From ad63436dd508b4e41ac772d6addbbd294d057917 Mon Sep 17 00:00:00 2001
From: Brett <brettgraham@gmail.com>
Date: Tue, 9 Jul 2024 10:13:04 -0400
Subject: [PATCH] change back to old abs_deriv

---
 src/stcal/outlier_detection/utils.py  | 28 +++++++++++++++++++++++++--
 tests/outlier_detection/test_utils.py | 16 +++++++++++----
 2 files changed, 38 insertions(+), 6 deletions(-)

diff --git a/src/stcal/outlier_detection/utils.py b/src/stcal/outlier_detection/utils.py
index 638f8790d..036823f28 100644
--- a/src/stcal/outlier_detection/utils.py
+++ b/src/stcal/outlier_detection/utils.py
@@ -73,9 +73,33 @@ def compute_weight_threshold(weight, maskpt):
     return weight_threshold
 
 
-def _abs_deriv(array):
+def _valid_abs_deriv(array):
     """Take the absolute derivate of a numpy array."""
-    # TODO is there a more efficient way to do this?
+    out = np.zeros_like(array)  # use same dtype as input
+
+    # compute row-wise absolute diffference
+    d = np.abs(np.diff(array, axis=0))
+    out[1:] = d  # no need to do max yet
+    # since these are absolute differences |r0-r1| = |r1-r0|
+    # make a view of the target portion of the array
+    v = out[:-1]
+    # compute an in-place maximum
+    np.putmask(v, d > v, d)
+
+    # compute col-wise absolute difference
+    d = np.abs(np.diff(array, axis=1))
+    v = out[:, 1:]
+    np.putmask(v, d > v, d)
+    v = out[:, :-1]
+    np.putmask(v, d > v, d)
+    return out
+
+
+def _abs_deriv(array):
+    # FIXME this assumes off-edge pixels are 0
+    # FIXME this upcasts to float64
+    # FIXME _valid_abs_deriv fixes the above issues and is more efficient
+    # but fixing the bugs will likely change the output
     tmp = np.zeros(array.shape, dtype=np.float64)
     out = np.zeros(array.shape, dtype=np.float64)
 
diff --git a/tests/outlier_detection/test_utils.py b/tests/outlier_detection/test_utils.py
index c35222487..6e0302119 100644
--- a/tests/outlier_detection/test_utils.py
+++ b/tests/outlier_detection/test_utils.py
@@ -12,20 +12,28 @@
 
 
 @pytest.mark.parametrize("shape,diff", [
-    ([5, 5], 100),
-    ([7, 7], 200),
+    ([5, 7], 100),
+    ([17, 13], -200),
 ])
-def test_abs_deriv(shape, diff):
+def test_abs_deriv_single_value(shape, diff):
     arr = np.zeros(shape)
     # put diff at the center
     np.put(arr, arr.size // 2, diff)
     # since abs_deriv with a single non-zero value is the same as a
     # convolution with a 3x3 cross kernel use it to test the result
-    expected = scipy.signal.convolve2d(arr, [[0, 1, 0], [1, 1, 1], [0, 1, 0]], mode='same')
+    expected = scipy.signal.convolve2d(np.abs(arr), [[0, 1, 0], [1, 1, 1], [0, 1, 0]], mode='same')
     result = _abs_deriv(arr)
     np.testing.assert_allclose(result, expected)
 
 
+@pytest.mark.skip(reason="_abs_deriv has edge effects due to treating off-edge pixels as 0")
+@pytest.mark.parametrize("nrows,ncols", [(5, 5), (7, 11), (17, 13)])
+def test_abs_deriv_range(nrows, ncols):
+    arr = np.arange(nrows * ncols).reshape(nrows, ncols)
+    result = _abs_deriv(arr)
+    np.testing.assert_allclose(result, ncols)
+
+
 @pytest.mark.parametrize("shape,mean,maskpt,expected", [
     ([5, 5], 11, 0.5, 5.5),
     ([5, 5], 11, 0.25, 2.75),