def _prepare_data_for_window(self, xlow, xhigh, ylow, yhigh):
'''
This is called every time the chaco window is rendered and it dynamically
computes and returns the data corresponding to the window limits. This allows
rendering to stay relatively fast since data is binned down to a resolution
roughly matching what the window can usefully display.
'''
#if not self.update_content:
if not self.loop1:
return self.zi
stack = self.dataset_stack.copy()
# deal with restricted y-range
if ylow < 1 or ylow >= yhigh:
ylow = 1
if yhigh > stack.shape[0] or yhigh <= ylow:
yhigh = stack.shape[0]
stack = stack[int(np.round(ylow - 1)):int(np.round(yhigh))]
if stack[0, 0, 0] > stack[0, -1, 0]:
stack = stack[:, ::-1, :]
# get a typical interval
interval = np.median(np.diff(stack[0, :10, 0]))
# get a region a couple of samples bigger on each side of the window to allow for
# misalignment
xs = stack[:, :, 0]
xlow_expanded = xs[0, 0] - interval * 2
xhigh_expanded = xs[0, -1] + interval * 2
column_mask = (xs[0] >= xlow_expanded) & (xs[0] <= xhigh_expanded)
stack = stack[:, column_mask][np.newaxis].reshape(xs.shape[0], -1, 3)
# regrid all rows - use an interval half that of the original interval to minimise interpolation errors
first_row = regrid_data(stack[0],
start=xlow_expanded,
end=xhigh_expanded,
interval=interval / 2)
new_stack = np.empty((xs.shape[0], first_row.shape[0], 3))
new_stack[0] = first_row
for i, stack_row in enumerate(stack[1:]):
new_stack[i + 1] = regrid_data(stack_row,
start=xlow_expanded,
end=xhigh_expanded,
interval=interval / 2)
# new_stack is the regridded version so window it properly
xs = new_stack[:, :, 0]
zs = new_stack[:, :, 1]
mask = (xs[0] >= xlow) & (xs[0] <= xhigh)
zs = zs[:, mask].reshape(zs.shape[0], -1)
YBINS = zs.shape[0] * 10
BINS = min(5000, zs.shape[1])
zi = congrid(zs, (YBINS, BINS), method='neighbour', minusone=True)
zi = np.clip(zi, 1, zi.max())
self.zi = zi
self.loop1 = False
return zi
def _prepare_data_for_window(self, xlow, xhigh, ylow, yhigh):
'''
This is called every time the chaco window is rendered and it dynamically
computes and returns the data corresponding to the window limits. This allows
rendering to stay relatively fast since data is binned down to a resolution
roughly matching what the window can usefully display.
'''
#if not self.update_content:
if not self.loop1:
return self.zi
stack = self.dataset_stack.copy()
# deal with restricted y-range
if ylow < 1 or ylow >= yhigh:
ylow = 1
if yhigh > stack.shape[0] or yhigh <= ylow:
yhigh = stack.shape[0]
stack = stack[int(np.round(ylow-1)):int(np.round(yhigh))]
if stack[0,0,0] > stack[0,-1,0]:
stack = stack[:,::-1,:]
# get a typical interval
interval = np.median(np.diff(stack[0,:10,0]))
# get a region a couple of samples bigger on each side of the window to allow for
# misalignment
xs = stack[:,:,0]
xlow_expanded = xs[0,0] - interval*2
xhigh_expanded = xs[0,-1] + interval*2
column_mask = (xs[0]>=xlow_expanded) & (xs[0]<=xhigh_expanded)
stack = stack[:,column_mask][np.newaxis].reshape(xs.shape[0],-1,3)
# regrid all rows - use an interval half that of the original interval to minimise interpolation errors
first_row = regrid_data(stack[0], start=xlow_expanded, end=xhigh_expanded, interval=interval/2)
new_stack = np.empty((xs.shape[0], first_row.shape[0], 3))
new_stack[0] = first_row
for i, stack_row in enumerate(stack[1:]):
new_stack[i+1] = regrid_data(stack_row, start=xlow_expanded, end=xhigh_expanded, interval=interval/2)
# new_stack is the regridded version so window it properly
xs = new_stack[:,:,0]
zs = new_stack[:,:,1]
mask = (xs[0]>=xlow) & (xs[0]<=xhigh)
zs = zs[:,mask].reshape(zs.shape[0],-1)
YBINS = zs.shape[0]*10
BINS = min(5000, zs.shape[1])
zi = congrid(zs, (YBINS, BINS), method='neighbour', minusone=True)
zi = np.clip(zi, 1, zi.max())
self.zi = zi
self.loop1=False
return zi
def regrid_data_test2(self):
# test that regridding preserves the y-data if resampled near the x-data points
np.random.seed(42)
POINTS = 5
data = np.c_[np.arange(POINTS), np.random.rand(POINTS), np.arange(POINTS)]
newdata = processing.regrid_data(data, start=0.000001, end=data[:,0][-1]+.000001, interval=1.0)
self.assertTrue(np.allclose(data, newdata, atol=1e-4))
def regrid_data_test(self):
# test that regridding preserves the y-data if resampled at the x-data points
np.random.seed(42)
POINTS = 50000
data = np.c_[np.arange(POINTS), np.random.rand(POINTS), np.arange(POINTS)]
newdata = processing.regrid_data(data, interval=1.0)
self.assertTrue(np.allclose(data, newdata))
def regrid_data_test(self):
# test that regridding preserves the y-data if resampled at the x-data points
np.random.seed(42)
POINTS = 50000
data = np.c_[np.arange(POINTS),
np.random.rand(POINTS),
np.arange(POINTS)]
newdata = processing.regrid_data(data, interval=1.0)
self.assertTrue(np.allclose(data, newdata))
def regrid_data_test2(self):
# test that regridding preserves the y-data if resampled near the x-data points
np.random.seed(42)
POINTS = 5
data = np.c_[np.arange(POINTS),
np.random.rand(POINTS),
np.arange(POINTS)]
newdata = processing.regrid_data(data,
start=0.000001,
end=data[:, 0][-1] + .000001,
interval=1.0)
self.assertTrue(np.allclose(data, newdata, atol=1e-4))