run.py

"""
Aim
---
Process an MSC dataset, given a set of parameters.

Example
-------
python -m csst_dag.cli.run -h

python -m csst_dag.cli.run \
    --dags csst-msc-l1-mbi \
    --dataset=csst-msc-c9-25sqdeg-v3 \
    --instrument=MSC \
    --obs-type=WIDE \
    --obs-group=W2 \
    --obs-id=10100232366 \
    --detector=09 \
    --batch-id=test-b1 \
    --priority=1 \
    --pmapname=csst_000070.pmap \
    --ref-cat=trilegal_093

# 25平方度宽场
python -m csst_dag.cli.run \
    --dags csst-msc-l1-mbi \
    --dataset=csst-msc-c9-25sqdeg-v3 \
    --instrument=MSC \
    --obs-type=WIDE \
    --prc-status=-2 \
    --batch-id=25sqdeg-test-b2 \
    --priority=1 \
    --pmapname=csst_000070.pmap \
    --ref-cat=trilegal_093 \
    --submit

"""

import argparse
import os
import json

import joblib

from csst_dag import CSST_DAGS, Dispatcher, BaseDAG, dfs

parser = argparse.ArgumentParser(
    description="Scheduler for CSST L1 pipeline.",
    formatter_class=argparse.ArgumentDefaultsHelpFormatter,
)

# data parameters
parser.add_argument("--dataset", type=str, help="Dataset name")
parser.add_argument("--instrument", type=str, help="Instrument name", default=None)
parser.add_argument("--obs-type", type=str, help="Observation type", default=None)
parser.add_argument("--obs-group", type=str, help="Observation group", default=None)
parser.add_argument("--obs-id", type=str, help="Observation ID", default=None)
parser.add_argument("--detector", type=str, help="Detector name", default=None)
parser.add_argument(
    "--prc-status", type=int, help="Initial processing status", default=None
)
parser.add_argument(
    "--qc-status", type=int, help="Initial QC status", default=None
)
# task parameters
parser.add_argument("--batch-id", type=str, help="Batch ID", default="test-batch")
parser.add_argument("--priority", type=str, help="Task priority", default="1")
# DAG parameters
parser.add_argument("--pmapname", type=str, help="CCDS pmapname", default="")
parser.add_argument(
    "--ref-cat", type=str, help="Reference catalog", default="trilegal_093"
)
# submit
parser.add_argument("--submit", action="store_true", help="Push results", default=False)
# post-processing parameters
parser.add_argument(
    "--final-prc-status", type=int, help="Final processing status", default=-2
)
# additional options
parser.add_argument("--force", action="store_true", help="Force success", default=False)
parser.add_argument("--verbose", action="store_true", help="Force success", default=False)
# submit top N
parser.add_argument(
    "--top-n", type=int, help="Submit top N tasks", default=-1
)
# select DAGs
parser.add_argument('--dags', nargs='+', type=str, help="DAGs to select", default=None)


args = parser.parse_args()


# from csst_dag import DotDict
#
# args = DotDict(
#     dataset="csst-msc-c9-25sqdeg-v3",
#     instrument="MSC",
#     obs_type="WIDE",
#     obs_group="W2",
#     obs_id="10100232366",
#     detector=None,
#     prc_status=None,
#     batch_id="test-batch",
#     priority=1,
#     pmapname="csst_000070.pmap",
#     ref_cat="trilegal_093",
#     submit=False,
# )

print("CLI parameters: ", args)

plan_basis, data_basis = Dispatcher.find_plan_level0_basis(
    dataset=args.dataset,
    instrument=args.instrument,
    obs_type=args.obs_type,
    obs_group=args.obs_group,
    obs_id=args.obs_id,
    detector=args.detector,
    prc_status=args.prc_status,
    qc_status=args.qc_status,
)
print(f"{len(plan_basis)} plan basis, {len(data_basis)} data basis found")

# generate DAG group run
dag_group_run = BaseDAG.gen_dag_group_run(
    dag_group="csst-l1-pipeline",
    batch_id=args.batch_id,
    priority=args.priority,
)

# generate DAG run list
print("\n")
print(">>> Matching DAGs ...")
dag_run_list = []
data_id_list = []
n_dag_run_all = 0
n_dag_run_success = 0

DEFAULT_DAGS = {
    "csst-msc-l1-mbi",
    "csst-msc-l1-ast",
    "csst-msc-l1-sls",
    "csst-msc-l1-qc0",
    "csst-msc-l1-ooc",
    "csst-mci-l1-qc0",
    "csst-mci-l1",
    "csst-ifs-l1",
    "csst-cpic-l1-qc0",
    "csst-cpic-l1",
    "csst-hstdm-l1",
}
assert args.dags is None or set(args.dags).issubset(DEFAULT_DAGS), f"Selected DAGs: {args.dags}"
SELECTED_DAGS = DEFAULT_DAGS.intersection(args.dags) if args.dags is not None else DEFAULT_DAGS
print("Selected DAGs: ", SELECTED_DAGS)

for dag in SELECTED_DAGS:
    this_task_list = CSST_DAGS[dag].schedule(
        dag_group_run=dag_group_run,
        plan_basis=plan_basis,
        data_basis=data_basis,
        pmapname=args.pmapname,
        ref_cat=args.ref_cat,
        force_success=args.force,
    )
    this_dag_run_list = [this_task["dag_run"] for this_task in this_task_list if this_task["dag_run"] is not None]
    this_data_id_list = [this_task["relevant_data_id_list"] for this_task in this_task_list if
                         this_task["dag_run"] is not None]

    dag_run_list.extend(this_dag_run_list)
    data_id_list.extend(this_data_id_list)
    this_n_dag_run_all = len(this_task_list)
    this_n_dag_run_success = len(this_dag_run_list)
    n_dag_run_all += this_n_dag_run_all
    n_dag_run_success += this_n_dag_run_success
    print(f"- `{dag}` : [ {this_n_dag_run_success} / {this_n_dag_run_all} ] dag_runs")

# print dag_group_run and dag_run_list
if args.verbose:
    print("\n")
    print(">>> `dag_group_run` :")
    print(f"\t- {json.dumps(dag_group_run, separators=(',', ':'))}")
    print(f">>> `dag_run_list` : [ {n_dag_run_success} / {n_dag_run_all} ]")
    if len(dag_run_list) > 0:
        for dag_run in dag_run_list:
            try:
                print(f"\t- {json.dumps(dag_run, separators=(',', ':'))}")
            except:
                print(f"\t- {dag_run}")

# dump dag_group_run
joblib.dump(
    dict(
        plan_basis=plan_basis,
        data_basis=data_basis,
        dag_group_run=dag_group_run,
        dag_run_list=dag_run_list,
    ),
    os.path.join(
        os.getenv("HOME"),
        "csst_dag",
        f"{dag_group_run['dag_group_run']}.joblib",
    ),
)
# submit DAG group run
if args.submit:
    res = dfs.dag.new_dag_group_run(
        dag_group_run=dag_group_run,
        dag_run_list=dag_run_list,
    )
    print(res)