scraping and data

finder/test_transform.py

@@ -1,13 +1,19 @@
 from transform import (
+    build_register_address,
     clean_listing_address,
     extract_full_postcode,
+    extract_outcode,
+    resolve_listing_postcode,
     transform_property,
 )
 
 
 class StubPostcodeIndex:
+    def __init__(self, postcode: str = "SW1A 9ZZ") -> None:
+        self._postcode = postcode
+
     def nearest(self, lat: float, lng: float) -> str:
-        return "SW1A 9ZZ"
+        return self._postcode
 
 
 def test_extract_full_postcode_normalizes_spacing() -> None:
@@ -24,6 +30,46 @@ def test_clean_listing_address_removes_postcode_and_outcode_suffixes() -> None:
     assert clean_listing_address("Kings Avenue, Bromley") == "Kings Avenue, Bromley"
 
 
+def test_build_register_address_prepends_house_number_or_name() -> None:
+    # House number/name prepended, with the trailing outcode/postcode stripped.
+    assert (
+        build_register_address("South Street, Bromley BR1", "12")
+        == "12, South Street, Bromley"
+    )
+    assert (
+        build_register_address("Riverside, Martham NR29", "Martham Mill")
+        == "Martham Mill, Riverside, Martham"
+    )
+    # No number/name -> identical to the plain cleaned address.
+    assert build_register_address("Kings Avenue, Bromley", None) == "Kings Avenue, Bromley"
+    # Already starts with the number/name -> no duplication.
+    assert (
+        build_register_address("12 South Street, Bromley", "12")
+        == "12 South Street, Bromley"
+    )
+    # Empty/whitespace number/name is ignored.
+    assert build_register_address("Kings Avenue, Bromley", "  ") == "Kings Avenue, Bromley"
+
+
+def test_extract_outcode() -> None:
+    assert extract_outcode("SW1A 2AA") == "SW1A"
+    assert extract_outcode("n4 2ha") == "N4"
+    assert extract_outcode("SW1A2AA") == "SW1A"
+    assert extract_outcode(None) is None
+    assert extract_outcode("") is None
+
+
+def test_resolve_listing_postcode() -> None:
+    # Outcode matches -> trust the more precise extracted postcode.
+    assert resolve_listing_postcode("SW1A 2AA", "SW1A 9ZZ") == ("SW1A 2AA", "address")
+    # Outcode mismatch -> fall back to the spatially-correct inferred postcode.
+    assert resolve_listing_postcode("E14 9SS", "SW1A 9ZZ") == ("SW1A 9ZZ", "coordinates")
+    # Well-formed but fabricated postcode in a different outcode is rejected.
+    assert resolve_listing_postcode("ZZ9 9ZZ", "SW1A 9ZZ") == ("SW1A 9ZZ", "coordinates")
+    # No extracted postcode -> inferred is authoritative.
+    assert resolve_listing_postcode(None, "SW1A 9ZZ") == ("SW1A 9ZZ", "coordinates")
+
+
 def test_rightmove_transform_prefers_postcode_from_display_address() -> None:
     prop = {
         "id": "123",
@@ -46,3 +92,84 @@ def test_rightmove_transform_prefers_postcode_from_display_address() -> None:
     assert result["Inferred postcode"] == "SW1A 9ZZ"
     assert result["Listing raw address"] == "Flat 2, 10 Downing Street, SW1A 2AA"
     assert result["Address per Property Register"] == "Flat 2, 10 Downing Street"
+
+
+def test_rightmove_transform_rejects_postcode_from_wrong_outcode() -> None:
+    prop = {
+        "id": "124",
+        "location": {"latitude": 51.5, "longitude": -0.1},
+        "price": {"amount": 750000, "displayPrices": []},
+        "propertySubType": "Terraced",
+        "bedrooms": 3,
+        "bathrooms": 1,
+        "keyFeatures": [],
+        "propertyUrl": "/properties/124",
+        # Address postcode is in a different outcode than the coordinate-nearest one.
+        "displayAddress": "10 Downing Street, E14 9SS",
+    }
+
+    result = transform_property(prop, "SW1A", StubPostcodeIndex())
+
+    assert result is not None
+    # The spatially-correct inferred postcode wins over the mismatching extracted one.
+    assert result["Postcode"] == "SW1A 9ZZ"
+    assert result["Postcode source"] == "coordinates"
+    assert result["Extracted postcode"] == "E14 9SS"
+
+
+def _rightmove_prop() -> dict:
+    return {
+        "id": "200",
+        "location": {"latitude": 51.5, "longitude": -0.1},
+        "price": {"amount": 750000, "displayPrices": []},
+        "propertySubType": "Terraced",
+        "bedrooms": 3,
+        "bathrooms": 1,
+        "keyFeatures": [],
+        "propertyUrl": "/properties/200",
+        # Search API only ever exposes the outcode in the display address.
+        "displayAddress": "Caldwell Street, Stockwell, SW9",
+    }
+
+
+def test_rightmove_transform_prefers_detail_postcode() -> None:
+    # The detail page's true full postcode (same outcode as the location) is
+    # preferred over the coordinate-nearest guess.
+    result = transform_property(
+        _rightmove_prop(),
+        "SW9",
+        StubPostcodeIndex("SW9 7AA"),
+        detail_postcode="SW9 0HD",
+    )
+
+    assert result is not None
+    assert result["Postcode"] == "SW9 0HD"
+    assert result["Postcode source"] == "detail_address"
+    # The coordinate inference is still surfaced separately.
+    assert result["Inferred postcode"] == "SW9 7AA"
+
+
+def test_rightmove_transform_rejects_detail_postcode_from_wrong_outcode() -> None:
+    # A detail postcode whose outcode disagrees with the location must not
+    # relocate the listing; the coordinate postcode wins instead.
+    result = transform_property(
+        _rightmove_prop(),
+        "SW9",
+        StubPostcodeIndex("SW9 7AA"),
+        detail_postcode="E14 9SS",
+    )
+
+    assert result is not None
+    assert result["Postcode"] == "SW9 7AA"
+    assert result["Postcode source"] == "coordinates"
+
+
+def test_rightmove_transform_without_detail_keeps_coordinate_logic() -> None:
+    # No detail postcode -> behaviour is unchanged (coordinate-nearest).
+    result = transform_property(
+        _rightmove_prop(), "SW9", StubPostcodeIndex("SW9 7AA")
+    )
+
+    assert result is not None
+    assert result["Postcode"] == "SW9 7AA"
+    assert result["Postcode source"] == "coordinates"