all good

pipeline/download/test_places.py

@@ -1,6 +1,9 @@
 import polars as pl
+from pyproj import Transformer
 
 from pipeline.download.places import (
+    _assign_london_display_city,
+    _display_city_from_tags,
     _is_dlr_station,
     _is_tram_station,
     _naptan_dlr_stations,
@@ -9,6 +12,22 @@ from pipeline.download.places import (
     _station_display_name,
 )
 
+WGS84_TO_BNG = Transformer.from_crs("EPSG:4326", "EPSG:27700", always_xy=True)
+
+
+def _postcode_row(postcode: str, lat: float, lon: float, *, london: bool) -> dict:
+    easting, northing = WGS84_TO_BNG.transform(lon, lat)
+    return {
+        "pcds": postcode,
+        "doterm": None,
+        "ctry25cd": "E92000001",
+        "east1m": int(round(easting)),
+        "north1m": int(round(northing)),
+        "rgn25cd": "E12000007" if london else "E12000008",
+        "lad25cd": "E09000008" if london else "E07000208",
+        "cty25cd": "E13000002" if london else "E10000030",
+    }
+
 
 def test_dlr_light_rail_is_not_treated_as_tram():
     dlr_tags = {
@@ -144,5 +163,56 @@ def test_ofs_universities_extracts_university_title_rows_with_postcode_coords():
             "lon": -1.2643,
             "population": 0,
             "travel_destination": True,
+            "display_city": None,
         }
     ]
+
+
+def test_display_city_from_tags_uses_explicit_london_context():
+    assert _display_city_from_tags({"is_in": "Croydon, London, UK"}) == "London"
+    assert _display_city_from_tags({"is_in": "Croydon, Cambridgeshire, UK"}) is None
+
+
+def test_assign_london_display_city_uses_nearest_active_postcode_admin(tmp_path):
+    postcodes = tmp_path / "postcodes.parquet"
+    pl.DataFrame(
+        [
+            _postcode_row("CR0 1SZ", 51.371273, -0.101793, london=True),
+            _postcode_row("KT19 8AG", 51.3326, -0.2678, london=False),
+        ]
+    ).write_parquet(postcodes)
+
+    places = [
+        {
+            "name": "Croydon",
+            "place_type": "town",
+            "lat": 51.3713049,
+            "lon": -0.101957,
+            "population": 173314,
+            "travel_destination": False,
+            "display_city": None,
+        },
+        {
+            "name": "East Croydon railway station",
+            "place_type": "station",
+            "lat": 51.375845,
+            "lon": -0.092732,
+            "population": 0,
+            "travel_destination": True,
+            "display_city": None,
+        },
+        {
+            "name": "Epsom",
+            "place_type": "town",
+            "lat": 51.3326,
+            "lon": -0.2678,
+            "population": 31489,
+            "travel_destination": False,
+            "display_city": None,
+        },
+    ]
+
+    assigned = _assign_london_display_city(places, postcodes)
+
+    assert assigned == 2
+    assert [place["display_city"] for place in places] == ["London", "London", None]