Testing of how services handle find_existing = False

qcfractal/qcfractal/components/gridoptimization/test_record_socket.py

@@ -176,11 +176,12 @@ def test_gridoptimization_socket_run(
     meta_1, id_1 = storage_socket.records.gridoptimization.add(
         [molecules_1], input_spec_1, "test_tag", PriorityEnum.low, "submit_user", "group1", True
     )
+    id_1 = id_1[0]
     assert meta_1.success
 
     time_0 = datetime.utcnow()
     finished, n_optimizations = run_service(
-        storage_socket, activated_manager_name, id_1[0], generate_task_key, result_data_1, 100
+        storage_socket, activated_manager_name, id_1, generate_task_key, result_data_1, 100
     )
     time_1 = datetime.utcnow()
 
@@ -196,7 +197,7 @@ def test_gridoptimization_socket_run(
     assert time_0 < rec.compute_history[-1].modified_on < time_1
     assert rec.service is None
 
-    desc_info = storage_socket.records.get_short_descriptions(id_1)[0]
+    desc_info = storage_socket.records.get_short_descriptions([id_1])[0]
     short_desc = desc_info["description"]
     assert desc_info["record_type"] == rec.record_type
     assert desc_info["created_on"] == rec.created_on
@@ -213,3 +214,39 @@ def test_gridoptimization_socket_run(
     for o in rec.optimizations:
         optr = session.get(OptimizationRecordORM, o.optimization_id)
         assert optr.energies[-1] == o.energy
+
+
+def test_gridoptimization_socket_run_duplicate(
+    storage_socket: SQLAlchemySocket,
+    session: Session,
+    activated_manager_name: ManagerName,
+):
+    input_spec_1, molecules_1, result_data_1 = load_test_data("go_H2O2_psi4_b3lyp")
+
+    meta_1, id_1 = storage_socket.records.gridoptimization.add(
+        [molecules_1], input_spec_1, "test_tag", PriorityEnum.low, None, None, True
+    )
+    id_1 = id_1[0]
+    assert meta_1.success
+
+    run_service(storage_socket, activated_manager_name, id_1, generate_task_key, result_data_1, 100)
+
+    rec_1 = session.get(GridoptimizationRecordORM, id_1)
+    assert rec_1.status == RecordStatusEnum.complete
+    opt_ids_1 = [x.optimization_id for x in rec_1.optimizations]
+
+    # Submit again, without duplicate checking
+    meta_2, id_2 = storage_socket.records.gridoptimization.add(
+        [molecules_1], input_spec_1, "test_tag", PriorityEnum.low, None, None, False
+    )
+    id_2 = id_2[0]
+    assert meta_2.success
+    assert id_2 != id_1
+
+    run_service(storage_socket, activated_manager_name, id_2, generate_task_key, result_data_1, 1000)
+
+    rec_2 = session.get(GridoptimizationRecordORM, id_2)
+    assert rec_2.status == RecordStatusEnum.complete
+    opt_ids_2 = [x.optimization_id for x in rec_2.optimizations]
+
+    assert set(opt_ids_1).isdisjoint(opt_ids_2)

qcfractal/qcfractal/components/manybody/test_record_socket.py

@@ -98,11 +98,12 @@ def test_manybody_socket_run(
     meta_1, id_1 = storage_socket.records.manybody.add(
         [molecules_1], input_spec_1, "test_tag", PriorityEnum.low, "submit_user", "group1", True
     )
+    id_1 = id_1[0]
     assert meta_1.success
 
     time_0 = datetime.utcnow()
     finished, n_singlepoints = run_service(
-        storage_socket, activated_manager_name, id_1[0], generate_task_key, result_data_1, 100
+        storage_socket, activated_manager_name, id_1, generate_task_key, result_data_1, 100
     )
     time_1 = datetime.utcnow()
 
@@ -118,7 +119,7 @@ def test_manybody_socket_run(
     assert time_0 < rec.compute_history[-1].modified_on < time_1
     assert rec.service is None
 
-    desc_info = storage_socket.records.get_short_descriptions(id_1)[0]
+    desc_info = storage_socket.records.get_short_descriptions([id_1])[0]
     short_desc = desc_info["description"]
     assert desc_info["record_type"] == rec.record_type
     assert desc_info["created_on"] == rec.created_on
@@ -130,3 +131,39 @@ def test_manybody_socket_run(
     assert "All manybody singlepoint computations are complete" in out
 
     assert len(rec.clusters) == n_singlepoints
+
+
+def test_manybody_socket_run_duplicate(
+    storage_socket: SQLAlchemySocket,
+    session: Session,
+    activated_manager_name: ManagerName,
+):
+    input_spec_1, molecules_1, result_data_1 = load_test_data("mb_cp_he4_psi4_mp2")
+
+    meta_1, id_1 = storage_socket.records.manybody.add(
+        [molecules_1], input_spec_1, "test_tag", PriorityEnum.low, None, None, True
+    )
+    id_1 = id_1[0]
+    assert meta_1.success
+
+    run_service(storage_socket, activated_manager_name, id_1, generate_task_key, result_data_1, 1000)
+
+    rec_1 = session.get(ManybodyRecordORM, id_1)
+    assert rec_1.status == RecordStatusEnum.complete
+    sp_ids_1 = [x.singlepoint_id for x in rec_1.clusters]
+
+    # Submit again, without duplicate checking
+    meta_2, id_2 = storage_socket.records.manybody.add(
+        [molecules_1], input_spec_1, "test_tag", PriorityEnum.low, None, None, False
+    )
+    id_2 = id_2[0]
+    assert meta_2.success
+    assert id_2 != id_1
+
+    run_service(storage_socket, activated_manager_name, id_2, generate_task_key, result_data_1, 1000)
+
+    rec_2 = session.get(ManybodyRecordORM, id_2)
+    assert rec_2.status == RecordStatusEnum.complete
+    sp_ids_2 = [x.singlepoint_id for x in rec_2.clusters]
+
+    assert set(sp_ids_1).isdisjoint(sp_ids_2)

qcfractal/qcfractal/components/neb/test_record_socket.py

@@ -245,3 +245,41 @@ def test_neb_socket_run(
 
     out = rec.compute_history[-1].outputs["stdout"].get_output()
     assert "NEB calculation is completed" in out
+
+
+def test_neb_socket_run_duplicate(
+    storage_socket: SQLAlchemySocket,
+    session: Session,
+    activated_manager_name: ManagerName,
+):
+    input_spec_1, molecules_1, result_data_1 = load_test_data("neb_HCN_psi4_pbe_opt_diff")
+
+    meta_1, id_1 = storage_socket.records.neb.add(
+        [molecules_1], input_spec_1, "test_tag", PriorityEnum.low, None, None, True
+    )
+    id_1 = id_1[0]
+    assert meta_1.success
+
+    run_service(storage_socket, activated_manager_name, id_1, generate_task_key, result_data_1, 100)
+
+    rec_1 = session.get(NEBRecordORM, id_1)
+    assert rec_1.status == RecordStatusEnum.complete
+    sp_ids_1 = [x.singlepoint_id for x in rec_1.singlepoints]
+    opt_ids_1 = [x.optimization_id for x in rec_1.optimizations]
+
+    # Submit again, without duplicate checking
+    meta_2, id_2 = storage_socket.records.neb.add(
+        [molecules_1], input_spec_1, "test_tag", PriorityEnum.low, None, None, False
+    )
+    id_2 = id_2[0]
+    assert meta_2.success
+    assert id_2 != id_1
+
+    run_service(storage_socket, activated_manager_name, id_2, generate_task_key, result_data_1, 1000)
+
+    rec_2 = session.get(NEBRecordORM, id_2)
+    assert rec_2.status == RecordStatusEnum.complete
+    sp_ids_2 = [x.singlepoint_id for x in rec_2.singlepoints]
+    opt_ids_2 = [x.optimization_id for x in rec_2.optimizations]
+
+    assert set(sp_ids_1 + opt_ids_1).isdisjoint(sp_ids_2 + opt_ids_2)

qcfractal/qcfractal/components/reaction/test_record_socket.py

@@ -6,7 +6,7 @@
 import pytest
 
 from qcarchivetesting import load_molecule_data
-from qcfractal.components.record_db_models import BaseRecordORM
+from qcfractal.components.reaction.record_db_models import ReactionRecordORM
 from qcfractal.db_socket import SQLAlchemySocket
 from qcfractal.testing_helpers import run_service
 from qcportal.auth import UserInfo, GroupInfo
@@ -34,7 +34,7 @@ def test_reaction_socket_add_get(storage_socket: SQLAlchemySocket, session: Sess
     time_1 = datetime.utcnow()
     assert meta.success
 
-    recs = [session.get(BaseRecordORM, i) for i in ids]
+    recs = [session.get(ReactionRecordORM, i) for i in ids]
     assert len(recs) == 2
 
     for r in recs:
@@ -116,17 +116,18 @@ def test_reaction_socket_run(
     meta_1, id_1 = storage_socket.records.reaction.add(
         [stoich_1], input_spec_1, "test_tag", PriorityEnum.low, "submit_user", "group1", True
     )
+    id_1 = id_1[0]
     assert meta_1.success
 
     time_0 = datetime.utcnow()
     finished, n_singlepoints = run_service(
-        storage_socket, activated_manager_name, id_1[0], generate_task_key, result_data_1, 100
+        storage_socket, activated_manager_name, id_1, generate_task_key, result_data_1, 100
     )
     time_1 = datetime.utcnow()
 
     assert finished is True
 
-    rec = session.get(BaseRecordORM, id_1)
+    rec = session.get(ReactionRecordORM, id_1)
 
     assert rec.status == RecordStatusEnum.complete
     assert time_0 < rec.modified_on < time_1
@@ -136,7 +137,7 @@ def test_reaction_socket_run(
     assert time_0 < rec.compute_history[-1].modified_on < time_1
     assert rec.service is None
 
-    desc_info = storage_socket.records.get_short_descriptions(id_1)[0]
+    desc_info = storage_socket.records.get_short_descriptions([id_1])[0]
     short_desc = desc_info["description"]
     assert desc_info["record_type"] == rec.record_type
     assert desc_info["created_on"] == rec.created_on
@@ -152,3 +153,41 @@ def test_reaction_socket_run(
     assert "All reaction components are complete" in out
 
     assert rec.total_energy < 0.0
+
+
+def test_reaction_socket_run_duplicate(
+    storage_socket: SQLAlchemySocket,
+    session: Session,
+    activated_manager_name: ManagerName,
+):
+    input_spec_1, molecules_1, result_data_1 = load_test_data("rxn_H2O_psi4_mp2_optsp")
+
+    meta_1, id_1 = storage_socket.records.reaction.add(
+        [molecules_1], input_spec_1, "test_tag", PriorityEnum.low, None, None, True
+    )
+    id_1 = id_1[0]
+    assert meta_1.success
+
+    run_service(storage_socket, activated_manager_name, id_1, generate_task_key, result_data_1, 1000)
+
+    rec_1 = session.get(ReactionRecordORM, id_1)
+    assert rec_1.status == RecordStatusEnum.complete
+    opt_ids_1 = [x.optimization_id for x in rec_1.components]
+    sp_ids_1 = [x.singlepoint_id for x in rec_1.components]
+
+    # Submit again, without duplicate checking
+    meta_2, id_2 = storage_socket.records.reaction.add(
+        [molecules_1], input_spec_1, "test_tag", PriorityEnum.low, None, None, False
+    )
+    id_2 = id_2[0]
+    assert meta_2.success
+    assert id_2 != id_1
+
+    run_service(storage_socket, activated_manager_name, id_2, generate_task_key, result_data_1, 1000)
+
+    rec_2 = session.get(ReactionRecordORM, id_2)
+    assert rec_2.status == RecordStatusEnum.complete
+    opt_ids_2 = [x.optimization_id for x in rec_2.components]
+    sp_ids_2 = [x.singlepoint_id for x in rec_2.components]
+
+    assert set(opt_ids_1 + sp_ids_1).isdisjoint(opt_ids_2 + sp_ids_2)

qcfractal/qcfractal/components/torsiondrive/test_record_socket.py

@@ -293,11 +293,12 @@ def test_torsiondrive_socket_run(
     meta_1, id_1 = storage_socket.records.torsiondrive.add(
         [molecules_1], input_spec_1, True, "test_tag", PriorityEnum.low, "submit_user", "group1", True
     )
+    id_1 = id_1[0]
     assert meta_1.success
 
     time_0 = datetime.utcnow()
     finished, n_optimizations = run_service(
-        storage_socket, activated_manager_name, id_1[0], generate_task_key, result_data_1, 1000
+        storage_socket, activated_manager_name, id_1, generate_task_key, result_data_1, 1000
     )
     time_1 = datetime.utcnow()
 
@@ -311,7 +312,7 @@ def test_torsiondrive_socket_run(
     assert time_0 < rec.compute_history[-1].modified_on < time_1
     assert rec.service is None
 
-    desc_info = storage_socket.records.get_short_descriptions(id_1)[0]
+    desc_info = storage_socket.records.get_short_descriptions([id_1])[0]
     short_desc = desc_info["description"]
     assert desc_info["record_type"] == rec.record_type
     assert desc_info["created_on"] == rec.created_on
@@ -324,3 +325,39 @@ def test_torsiondrive_socket_run(
     assert "Job Finished" in out
 
     assert len(rec.optimizations) == n_optimizations
+
+
+def test_torsiondrive_socket_run_duplicate(
+    storage_socket: SQLAlchemySocket,
+    session: Session,
+    activated_manager_name: ManagerName,
+):
+    input_spec_1, molecules_1, result_data_1 = load_test_data("td_H2O2_mopac_pm6")
+
+    meta_1, id_1 = storage_socket.records.torsiondrive.add(
+        [molecules_1], input_spec_1, True, "test_tag", PriorityEnum.low, None, None, True
+    )
+    id_1 = id_1[0]
+    assert meta_1.success
+
+    run_service(storage_socket, activated_manager_name, id_1, generate_task_key, result_data_1, 1000)
+
+    rec_1 = session.get(TorsiondriveRecordORM, id_1)
+    assert rec_1.status == RecordStatusEnum.complete
+    opt_ids_1 = [x.optimization_id for x in rec_1.optimizations]
+
+    # Submit again, without duplicate checking
+    meta_2, id_2 = storage_socket.records.torsiondrive.add(
+        [molecules_1], input_spec_1, True, "test_tag", PriorityEnum.low, None, None, False
+    )
+    id_2 = id_2[0]
+    assert meta_2.success
+    assert id_2 != id_1
+
+    run_service(storage_socket, activated_manager_name, id_2, generate_task_key, result_data_1, 1000)
+
+    rec_2 = session.get(TorsiondriveRecordORM, id_2)
+    assert rec_2.status == RecordStatusEnum.complete
+    opt_ids_2 = [x.optimization_id for x in rec_2.optimizations]
+
+    assert set(opt_ids_1).isdisjoint(opt_ids_2)