def __init__()

in baselines.py [0:0]

• 35 lines of code
• 5 McCabe index (conditional complexity)

    def __init__(self, tag: str, cloud_target: int, rep_rate: float):
        self.parent_dir = "baselines"
        os.makedirs(self.parent_dir, exist_ok=True)

        self.tag = tag
        my_dir_name = f"{tag}_c{cloud_target}_rep{rep_rate:.3f}_{datetime.now().strftime('%Y%m%d_%H%M%S')}"
        self.my_dir = os.path.join(self.parent_dir, my_dir_name)
        os.makedirs(self.my_dir, exist_ok=True)

        self.cloud_target = cloud_target
        assert 0 <= cloud_target <= 100, "cloud_target must be in [0, 100]"

        self.rep_rate = rep_rate

        self.table_size_lookup = self._load_table_size()
        self.total_data_size = sum(self.table_size_lookup.values())
        self.on_prem_data_size = 0
        self.on_prem_capacity = self.total_data_size * (100 - cloud_target) / 100

        # header: abstractFingerPrint,db_name,table_name,inputDataSize,outputDataSize
        self.workload = self._load_workload()

        self.placement = {} # db_table name -> Status
        if tag == "rep_x_month":
            assert rep_rate in [0.21] # 3M
            self.on_prem_data_size = self.total_data_size * rep_rate
            assert self.on_prem_data_size <= self.on_prem_capacity, "Not enough capacity for replication"
            self.rep_x_month_placement()
        elif tag == "rep_rtd":
            self.preselect_replication()
            self.data_placement_random()
        elif tag == "MoiJob":
            self.preselect_replication()
            df = self.get_moirai_job_distribution()
            self.data_placement_by_compute_distribution(df)
            #self.volley_placement()
        elif tag == "volley_new":
            self.preselect_replication()
            df = self.get_random_project_distribution()
            self.data_placement_by_compute_distribution(df)
        else:
            raise ValueError(f"Unknown baseline: {tag}")

        self.persist_placement()