test for exclusion mask (#104)

* test for exclusion mask * add edge cases for exclusion testing * fix edge case dict key * sorted geometries for exclusion -- DF index varied across OS * try sorted geometries apalachicola standard -- DF index varied across OS * reup apalachicola edge cases -- #106 * add none clause
uscuni · Nov 24, 2024 · 2a5d6eb · 2a5d6eb
1 parent ba343d0
commit 2a5d6eb
Show file tree

Hide file tree

Showing 4 changed files with 54 additions and 42 deletions.
diff --git a/sgeop/tests/conftest.py b/sgeop/tests/conftest.py
@@ -24,6 +24,8 @@
 
 ####################################################
 # see:
+#   - gh#106
+#   - gh#102
 #   - gh#77
 #   - gh#75
 #   - gh#74
@@ -34,38 +36,8 @@
     "douala_809": [],
     "liege_1656": [921],
     "slc_4881": [1144, 1146],
-    "apalachicola": [
-        468,
-        469,
-        470,
-        471,
-        472,
-        473,
-        474,
-        475,
-        476,
-        477,
-        478,
-        479,
-        480,
-        481,
-        482,
-        494,
-        495,
-        496,
-        497,
-        498,
-        499,
-        500,
-        501,
-        502,
-        503,
-        504,
-        505,
-        506,
-        507,
-        508,
-    ],
+    "apalachicola_standard": [324],
+    "apalachicola_exclusion_mask": [],
 }
 ####################################################
 
@@ -105,6 +77,12 @@ def geom_test(
     geoms1 = collection1.geometry.normalize()
     geoms2 = collection2.geometry.normalize()
 
+    if aoi and aoi.startswith("apalachicola"):
+        # Varied index order across OSs.
+        # See [https://github.com/uscuni/sgeop/pull/104#issuecomment-2495572388]
+        geoms1 = geoms1.sort_values().reset_index(drop=True)
+        geoms2 = geoms2.sort_values().reset_index(drop=True)
+
     try:
         assert shapely.equals_exact(geoms1, geoms2, tolerance=tolerance).all()
     except AssertionError:
@@ -120,7 +98,8 @@ def geom_test(
                     "n_coords": {
                         "g1": shapely.get_coordinates(g1).shape[0],
                         "g2": shapely.get_coordinates(g2).shape[0],
-                    }
+                    },
+                    "length": {"g1": g1.length, "g2": g2.length},
                 }
         if unexpected_bad:
             raise AssertionError(

diff --git a/sgeop/tests/data/apalachicola_simplified_exclusion_mask.parquet b/sgeop/tests/data/apalachicola_simplified_exclusion_mask.parquet
diff --git a/...ests/data/apalachicola_simplified.parquet → .../apalachicola_simplified_standard.parquet b/...ests/data/apalachicola_simplified.parquet → .../apalachicola_simplified_standard.parquet
diff --git a/sgeop/tests/test_simplify.py b/sgeop/tests/test_simplify.py
@@ -3,6 +3,7 @@
 import geopandas
 import numpy
 import pytest
+import shapely
 from pandas.testing import assert_series_equal
 
 import sgeop
@@ -13,28 +14,60 @@
 ci_artifacts = pathlib.Path("ci_artifacts")
 
 
-def test_simplify_network_small():
+@pytest.mark.parametrize(
+    "scenario,tol,known_length",
+    [
+        ("standard", 1.5, 64566.0),
+        ("exclusion_mask", 1.05, 65765.0),
+    ],
+)
+def test_simplify_network_small(scenario, tol, known_length):
     ac = "apalachicola"
-    known = geopandas.read_parquet(test_data / f"{ac}_simplified.parquet")
-    known_length = 64566.0
 
-    observed = sgeop.simplify_network(
-        geopandas.read_parquet(test_data / f"{ac}_original.parquet")
-    )
+    original = geopandas.read_parquet(test_data / f"{ac}_original.parquet")
+
+    known = geopandas.read_parquet(test_data / f"{ac}_simplified_{scenario}.parquet")
+
+    if scenario == "exclusion_mask":
+        exclusion_mask = [
+            shapely.Polygon(
+                (
+                    (-9461361.807208396, 3469029.2708674935),
+                    (-9461009.046874022, 3469029.2708674935),
+                    (-9461009.046874022, 3469240.1785251377),
+                    (-9461361.807208396, 3469240.1785251377),
+                    (-9461361.807208396, 3469029.2708674935),
+                )
+            ),
+            shapely.Polygon(
+                (
+                    (-9461429.266819818, 3469157.7482423405),
+                    (-9461361.807208396, 3469157.7482423405),
+                    (-9461361.807208396, 3469240.1785251377),
+                    (-9461429.266819818, 3469240.1785251377),
+                    (-9461429.266819818, 3469157.7482423405),
+                )
+            ),
+        ]
+        exclusion_mask = geopandas.GeoSeries(exclusion_mask, crs=original.crs)
+    else:
+        exclusion_mask = None
+
+    observed = sgeop.simplify_network(original, exclusion_mask=exclusion_mask)
     observed_length = observed.geometry.length.sum()
 
     # storing GH artifacts
     artifact_dir = ci_artifacts / ac
     artifact_dir.mkdir(parents=True, exist_ok=True)
-    observed.to_parquet(artifact_dir / "simplified.parquet")
+    observed.to_parquet(artifact_dir / f"simplified_{scenario}.parquet")
 
-    assert pytest.approx(observed_length, rel=0.00001) == known_length
+    assert pytest.approx(observed_length, rel=0.0001) == known_length
     assert observed.index.dtype == numpy.dtype("int64")
 
     assert observed.shape == known.shape
     assert_series_equal(known._status, observed._status)
 
-    pytest.geom_test(known, observed, tolerance=1.5, aoi=ac)
+    pytest.geom_test(known, observed, tolerance=tol, aoi=f"{ac}_{scenario}")
 
 
 @pytest.mark.parametrize(