def normalise(self, data, uri, view_date):
ans = super().normalise(data, uri, view_date)
salary_raw = data["baseSalary"]["value"].get("value")
salary = get_salary_data(salary_raw)
location_raw = location_jsonld(data)
return {
**ans,
**salary,
"location_raw": location_raw,
**AU_GEOCODER.geocode(location_raw),
}
def normalise(self, *args, **data) -> Dict[str, Any]:
return {
"title": data["job_title"],
"description": data["job_description"],
"uri": data["page_url"],
# Not quite true; this is date added
"view_date": datetime.datetime.strptime(data["date_added"],
"%d/%m/%Y"),
**get_salary_data(data["salary"]),
"location_raw": data["location"],
**AU_GEOCODER.geocode(data["location"]),
}
def normalise(self, data, uri, view_date):
metadata = {row["value"]: row["name"] for row in data["mainAttributes"]}
salary_raw = metadata.get("Salary Detail")
salary_data = get_salary_data(salary_raw)
return {
"title": data["title"],
"description": html2plain(data["description"]),
"uri": uri,
"view_date": datetime_from_iso_utc(view_date),
"org": None,
**salary_data,
"location_raw": data["mapAddress"],
**AU_GEOCODER.geocode(data["mapAddress"]),
}
def normalise(self, title, description, metadata, uri, view_date):
salary = get_salary_data(
metadata.get("Total Remuneration Package:") or "")
location_raw = metadata["Job Location:"]
return {
"title": title,
"description": html2plain(description),
"uri": uri,
"view_date": datetime_from_iso_utc(view_date),
"org": metadata["Organisation/Entity:"],
**salary,
"location_raw": location_raw,
**AU_GEOCODER.geocode(fixup_iworkfornsw_loc(location_raw)),
}
def normalise(self, title, description, metadata, uri, view_date):
salary_data = get_salary_data(
metadata.get("Salary:") or metadata["Salary Range:"]
)
location_raw = metadata.get("Location:") or metadata["Work location:"]
return {
"title": title,
"description": html2plain(description),
"uri": uri,
"view_date": datetime_from_iso_utc(view_date),
"org": metadata.get("Organisation:"),
**salary_data,
"location_raw": location_raw,
**AU_GEOCODER.geocode(fixup_careers_vic_location(location_raw)),
}
def normalise(
self, title, description, organisation_description, metadata, uri, view_date
):
salary_text = metadata.get("Salary :")
salary_data = get_salary_data(salary_text)
location_raw = metadata["Location :"]
return {
"title": title,
"description": html2plain(description),
"uri": uri,
"view_date": datetime_from_iso_utc(view_date),
"org": metadata.get("Organisation :"),
**salary_data,
"location_raw": location_raw,
**AU_GEOCODER.geocode(fix_probono_location(location_raw)),
}
def normalise(self, data, uri, view_date):
salary_text = data["salary"]
location = data["locationHierarchy"]
location_text = ", ".join([
location["suburb"],
location["city"],
location["state"],
location["nation"],
])
return {
"title": data["title"],
"description": html2plain(data["mobileAdTemplate"]),
"uri": uri,
"view_date": datetime_from_iso_utc(view_date),
"org": data["advertiser"]["description"],
**get_salary_data(salary_text),
"location_raw": location_text,
**AU_GEOCODER.geocode(location_text),
}
def normalise(self, *args, **data) -> Dict[str, Any]:
location_text = ", ".join([
data["city"],
data["state"],
data.get("country") or data["inferred_country"],
])
salary_text = data.get("salary_offered")
return {
"title":
data["job_title"],
"description":
data["job_description"],
"uri":
data["url"],
"view_date":
datetime.datetime.strptime(data["crawl_timestamp"],
"%Y-%m-%d %H:%M:%S +0000"),
"org":
data["company_name"],
**get_salary_data(salary_text),
"location_raw":
location_text,
**AU_GEOCODER.geocode(location_text),
}
def normalise(self, *args, **data) -> Dict[str, Any]:
parts = [
x for x in [data.get("city"),
data.get("state"),
data.get("geo")] if x
]
location_text = ", ".join(parts)
return {
"title":
data["job_title"],
"description":
data["job_description"],
"uri":
data["pageurl"],
"view_date":
datetime.datetime.strptime(data["crawl_timestamp"],
"%Y-%m-%d %H:%M:%S +0000"),
"org":
data["company_name"],
**get_salary_data(data.get("salary_offered")),
"location_raw":
location_text,
**AU_GEOCODER.geocode(location_text),
}