Better merge

pipeline/transform/test_merge.py

@@ -1129,6 +1129,70 @@ def test_match_direct_epc_street_fallback_recovers_numbered_listing() -> None:
     assert matches["_direct_epc_match_method"].to_list() == ["street"]
 
 
+def test_match_direct_epc_street_fallback_nulls_year_on_mixed_era_street() -> None:
+    # A number-less listing is matched to the *street*, never the exact property.
+    # When the same-street certificates span construction eras (a Victorian house
+    # and 2007 infill here), no single year represents the unidentified property,
+    # so the street-representative construction year is nulled rather than letting
+    # a new-build inherit an 1890 build date (the Enderby Street, Greenwich bug).
+    # Other EPC facts stay as street-representative estimates.
+    matches = _match_direct_epc(
+        _listing_matches([{"_listing_match_address": "EXAMPLE ROAD BROMLEY"}]),
+        _direct_epc_candidates(
+            [
+                {
+                    "_direct_epc_match_address": "8 EXAMPLE ROAD",
+                    "_direct_epc_address": "8, Example Road",
+                    "_direct_construction_age_band": 1890,
+                },
+                {
+                    "_direct_epc_row": 1,
+                    "_direct_epc_match_address": "15 EXAMPLE ROAD",
+                    "_direct_epc_address": "15, Example Road",
+                    "_direct_construction_age_band": 2007,
+                },
+            ]
+        ),
+    )
+
+    assert matches.height == 1
+    assert matches["_direct_epc_match_method"].to_list() == ["street"]
+    assert matches["_direct_construction_age_band"].to_list() == [None]
+    assert matches["_direct_is_construction_date_approximate"].to_list() == [None]
+    # Energy rating remains a street-representative estimate (only the year is
+    # categorically wrong on a mixed street; the rest stay best-effort).
+    assert matches["_direct_current_energy_rating"].to_list() == ["C"]
+
+
+def test_match_direct_epc_street_fallback_keeps_year_on_uniform_street() -> None:
+    # When the street is era-homogeneous (one development), the
+    # street-representative construction year IS meaningful and is kept. Two
+    # adjacent EPC bands a few years apart count as uniform.
+    matches = _match_direct_epc(
+        _listing_matches([{"_listing_match_address": "EXAMPLE ROAD BROMLEY"}]),
+        _direct_epc_candidates(
+            [
+                {
+                    "_direct_epc_match_address": "8 EXAMPLE ROAD",
+                    "_direct_epc_address": "8, Example Road",
+                    "_direct_construction_age_band": 2007,
+                },
+                {
+                    "_direct_epc_row": 1,
+                    "_direct_epc_match_address": "15 EXAMPLE ROAD",
+                    "_direct_epc_address": "15, Example Road",
+                    "_direct_construction_age_band": 2009,
+                },
+            ]
+        ),
+    )
+
+    assert matches.height == 1
+    assert matches["_direct_epc_match_method"].to_list() == ["street"]
+    assert matches["_direct_construction_age_band"].to_list() == [2007]
+    assert matches["_direct_is_construction_date_approximate"].to_list() == [1]
+
+
 def test_match_direct_epc_street_fallback_rejects_town_only_address() -> None:
     # A town-only listing address ("COULSDON SURREY") shares only the locality
     # suffix that most street keys in the outcode carry; without a street-name