Merge pull request #35 from lamalab-org/fix/qm9-normalization-roundtrip

fix: H-assigment

Author: r-fedorov

src/openclatura/additive.py

@@ -53,7 +53,10 @@ def add_indicated_hydrogens(mol: Molecule, parts: AssemblyParts, numbered_path:
                 and len(ring_bonds) == 3
                 and atom.explicit_h_count + atom.total_h_count > 0
             )
-            if sum(b.order for b in ring_bonds) == 2 or fusion_carbon_h:
+            indicated_h_site = sum(b.order for b in ring_bonds) == 2 and (
+                not atom.is_carbon or atom.explicit_h_count + atom.total_h_count > 0
+            )
+            if indicated_h_site or fusion_carbon_h:
                 candidates.append((locant, idx))
 
     # A hydrogenated fusion carbon changes the retained-parent hydride state.

src/openclatura/retained_fused_templates.py

@@ -109,16 +109,15 @@ def retained_fused_graph_templates(*, include_disabled: bool = False) -> tuple[R
 
 @cache
 def retained_parent_metadata(parent_name: str) -> RetainedParentMetadata | None:
-    """Return graph-template metadata for a retained parent spelling."""
+    """Return metadata only when the spelling identifies one exact hydride.
+
+    Bare aliases such as ``isoindole`` do not identify the indicated-hydrogen
+    tautomer.  Those sites must be derived from the selected molecular graph;
+    production graph-template matches pass their metadata directly.
+    """
 
     for template in retained_fused_graph_templates(include_disabled=True):
-        spellings = {template.name, *template.aliases}
-        spellings.update(
-            f"{locant}H-{template.name}"
-            for locant in template.default_indicated_h
-            if not template.name.startswith(f"{locant}H-")
-        )
-        if parent_name in spellings:
+        if parent_name == template.name:
             return RetainedParentMetadata(
                 default_indicated_h=template.default_indicated_h,
                 fusion_locants=template.fusion_atoms,

src/openclatura/tests/test_analysis.py

@@ -142,6 +142,7 @@
     retained_fused_base_templates,
     retained_fused_graph_templates,
     retained_fused_template_from_data,
+    retained_parent_metadata,
     template_molecule,
 )
 from openclatura.retained_specs import retained_parent_spec
@@ -2968,6 +2969,40 @@ def test_retained_fused_base_skeletons_are_loaded_from_the_graph_template_table(
     assert base_templates["purinoid_9"]["mancude_double_bonds"] == 4
 
 
+def test_no_bare_retained_alias_infers_a_canonical_hydride_tautomer():
+    templates = retained_fused_graph_templates(include_disabled=True)
+    bare_hydride_aliases = {
+        alias
+        for template in templates
+        if template.default_indicated_h
+        for alias in template.aliases
+        if "H-" not in alias
+    }
+
+    # This registry-wide invariant automatically covers future retained
+    # hydrides whose bare alias omits the tautomer-defining H locant.
+    assert {"phenalene", "perimidine", "isoindole", "indole"} <= bare_hydride_aliases
+    assert all(retained_parent_metadata(alias) is None for alias in bare_hydride_aliases)
+
+    exact_hydride_templates = [template for template in templates if "H-" in template.name]
+    assert exact_hydride_templates
+    for template in exact_hydride_templates:
+        metadata = retained_parent_metadata(template.name)
+        assert metadata is not None
+        assert metadata.default_indicated_h == template.default_indicated_h
+
+
+def test_indicated_hydrogen_follows_graph_tautomer_but_not_hydrogen_free_spiro_carbon():
+    cases = {
+        "c1ccc2c(c1)CN=C2C1=NCc2ccccc21": "3-(3H-isoindol-1-yl)-1H-isoindole",
+        "O=C(OCCNC1=NCc2ccccc21)c1ccccc1": "2-((3H-isoindol-1-yl)amino)ethyl benzoate",
+        "FN1CCC2(C=Nc3ccccc32)CC1": "1'-fluorospiro[indole-3,4'-piperidine]",
+    }
+
+    for smiles, expected in cases.items():
+        assert name_smiles(smiles) == expected
+
+
 def test_retained_fused_graph_template_schema_validates_locant_graphs():
     template = retained_fused_template_from_data(_naphthalene_graph_template_row())