all is well

finder/homecouk.py

@@ -19,7 +19,12 @@ from constants import (
     RETRY_BASE_DELAY,
 )
 from spatial import PostcodeSpatialIndex
-from transform import normalize_postcode, normalize_sub_type, validate_floor_area
+from transform import (
+    normalize_postcode,
+    normalize_sub_type,
+    parse_int_value,
+    validate_floor_area,
+)
 
 log = logging.getLogger("homecouk")
 
@@ -170,11 +175,19 @@ def parse_floor_area(description: str | None) -> float | None:
     """Try to extract floor area from description text like '789 sq.ft.' or '73 sq.m.'."""
     if not description:
         return None
-    m = re.search(r"([\d,]+(?:\.\d+)?)\s*sq\.?\s*ft", description, re.IGNORECASE)
+    m = re.search(
+        r"([\d,]+(?:\.\d+)?)\s*(?:sq\.?\s*ft|square\s+feet|ft(?:\^?2|²))",
+        description,
+        re.IGNORECASE,
+    )
     if m:
         sqft = float(m.group(1).replace(",", ""))
         return validate_floor_area(round(sqft * 0.092903, 1))
-    m = re.search(r"([\d,]+(?:\.\d+)?)\s*sq\.?\s*m", description, re.IGNORECASE)
+    m = re.search(
+        r"([\d,]+(?:\.\d+)?)\s*(?:sq\.?\s*m|square\s+met(?:er|re)s?|m(?:\^?2|²))",
+        description,
+        re.IGNORECASE,
+    )
     if m:
         return validate_floor_area(round(float(m.group(1).replace(",", "")), 1))
     return None
@@ -237,6 +250,15 @@ def map_property_type(raw_type: str | None) -> str:
     # Home.co.uk uses types like "House", "Flat", "Apartment", "Detached", etc.
     # Try common patterns
     lower = raw_type.lower()
+    excluded_flat_like = (
+        "block of apartment",
+        "house of multiple occupation",
+        "private halls",
+        "retirement",
+        "serviced apartment",
+    )
+    if any(term in lower for term in excluded_flat_like):
+        return "Other"
     if (
         "flat" in lower
         or "apartment" in lower
@@ -269,8 +291,10 @@ def transform_property(
         log.debug("Coords outside England: lat=%.4f lng=%.4f — skipping", lat, lng)
         return None
 
-    price = prop.get("price") or prop.get("latest_price")
-    if not price or int(price) <= 0:
+    price = parse_int_value(prop.get("price")) or parse_int_value(
+        prop.get("latest_price")
+    )
+    if not price or price <= 0:
         return None
 
     # Home.co.uk provides postcodes directly, but fall back to spatial index
@@ -281,10 +305,10 @@ def transform_property(
             log.debug("No postcode for property at %.4f, %.4f — skipping", lat, lng)
             return None
 
-    raw_beds = prop.get("bedrooms", 0) or 0
-    raw_baths = prop.get("bathrooms", 0) or 0
-    bedrooms = raw_beds if raw_beds <= MAX_BEDROOMS else 0
-    bathrooms = raw_baths if raw_baths <= MAX_BEDROOMS else 0
+    raw_beds = parse_int_value(prop.get("bedrooms")) or 0
+    raw_baths = parse_int_value(prop.get("bathrooms")) or 0
+    bedrooms = raw_beds if 0 <= raw_beds <= MAX_BEDROOMS else 0
+    bathrooms = raw_baths if 0 <= raw_baths <= MAX_BEDROOMS else 0
     if raw_beds > MAX_BEDROOMS or raw_baths > MAX_BEDROOMS:
         log.warning(
             "home.co.uk %s: implausible beds=%d baths=%d (capped to 0)",
@@ -318,7 +342,7 @@ def transform_property(
         "Leasehold/Freehold": parse_tenure(prop),
         "Property type": map_property_type(listing_type),
         "Property sub-type": normalize_sub_type(listing_type),
-        "price": int(price),
+        "price": price,
         "price_frequency": "",
         "Price qualifier": price_qualifier,
         "Total floor area (sqm)": parse_floor_area(prop.get("description")),
@@ -362,7 +386,16 @@ def search_outcode(
             break
 
         for prop in raw_props:
-            transformed = transform_property(prop, pc_index)
+            try:
+                transformed = transform_property(prop, pc_index)
+            except Exception as exc:
+                log.warning(
+                    "home.co.uk %s property %s failed to transform: %s",
+                    outcode,
+                    prop.get("listing_id") or prop.get("property_id") or "?",
+                    exc,
+                )
+                continue
             if transformed:
                 properties.append(transformed)
                 if max_properties is not None and len(properties) >= max_properties: