[Fix] Iuclid export produce only one root substance (#429)

Reviewed-on: enviPath/enviPy#429
Co-authored-by: Tobias O <tobias.olenyi@envipath.com>
Co-committed-by: Tobias O <tobias.olenyi@envipath.com>

epiuclid/serializers/pathway_mapper.py

@@ -120,13 +120,6 @@ class PathwayMapper:
             )
             bundle.reference_substances.append(ref_sub)
 
-            sub = IUCLIDSubstanceData(
-                uuid=sub_uuid,
-                name=compound.name,
-                reference_substance_uuid=ref_sub_uuid,
-            )
-            bundle.substances.append(sub)
-
         if not export.compounds:
             return bundle
 
@@ -145,6 +138,16 @@ class PathwayMapper:
         if not root_compound_pks:
             return bundle
 
+        for root_pk in root_compound_pks:
+            root_sub_uuid, root_ref_uuid = seen_compounds[root_pk]
+            bundle.substances.append(
+                IUCLIDSubstanceData(
+                    uuid=root_sub_uuid,
+                    name=compound_names[root_pk],
+                    reference_substance_uuid=root_ref_uuid,
+                )
+            )
+
         edge_templates: list[tuple[UUID, frozenset[int], tuple[int, ...], tuple[UUID, ...]]] = []
         for edge in sorted(export.edges, key=lambda item: str(item.edge_uuid)):
             parent_compound_pks = sorted(

epiuclid/tests/test_api.py

@@ -70,8 +70,7 @@ class IUCLIDExportAPITest(TestCase):
             names = zf.namelist()
             self.assertIn("manifest.xml", names)
             i6d_files = [n for n in names if n.endswith(".i6d")]
-            # 2 substances + 2 ref substances + 1 ESR = 5 i6d files
-            self.assertEqual(len(i6d_files), 5)
+            self.assertEqual(len(i6d_files), 4)
 
     def test_anonymous_returns_401(self):
         self.client.logout()

epiuclid/tests/test_i6z.py

@@ -7,6 +7,11 @@ from uuid import uuid4
 
 from django.test import SimpleTestCase, tag
 
+from epapi.v1.interfaces.iuclid.dto import (
+    PathwayCompoundDTO,
+    PathwayEdgeDTO,
+    PathwayExportDTO,
+)
 from epiuclid.serializers.i6z import I6ZSerializer
 from epiuclid.serializers.pathway_mapper import (
     IUCLIDDocumentBundle,
@@ -14,9 +19,24 @@ from epiuclid.serializers.pathway_mapper import (
     IUCLIDReferenceSubstanceData,
     IUCLIDSubstanceData,
     IUCLIDTransformationProductEntry,
+    PathwayMapper,
 )
 
 
+def _unlinked_documents(manifest_xml: str) -> list[tuple[str | None, str]]:
+    ns = "http://iuclid6.echa.europa.eu/namespaces/manifest/v1"
+    root = ET.fromstring(manifest_xml)
+    base = root.findtext(f"{{{ns}}}base-document-uuid")
+    linked_targets: set[str | None] = {base}
+    docs: dict[str, str | None] = {}
+    for doc in root.findall(f".//{{{ns}}}document"):
+        uuid = doc.findtext(f"{{{ns}}}uuid")
+        docs[uuid] = doc.findtext(f"{{{ns}}}type")
+        for link in doc.findall(f"{{{ns}}}links/{{{ns}}}link"):
+            linked_targets.add(link.findtext(f"{{{ns}}}ref-uuid"))
+    return [(doc_type, uuid) for uuid, doc_type in docs.items() if uuid not in linked_targets]
+
+
 def _make_bundle() -> IUCLIDDocumentBundle:
     ref_uuid = uuid4()
     sub_uuid = uuid4()
@@ -197,3 +217,29 @@ class I6ZSerializerTest(SimpleTestCase):
             }
             self.assertIn(parent_ref_key, reference_links)
             self.assertIn(product_ref_key, reference_links)
+
+    def test_multi_compound_pathway_has_no_unlinked_documents(self):
+        compounds = [
+            PathwayCompoundDTO(pk=1, name="Root", smiles="c1ccccc1"),
+            PathwayCompoundDTO(pk=2, name="P1", smiles="CCO"),
+            PathwayCompoundDTO(pk=3, name="P2", smiles="CCN"),
+            PathwayCompoundDTO(pk=4, name="P3", smiles="CCC"),
+        ]
+        export = PathwayExportDTO(
+            pathway_uuid=uuid4(),
+            pathway_name="Regression Pathway",
+            compounds=compounds,
+            edges=[
+                PathwayEdgeDTO(edge_uuid=uuid4(), start_compound_pks=[1], end_compound_pks=[2]),
+                PathwayEdgeDTO(edge_uuid=uuid4(), start_compound_pks=[1], end_compound_pks=[3]),
+                PathwayEdgeDTO(edge_uuid=uuid4(), start_compound_pks=[2], end_compound_pks=[4]),
+            ],
+            root_compound_pks=[1],
+        )
+        bundle = PathwayMapper().map(export)
+        data = I6ZSerializer().serialize(bundle)
+
+        with zipfile.ZipFile(io.BytesIO(data)) as zf:
+            manifest_xml = zf.read("manifest.xml").decode("utf-8")
+
+        self.assertEqual(_unlinked_documents(manifest_xml), [])

epiuclid/tests/test_pathway_mapper.py

@@ -31,7 +31,7 @@ class PathwayMapperTest(SimpleTestCase):
         )
         bundle = PathwayMapper().map(export)
 
-        self.assertEqual(len(bundle.substances), 2)
+        self.assertEqual(len(bundle.substances), 1)
         self.assertEqual(len(bundle.reference_substances), 2)
         self.assertEqual(len(bundle.endpoint_study_records), 1)
 
@@ -49,8 +49,7 @@ class PathwayMapperTest(SimpleTestCase):
         )
         bundle = PathwayMapper().map(export)
 
-        # 2 unique compounds -> 2 substances, 2 ref substances
-        self.assertEqual(len(bundle.substances), 2)
+        self.assertEqual(len(bundle.substances), 1)
         self.assertEqual(len(bundle.reference_substances), 2)
         # One endpoint study record per pathway
         self.assertEqual(len(bundle.endpoint_study_records), 1)