def __init__()

in scheduler.py [0:0]

• 32 lines of code
• 9 McCabe index (conditional complexity)

    def __init__(self, dir_path, table_size_path, weight_lookup=None,
                 yugong=False, ownership=None):
        assert os.path.exists(dir_path), f"dir_path={dir_path} not found"
        self.dir_path = dir_path
        logging.info(f"dir_path: {dir_path}, yugong mode: {yugong}")

        self.yugong = yugong
        if self.yugong:
            assert ownership is not None, "ownership is None"
            self.ownership = ownership

        # load template and dataset placements
        # query_kv = self._load_query_placement()
        # self.query_map = Cache(maxsize=len(query_kv) // 100, kv_store=query_kv)
        # logging.info(f"# of query placements: {len(query_kv)}")
        self.query_map = self._load_query_placement()
        logging.info(f"# of query placements: {len(self.query_map)}")
        self.dataset_map = self._load_dataset_placement()
        logging.info(f"# of dataset placements: {len(self.dataset_map)}")

        # load table sizes
        logging.info(f"Loaded table size: {table_size_path}")

        self.table_size_map = self._load_table_sizes(table_size_path)

        self.size_lookup = None
        self.db_table_size = None

        self.weight_lookup = weight_lookup

        # historical workload
        self.stat_cloud_query = Stat("cloud queries")
        self.stat_on_prem_query = Stat("on-prem queries")

        self.stat_categories = {
            "both_sides": Stat("Old queries with tables on both sides"), # all_table_local & all_table_cloud == True
            "only_cloud": Stat("Old queries with all tables on cloud"), # all_table_local == False and all_table_cloud == True
            "only_onprem": Stat("Old queries with all tables on-prem"), # all_table_local == True and all_table_cloud == False
            "needs_transfer": Stat("Old queries requiring ingress/egress"), # all_table_local == False and all_table_cloud == False
        }

        self.stat_categories_new = {
            "both_sides": Stat("New queries with tables on both sides"),  # all_table_local & all_table_cloud == True
            "only_cloud": Stat("New queries with all tables on cloud"),
            # all_table_local == False and all_table_cloud == True
            "only_onprem": Stat("New queries with all tables on-prem"),
            # all_table_local == True and all_table_cloud == False
            "needs_transfer": Stat("New queries requiring ingress/egress"),
            # all_table_local == False and all_table_cloud == False
        }