added unit tests for floors

backend/onboarders/mappings/as_built_floor_classifiers.py

@@ -0,0 +1,46 @@
+from backend.onboarders.epc_descriptions import EpcConstructionAgeBand, EpcFloorDescriptions
+
+
+def unknown_floor_as_built(age_band: EpcConstructionAgeBand) -> EpcFloorDescriptions:
+    year = age_band.start_year()
+
+    if year >= 2003:
+        return EpcFloorDescriptions.solid_insulated_assumed
+
+    if year >= 1996:
+        return EpcFloorDescriptions.solid_limited_insulation_assumed
+
+    if year >= 1930:
+        return EpcFloorDescriptions.solid_no_insulation_assumed
+
+    return EpcFloorDescriptions.suspended_no_insulation_assumed
+
+
+def unknown_floor_retrofitted(age_band: EpcConstructionAgeBand) -> EpcFloorDescriptions:
+    year = age_band.start_year()
+
+    if year >= 1930:
+        return EpcFloorDescriptions.solid_insulated
+
+    return EpcFloorDescriptions.suspended_insulated
+
+
+def solid_floor_as_built(age_band: EpcConstructionAgeBand) -> EpcFloorDescriptions:
+    year = age_band.start_year()
+
+    if year >= 2003:
+        return EpcFloorDescriptions.solid_insulated_assumed
+    if year >= 1996:
+        return EpcFloorDescriptions.solid_limited_insulation_assumed
+    return EpcFloorDescriptions.solid_no_insulation_assumed
+
+
+def suspended_floor_as_built(age_band: EpcConstructionAgeBand) -> EpcFloorDescriptions:
+    year = age_band.start_year()
+
+    if year >= 2003:
+        return EpcFloorDescriptions.suspended_insulated_assumed
+    if year >= 1996:
+        return EpcFloorDescriptions.suspended_limited_insulation_assumed
+
+    return EpcFloorDescriptions.suspended_no_insulation_assumed

backend/onboarders/parity.py

@@ -9,6 +9,8 @@ from backend.onboarders.epc_descriptions import EpcWallDescriptions, EpcConstruc
     WALL_DESCRIPTION_EFFICIENCIES, EpcRoofDescriptions, resolve_roof_efficiency, EpcFloorDescriptions
 from backend.onboarders.mappings.as_built_wall_classifiers import AS_BUILT_WALL_CLASSIFIERS
 from backend.onboarders.mappings.as_built_roof_classifiers import AS_BUILT_ROOF_CLASSIFIERS
+from backend.onboarders.mappings.as_built_floor_classifiers import unknown_floor_as_built, unknown_floor_retrofitted, \
+    solid_floor_as_built, suspended_floor_as_built
 
 tqdm.pandas()
 
@@ -379,63 +381,6 @@ floor_mapping = {
     ('Unknown', 'AsBuilt'): None,  # Mapped unknown_floor_as_built
 }
 
-
-# Unknown floor, as built
-# Before 1900, 1900 - 1929 -> Suspended, no insulation (assumed)
-# 1930-1949, 1950 - 1966, 1967 - 1975, 1976-1982, 1983-1990, 1991-1995,   -> Solid, no insulation (assumed)
-# 1996 - 2002, Solid, limited insulation (assumed)
-# 2003 onwards -> Solid, insulated (assumed)
-
-def unknown_floor_as_built(age_band: EpcConstructionAgeBand) -> EpcFloorDescriptions:
-    year = age_band.start_year()
-
-    if year >= 2003:
-        return EpcFloorDescriptions.solid_insulated_assumed
-
-    if year >= 1930:
-        return EpcFloorDescriptions.solid_no_insulation_assumed
-
-    return EpcFloorDescriptions.suspended_no_insulation_assumed
-
-
-# before 1900, 1900-1929 -> Suspended, insulated
-# Thereafter, 1930 onwards -> Solid, insulated
-def unknown_floor_retrofitted(age_band: EpcConstructionAgeBand) -> EpcFloorDescriptions:
-    year = age_band.start_year()
-
-    if year >= 1930:
-        return EpcFloorDescriptions.solid_insulated
-
-    return EpcFloorDescriptions.suspended_insulated
-
-
-# 2003 - 2006, 2023 onwards -> Solid, insulated (assumed)
-# 1996 - 2022 -> Solid, limited insulation (assumed)
-# 1983 - 1990, 1991 - 1995 -> Solid, no insulation (assumed)
-def solid_floor_as_built(age_band: EpcConstructionAgeBand) -> EpcFloorDescriptions:
-    year = age_band.start_year()
-
-    if year >= 2003:
-        return EpcFloorDescriptions.solid_insulated_assumed
-    if year >= 1996:
-        return EpcFloorDescriptions.solid_limited_insulation_assumed
-    return EpcFloorDescriptions.solid_no_insulation_assumed
-
-
-# 2003 -> 2006 -> Suspended, insulated (assumed)
-# 1996 - 2022 -> Suspended, limited insulation (assumed)
-# 1983 - 1995 -> Suspended, no insulation (assumed)
-def suspended_floor_as_built(age_band: EpcConstructionAgeBand) -> EpcFloorDescriptions:
-    year = age_band.start_year()
-
-    if year >= 2003:
-        return EpcFloorDescriptions.suspended_insulated_assumed
-    if year >= 1996:
-        return EpcFloorDescriptions.suspended_limited_insulation_assumed
-
-    return EpcFloorDescriptions.suspended_no_insulation_assumed
-
-
 data["landlord_floor_description"] = (
     data[["Floor Construction", "Floor Insulation"]]
     .progress_apply(tuple, axis=1)
@@ -478,6 +423,13 @@ data["landlord_floor_description"] = data.progress_apply(
     axis=1,
 )
 
+# All values should be remapped now
+assert data["landlord_floor_description"].isnull().sum() == 0, (
+    "Some floor descriptions could not be resolved"
+)
+
+# TODO: Convert everything to values
+
 # Variables we want to map
 # 'Org Ref', 'Address 1', 'Address 2', 'Address 3', 'Postcode',
 #        'Floor Construction', 'Floor Insulation', 'Glazing', 'Heating',

backend/onboarders/tests/test_floor_remapping.py

@@ -0,0 +1,98 @@
+import pytest
+
+from backend.onboarders.epc_descriptions import (
+    EpcConstructionAgeBand,
+    EpcFloorDescriptions,
+)
+from backend.onboarders.mappings.as_built_floor_classifiers import (
+    unknown_floor_as_built,
+    unknown_floor_retrofitted,
+    solid_floor_as_built,
+    suspended_floor_as_built,
+)
+
+
+@pytest.mark.parametrize(
+    "age_band,expected",
+    [
+        # Before 1900 / 1900–1929 → suspended, no insulation
+        (EpcConstructionAgeBand.before_1900, EpcFloorDescriptions.suspended_no_insulation_assumed),
+        (EpcConstructionAgeBand.from_1900_to_1929, EpcFloorDescriptions.suspended_no_insulation_assumed),
+
+        # 1930–1995 → solid, no insulation
+        (EpcConstructionAgeBand.from_1930_to_1949, EpcFloorDescriptions.solid_no_insulation_assumed),
+        (EpcConstructionAgeBand.from_1950_to_1966, EpcFloorDescriptions.solid_no_insulation_assumed),
+        (EpcConstructionAgeBand.from_1967_to_1975, EpcFloorDescriptions.solid_no_insulation_assumed),
+        (EpcConstructionAgeBand.from_1976_to_1982, EpcFloorDescriptions.solid_no_insulation_assumed),
+        (EpcConstructionAgeBand.from_1983_to_1990, EpcFloorDescriptions.solid_no_insulation_assumed),
+        (EpcConstructionAgeBand.from_1991_to_1995, EpcFloorDescriptions.solid_no_insulation_assumed),
+
+        # 1996–2002 → solid, limited insulation
+        (EpcConstructionAgeBand.from_1996_to_2002, EpcFloorDescriptions.solid_limited_insulation_assumed),
+
+        # 2003+ → solid, insulated
+        (EpcConstructionAgeBand.from_2003_to_2006, EpcFloorDescriptions.solid_insulated_assumed),
+        (EpcConstructionAgeBand.from_2012_to_2022, EpcFloorDescriptions.solid_insulated_assumed),
+        (EpcConstructionAgeBand.from_2023_onwards, EpcFloorDescriptions.solid_insulated_assumed),
+    ],
+)
+def test_unknown_floor_as_built(age_band, expected):
+    assert unknown_floor_as_built(age_band) == expected
+
+
+@pytest.mark.parametrize(
+    "age_band,expected",
+    [
+        # Pre-1930 → suspended, insulated
+        (EpcConstructionAgeBand.before_1900, EpcFloorDescriptions.suspended_insulated),
+        (EpcConstructionAgeBand.from_1900_to_1929, EpcFloorDescriptions.suspended_insulated),
+
+        # 1930+ → solid, insulated
+        (EpcConstructionAgeBand.from_1930_to_1949, EpcFloorDescriptions.solid_insulated),
+        (EpcConstructionAgeBand.from_1950_to_1966, EpcFloorDescriptions.solid_insulated),
+        (EpcConstructionAgeBand.from_1976_to_1982, EpcFloorDescriptions.solid_insulated),
+        (EpcConstructionAgeBand.from_2023_onwards, EpcFloorDescriptions.solid_insulated),
+    ],
+)
+def test_unknown_floor_retrofitted(age_band, expected):
+    assert unknown_floor_retrofitted(age_band) == expected
+
+
+@pytest.mark.parametrize(
+    "age_band,expected",
+    [
+        # 1983–1995 → no insulation
+        (EpcConstructionAgeBand.from_1983_to_1990, EpcFloorDescriptions.solid_no_insulation_assumed),
+        (EpcConstructionAgeBand.from_1991_to_1995, EpcFloorDescriptions.solid_no_insulation_assumed),
+
+        # 1996–2002 → limited insulation
+        (EpcConstructionAgeBand.from_1996_to_2002, EpcFloorDescriptions.solid_limited_insulation_assumed),
+
+        # 2003+ → insulated
+        (EpcConstructionAgeBand.from_2003_to_2006, EpcFloorDescriptions.solid_insulated_assumed),
+        (EpcConstructionAgeBand.from_2012_to_2022, EpcFloorDescriptions.solid_insulated_assumed),
+        (EpcConstructionAgeBand.from_2023_onwards, EpcFloorDescriptions.solid_insulated_assumed),
+    ],
+)
+def test_solid_floor_as_built(age_band, expected):
+    assert solid_floor_as_built(age_band) == expected
+
+
+@pytest.mark.parametrize(
+    "age_band,expected",
+    [
+        # 1983–1995 → no insulation
+        (EpcConstructionAgeBand.from_1983_to_1990, EpcFloorDescriptions.suspended_no_insulation_assumed),
+        (EpcConstructionAgeBand.from_1991_to_1995, EpcFloorDescriptions.suspended_no_insulation_assumed),
+
+        # 1996–2002 → limited insulation
+        (EpcConstructionAgeBand.from_1996_to_2002, EpcFloorDescriptions.suspended_limited_insulation_assumed),
+
+        # 2003+ → insulated
+        (EpcConstructionAgeBand.from_2003_to_2006, EpcFloorDescriptions.suspended_insulated_assumed),
+        (EpcConstructionAgeBand.from_2012_to_2022, EpcFloorDescriptions.suspended_insulated_assumed),
+        (EpcConstructionAgeBand.from_2023_onwards, EpcFloorDescriptions.suspended_insulated_assumed),
+    ],
+)
+def test_suspended_floor_as_built(age_band, expected):
+    assert suspended_floor_as_built(age_band) == expected