add support for mode median

Author: Vahid Tavanashad

dpnp/dpnp_iface_manipulation.py

@@ -1874,6 +1874,9 @@ def pad(array, pad_width, mode="constant", **kwargs):
         "mean"
             Pads with the mean value of all or part of the
             vector along each axis.
+        "median"
+            Pads with the median value of all or part of the
+            vector along each axis.
         "minimum"
             Pads with the minimum value of all or part of the
             vector along each axis.
@@ -1894,10 +1897,10 @@ def pad(array, pad_width, mode="constant", **kwargs):
             Padding function, see Notes.
         Default: ``"constant"``.
     stat_length : {None, int, sequence of ints}, optional
-        Used in ``"maximum"``, ``"mean"``, and ``"minimum"``. Number of
-        values at edge of each axis used to calculate the statistic value.
-        ``((before_1, after_1), ... (before_N, after_N))`` unique statistic
-        lengths for each axis.
+        Used in ``"maximum"``, ``"mean"``, ``"median"``, and ``"minimum"``.
+        Number of values at edge of each axis used to calculate the statistic
+        value. ``((before_1, after_1), ... (before_N, after_N))`` unique
+        statistic lengths for each axis.
         ``(before, after)`` or ``((before, after),)`` yields same before
         and after statistic lengths for each axis.
         ``(stat_length,)`` or ``int`` is a shortcut for
@@ -1936,11 +1939,6 @@ def pad(array, pad_width, mode="constant", **kwargs):
         Padded array of rank equal to `array` with shape increased
         according to `pad_width`.
 
-    Limitations
-    -----------
-    Parameter `mode` as ``"median"`` is not currently supported and
-    ``NotImplementedError`` exception will be raised.
-
     Notes
     -----
     For an array with rank greater than 1, some of the padding of later
@@ -1988,7 +1986,7 @@ def pad(array, pad_width, mode="constant", **kwargs):
     array([3, 3, 1, 2, 3, 4, 5, 3, 3])
 
     >>> np.pad(a, (2,), 'median')
-    NotImplementedError: Keyword argument `mode` does not support 'median'
+    array([3, 3, 1, 2, 3, 4, 5, 3, 3])
 
     >>> a = np.array([[1, 2], [3, 4]])
     >>> np.pad(a, ((3, 2), (2, 3)), 'minimum')
@@ -2040,10 +2038,6 @@ def pad(array, pad_width, mode="constant", **kwargs):
     """
 
     dpnp.check_supported_arrays_type(array)
-    if mode == "median":
-        raise NotImplementedError(
-            "Keyword argument `mode` does not support 'median'"
-        )
     return dpnp_pad(array, pad_width, mode=mode, **kwargs)
 
 

dpnp/dpnp_utils/dpnp_utils_pad.py

@@ -659,7 +659,7 @@ def dpnp_pad(array, pad_width, mode="constant", **kwargs):
         "linear_ramp": ["end_values"],
         "maximum": ["stat_length"],
         "mean": ["stat_length"],
-        # "median": ["stat_length"],  # TODO: dpnp.median is not implemented
+        "median": ["stat_length"],
         "minimum": ["stat_length"],
         "reflect": ["reflect_type"],
         "symmetric": ["reflect_type"],
@@ -684,11 +684,11 @@ def dpnp_pad(array, pad_width, mode="constant", **kwargs):
             # faster path for 1d arrays or small n-dimensional arrays
             return _pad_simple(array, pad_width, 0)[0]
 
-    # TODO: add "median": dpnp.median when dpnp.median is implemented
     stat_functions = {
         "maximum": dpnp.amax,
         "minimum": dpnp.amin,
         "mean": dpnp.mean,
+        "median": dpnp.median,
     }
 
     # Create array with final shape and original values

tests/test_arraypad.py

@@ -22,6 +22,7 @@ class TestPad:
         "linear_ramp": {"end_values": 0},
         "maximum": {"stat_length": None},
         "mean": {"stat_length": None},
+        "median": {"stat_length": None},
         "minimum": {"stat_length": None},
         "reflect": {"reflect_type": "even"},
         "symmetric": {"reflect_type": "even"},
@@ -303,7 +304,7 @@ def test_linear_ramp_negative_diff(self, dtype, data, end_values):
         assert_array_equal(result, expected)
 
     @pytest.mark.parametrize("pad_width", [5, (25, 20)])
-    @pytest.mark.parametrize("mode", ["maximum", "minimum", "mean"])
+    @pytest.mark.parametrize("mode", ["maximum", "minimum", "mean", "median"])
     @pytest.mark.parametrize("stat_length", [10, (2, 3)])
     def test_stat_func_1d(self, pad_width, mode, stat_length):
         a_np = numpy.arange(100)
@@ -315,7 +316,7 @@ def test_stat_func_1d(self, pad_width, mode, stat_length):
         assert_array_equal(result, expected)
 
     @pytest.mark.parametrize("pad_width", [((1,), (2,)), ((2, 3), (3, 2))])
-    @pytest.mark.parametrize("mode", ["maximum", "minimum", "mean"])
+    @pytest.mark.parametrize("mode", ["maximum", "minimum", "mean", "median"])
     @pytest.mark.parametrize("stat_length", [(3,), (2, 3)])
     def test_stat_func_2d(self, pad_width, mode, stat_length):
         a_np = numpy.arange(30).reshape(6, 5)
@@ -326,7 +327,7 @@ def test_stat_func_2d(self, pad_width, mode, stat_length):
         result = dpnp.pad(a_dp, pad_width, mode=mode, stat_length=stat_length)
         assert_array_equal(result, expected)
 
-    @pytest.mark.parametrize("mode", ["mean", "minimum", "maximum"])
+    @pytest.mark.parametrize("mode", ["mean", "minimum", "maximum", "median"])
     def test_same_prepend_append(self, mode):
         """Test that appended and prepended values are equal"""
         a = dpnp.array([-1, 2, -1]) + dpnp.array(
@@ -335,14 +336,15 @@ def test_same_prepend_append(self, mode):
         result = dpnp.pad(a, (1, 1), mode)
         assert_equal(result[0], result[-1])
 
-    def test_mean_with_zero_stat_length(self):
+    @pytest.mark.parametrize("mode", ["mean", "median"])
+    def test_zero_stat_length_valid(self):
         a_np = numpy.array([1.0, 2.0])
         a_dp = dpnp.array(a_np)
-        expected = numpy.pad(a_np, (1, 2), "mean")
-        result = dpnp.pad(a_dp, (1, 2), "mean")
+        expected = numpy.pad(a_np, (1, 2), mode, stat_length=0)
+        result = dpnp.pad(a_dp, (1, 2), mode, stat_length=0)
         assert_array_equal(result, expected)
 
-    @pytest.mark.parametrize("mode", ["mean", "minimum", "maximum"])
+    @pytest.mark.parametrize("mode", ["mean", "minimum", "maximum", "median"])
     @pytest.mark.parametrize(
         "stat_length", [-2, (-2,), (3, -1), ((5, 2), (-2, 3)), ((-4,), (2,))]
     )