better "extras" passing in Molecule for EFP

Author: loriab

qcelemental/models/molecule.py

@@ -724,9 +724,9 @@ def from_data(cls,
                 raise TypeError("Input type not understood, please supply the 'dtype' kwarg.")
 
         if dtype in ["string", "psi4", "psi4+", "xyz", "xyz+"]:
-            mol_dict = from_string(data)
-            assert isinstance(mol_dict, dict)
-            input_dict = to_schema(mol_dict["qm"], dtype=2)
+            molrec = from_string(data, enable_qm=True, missing_enabled_return_qm='minimal')
+            assert isinstance(molrec, dict)
+            input_dict = to_schema(molrec["qm"], dtype=2)
             validate = True
         elif dtype == "numpy":
             data = np.asarray(data)

qcelemental/molparse/to_schema.py

@@ -94,6 +94,14 @@ def to_schema(molrec: Dict[str, Any],
         if 'connectivity' in molrec:
             molecule['connectivity'] = deepcopy(molrec['connectivity'])
 
+        # EFP extras
+        if 'fragment_files' in molrec:
+            molecule['extras'] = {
+                'fragment_files': molrec['fragment_files'],
+                'hint_types': molrec['hint_types'],
+                'geom_hints': molrec['geom_hints'],
+            }
+
         if dtype == 1:
             qcschema = {'schema_name': 'qcschema_input', 'schema_version': 1, 'molecule': molecule}
         elif dtype == 2:

qcelemental/tests/test_utils.py

@@ -55,7 +55,8 @@ def test_unique_everseen(inp, expected):
 @pytest.mark.parametrize("inp,expected", [
     (({1:{"a":"A"},2:{"b":"B"}}, {2:{"c":"C"},3:{"d":"D"}}), {1:{"a":"A"},2:{"b":"B","c":"C"},3:{"d":"D"}}),
     (({1:{"a":"A"},2:{"b":"B","c":None}}, {2:{"c":"C"},3:{"d":"D"}}), {1:{"a":"A"},2:{"b":"B","c":"C"},3:{"d":"D"}}),
-    (({1: [None, 1]}, {1: [2, 1],3:{"d":"D"}}), {1:[2, 1], 3:{"d":"D"}})
+    (({1: [None, 1]}, {1: [2, 1],3:{"d":"D"}}), {1:[2, 1], 3:{"d":"D"}}),
+    (({1: True, 'extras':{'aa': [True]}}, {2: None, 'extras':{'bb': 5}}), {1: True, 2: None, 'extras':{'aa': [True], 'bb':5}}),
 ]) # yapf: disable
 def test_updatewitherror(inp, expected):
     assert compare_recursive(expected, qcel.util.update_with_error(inp[0], inp[1]))