Merge branch 'dev' into f-dict-heuristic

src/caoscrawler/identifiable_adapters.py

@@ -236,10 +236,11 @@ startswith: bool, optional
                             refs.extend(referencing_entities[rtname])
                             found = True
             if not found:
-                raise NotImplementedError(
+                raise RuntimeError(
                     f"Could not find referencing entities of type(s): {prop.value}\n"
                     f"for registered identifiable:\n{registered_identifiable}\n"
-                    f"There were {len(referencing_entities)} referencing entities to choose from."
+                    f"There were {len(referencing_entities)} referencing entities to choose from.\n"
+                    f"This error can also occur in case of merge conflicts in the referencing entities."
                 )
         return refs
 

unittests/test_crawler.py

@@ -405,7 +405,6 @@ def test_split_into_inserts_and_updates_with_copy_attr(crawler_mocked_identifiab
     crawler.identifiableAdapter.retrieve_identified_record_for_identifiable.assert_called()
 
 
-@pytest.mark.xfail(reason="https://gitlab.com/linkahead/linkahead-crawler/-/issues/88")
 @patch("caoscrawler.identifiable_adapters.cached_query",
        new=Mock(side_effect=mock_cached_only_rt))
 def test_split_iiau_with_unmergeable_list_items():
@@ -429,12 +428,16 @@ Identifiables
 
 id_A: [prop_ident]
 id_B: [prop_ident, "is_referenced_by: A"]
+id_C: [prop_other, "is_referenced_by: B"]
 
 Data
 ----
 
-b1: ("same", 23)
-b2: ("same", 42)
+c1: (23)
+c2: (42)
+
+b1: ("same", c1)
+b2: ("same", c2)
 
 a: ([b1, b2])
     """
@@ -469,18 +472,13 @@ a: ([b1, b2])
 
     crawler = Crawler(identifiableAdapter=ident_adapter)
 
-    # This should give a merge conflict, and not
-    # "Could not find referencing entities of type(s): A"
-
-    # from IPython import embed; embed()
     with raises(RuntimeError) as rte:
         crawler.synchronize(commit_changes=False,
                             crawled_data=[rec_a, *rec_b, *rec_c])
     assert not isinstance(rte.value, NotImplementedError), \
         "Exception must not be NotImplementedError, but plain RuntimeError."
-    assert "Could not find referencing entities" not in rte.value.args[0]
-    assert "merging impossible" in rte.something
-    # crawler.split_into_inserts_and_updates(ent_list=[rec_a, *rec_b, *rec_c])
+    assert "Could not find referencing entities" in rte.value.args[0]
+    assert "merge conflicts in the referencing" in rte.value.args[0]
 
 
 def test_has_missing_object_in_references():
@@ -785,8 +783,8 @@ def test_split_into_inserts_and_updates_backref(crawler_mocked_for_backref_test)
     entlist = [referenced, db.Record(name="A").add_parent("BR").add_property("ref", referenced), ]
 
     # Test without referencing object
-    # currently a NotImplementedError is raised if necessary properties are missing.
-    with raises(NotImplementedError):
+    # currently a RuntimeError is raised if necessary properties are missing.
+    with raises(RuntimeError):
         crawler.split_into_inserts_and_updates([db.Record(name="B").add_parent("C")])
 
     # identifiables were not yet checked