Add an example showing how to dynamically resample for plotting using event handling.

dopplershift · dopplershift · commit 65ebe493af64 · 2009-08-21T17:39:46.000Z
svn path=/trunk/matplotlib/; revision=7515
diff --git a/examples/event_handling/resample.py b/examples/event_handling/resample.py
@@ -0,0 +1,53 @@
+import numpy as np
+import matplotlib.pyplot as plt
+from scikits.audiolab import wavread
+
+# A class that will downsample the data and recompute when zoomed.
+class DataDisplayDownsampler(object):
+    def __init__(self, xdata, ydata):
+        self.origYData = ydata
+        self.origXData = xdata
+        self.numpts = 3000
+        self.delta = xdata[-1] - xdata[0]
+
+    def resample(self, xstart, xend):
+        # Very simple downsampling that takes the points within the range
+        # and picks every Nth point
+        mask = (self.origXData > xstart) & (self.origXData < xend)
+        xdata = self.origXData[mask]
+        ratio = int(xdata.size / self.numpts) + 1
+        xdata = xdata[::ratio]
+
+        ydata = self.origYData[mask]
+        ydata = ydata[::ratio]
+
+        return xdata, ydata
+
+    def update(self, ax):
+        # Update the line
+        lims = ax.viewLim
+        if np.abs(lims.width - self.delta) > 1e-8:
+            self.delta = lims.width
+            xstart, xend = lims.intervalx
+            self.line.set_data(*self.downsample(xstart, xend))
+            ax.figure.canvas.draw_idle()
+
+# Read data
+data = wavread('/usr/share/sounds/purple/receive.wav')[0]
+ydata = np.tile(data[:, 0], 100)
+xdata = np.arange(ydata.size)
+
+d = DataDisplayDownsampler(xdata, ydata)
+
+fig = plt.figure()
+ax = fig.add_subplot(1, 1, 1)
+
+#Hook up the line
+xdata, ydata = d.downsample(xdata[0], xdata[-1])
+d.line, = ax.plot(xdata, ydata)
+ax.set_autoscale_on(False) # Otherwise, infinite loop
+
+# Connect for changing the view limits
+ax.callbacks.connect('xlim_changed', d.update)
+
+plt.show()
