Merge pull request #103 from volkamerlab/fix-klifs-parallelization

KLIFS module: RecursionError with parallelization... again!

opencadd/databases/klifs/local.py

@@ -23,7 +23,8 @@
     FIELDS,
     POCKET_KLIFS_REGIONS,
 )
-from .utils import KLIFS_CLIENT, PATH_DATA, metadata_to_filepath, filepath_to_metadata
+from .utils import PATH_DATA, metadata_to_filepath, filepath_to_metadata
+from .remote import KLIFS_CLIENT
 from .exceptions import KlifsPocketIncompleteError, KlifsPocketUnequalSequenceStructure
 from opencadd.io import DataFrame, Rdkit
 

opencadd/databases/klifs/remote.py

@@ -44,6 +44,12 @@ def __getstate__(self, *args):
         }
 
 
+KLIFS_API_DEFINITIONS = "https://dev.klifs.net/swagger_v2/swagger.json"
+KLIFS_CLIENT = SerializableSwaggerClient.from_url(
+    KLIFS_API_DEFINITIONS, config={"validate_responses": False}
+)
+
+
 class RemoteInitializer:
     """
     Base class used to define __init__ for all remote classes.

opencadd/databases/klifs/session.py

@@ -10,7 +10,7 @@
 
 from . import remote
 from . import local
-from .utils import KLIFS_CLIENT
+from .remote import KLIFS_CLIENT
 
 
 class Session:

opencadd/databases/klifs/utils.py

@@ -9,13 +9,8 @@
 from pathlib import Path
 import re
 
-from bravado.client import SwaggerClient
-
 _logger = logging.getLogger(__name__)
 
-KLIFS_API_DEFINITIONS = "https://dev.klifs.net/swagger_v2/swagger.json"
-KLIFS_CLIENT = SwaggerClient.from_url(KLIFS_API_DEFINITIONS, config={"validate_responses": False})
-
 PATH_DATA = Path(__file__).parent / ".." / ".." / "data"
 
 

opencadd/tests/databases/test_klifs_remote_parallelization.py

@@ -0,0 +1,35 @@
+"""
+Tests for opencadd.databases.klifs.remote parallelization!
+"""
+
+from itertools import repeat
+from multiprocessing import Pool
+
+import pytest
+
+from opencadd.databases.klifs import setup_remote
+
+# Set local session
+REMOTE = setup_remote()
+
+
+@pytest.mark.parametrize(
+    "klifs_session, structure_klifs_ids, n_cores",
+    [(REMOTE, [111, 113], 2)],
+)
+def test_remote_parallelization(klifs_session, structure_klifs_ids, n_cores):
+
+    pool = Pool(processes=n_cores)
+
+    pool.starmap(_parallelize_klifs_session, zip(structure_klifs_ids, repeat(klifs_session)))
+    pool.close()
+    pool.join()
+
+
+def _parallelize_klifs_session(structure_klifs_id, klifs_session):
+    """
+    Dummy function that simulates a function that takes the KLIFS session as input;
+    shall be used in our test where we want to pass a single klifs_session to multiple functions
+    that are called in a parallelized process.
+    """
+    return klifs_session.structures.by_structure_klifs_id(structure_klifs_id)

opencadd/tests/databases/test_sync_klifs_and_opencadd.py

@@ -3,7 +3,7 @@
 If errors are raised, it is time to update opencadd.
 """
 
-from opencadd.databases.klifs.utils import KLIFS_CLIENT
+from opencadd.databases.klifs.remote import KLIFS_CLIENT
 from opencadd.databases.klifs.schema import FIELDS