junte/survey-extraction
1
0
Fork 0
mirror of https://github.com/Hestia-Homes/survey-extraction.git synced 2026-06-30 13:10:56 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

doors finished

Browse source
This commit is contained in:
Jun-te Kim 2025-03-14 16:31:20 +00:00
parent ba6cf1402d
commit a724c37c3a
3 changed files with 46 additions and 18 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
etl/development.py
View file

@ -15,7 +15,7 @@ pdfReader = pdfReaderToText(DATA_LOC_1)
doc2 = pdfReader.get_reader()
pdfReader2 = pdfReaderToText(DATA_LOC_2)
doc1 = pdfReader2.get_reader()
print(doc1.property_description.main_property)
print(doc1.property_description.door)
# Transform

48
etl/pdfReader/sitenotes.py
View file

@ -2,7 +2,7 @@ from etl.pdfReader.reportType import ReportType
from transform.types import (
CompanyInfo, SurverySummaryInfo, AssessorInfo,
PropertyDescription, PropertyDetail, Dimension,
Walls, Roofs,
Walls, Roofs, Floors, Door
)
from datetime import datetime
@ -36,8 +36,6 @@ class QuidosSiteNotesExtractor(SiteNotesExtractor):
"""
self.transform_summary_information()
self.transform_sections()
# self.get_section_9()
# self.get_section_10()
# self.get_section_11()
# self.get_section_12()
# self.get_section_13()
@ -177,7 +175,7 @@ class QuidosSiteNotesExtractor(SiteNotesExtractor):
# Section 3
age_bands = self.get_age_band()
print()
# Section 4
no_of_main_property = int(get_value("Main Property"))
no_of_extension_1 = int(get_value('Extension 1') or 0)
@ -202,6 +200,12 @@ class QuidosSiteNotesExtractor(SiteNotesExtractor):
# Section 8
roofs = self.get_roof()
# Section 9
floors = self.get_floors()
# Section 10
door = self.get_door()
self.property_description = PropertyDescription(
built_form = get_value("Built Form"),
detachment_or_position = get_value("Detachment/Position"),
@ -221,26 +225,31 @@ class QuidosSiteNotesExtractor(SiteNotesExtractor):
dimensions=dimensions["main"] if "main" in dimensions else [],
wall=walls[0],
roof=roofs[0],
floor=floors[0],
)if no_of_main_property > 0 else None,
ex1_property=PropertyDetail(
age_band= age_bands[1],
dimensions= dimensions["ex1"] if "ex1" in dimensions else [],
wall=walls[1],
roof=roofs[1],
floor=floors[1],
)if no_of_extension_1 > 0 else None,
ex2_property=PropertyDetail(
age_band= age_bands[2],
dimensions= dimensions["ex2"] if "ex2" in dimensions else [],
wall=walls[2],
roof=roofs[2],
floor=floors[2],
)if no_of_extension_2 > 0 else None,
ex3_property=PropertyDetail(
age_band= age_bands[3],
dimensions=dimensions["ex3"] if "ex3" in dimensions else [],
wall=walls[3],
roof=roofs[3],
floor=floors[3],
)if no_of_extension_4 > 0 else None,
conservatory=conservatory,
door=door,
)
@ -446,7 +455,7 @@ class QuidosSiteNotesExtractor(SiteNotesExtractor):
return group_data
def get_section_9(self):
def get_floors(self):
data = self.raw_data[self.raw_data.index('9.0 Floors'): self.raw_data.index('10.0 Doors')]
sub_titles = [
"Floor Type",
@ -465,9 +474,19 @@ class QuidosSiteNotesExtractor(SiteNotesExtractor):
]
self.two_column_with_extension_processor(data, sub_titles, titles, 9)
lst = self.two_column_with_extension_processor(data, sub_titles, titles)
floors = []
for floor in lst:
floors.append(Floors(
floor_type=floor.get("floor_type", ""),
ground_floor_construction=floor.get("get_floor_construction", ""),
ground_floor_insulation_type=floor.get("ground_floor_insulation_type", ""),
floor_insulation_thickness_mm=floor.get("floor_insulation_thickness_(mm)", -1) if floor.get("floor_insulation_thickness_(mm)", -1) is not None else -1,
u_value_known=True if floor.get("u_value_known").upper() == "YES" else False,
))
return floors
def get_section_10(self):
def get_door(self):
data = self.raw_data[self.raw_data.index("10.0 Doors"): self.raw_data.index("11.0 Windows")]
avoid = [
"10.0 Doors",
@ -478,9 +497,15 @@ class QuidosSiteNotesExtractor(SiteNotesExtractor):
"Number of Insulated Doors",
"U-value (W/m²K)",
]
self.two_columns_processor(data, sub_titles, avoid, 10)
dict_ = self.two_columns_processor(data, sub_titles, avoid)
def two_columns_processor(self, data, sub_titles_to_gather, avoid, section, indexAdd = 1):
return Door(
no_of_doors=int(dict_.get("no_of_doors", -1)),
no_of_insulated_doors=int(dict_.get("number_of_insulated_doors", -1)),
u_value_w_m2_k=dict_.get("u_value_(w/m²k)", "") if dict_.get("u_value_(w/m²k), '')") is not None else "",
)
def two_columns_processor(self, data, sub_titles_to_gather, avoid, indexAdd = 1):
def get_value(key):
try:
index = data.index(key)
@ -488,12 +513,13 @@ class QuidosSiteNotesExtractor(SiteNotesExtractor):
return None if value in avoid else value
except (ValueError, IndexError):
return None
dict_ = {}
for items in data:
if items in avoid:
continue
elif items in sub_titles_to_gather:
setattr(self, f"section_{section}_{items.lower().replace('-', '_').replace(' ','_')}", get_value(items))
dict_.update({f"{items.lower().replace('-', '_').replace(' ','_')}":get_value(items)})
return dict_
def get_section_11(self):
data = self.get_data_between("Window Location", "12.0 Ventilation & Cooling")

14
etl/transform/types.py
View file

@ -62,13 +62,13 @@ class Floors(BaseModel):
floor_type: str
ground_floor_construction: str
ground_floor_insulation_type: str
floor_insulation_thickness_mm: Optional[float]
floor_insulation_thickness_mm: Optional[float] = -1
u_value_known: bool
class Doors(BaseModel):
no_of_doofs: int
nu_of_insulated_doors: int
u_value_w_m2_k: Optional[str] = None
class Door(BaseModel):
no_of_doors: int
no_of_insulated_doors: int
u_value_w_m2_k: Optional[str]
class AssessorInfo(BaseModel):
accreditation_number: str
@ -115,6 +115,7 @@ class PropertyDetail(BaseModel):
dimensions: List[Dimension] = []
wall: Optional[Walls] = None
roof: Optional[Roofs] = None
floor: Optional[Floors] = None
class PropertyDescription(BaseModel):
built_form: str
@ -135,4 +136,5 @@ class PropertyDescription(BaseModel):
ex2_property: Optional[PropertyDetail] = None
ex3_property: Optional[PropertyDetail] = None
ex4_property: Optional[PropertyDetail] = None
conservatory: bool
conservatory: bool
door: Optional[Door]