From 69b3ec7961eda55e7b2fe36d17353f215a1bf068 Mon Sep 17 00:00:00 2001
From: Khalim Conn-Kowlessar <kconnkowlessar@gmail.com>
Date: Wed, 13 Nov 2024 23:35:02 +0000
Subject: [PATCH] fixing alternative wall extracction for summary report

---
 .../stonewater/Wave 3 Preparation.py          | 50 +++++++++++--------
 1 file changed, 30 insertions(+), 20 deletions(-)

diff --git a/etl/customers/stonewater/Wave 3 Preparation.py b/etl/customers/stonewater/Wave 3 Preparation.py
index 0af3ffbb..8791912a 100644
--- a/etl/customers/stonewater/Wave 3 Preparation.py	
+++ b/etl/customers/stonewater/Wave 3 Preparation.py	
@@ -50,18 +50,23 @@ def extract_wall_details_summary(text):
 
     # Define pattern to match each building part's wall entry within the section
     building_part_pattern = re.compile(
-        r"(Main Property|1st Extension|2nd Extension|[\w\s]+)\n"  # Matches each building part
+        r"(Main Property|1st Extension|2nd Extension|[\w\s]+)\n"  # Matches each building part label
         r"Type\s+(.*?)\n"  # Matches main wall Type
         r"Insulation\s+(.*?)\n"  # Matches main wall Insulation
-        r"(Dry-lining\s+(.*?)\n)?"  # Optional Dry-lining
+        r"(Dry-lining\s+(.*?)\n)?"  # Optional main wall Dry-lining
         r"Wall Thickness Unknown\s+(.*?)\n"  # Matches main wall Thickness Unknown
-        r"Wall Thickness \[mm\]\s+(\d+)"  # Matches main wall Thickness
-        r"(?:\nAlternative Wall Area.*?\n"  # Starts matching alternative wall section if present
+        r"Wall Thickness \[mm\]\s+(\d+)",  # Matches main wall Thickness
+        re.DOTALL
+    )
+
+    # Define pattern to capture alternative wall details, if present
+    alternative_wall_pattern = re.compile(
+        r"Alternative Wall Area.*?\n"  # Matches start of alternative wall section
         r"Alternative Type\s+(.*?)\n"  # Matches alternative wall Type
         r"Alternative Insulation\s+(.*?)\n"  # Matches alternative wall Insulation
         r"(Alternative Dry-lining\s+(.*?)\n)?"  # Optional Alternative Dry-lining
         r"Alternative Wall Thickness Unknown\s+(.*?)\n"  # Matches alternative wall Thickness Unknown
-        r"Alternative Wall Thickness\s+(\d+))?",  # Matches alternative wall Thickness
+        r"Alternative Wall Thickness\s+(\d+)",  # Matches alternative wall Thickness
         re.DOTALL
     )
 
@@ -74,27 +79,32 @@ def extract_wall_details_summary(text):
         main_wall_thickness_unknown = match.group(6).strip()
         main_wall_thickness = int(match.group(7))
 
-        # Optional alternative wall fields
-        alt_wall_type = match.group(8).strip() if match.group(8) else None
-        alt_wall_insulation = match.group(9).strip() if match.group(9) else None
-        alt_wall_dry_lining = match.group(10).strip() if match.group(10) else None
-        alt_wall_thickness_unknown = match.group(11).strip() if match.group(11) else None
-        alt_wall_thickness = int(match.group(12)) if match.group(12) else None
-
-        # Append each building part as a dictionary in the wall_data list
-        wall_data.append({
+        # Initialize dictionary for this wall entry
+        wall_entry = {
             "Building Part": wall_label,
             "Wall Type": main_wall_type,
             "Wall Insulation": main_wall_insulation,
             "Wall Dry-lining": main_wall_dry_lining,
             "Wall Thickness Unknown": main_wall_thickness_unknown,
             "Wall Thickness (mm)": main_wall_thickness,
-            "Alternative Wall Type": alt_wall_type,
-            "Alternative Wall Insulation": alt_wall_insulation,
-            "Alternative Wall Dry-lining": alt_wall_dry_lining,
-            "Alternative Wall Thickness Unknown": alt_wall_thickness_unknown,
-            "Alternative Wall Thickness (mm)": alt_wall_thickness,
-        })
+            "Alternative Wall Type": None,
+            "Alternative Wall Insulation": None,
+            "Alternative Wall Dry-lining": "N/A",
+            "Alternative Wall Thickness Unknown": None,
+            "Alternative Wall Thickness (mm)": None,
+        }
+
+        # Check if there's an alternative wall section following this wall entry
+        alt_match = alternative_wall_pattern.search(wall_section, match.end())
+        if alt_match:
+            wall_entry["Alternative Wall Type"] = alt_match.group(1).strip()
+            wall_entry["Alternative Wall Insulation"] = alt_match.group(2).strip()
+            wall_entry["Alternative Wall Dry-lining"] = alt_match.group(4).strip() if alt_match.group(4) else "N/A"
+            wall_entry["Alternative Wall Thickness Unknown"] = alt_match.group(5).strip()
+            wall_entry["Alternative Wall Thickness (mm)"] = int(alt_match.group(6))
+
+        # Append each building part as a dictionary in the wall_data list
+        wall_data.append(wall_entry)
 
     return wall_data