schedule returns task_list instead of dag_run_list

csst_dag/cli/csst_hstdm_l1.py

@@ -83,20 +83,22 @@ dag_group_run = BaseDAG.gen_dag_group_run(
 print("\n"*2)
 print(">>> Matching DAGs ...")
 dag_run_list = []
-data_id_set = set()
+data_id_list = []
 for dag in [
     "csst-hstdm-l1",
 ]:
-    this_dag_run_list, this_data_id_set = CSST_DAGS[dag].schedule(
+    this_task_list = CSST_DAGS[dag].schedule(
         dag_group_run=dag_group_run,
         plan_basis=plan_basis,
         data_basis=data_basis,
         pmapname=args.pmapname,
         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"] for this_task in this_task_list if this_task["dag_run"] is not None]
     print(f"- [{dag}] : {len(this_dag_run_list)} dag_runs")
     dag_run_list.extend(this_dag_run_list)
-    data_id_set.union(this_data_id_set)
+    data_id_list.extend(this_data_id_list)
 
 # print dag_group_run and dag_run_list
 print("\n")

csst_dag/cli/csst_mci_l1.py

+"""
+Aim
+---
+Process an MCI dataset.
+
+Example
+-------
+python -m csst_dag.cli.csst_mci_l1 -h
+
+python -m csst_dag.cli.csst_mci_l1 \
+    --dataset=csst-cpic-c11-hip71681-v1 \
+    --instrument=CPIC \
+    --obs-type=SCI \
+    --obs-group=hip71681 \
+    --obs-id=40100000001 \
+    --detector=VIS \
+    --prc-status=-1024 \
+    --batch-id=test-b1 \
+    --priority=1
+"""
+
+import argparse
+import os
+
+import joblib
+
+from csst_dag import CSST_DAGS, Dispatcher, BaseDAG, dfs
+
+parser = argparse.ArgumentParser(
+    description="Scheduler for CSST CPIC 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
+)
+# 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
+)
+
+args = parser.parse_args()
+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,
+)
+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-cpic-l1",
+    batch_id=args.batch_id,
+    priority=args.priority,
+)
+
+# generate DAG run list
+print("\n"*2)
+print(">>> Matching DAGs ...")
+dag_run_list = []
+data_id_list = []
+for dag in [
+    "csst-cpic-l1",
+    "csst-cpic-l1-qc0",
+]:
+    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,
+    )
+    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"] for this_task in this_task_list if this_task["dag_run"] is not None]
+    print(f"- [{dag}] : {len(this_dag_run_list)} dag_runs")
+    dag_run_list.extend(this_dag_run_list)
+    data_id_list.extend(this_data_id_list)
+
+# print dag_group_run and dag_run_list
+print("\n")
+print(">>> dag_group_run:")
+print(f"\t- {dag_group_run}")
+print(f">>> dag_run_list[{len(dag_run_list)}]:")
+if len(dag_run_list) > 0:
+    for dag_run in dag_run_list:
+        print(f"\t- {dag_run}")
+
+# dump dag_group_run
+joblib.dump(
+    dict(
+        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)

csst_dag/cli/csst_msc_l1.py

@@ -32,14 +32,6 @@ python -m csst_dag.cli.csst_msc_l1 \
     --ref-cat=trilegal_093 \
     --submit
 
-# 1000 平方度宽场测试天区 50个指向
-python -m csst_dag.cli.csst_msc_l1 \
-    --dataset=csst-msc-c11-1000sqdeg-wide-test-v1 \
-    --instrument=MSC \
-    --prc-status=-1024 \
-    --batch-id=1000sqdeg-test-b1 \
-    --ref-cat=trilegal_1000_w1 \
-    --submit
 """
 
 import argparse
@@ -123,23 +115,25 @@ dag_group_run = BaseDAG.gen_dag_group_run(
 print("\n"*2)
 print(">>> Matching DAGs ...")
 dag_run_list = []
-data_id_set = set()
+data_id_list = []
 for dag in [
     "csst-msc-l1-mbi",
     "csst-msc-l1-ast",
     "csst-msc-l1-sls",
     "csst-msc-l1-qc0",
 ]:
-    this_dag_run_list, this_data_id_set = CSST_DAGS[dag].schedule(
+    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,
     )
+    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"] for this_task in this_task_list if this_task["dag_run"] is not None]
     print(f"- [{dag}] : {len(this_dag_run_list)} dag_runs")
     dag_run_list.extend(this_dag_run_list)
-    data_id_set.union(this_data_id_set)
+    data_id_list.extend(this_data_id_list)
 
 # print dag_group_run and dag_run_list
 print("\n")

csst_dag/dag/_base_dag.py

@@ -103,9 +103,6 @@ class BaseDAG:
         )
         # dispatch tasks
         task_list = self.dispatcher(filtered_plan_basis, filtered_data_basis)
-        # convert tasks to dag_run_list
-        dag_run_list = []  # each element is a dag_run (dict)
-        data_id_set = set()  # each element is a data_id (str)
         for this_task in task_list:
             # only convert success tasks
             if force_success or this_task["success"]:
@@ -114,9 +111,10 @@ class BaseDAG:
                     **this_task["task"],
                     **kwargs,
                 )
-                dag_run_list.append(dag_run)
-                data_id_set.union(this_task["relevant_data"]["_id"])
-        return dag_run_list, data_id_set
+                this_task["dag_run"] = dag_run
+            else:
+                this_task["dag_run"] = dict()
+        return task_list
 
     @staticmethod
     def generate_sha1():