def _create_features(self):
return [
FeatureF(),
FeatureF([1, 1], 1, 2, 1),
FeatureD(),
FeatureD([1, 1], 1, 2, 1),
]
def test_get_default_color(self):
dflt_color = RGBColor()
f = FeatureD()
nt.assert_equal(f.color, dflt_color)
c = RGBColor(5, 32, 10)
f = FeatureD([1, 1], rgb_color=c)
nt.assert_equal(f.color, c)
def test_get_default_location(self):
f = FeatureD()
np.testing.assert_almost_equal(f.location, [0, 0])
expected = [12.3, 643]
f = FeatureD(loc=expected)
np.testing.assert_almost_equal(f.location, expected)
# iterable form
f = FeatureD(loc=(12.3, 643))
np.testing.assert_almost_equal(f.location, expected)
def test_set_and_get_color(self):
expected = RGBColor(4, 20, 0)
f = FeatureD()
f.color = expected
nt.assert_equal(f.color, expected)
f = FeatureF()
f.color = expected
nt.assert_equal(f.color, expected)
def test_set_and_get_angle(self):
f = FeatureD()
expected = random.random()
f.angle = expected
nt.assert_almost_equal(f.angle, expected, 16)
f = FeatureF()
expected = random.random()
f.angle = expected
nt.assert_almost_equal(f.angle, expected, 6)
def test_set_and_get_location(self):
f = FeatureD()
expected = [random.random(), random.random()]
f.location = expected
# making sure that we went through the setter, and not just setting the
# exact value to the property
np.testing.assert_array_almost_equal(f.location, expected, 16)
f = FeatureF()
expected = [random.random(), random.random()]
f.location = expected
np.testing.assert_array_almost_equal(f.location, expected, 6)
def test_clone(self):
features = [
FeatureD(),
FeatureD([1, 1], 1, 2, 1),
FeatureF(),
FeatureF([1, 1], 1, 2, 1),
]
for f in features:
f_clone = f.clone()
nt.ok_(f == f_clone)
# Changing one doesn't reflect in the other
f_clone.scale = f.scale + 1
nt.ok_(f != f_clone)
def test_to_str(self):
expected_loc = np.array([random.random(), random.random()])
expected_color = RGBColor(50, 60, 70)
features = [
FeatureD(expected_loc, 1.5, 2.5, 1.7, expected_color),
FeatureF(expected_loc, 1.5, 2.5, 1.7, expected_color),
]
for f in features:
s = str(f).split()
# Notice 6 is used here, as that is the
# default precision used for C++ stringstreams
np.testing.assert_almost_equal(float(s[0]), expected_loc[0],
6) # loc[0]
np.testing.assert_almost_equal(float(s[1]), expected_loc[1],
6) # loc[1]
np.testing.assert_almost_equal(float(s[2]), 1.5) # mag
np.testing.assert_almost_equal(float(s[3]), 2.5) # scale
np.testing.assert_almost_equal(float(s[4]), 1.7) # angle
nt.assert_equals(int(s[5]), expected_color.r) # color.r
nt.assert_equals(int(s[6]), expected_color.g) # color.g
nt.assert_equals(int(s[7]), expected_color.b) # color.b
print("Non default feature representation:", features[0], sep="\n")
def test_set_and_get_covar(self):
f = FeatureD()
expected = [[1, 2], [3, 4]]
c = Covar2d(expected)
f.covariance = c
np.testing.assert_array_almost_equal(f.covariance.matrix(), c.matrix(),
16)
# And for floats...
f = FeatureF()
c = Covar2f(expected)
f.covariance = c
np.testing.assert_array_almost_equal(f.covariance.matrix(), c.matrix(),
6)
def test_to_str_default(self):
features = [FeatureD(), FeatureF()]
for f in features:
s = str(f).split()
nt.assert_equals(s[0], "0") # loc[0]
nt.assert_equals(s[1], "0") # loc[1]
nt.assert_equals(s[2], "0") # mag
nt.assert_equals(s[3], "1") # scale
nt.assert_equals(s[4], "0") # angle
nt.assert_equals(s[5], "255") # color.r
nt.assert_equals(s[6], "255") # color.g
nt.assert_equals(s[7], "255") # color.b
print("Default feature representation:", features[0], sep="\n")
def test_copy_constructor(self):
double_def = FeatureD()
double_nondef = FeatureD([1, 1], 1, 2, 1)
float_def = FeatureF()
float_nondef = FeatureF([1, 1], 1, 2, 1)
nt.ok_(FeatureD(double_def) == double_def)
nt.ok_(FeatureD(double_nondef) == double_nondef)
nt.ok_(FeatureF(double_def) != double_def)
nt.ok_(FeatureF(double_nondef) != double_nondef)
nt.ok_(FeatureF(float_def) == float_def)
nt.ok_(FeatureF(float_nondef) == float_nondef)
nt.ok_(FeatureD(float_def) != float_def)
nt.ok_(FeatureD(float_nondef) != float_nondef)
def test_get_default_mag(self):
f = FeatureD()
nt.assert_equal(f.magnitude, 0)
f = FeatureD([1, 1], mag=1.1)
nt.assert_equal(f.magnitude, 1.1)
def test_get_default_scale(self):
f = FeatureD()
nt.assert_equal(f.scale, 1)
f = FeatureD([1, 1], scale=2.1)
nt.assert_equal(f.scale, 2.1)
def test_comparisons(self):
nt.assert_false(FeatureD() == FeatureF())
ctors = [FeatureD, FeatureF]
for ctor in ctors:
f1, f2 = ctor(), ctor()
self.comparison_helper(f1, f2, True, True)
# location
f2.location = f1.location + 1
self.comparison_helper(f1, f2)
f2.location = f1.location
self.comparison_helper(f1, f2, True, True)
# magnitude
f2.magnitude = f1.magnitude + 1
self.comparison_helper(f1, f2)
f2.magnitude = f1.magnitude
self.comparison_helper(f1, f2, True, True)
# scale
f2.scale = f1.scale + 1
self.comparison_helper(f1, f2)
f2.scale = f1.scale
self.comparison_helper(f1, f2, True, True)
# angle
f2.angle = f1.angle + 1
self.comparison_helper(f1, f2, eq_except_angle=True)
f2.angle = f1.angle
self.comparison_helper(f1, f2, True, True)
# covariance
f2.covariance[0, 0] = f1.covariance[0, 0] + 1
self.comparison_helper(f1, f2)
f2.covariance[0, 0] = f1.covariance[0, 0]
self.comparison_helper(f1, f2, True, True)
# color
f2.color = RGBColor(r=f1.color.r + 1)
self.comparison_helper(f1, f2)
f2.color = f1.color
self.comparison_helper(f1, f2, True, True)
# Try many at once
f2.location = f1.location + 1
self.comparison_helper(f1, f2)
f2.magnitude = f1.magnitude + 1
self.comparison_helper(f1, f2)
f2.scale = f1.scale + 1
self.comparison_helper(f1, f2)
f2.angle = f1.angle + 1
self.comparison_helper(f1, f2)
f2.covariance[0, 0] = f1.covariance[0, 0] + 1
self.comparison_helper(f1, f2)
f2.color = RGBColor(r=f1.color.r + 1)
self.comparison_helper(f1, f2)
def test_get_typename(self):
f = FeatureD()
nt.assert_equal(f.type_name, "d")
f = FeatureF()
nt.assert_equal(f.type_name, "f")
def test_new(self):
f1 = FeatureF([1, 1], 1, 2, 1)
f2 = FeatureF()
f3 = FeatureD([1, 1], 1, 2, 1)
f4 = FeatureD()
def test_get_default_angle(self):
f = FeatureD()
nt.assert_equal(f.angle, 0)
f = FeatureD([1, 1], angle=1.1)
nt.assert_equal(f.angle, 1.1)
def test_overriden_features(self):
nt.assert_equals(ConcreteFeatureSet().features(), [FeatureD([1, 2], 3, 4, 5)])
def test_get_default_covar(self):
dflt_covar = Covar2d()
f = FeatureD()
np.testing.assert_array_equal(f.covariance.matrix(),
dflt_covar.matrix())
def features(self):
return [FeatureD([1, 2], 3, 4, 5)]
