extract window frame details from elmhurst site notes 🟩

backend/documents_parser/elmhurst_extractor.py

@@ -230,7 +230,7 @@ class ElmhurstSiteNotesExtractor:
                 i += 1
                 continue
             i += 3
-            # Collect glazing type until frame_factor (0 < v ≤ 1.0)
+            # Collect glazing type tokens until frame_factor (0 < v ≤ 1.0)
             glazing_parts: List[str] = []
             while i < len(tokens):
                 try:
@@ -241,10 +241,21 @@ class ElmhurstSiteNotesExtractor:
                 except ValueError:
                     glazing_parts.append(tokens[i])
                 i += 1
+            # If last glazing token is a single word (no spaces, not numeric) it's the frame_type
+            frame_type: Optional[str] = None
+            if glazing_parts and " " not in glazing_parts[-1] and not glazing_parts[-1].replace(".", "").isdigit():
+                frame_type = glazing_parts.pop()
             glazing_type = " ".join(glazing_parts).strip()
             if i >= len(tokens):
                 break
             frame_factor = float(tokens[i]); i += 1
+            # Consume glazing_gap if present ("mm" token, possibly multi-token e.g. "16 mm or more")
+            glazing_gap: Optional[str] = None
+            if i < len(tokens) and "mm" in tokens[i]:
+                gap_parts = [tokens[i]]; i += 1
+                while i < len(tokens) and tokens[i].lower() in {"or", "more"}:
+                    gap_parts.append(tokens[i]); i += 1
+                glazing_gap = " ".join(gap_parts)
             building_part = tokens[i]; i += 1
             location = tokens[i]; i += 1
             orientation = tokens[i]; i += 1
@@ -268,6 +279,8 @@ class ElmhurstSiteNotesExtractor:
                     g_value=g_value,
                     draught_proofed=draught_proofed,
                     permanent_shutters=permanent_shutters,
+                    frame_type=frame_type,
+                    glazing_gap=glazing_gap,
                 )
             )
         return windows
@@ -296,12 +309,17 @@ class ElmhurstSiteNotesExtractor:
         )
 
     def _extract_lighting(self) -> Lighting:
+        led_cfl_count_known = self._bool_val("Number of LED and CFL Known")
         return Lighting(
             total_bulbs=self._int_val("Total number of bulbs"),
-            led_cfl_count_known=self._bool_val("Number of LED and CFL Known"),
+            led_cfl_count_known=led_cfl_count_known,
             led_count=self._int_val("Number of LED lights"),
             cfl_count=self._int_val("Number of CFL lights"),
             incandescent_count=self._int_val("Total number of incandescents"),
+            low_energy_count=(
+                0 if led_cfl_count_known
+                else self._int_val("Total number of Low Energy")
+            ),
         )
 
     def _extract_main_heating(self) -> MainHeating:

datatypes/epc/surveys/elmhurst_site_notes.py

@@ -121,6 +121,7 @@ class Lighting:
     led_count: int
     cfl_count: int
     incandescent_count: int
+    low_energy_count: int = 0
 
 
 @dataclass