lgtm

pipeline/transform/test_merge.py

@@ -14,6 +14,7 @@ from pipeline.transform.merge import (
     _canonical_postcode_expr,
     _best_listing_match,
     _coalesce_direct_epc_columns,
+    _epc_council_by_postcode,
     _fill_property_level_no_defaults,
     _join_area_side_tables,
     _finalize_listings,
@@ -117,6 +118,86 @@ def test_tree_density_is_area_level_and_survives_the_split() -> None:
     assert TREE_DENSITY_FEATURE not in properties_df.columns
 
 
+def test_epc_council_by_postcode_aggregates_social_tenure_shares() -> None:
+    # 4 dwellings in one postcode, 2 ever council (was_council_house "Yes"). One
+    # of the two is STILL social (latest_tenure_status "Rented (social)"); the
+    # other has been sold off (latest cert owner-occupied) and is ex-council.
+    # % Council housing = 2/4 = 50.0; % Ex-council = 1/4 = 25.0.
+    wide = pl.LazyFrame(
+        {
+            "postcode": ["AB1 2CD", "AB1 2CD", "AB1 2CD", "AB1 2CD"],
+            "was_council_house": ["Yes", "Yes", "No", "No"],
+            "latest_tenure_status": [
+                "Rented (social)",  # still social -> not ex-council
+                "Owner-occupied",  # sold off -> ex-council
+                "Owner-occupied",
+                None,  # never council; a null latest tenure is harmless here
+            ],
+        }
+    )
+
+    result = _epc_council_by_postcode(wide).collect()
+
+    assert result.to_dicts() == [
+        {"postcode": "AB1 2CD", "% Council housing": 50.0, "% Ex-council": 25.0}
+    ]
+
+
+def test_epc_council_by_postcode_null_latest_tenure_counts_as_ex_council() -> None:
+    # An ever-council dwelling whose latest certificate has no recognised tenure
+    # (latest_tenure_status null) is treated as no-longer-social, so it counts
+    # toward % Ex-council (null currently_social -> False, per the spec).
+    wide = pl.LazyFrame(
+        {
+            "postcode": ["AB1 2CD", "AB1 2CD"],
+            "was_council_house": ["Yes", "No"],
+            "latest_tenure_status": [None, None],
+        }
+    )
+
+    result = _epc_council_by_postcode(wide).collect()
+
+    assert result.to_dicts() == [
+        {"postcode": "AB1 2CD", "% Council housing": 50.0, "% Ex-council": 50.0}
+    ]
+
+
+def test_epc_council_columns_are_area_level_and_survive_the_split() -> None:
+    # The EPC council shares are postcode-level AREA columns: they must route to
+    # the postcode output and NOT appear in the property output. The per-property
+    # "Former council house" flag is unaffected and stays property-level.
+    assert "% Council housing" in _AREA_COLUMNS
+    assert "% Ex-council" in _AREA_COLUMNS
+
+    df = pl.DataFrame(
+        {
+            "Postcode": ["AA1 1AA"],
+            "Last known price": [250_000],
+            "Former council house": ["Yes"],
+        }
+    )
+    postcode_features = pl.DataFrame(
+        {
+            "Postcode": ["AA1 1AA", "BB1 1BB"],
+            "lat": [51.0, 52.0],
+            "lon": [-0.1, -0.2],
+            "ctry25cd": ["E92000001", "E92000001"],
+            "% Council housing": [50.0, 0.0],
+            "% Ex-council": [25.0, 0.0],
+        }
+    )
+
+    postcode_df, properties_df = _split_normal_outputs(
+        df, postcode_features, expected_postcode_count=2
+    )
+
+    assert postcode_df["% Council housing"].to_list() == [50.0, 0.0]
+    assert postcode_df["% Ex-council"].to_list() == [25.0, 0.0]
+    assert "% Council housing" not in properties_df.columns
+    assert "% Ex-council" not in properties_df.columns
+    assert "Former council house" in properties_df.columns
+
+
 def test_crime_columns_are_average_annual_counts() -> None:
     # Crime is the average annual recorded incident count (incidents/yr) over
     # 7-year and 2-year windows; the old per-1,000 "(per 1k/yr, …)" rate columns
@@ -955,7 +1036,7 @@ def test_match_direct_epc_matches_by_uprn_across_postcodes() -> None:
 
 def test_match_direct_epc_ignores_nonunique_building_uprn() -> None:
     # A parent/building UPRN that resolves to several distinct (address,
-    # postcode) flats cannot act as a 1:1 exact-match key — it would mis-link
+    # postcode) flats cannot act as a 1:1 exact-match key: it would mis-link
     # the listing to one arbitrary flat. The listing must fall through to the
     # street-address matcher, which resolves the specific flat.
     matches = _match_direct_epc(
@@ -1490,7 +1571,7 @@ def test_match_listing_properties_uprn_wins_dedup_tie() -> None:
 
 
 def test_match_direct_epc_does_not_match_other_outcode_without_uprn() -> None:
-    # Matching is by postcode/UPRN/street — never by coordinate proximity — and
+    # Matching is by postcode/UPRN/street (never by coordinate proximity), and
     # the street fallback is outcode-scoped, so a same-street EPC in a different
     # OUTCODE with no shared UPRN is skipped.
     matches = _match_direct_epc(
@@ -1909,7 +1990,7 @@ def test_finalize_listings_dedupes_fanned_out_listing_rows() -> None:
             "Leasehold/Freehold": ["Leasehold", "Leasehold"],
             "Last known price": [500_000, 480_000],
             "Tree canopy density percentile": [42.0, 42.0],
-            # Same listing URL on both collapsed rows — the fan-out to fix.
+            # Same listing URL on both collapsed rows: the fan-out to fix.
             "_actual_listing_url": ["url0", "url0"],
             "_actual_asking_price": [600_000, 600_000],
             "_actual_asking_price_per_sqm": [5_000, 5_000],