Example #1
0
    def process_facilities(self, records):
        s = self.session()

        # Store a list of all existing duids
        all_duids = list(
            set([
                i[0] for i in s.query(Facility.network_code).filter(
                    Facility.network_code != None).all()
            ]))

        for _, facility_records in records.items():
            facility_index = 1
            facility_station = None
            created_station = False

            station_network_name = record_get_station_name(facility_records)
            station_name = station_name_cleaner(station_network_name)

            duid_unique = has_unique_duid(facility_records)
            facility_count = len(facility_records)

            # Step 1. Find the station
            # First by duid if it's unique
            duid = get_unique_duid(facility_records)

            # all GI records should have a region
            station_network_region = get_unique_reqion(facility_records)

            # This is the most suitable unit record to use for the station
            # see helper above
            facility_station_record = get_station_record_from_facilities(
                facility_records)

            if duid and duid_unique and facility_count == 1:

                facility_lookup = None

                try:
                    facility_lookup = (s.query(Facility).filter(
                        Facility.network_code == duid).filter(
                            Facility.network_region ==
                            station_network_region).one_or_none())
                except MultipleResultsFound:
                    logger.error(
                        "Found multiple duid for station with code {}".format(
                            duid))
                    continue

                if facility_lookup and facility_lookup.station:
                    facility_station = facility_lookup.station

            if (duid and (duid_unique and facility_count > 1)
                    or not duid_unique):

                facility_lookup = (s.query(Facility).filter(
                    Facility.network_code == duid).filter(
                        Facility.network_region ==
                        station_network_region).first())

                if facility_lookup and facility_lookup.station:
                    facility_station = facility_lookup.station

            if not facility_station and facility_station_join_by_name(
                    station_name):
                try:
                    facility_station = (s.query(Station).filter(
                        Station.name == station_name).one_or_none())
                except MultipleResultsFound:
                    logger.warning(
                        "Multiple results found for station name : {}".format(
                            station_name))
                    facility_station = None

            # If we have a station name, and no duid, and it's ok to join by name
            # then find the station (make sure to region lock)
            if (station_name and not duid and not facility_station
                    and facility_station_join_by_name(station_name)):
                facility = (s.query(Facility).join(Facility.station).filter(
                    Facility.network_region == station_network_region).filter(
                        Station.name == station_name).first())

                if facility:
                    facility_station = facility.station

            # Create one as it doesn't exist
            if not facility_station:
                facility_station = Station(
                    name=station_name,
                    network_name=name_normalizer(
                        facility_station_record["station_name"]),
                    network_id="NEM",
                    created_by="pipeline.aemo.general_information",
                )

                s.add(facility_station)
                s.commit()

                created_station = True
            else:
                facility_station.updated_by = (
                    "pipeline.aemo.general_information")

            for facility_record in facility_records:
                if facility_record["FuelType"] in ["Natural Gas Pipeline"]:
                    continue

                # skip these statuses too
                if facility_record["UnitStatus"] in FACILITY_INVALID_STATUS:
                    continue

                facility = None
                created_facility = False

                facility_network_name = name_normalizer(
                    facility_record["station_name"])
                facility_name = station_name_cleaner(
                    facility_record["station_name"])
                duid = normalize_duid(facility_record["duid"])
                reg_cap = clean_capacity(facility_record["NameCapacity"])

                units_num = facility_record["Units"] or 1
                unit_id = facility_index + (units_num - 1)

                unit = parse_unit_duid(unit_id, duid)
                unit_size = clean_capacity(facility_record["unit_capacity"])
                unit_code = get_unit_code(unit, duid,
                                          facility_record["station_name"])

                facility_comissioned = facility_record["SurveyEffective"]
                facility_comissioned_dt = None

                if type(facility_comissioned) is datetime:
                    facility_comissioned_dt = facility_comissioned

                try:
                    if type(facility_comissioned) is str:
                        facility_comissioned_dt = datetime.strptime(
                            facility_comissioned, "%d/%m/%y")
                except ValueError:
                    logger.error(
                        "Error parsing date: {}".format(facility_comissioned))

                facility_status = map_aemo_facility_status(
                    facility_record["UnitStatus"])
                facility_network_region = normalize_aemo_region(
                    facility_record["Region"])
                facility_fueltech = (lookup_fueltech(
                    facility_record["FuelType"],
                    techtype=facility_record["TechType"],
                ) if ("FuelType" in facility_record
                      and facility_record["FuelType"]) else None)

                if not facility_fueltech:
                    logger.error("Error looking up fueltech: {} {} ".format(
                        facility_record["FuelType"],
                        facility_record["TechType"],
                    ))

                # check if we have it by ocode first
                facility = (s.query(Facility).filter(
                    Facility.code == unit_code).one_or_none())

                if not facility and duid:
                    try:
                        facility = (
                            s.query(Facility).filter(
                                Facility.network_code == duid).filter(
                                    Facility.network_region ==
                                    facility_network_region)
                            # .filter(Facility.nameplate_capacity != None)
                            .one_or_none())
                    except MultipleResultsFound:
                        logger.warn(
                            "Multiple results found for duid : {}".format(
                                duid))

                    if facility:
                        if facility.station and not facility_station:
                            facility_station = facility.station

                        logger.info(
                            "GI: Found facility by DUID: code {} station {}".
                            format(
                                facility.code,
                                facility.station.name
                                if facility.station else None,
                            ))

                # Done trying to find existing
                if not facility:
                    facility = Facility(
                        code=unit_code,
                        network_code=duid,
                        created_by="pipeline.aemo.general_information",
                    )
                    facility.station = facility_station

                    created_facility = True

                if duid and not facility.network_code:
                    facility.network_code = duid
                    facility.updated_by = "pipeline.aemo.general_information"

                if not facility.network_region:
                    facility.network_region = facility_network_region
                    facility.updated_by = "pipeline.aemo.general_information"

                if not facility.network_name:
                    facility.network_name = facility_network_name
                    facility.updated_by = "pipeline.aemo.general_information"

                if not facility.fueltech_id and facility_fueltech:
                    facility.fueltech_id = facility_fueltech
                    facility.updated_by = "pipeline.aemo.general_information"

                if not facility.capacity_registered or (
                        facility.status and facility.status != "operating"):
                    facility.capacity_registered = reg_cap
                    facility.updated_by = "pipeline.aemo.general_information"

                # @TODO work this out
                # facility.dispatch_type = facility_dispatch_type

                if not facility.unit_id:
                    facility.unit_id = unit.id
                    facility.unit_number = unit.number
                    facility.unit_size = unit_size
                    facility.unit_alias = unit.alias

                if not facility.unit_capacity or (
                        facility.status and facility.status != "operating"):
                    facility.unit_capacity = unit_size
                    facility.updated_by = "pipeline.aemo.general_information"

                # if not facility.status_id:
                facility.status_id = facility_status
                # facility.updated_by = "pipeline.aemo.general_information"

                if not facility.registered and facility_comissioned_dt:
                    facility.registered = facility_comissioned_dt
                    facility.updated_by = "pipeline.aemo.general_information"

                facility.station = facility_station

                if facility.fueltech_id is None:
                    logger.warning("Could not find fueltech for: {} {}".format(
                        facility.code, facility.network_code))

                # facility.status_id = facility_status

                if facility_station and not facility.station:
                    facility.station = facility_station

                if facility.status_id is None:
                    raise Exception(
                        "GI: Failed to map status ({}) on row: {}".format(
                            facility.status_id, facility_record))

                s.add(facility)
                s.commit()

                facility_index += units_num

                if created_station:
                    logger.info("GI: {} station with name {} ".format(
                        "Created" if created_station else "Updated",
                        station_name,
                        # facility_station.id,
                    ))

                if created_facility:
                    logger.info(
                        "GI: {} facility with duid {} to station {}".format(
                            "Created" if created_facility else "Updated",
                            duid,
                            station_name,
                        ))

        try:
            s.commit()
        except Exception as e:
            logger.error(e)
            raise e
        finally:
            s.close()
Example #2
0
def gi_grouper(records, station_code_map):

    # filter out records we don't want
    records = list(filter(gi_filter, records))

    records = [{
        "name": station_name_cleaner(i["station_name"]),
        **i
    } for i in records]

    grouped_records = {}

    for k, v in groupby(records,
                        key=lambda v: (v["name"].strip(), v["owner"].strip())):
        v = list(v)

        key = k[0]

        if key not in grouped_records:
            grouped_records[key] = []

        grouped_records[key] += v

    records_parsed = []

    for station_name, facilities in grouped_records.items():
        facility_index = 0

        for i in facilities:
            name = station_name_cleaner(i["station_name"])
            duid = normalize_duid(i["duid"])

            unit = parse_unit_duid(i["unit_num"], duid)
            units_num = i["unit_num"] or 1
            unit_id = facility_index + (units_num - 1)
            unit = parse_unit_duid(unit_id, duid)

            fueltech = lookup_fueltech(i["FuelType"], techtype=i["TechType"])

            if not duid:
                duid = get_unit_code(unit, duid, i["station_name"])

            facility_duids = [duid]
            station_code = lookup_station_code(facility_duids, station_name,
                                               station_code_map)

            records_parsed.append({
                # not a real station id
                # "id": _id,
                "name":
                name,
                "code":
                duid,
                "network_code":
                duid,
                "station_code":
                station_code,
                "network_region":
                normalize_aemo_region(i["Region"]),
                "network_name":
                i["station_name"].strip(),
                "dispatch_type":
                "GENERATOR",
                "fueltech":
                parse_facility_fueltech(fueltech) if fueltech else None,
                "status":
                parse_facility_status(map_aemo_facility_status(i["status"])),
                "registered":
                parse_comissioned_date(i["SurveyEffective"]),
                **get_capacities(i),
            })

            facility_index += 1

    coded_records = {}
    _id = 3000

    for station_code, facilities in groupby(records_parsed,
                                            key=lambda x: x["station_code"]):

        # station_name = facilities[0]["name"]
        facilities = list(facilities)

        if not station_code:
            raise Exception("Unmapped station {}: {}".format(
                station_code, facilities))

        if station_code not in coded_records:
            coded_records[station_code] = {
                "id": _id,
                "code": station_code,
                # "name": station_name_cleaner(station_name),
                # "network_name": station_name,
                "facilities": [],
            }
            _id += 1

        coded_records[station_code]["facilities"] += facilities

    grouped_records = {}

    for station_code, station_record in coded_records.items():
        grouped_records[station_code] = {
            "id": station_record.get("id"),
            "code": station_code,
            "name": station_name_cleaner(station_name),
            "network_name": station_name,
            "facilities": [],
        }

        for facility in station_record["facilities"]:
            grouped_records[station_code]["name"] = facility["name"]
            grouped_records[station_code]["network_name"] = facility[
                "network_name"]
            facility.pop("name")
            facility.pop("network_name")
            grouped_records[station_code]["facilities"].append(facility)

    return grouped_records