94 lines
2.7 KiB
Python
94 lines
2.7 KiB
Python
"""Helper function to parse rows into cleaner data that can be inserted into
|
|
the destination database"""
|
|
|
|
import logging
|
|
from typing import TypedDict, Dict, Optional, List
|
|
|
|
from helpers.parsers.parse_xml import parse_raw_specs
|
|
from helpers.parsers.materials import parse_materials
|
|
from helpers.parsers.origin import clean_origin_name
|
|
from helpers.parsers.dimensions import parse_dimensions
|
|
from helpers.parsers.weight import parse_weight, dimensional_weight
|
|
|
|
|
|
class CleanRow(TypedDict):
|
|
"""Type to represent clean rows to be inserted in the database"""
|
|
|
|
gtin13: str
|
|
tcin: str
|
|
primary_category: str
|
|
materials: Optional[List[str]]
|
|
packaging: int
|
|
origin: str
|
|
weight: Optional[float]
|
|
height: Optional[float]
|
|
width: Optional[float]
|
|
depth: Optional[float]
|
|
|
|
|
|
def parse_row(element: Dict[str, str]) -> Optional[CleanRow]:
|
|
"""Parse a dictionary representing a row for the CSV input into a cleaner
|
|
dictionary representing a row to be inserted in the database"""
|
|
|
|
# gtin13 should always be there
|
|
try:
|
|
gtin13 = element["gtin13"]
|
|
except KeyError:
|
|
logging.error("gtin13 missing")
|
|
return None
|
|
|
|
# primary category should always be there
|
|
try:
|
|
primary_category = element["primary_category"]
|
|
except KeyError:
|
|
logging.error("primary_category missing")
|
|
return None
|
|
|
|
specifications = parse_raw_specs(element["raw_specifications"])
|
|
if specifications is None:
|
|
logging.error("could not parse raw_specifications")
|
|
return None
|
|
|
|
# TCIN should be a mandatory field
|
|
try:
|
|
tcin = specifications["tcin"]
|
|
except KeyError:
|
|
logging.error("TCIN missing")
|
|
return None
|
|
|
|
materials = parse_materials(specifications.get("materials"))
|
|
|
|
# if packaging is not specified, assume only one unit is found in the
|
|
# package
|
|
packaging = specifications.get("packaging", 1)
|
|
try:
|
|
packaging = int(packaging)
|
|
except ValueError:
|
|
logging.error("could not cast packaging %s into an integer")
|
|
packaging = 1
|
|
|
|
origin = clean_origin_name(specifications.get("origin"))
|
|
|
|
weight = parse_weight(specifications.get("weight"))
|
|
|
|
dimensions = parse_dimensions(specifications.get("dimensions"))
|
|
height = dimensions["height"]
|
|
width = dimensions["width"]
|
|
depth = dimensions["depth"]
|
|
|
|
if weight is None:
|
|
weight = dimensional_weight(height=height, width=width, depth=depth)
|
|
|
|
return {
|
|
"gtin13": gtin13,
|
|
"tcin": tcin,
|
|
"primary_category": primary_category,
|
|
"materials": materials,
|
|
"packaging": packaging,
|
|
"origin": origin,
|
|
"weight": weight,
|
|
"height": height,
|
|
"width": width,
|
|
"depth": depth,
|
|
}
|