idk2

pipeline/utils/test_postcode_mapping.py

@@ -0,0 +1,38 @@
+import polars as pl
+
+from pipeline.utils.postcode_mapping import (
+    MAX_REMAP_DISTANCE_M,
+    build_postcode_mapping,
+)
+
+
+def test_remap_drops_terminated_postcodes_beyond_distance_cap(tmp_path):
+    # One active England postcode at the origin.
+    # One terminated postcode 100m away (legitimate adjacent remap -> mapped).
+    # One terminated postcode 5km away (gross misattribution -> dropped).
+    arcgis = pl.DataFrame(
+        {
+            "pcds": ["AB1 1AA", "AB1 1AB", "ZZ9 9ZZ"],
+            "ctry25cd": ["E92000001", "E92000001", "E92000001"],
+            "doterm": [None, "202001", "202001"],
+            "east1m": [500000.0, 500100.0, 505000.0],
+            "north1m": [200000.0, 200000.0, 200000.0],
+        }
+    )
+    arcgis_path = tmp_path / "arcgis.parquet"
+    arcgis.write_parquet(arcgis_path)
+
+    mapping = build_postcode_mapping(arcgis_path)
+
+    # The nearby terminated postcode is remapped onto the active one.
+    assert mapping.filter(pl.col("old_postcode") == "AB1 1AB")[
+        "new_postcode"
+    ].to_list() == ["AB1 1AA"]
+
+    # The far (5km > 1km cap) terminated postcode must NOT appear in the mapping.
+    assert "ZZ9 9ZZ" not in mapping["old_postcode"].to_list()
+    assert mapping.height == 1
+
+
+def test_max_remap_distance_constant_is_one_kilometre():
+    assert MAX_REMAP_DISTANCE_M == 1000.0