Fixed JobStore.from_dict_spec so that the original dict_spec is not modified

src/jobflow/core/store.py

@@ -672,12 +672,21 @@ def all_subclasses(cl):
 
 
 def _construct_store(spec_dict, valid_stores):
-    """Parse the dict containing {"type": <StoreType>} recursively."""
-    store_type = spec_dict.pop("type")
-    for k, v in spec_dict.items():
+    """Parse the dict containing {"type": <StoreType>} recursively.
+
+    Note: the spec is copied so that if this is used programmatically, the type
+    is not removed from the original spec_dict, which may be used afterwards,
+    (e.g. stored in a database), or if there is some validation by a pydantic
+    schema.
+    This does not occur when using jobflow's standard loading of a config file
+    as it is only called once.
+    """
+    _spec_dict = dict(spec_dict)
+    store_type = _spec_dict.pop("type")
+    for k, v in _spec_dict.items():
         if isinstance(v, dict) and "type" in v:
-            spec_dict[k] = _construct_store(v, valid_stores)
-    return valid_stores[store_type](**spec_dict)
+            _spec_dict[k] = _construct_store(v, valid_stores)
+    return valid_stores[store_type](**_spec_dict)
 
 
 def _prepare_load(

tests/core/test_store.py

@@ -434,6 +434,45 @@ def test_from_db_file(test_data):
         JobStore.from_file(test_data / "db_bad.yaml")
 
 
+def test_from_dict_spec():
+    from jobflow import JobStore
+
+    dict_spec = {
+        "docs_store": {
+            "type": "MongoStore",
+            "database": "jobflow_unittest",
+            "collection_name": "outputs",
+            "host": "localhost",
+            "port": 27017,
+        }
+    }
+    JobStore.from_dict_spec(dict_spec)
+    assert "type" in dict_spec["docs_store"]
+    dict_spec = {
+        "docs_store": {
+            "type": "MongoStore",
+            "database": "jobflow_unittest",
+            "collection_name": "outputs",
+            "host": "localhost",
+            "port": 27017,
+        },
+        "additional_stores": {
+            "data": {
+                "type": "GridFSStore",
+                "database": "jobflow_unittest",
+                "collection_name": "outputs_blobs",
+                "host": "localhost",
+                "port": 27017,
+            }
+        },
+    }
+    JobStore.from_dict_spec(dict_spec)
+    assert "type" in dict_spec["docs_store"]
+    assert dict_spec["docs_store"]["type"] == "MongoStore"
+    assert "type" in dict_spec["additional_stores"]["data"]
+    assert dict_spec["additional_stores"]["data"]["type"] == "GridFSStore"
+
+
 def test_ensure_index(memory_jobstore):
     assert memory_jobstore.ensure_index("test_key")
     # TODO: How to check for exception?