnts[3]) # adjacent_edge
dt.add_node(x=[3, 0]) # colinear
if plot:
plt.clf()
dt.plot_nodes(labeler=lambda n, nrec: str(n))
dt.plot_edges(alpha=0.5, lw=2)
dt.plot_cells(lw=13, facecolor='#ddddff', edgecolor='w', zorder=-5)
# This where topo sort is failing
plt.pause(0.1)
n = dt.add_node(x=[6, 2]) # into cell interior
if plot:
plt.clf()
dt.plot_nodes(labeler=lambda n, nrec: str(n))
dt.plot_edges(alpha=0.5, lw=2)
dt.plot_cells(lw=13, facecolor='#ddddff', edgecolor='w', zorder=-5)
def test_move1():
plot = False
dt = exact_delaunay.Triangulation()
pnts = [[0, 0], [5, 0], [10, 0], [5, 5]]
dt.add_node(x=pnts[0]) # This tests insert into empty
dt.add_node(x=pnts[1]) # adjacent_vertex
dt.add_node(x=pnts[2]) # adjacent_vertex
dt.add_node(x=pnts[3]) # adjacent_edge
dt.add_node(x=[3, 0]) # colinear
if plot:
plt.clf()
dt.plot_nodes()
dt.plot_edges(alpha=0.5, lw=2)
dt.plot_cells(lw=13, facecolor='#ddddff', edgecolor='w', zorder=-5)
n = dt.add_node(x=[6, 2]) # into cell interior
dt.modify_node(4, x=[3.01, 0.25])
dt.modify_node(4, x=[3.01, -0.25])
if plot:
plt.clf()
dt.plot_nodes(labeler=lambda n, nrec: str(n))
dt.plot_edges(alpha=0.5, lw=2)
dt.plot_cells(lw=13, facecolor='#ddddff', edgecolor='w', zorder=-5)
dt.check_global_delaunay()
return dt
def test_delete3():
plot = False
dt = exact_delaunay.Triangulation()
nodes = [dt.add_node(x=[10, i]) for i in range(10)]
dt.delete_node(nodes[0])
dt.delete_node(nodes[4])
if plot:
plt.cla()
dt.plot_nodes()
dt.plot_edges(alpha=0.5, lw=2)
dt.plot_cells(lw=13, facecolor='#ddddff', edgecolor='w', zorder=-5)
def test_test_dim_down():
dt = exact_delaunay.Triangulation()
n = dt.add_node(x=[0, 0])
for i in range(5):
other = dt.add_node(x=[10, i])
assert dt.test_delete_node_dim_down(n)
assert not dt.test_delete_node_dim_down(other)
friend = dt.add_node(x=[0, 2])
assert not dt.test_delete_node_dim_down(friend)
assert not dt.test_delete_node_dim_down(n)
def test_flip2():
plot = False
dt = exact_delaunay.Triangulation()
dt.add_node(x=[0, 0])
for i in range(5):
dt.add_node(x=[10, i])
# This one requires a flip:
n = dt.add_node(x=[5, 1])
if plot:
plt.cla()
dt.plot_nodes()
dt.plot_edges(alpha=0.5, lw=2)
dt.plot_cells(lw=13, facecolor='#ddddff', edgecolor='w', zorder=-5)
def test_constraints_dim1():
dt = exact_delaunay.Triangulation()
pnts = [[0, 0], [5, 0], [10, 0]]
for pnt in pnts:
dt.add_node(x=pnt)
dt.add_constraint(0, 1)
dt.add_constraint(1, 2)
dt.remove_constraint(0, 1)
dt.remove_constraint(1, 2)
try:
dt.add_constraint(0, 2)
assert False
except dt.ConstraintCollinearNode:
pass #
def test_flip1():
plot = False
dt = exact_delaunay.Triangulation()
pnts = [[0, 0], [8, 0], [10, 5], [5, 5]]
dt.add_node(x=pnts[0]) # This tests insert into empty
dt.add_node(x=pnts[1]) # adjacent_vertex
dt.add_node(x=pnts[2]) # adjacent_vertex
dt.add_node(x=pnts[3]) # adjacent_edge
dt.add_node(x=[3, 0]) # colinear
if plot:
plt.clf()
dt.plot_nodes()
dt.plot_edges(alpha=0.5, lw=2)
dt.plot_cells(lw=13, facecolor='#ddddff', edgecolor='w', zorder=-5)
def test_fuzz1():
plot = False
# Fuzzing, regular
x = np.arange(5)
y = np.arange(5)
X, Y = np.meshgrid(x, y)
xys = np.array([X.ravel(), Y.ravel()]).T
if plot:
plt.figure(1).clf()
fig, ax = plt.subplots(num=1)
ax.plot(xys[:, 0], xys[:, 1], 'go', alpha=0.4)
ax.axis([-1, 5, -1, 5])
idxs = np.zeros(len(xys), 'i8') - 1
dt = exact_delaunay.Triangulation()
# definitely slows down as the number of nodes gets larger.
# starting off with <1s per 100 operations, later more like 2s
for repeat in range(1):
print "Repeat: ", repeat
for step in range(1000):
if step % 200 == 0:
print " step: ", step
toggle = np.random.randint(len(idxs))
if idxs[toggle] < 0:
idxs[toggle] = dt.add_node(x=xys[toggle])
else:
dt.delete_node(idxs[toggle])
idxs[toggle] = -1
if plot:
del ax.lines[1:]
ax.texts = []
ax.collections = []
dt.plot_nodes(labeler=lambda n, nrec: str(n))
dt.plot_edges(alpha=0.5, lw=2)
dt.plot_cells(lw=7,
facecolor='#ddddff',
edgecolor='w',
zorder=-5)
plt.draw()
def test_delete4():
plot = False
dt = exact_delaunay.Triangulation()
dt.add_node(x=[0, 0])
dt.add_node(x=[5, 0])
dt.add_node(x=[10, 0])
dt.add_node(x=[-2, 3])
dt.add_node(x=[-2, -3])
# and now delete the one in the middle:
dt.delete_node(1)
if plot:
plt.cla()
dt.plot_nodes(labeler=lambda n, nrec: str(n))
dt.plot_edges(alpha=0.5, lw=2)
dt.plot_cells(lw=7, facecolor='#ddddff', edgecolor='w', zorder=-5)
def test_delete2():
plot = False
dt = exact_delaunay.Triangulation() # ExactDelaunay()
dt.add_node(x=[0, 0])
for i in range(5):
if 0: # sparser nodes for testing
if i > 0 and i < 4:
continue
dt.add_node(x=[10, i])
# This one requires a flip:
n = dt.add_node(x=[5, 1])
far = dt.add_node(x=[12, 4])
dt.delete_node(0)
if plot:
plt.cla()
dt.plot_nodes(labeler=lambda n, nr: str(n))
dt.plot_edges(alpha=0.5, lw=2)
dt.plot_cells(lw=6, facecolor='#ddddff', edgecolor='w', zorder=-5)
def test_flip2():
plot = False
# testing Delaunay flipping
dt = exact_delaunay.Triangulation() # ExactDelaunay()
dt.add_node(x=[0, 0])
for i in range(10):
dt.add_node(x=[10, i])
dt.add_node(x=[5, 1.0])
dt.add_node(x=[10, 10])
dt.add_node(x=[5, 2])
# and test a flip when the new vertex is outside the convex hull
dt.add_node(x=[5, -1])
dt.delete_node(11)
if plot:
plt.cla()
dt.plot_nodes(labeler=lambda n, nrec: str(n))
dt.plot_edges(alpha=0.5, lw=2)
dt.plot_cells(lw=13, facecolor='#ddddff', edgecolor='w', zorder=-5)
def test_delete1():
plot = False
dt = exact_delaunay.Triangulation() # ExactDelaunay()
dt.add_node(x=[0, 0])
for i in range(5):
dt.add_node(x=[10, i])
# This one requires a flip:
n = dt.add_node(x=[5, 1])
if plot:
plt.cla()
dt.plot_nodes(labeler=lambda n, nr: str(n))
dt.plot_edges(alpha=0.5, lw=2)
bad = (~dt.cells['deleted']) & (dt.cells_area() <= 0)
good = (~dt.cells['deleted']) & (dt.cells_area() > 0)
dt.plot_cells(lw=8,
facecolor='#ddddff',
edgecolor='w',
zorder=-5,
mask=good)
dt.plot_cells(lw=8,
facecolor='#ffdddd',
edgecolor='r',
zorder=-5,
mask=bad)
# deleting n work okay
dt.delete_node(n)
if plot:
plt.cla()
dt.plot_nodes(labeler=lambda n, nr: str(n))
dt.plot_edges(alpha=0.5, lw=2)
bad = (~dt.cells['deleted']) & (dt.cells_area() <= 0)
good = (~dt.cells['deleted']) & (dt.cells_area() > 0)
dt.plot_cells(lw=8,
facecolor='#ddddff',
edgecolor='w',
zorder=-5,
mask=good)
dt.plot_cells(lw=8,
facecolor='#ffdddd',
edgecolor='r',
zorder=-5,
mask=bad)
# delete a node which defines part of the convex hull:
dt.delete_node(5) # works, though may need to patch up edges['cells'] ?
dt.delete_node(2) # collinear portion of the convex hull
dt.delete_node(3)
# everything looks okay. Surprising that we didn't have to
# add code to deal with this case...
if plot:
plt.cla()
dt.plot_nodes(labeler=lambda n, nr: str(n))
dt.plot_edges(alpha=0.5, lw=2)
bad = (~dt.cells['deleted']) & (dt.cells_area() <= 0)
good = (~dt.cells['deleted']) & (dt.cells_area() > 0)
dt.plot_cells(lw=8,
facecolor='#ddddff',
edgecolor='w',
zorder=-5,
mask=good)
dt.plot_cells(lw=8,
facecolor='#ffdddd',
edgecolor='r',
zorder=-5,
mask=bad)
dt.delete_node(0)
if plot:
assert dt.dim() == 1
plt.cla()
dt.plot_nodes(labeler=lambda n, nr: str(n))
dt.plot_edges(alpha=0.5, lw=2)
bad = (~dt.cells['deleted']) & (dt.cells_area() <= 0)
good = (~dt.cells['deleted']) & (dt.cells_area() > 0)
dt.plot_cells(lw=8,
facecolor='#ddddff',
edgecolor='w',
zorder=-5,
mask=good)
dt.plot_cells(lw=8,
facecolor='#ffdddd',
edgecolor='r',
zorder=-5,
mask=bad)
poly = sources['geom'][39]
buff = 200
xyz_sub = xyz.clip_to_polygon(poly.buffer(buff))
##
dem1 = xyz_sub.to_grid(dx=5, dy=10, interp='linear')
##
# That's okay, although not that great.
# 1st step: build a triangulation, embed the polygon as
# constrained edges, see how that looks.
cdt = exact_delaunay.Triangulation(extra_node_fields=[('value', np.float64)])
cdt.bulk_init(xyz_sub.X)
cdt.nodes['value'] = xyz_sub.F
# Add in the nodes and edges of the polygon:
poly_pnts = np.array(poly.exterior)
if np.all(poly_pnts[-1] == poly_pnts[0]):
poly_pnts = poly_pnts[:-1]
new_values = xyz_sub.interpolate(poly_pnts, 'nearest')
poly_nodes = []
for pnt, value in zip(poly_pnts, new_values):
poly_nodes.append(cdt.add_node(x=pnt, value=value))
from stompy import utils
for a, b in utils.circular_pairs(poly_nodes):
