add linear filter utility function.

amodem/sigproc.py

@@ -27,6 +27,12 @@ class Filter(object):
             yield y
 
 
+def lfilter(b, a, x):
+    f = Filter(b=b, a=a)
+    y = list(f(x))
+    return np.array(y)
+
+
 def train(S, training):
     A = np.array([S[1:], S[:-1], training[:-1]]).T
     b = training[1:]

tests/test_full.py

@@ -36,18 +36,13 @@ def run(size, chan):
 
     assert rx_data == tx_data
 
-def apply_filter(b, a, x):
-    f = sigproc.Filter(b=b, a=a)
-    y = list(f(list(x)))
-    return np.array(y)
-
 
 def test_lowpass():
-    run(1024, lambda x: apply_filter(b=[0.9], a=[1.0, -0.1], x=x))
+    run(1024, lambda x: sigproc.lfilter(b=[0.9], a=[1.0, -0.1], x=x))
 
 
 def test_highpass():
-    run(1024, lambda x: apply_filter(b=[0.9], a=[1.0, 0.1], x=x))
+    run(1024, lambda x: sigproc.lfilter(b=[0.9], a=[1.0, 0.1], x=x))
 
 
 def test_small():

tests/test_sigproc.py

@@ -26,12 +26,10 @@ def test_linreg():
 
 
 def test_filter():
-    f = sigproc.Filter(b=[1], a=[1])
     x = range(10)
-    y = list(f(x))
-    assert [float(i) for i in x] == y
+    y = sigproc.lfilter(b=[1], a=[1], x=x)
+    assert (np.array(x) == y).all()
 
-    f = sigproc.Filter(b=[0.5], a=[1, -0.5])
     x = [1] + [0] * 10
-    y = list(f(x))
-    assert y == [0.5 ** (i+1) for i in range(len(x))]
+    y = sigproc.lfilter(b=[0.5], a=[1, -0.5], x=x)
+    assert list(y) == [0.5 ** (i+1) for i in range(len(x))]