Filter labels created after dated imagery

2026-07-27 03:10:04 +02:00
parent a31dce325f
commit 52c4c4e709
6 changed files with 98 additions and 1 deletions
@@ -88,3 +88,47 @@ def test_spatial_leakage_audit_fails_cross_split_neighbors() -> None:
    assert audit["status"] == "failed"
    assert audit["findings"][0]["left"] == "train-a"
    assert audit["findings"][0]["right"] == "val-a"
+
+
+def test_normalizer_rejects_features_created_after_dated_imagery(tmp_path: Path) -> None:
+    raster_path = tmp_path / "image.tif"
+    with rasterio.open(
+        raster_path,
+        "w",
+        driver="GTiff",
+        width=100,
+        height=100,
+        count=3,
+        dtype="uint8",
+        crs="EPSG:4326",
+        transform=from_origin(4.0, 51.0, 0.001, 0.001),
+    ) as dataset:
+        dataset.write(np.zeros((3, 100, 100), dtype="uint8"))
+    geometry = {
+        "type": "Polygon",
+        "coordinates": [[[4.01, 50.99], [4.02, 50.99], [4.02, 50.98], [4.01, 50.98], [4.01, 50.99]]],
+    }
+    reference_path = tmp_path / "reference.geojson"
+    reference_path.write_text(
+        json.dumps(
+            {
+                "type": "FeatureCollection",
+                "features": [
+                    {"type": "Feature", "id": "old", "properties": {"BEGINDATUM": "2024-01-01"}, "geometry": geometry},
+                    {"type": "Feature", "id": "new", "properties": {"BEGINDATUM": "2026-01-01"}, "geometry": geometry},
+                ],
+            }
+        ),
+        encoding="utf-8",
+    )
+    normalized, audit = module.normalize(
+        reference_path=reference_path,
+        raster_path=raster_path,
+        source_name="grb",
+        min_label_px=3,
+        imagery_observed_at="2025-01-01T00:00:00Z",
+        imagery_valid_to="2025-12-31T23:59:59Z",
+        reference_observed_at="2026-07-01T00:00:00Z",
+    )
+    assert len(normalized["features"]) == 1
+    assert audit["decision_counts"] == {"accepted": 1, "created_after_imagery_period": 1}