Esempio n. 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()
Esempio n. 2
0
def opennem_init() -> None:
    session = SessionLocal()
    station_data = load_data("opennem_stations.json", from_project=True)
    stations = StationSet()

    for s in station_data:
        stations.add_dict(s)

    for station in stations:
        logger.debug("Adding station: {}".format(station.code))

        station_model = session.query(Station).filter_by(
            code=station.code).one_or_none()

        if not station_model:
            station_model = Station(code=station.code)
            station_model.approved = True
            station_model.approved_at = datetime.now()
            station_model.approved_by = "opennem.init"
            station_model.created_by = "opennem.init"

        station_model.description = station.description
        station_model.name = station.name
        station_model.network_name = station.network_name

        if not station_model.location:
            station_model.location = Location()

        if station.location:
            station_model.location.locality = station.location.locality
            station_model.location.state = station.location.state
            station_model.location.postcode = station.location.postcode
            station_model.location.country = station.location.country

        if station.location.lat and station.location.lng:
            station_model.location.geom = "SRID=4326;POINT({} {})".format(
                station.location.lng, station.location.lat)

        for fac in station.facilities:
            facility_model = (session.query(Facility).filter_by(
                code=fac.code).filter_by(
                    network_id=fac.network.code).one_or_none())

            if not facility_model:
                facility_model = Facility(code=fac.code,
                                          network_id=fac.network.code)

            facility_model.network_region = fac.network_region
            facility_model.fueltech_id = fac.fueltech.code
            facility_model.status_id = fac.status.code
            facility_model.dispatch_type = fac.dispatch_type
            facility_model.capacity_registered = fac.capacity_registered
            facility_model.registered = fac.registered
            facility_model.unit_id = fac.unit_id
            facility_model.unit_number = fac.unit_number
            facility_model.unit_alias = fac.unit_alias
            facility_model.unit_capacity = fac.unit_capacity
            facility_model.emissions_factor_co2 = fac.emissions_factor_co2
            facility_model.approved = fac.approved
            facility_model.approved_by = fac.approved_by

            facility_model.created_by = "opennem.init"
            facility_model.approved_by = "opennem.init"

            session.add(facility_model)
            station_model.facilities.append(facility_model)
            logger.debug(" => Added facility {}".format(fac.code))

        print(station_model)
        session.add(station_model)
        session.commit()
Esempio n. 3
0
    def process_item(self, item, spider=None):

        s = self.session()

        records_updated = 0
        records_created = 0

        for record in item:
            created = False

            participant_code = normalize_duid(
                record["facilities"][0]["PARTICIPANTID"])

            # Step 1. Find participant by code or create
            participant = (s.query(ParticipantModel).filter(
                ParticipantModel.code == participant_code).one_or_none())

            if not participant:
                participant = ParticipantModel(
                    code=participant_code,
                    network_code=participant_code,
                    created_by="au.nem.mms.dudetail_summary",
                )
                logger.debug("Created participant {}".format(participant_code))
            else:
                participant.updated_by = "au.nem.mms.dudetail_summary"

            # Step 3. now create the facilities and associate
            for facility_record in record["facilities"]:
                duid = normalize_duid(facility_record["DUID"])
                station_code = facility_map_station(
                    duid, normalize_duid(record["id"]))

                network_region = normalize_aemo_region(
                    facility_record["REGIONID"])
                date_start = facility_record["date_start"]
                date_end = facility_record["date_end"]
                facility_state = "retired"

                # Step 2. Find station or create
                station = (s.query(Station).filter(
                    Station.network_code == station_code).one_or_none())

                if not station:
                    station = Station(
                        code=station_code,
                        network_code=station_code,
                        network_id="NEM",
                        created_by="au.nem.mms.dudetail_summary",
                    )
                    logger.debug("Created station {}".format(station_code))
                else:
                    station.updated_by = "au.nem.mms.dudetail_summary"

                station.participant = participant

                if date_end == None:
                    facility_state = "operating"

                if not "DISPATCHTYPE" in facility_record:
                    logger.error(
                        "MMS dudetailsummary: Invalid record: {}".format(
                            facility_record))
                    continue

                dispatch_type = parse_dispatch_type(
                    facility_record["DISPATCHTYPE"])

                facility = (s.query(Facility).filter(
                    Facility.network_code == duid).one_or_none())

                if not facility:

                    facility = Facility(
                        code=duid,
                        network_code=duid,
                        dispatch_type=dispatch_type,
                        created_by="au.nem.mms.dudetail_summary",
                    )

                    records_created += 1
                    created = True
                else:
                    facility.updated_by = "au.nem.mms.dudetail_summary"
                    records_updated += 1

                facility.network_region = network_region
                facility.deregistered = date_end
                facility.registered = date_start

                facility.status_id = facility_state

                if not facility.dispatch_type:
                    facility.dispatch_type = dispatch_type

                # Associations
                facility_station_id = facility_map_station(duid, station.id)

                facility.station_id = station.id

                try:
                    s.add(facility)
                    s.commit()
                except Exception as e:
                    logger.error(e)

            logger.debug(
                "MMS DudetailSummary:{} facility record with id {}".format(
                    "Created" if created else "Updated", duid))

        logger.info(
            "MMS DudetailSummary: Created {} facility records and updated {}".
            format(records_created, records_updated))
Esempio n. 4
0
    def process_item(self, item, spider):
        s = SessionLocal()

        records_added = 0
        csvreader = csv.DictReader(item["content"].split("\n"))

        for row in csvreader:

            if "Participant Code" not in row:
                logger.error("Invalid row")
                continue

            if row["Facility Type"] in [
                    "Demand Side Program",
                    "Non-Dispatchable Load",
                    "Network",
            ]:
                continue

            participant = None

            participant_name = participant_name_filter(row["Participant Name"])
            participant_network_name = normalize_string(
                row["Participant Name"])
            participant_code = normalize_duid(row["Participant Code"])

            participant = (s.query(Participant).filter(
                Participant.code == participant_code).one_or_none())

            if not participant:
                participant = Participant(
                    created_by=spider.name,
                    approved_at=datetime.now(),
                    code=participant_code,
                    name=participant_name,
                    network_name=participant_network_name,
                )
                s.add(participant)
                s.commit()

                logger.debug(
                    "Participant not found created new database entry: %s",
                    participant_code,
                )

            station = None
            facility = None

            facility_code = normalize_duid(row["Facility Code"])
            station_code = parse_wem_facility_code(facility_code)

            station = (s.query(Station).filter(
                Station.code == station_code).one_or_none())

            if not station:
                station = Station(
                    created_by="opennem.wem.facilities",
                    approved_at=datetime.now(),
                    code=station_code,
                    network_code=station_code,
                    participant=participant,
                )

                location = Location(state="WA")

                station.location = location

                logger.debug("Added WEM station: {}".format(station_code))

            facility = (s.query(Facility).filter(
                Facility.code == facility_code).one_or_none())

            if not facility:
                facility = Facility(
                    created_by=spider.name,
                    approved_at=datetime.now(),
                    code=facility_code,
                    network_id="WEM",
                    network_code=facility_code,
                    network_region="WEM",
                )

            capacity_registered = clean_capacity(row["Maximum Capacity (MW)"])
            capacity_unit = clean_capacity(row["Maximum Capacity (MW)"])
            registered_date = row["Registered From"]

            facility.status_id = "operating"
            facility.capacity_registered = capacity_registered
            facility.unit_id = 1
            facility.unit_number = 1
            facility.unit_capacity = capacity_unit

            if registered_date:
                registered_date_dt = datetime.strptime(registered_date,
                                                       "%Y-%m-%d %H:%M:%S")
                facility.registered = registered_date_dt

            facility.station = station

            s.add(facility)
            records_added += 1

        try:
            s.commit()
        except IntegrityError as e:
            logger.error(e)
            pass
        except Exception as e:
            logger.error("Error: {}".format(e))
        finally:
            s.close()

        return records_added
Esempio n. 5
0
    def process_item(self, item, spider=None):

        s = SessionLocal()

        records_added = 0
        csvreader = csv.DictReader(item["content"].split("\n"))

        for row in csvreader:

            if "PARTICIPANT_CODE" not in row:
                logger.error("Invalid row")
                continue

            participant = None

            participant_code = normalize_duid(row["PARTICIPANT_CODE"])

            participant = (s.query(Participant).filter(
                Participant.code == participant_code).one_or_none())

            if not participant:
                participant = Participant(
                    created_by=spider.name,
                    approved_by=spider.name,
                    approved_at=datetime.now(),
                    code=participant_code,
                    network_code=participant_code,
                    network="WEM",
                )

                s.add(participant)
                s.commit()

                logger.warning(
                    "Participant not found created new database entry: {}".
                    format(participant_code))

            station = None
            facility = None

            facility_code = normalize_duid(row["FACILITY_CODE"])
            station_code = parse_wem_facility_code(facility_code)
            station_name = station_name_cleaner(row["DISPLAY_NAME"])
            station_network_name = normalize_string(row["DISPLAY_NAME"])

            station = (s.query(Station).filter(
                Station.code == station_code).one_or_none())

            if not station:
                station = Station(
                    created_by=spider.name,
                    approved_by=spider.name,
                    approved_at=datetime.now(),
                    code=station_code,
                    network_code=station_code,
                    participant=participant,
                )

                location = Location(state="WA")
                s.add(location)

                station.location = location

                logger.debug("Added WEM station: {}".format(station_code))

            lat = row["LATITUDE"]
            lng = row["LONGITUDE"]

            station.name = station_name
            station.network_name = station_network_name

            if lat and lng and not station.location.geom:
                station.location.geom = "SRID=4326;POINT({} {})".format(
                    lat, lng)
                station.location.geocode_by = "aemo"
                station.location.geocode_approved = True

            facility = (s.query(Facility).filter(
                Facility.code == facility_code).one_or_none())

            if not facility:
                facility = Facility(
                    created_by="opennem.wem.live.facilities",
                    approved_at=datetime.now(),
                    network_id="WEM",
                    code=facility_code,
                    network_code=facility_code,
                    network_region="WEM",
                )

            registered_date = row["YEAR_COMMISSIONED"]

            if registered_date:
                registered_date_dt = None

                date_fmt = "%Y"

                try:
                    registered_date_dt = datetime.strptime(
                        registered_date, date_fmt)
                except Exception:
                    logger.error("Bad date: %s for format %s", registered_date,
                                 date_fmt)

                if registered_date_dt:
                    facility.registered = registered_date_dt

            fueltech = lookup_fueltech(fueltype=row["PRIMARY_FUEL"],
                                       techtype=row["FACILITY_TYPE"])

            if fueltech and not facility.fueltech:
                facility.fueltech_id = fueltech

            facility.status_id = "operating"
            facility.station = station

            s.add(station)
            s.add(facility)
            s.commit()
            records_added += 1

        try:
            s.commit()
        except IntegrityError as e:
            logger.error(e)
            pass
        except Exception as e:
            logger.error("Error: {}".format(e))
        finally:
            s.close()

        return records_added
Esempio n. 6
0
def import_station_set(stations: StationSet,
                       only_insert_facilities: bool = False) -> None:
    session = SessionLocal()

    for station in stations:
        add_or_update: str = "Updating"
        station_model = session.query(Station).filter_by(
            code=station.code).one_or_none()

        if not station_model:
            add_or_update = "Adding"
            station_model = Station(code=station.code)
            station_model.created_by = "opennem.init"

        logger.debug("{} station: {}".format(add_or_update, station.code))

        if station.description:
            station_model.description = station.description

        if station.name:
            station_model.name = station.name

        station_model.approved = station.approved

        if station.approved:
            station_model.approved = True
            station_model.approved_at = datetime.now()
            station_model.approved_by = "opennem.init"
        else:
            station_model.approved_at = None
            station_model.approved_by = None

        if station.website_url:
            station_model.website_url = station.website_url

        if station.network_name:
            station_model.network_name = station.network_name

        if not station_model.location:
            station_model.location = Location()

        if station.location:
            station_model.location.locality = station.location.locality
            station_model.location.state = station.location.state
            station_model.location.postcode = station.location.postcode
            station_model.location.country = station.location.country

        if station.location.lat and station.location.lng:
            station_model.location.geom = "SRID=4326;POINT({} {})".format(
                station.location.lng, station.location.lat)

        session.add(station_model)
        session.commit()

        for fac in station.facilities:
            facility_added = False

            facility_model = (session.query(Facility).filter_by(
                code=fac.code).filter_by(
                    network_id=fac.network.code).one_or_none())

            if facility_model and only_insert_facilities:
                logger.debug(" => skip updating {}".format(
                    facility_model.code))
                continue

            if not facility_model:
                facility_model = Facility(code=fac.code,
                                          network_id=fac.network.code)
                facility_added = True

            if facility_model.station_id != station_model.id:
                facility_model.station_id = station_model.id
                logger.debug(" => Reassigned facility {} to station {}".format(
                    facility_model.code, station_model.code))

            # fueltech
            if fac.fueltech:
                facility_model.fueltech_id = fac.fueltech.code

            if fac.fueltech_id:
                facility_model.fueltech_id = fac.fueltech_id

            # network
            if fac.network:
                facility_model.network_id = fac.network.code

            if fac.network_id:
                facility_model.network_id = fac.network_id

            # status
            if fac.status:
                facility_model.status_id = fac.status.code

            if fac.status_id:
                facility_model.status_id = fac.status_id

            # rest
            if fac.dispatch_type:
                facility_model.dispatch_type = fac.dispatch_type

            if fac.capacity_registered:
                facility_model.capacity_registered = fac.capacity_registered

            if fac.registered:
                facility_model.registered = fac.registered

            if fac.network_region:
                facility_model.network_region = fac.network_region

            facility_model.unit_id = fac.unit_id
            facility_model.unit_number = fac.unit_number
            facility_model.unit_alias = fac.unit_alias
            facility_model.unit_capacity = fac.unit_capacity

            if fac.emissions_factor_co2:
                facility_model.emissions_factor_co2 = fac.emissions_factor_co2

            if fac.approved:
                facility_model.approved = fac.approved

            if fac.approved:
                facility_model.approved_by = "opennem.importer"
            else:
                facility_model.approved_by = None

            if not facility_model.created_by:
                facility_model.created_by = "opennem.init"

            session.add(facility_model)
            station_model.facilities.append(facility_model)
            logger.debug(" => {} facility {} to {} {}".format(
                "Added" if facility_added else "Updated",
                fac.code,
                facility_model.network_id,
                facility_model.network_region,
            ))

        session.add(station_model)
        session.commit()