junte/Model
1
0
Fork 0
mirror of https://github.com/Hestia-Homes/Model.git synced 2026-06-30 13:10:47 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fixing alternative wall extracction for summary report

Browse source
This commit is contained in:
Khalim Conn-Kowlessar 2024-11-13 23:35:02 +00:00
parent 8da1aecb55
commit 69b3ec7961
1 changed files with 30 additions and 20 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

50
etl/customers/stonewater/Wave 3 Preparation.py
View file

@ -50,18 +50,23 @@ def extract_wall_details_summary(text):
# Define pattern to match each building part's wall entry within the section # Define pattern to match each building part's wall entry within the section
building_part_pattern = re.compile( 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"Type\s+(.*?)\n" # Matches main wall Type
r"Insulation\s+(.*?)\n" # Matches main wall Insulation 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 Unknown\s+(.*?)\n" # Matches main wall Thickness Unknown
r"Wall Thickness \[mm\]\s+(\d+)" # Matches main wall Thickness r"Wall Thickness \[mm\]\s+(\d+)", # Matches main wall Thickness
r"(?:\nAlternative Wall Area.*?\n" # Starts matching alternative wall section if present 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 Type\s+(.*?)\n" # Matches alternative wall Type
r"Alternative Insulation\s+(.*?)\n" # Matches alternative wall Insulation r"Alternative Insulation\s+(.*?)\n" # Matches alternative wall Insulation
r"(Alternative Dry-lining\s+(.*?)\n)?" # Optional Alternative Dry-lining 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 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 re.DOTALL
) )
@ -74,27 +79,32 @@ def extract_wall_details_summary(text):
main_wall_thickness_unknown = match.group(6).strip() main_wall_thickness_unknown = match.group(6).strip()
main_wall_thickness = int(match.group(7)) main_wall_thickness = int(match.group(7))
# Optional alternative wall fields # Initialize dictionary for this wall entry
alt_wall_type = match.group(8).strip() if match.group(8) else None wall_entry = {
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({
"Building Part": wall_label, "Building Part": wall_label,
"Wall Type": main_wall_type, "Wall Type": main_wall_type,
"Wall Insulation": main_wall_insulation, "Wall Insulation": main_wall_insulation,
"Wall Dry-lining": main_wall_dry_lining, "Wall Dry-lining": main_wall_dry_lining,
"Wall Thickness Unknown": main_wall_thickness_unknown, "Wall Thickness Unknown": main_wall_thickness_unknown,
"Wall Thickness (mm)": main_wall_thickness, "Wall Thickness (mm)": main_wall_thickness,
"Alternative Wall Type": alt_wall_type, "Alternative Wall Type": None,
"Alternative Wall Insulation": alt_wall_insulation, "Alternative Wall Insulation": None,
"Alternative Wall Dry-lining": alt_wall_dry_lining, "Alternative Wall Dry-lining": "N/A",
"Alternative Wall Thickness Unknown": alt_wall_thickness_unknown, "Alternative Wall Thickness Unknown": None,
"Alternative Wall Thickness (mm)": alt_wall_thickness, "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 return wall_data