TEST for #95: Merging two entities that are referenced by a third.

src/caoscrawler/identifiable_adapters.py

@@ -576,6 +576,11 @@ class CaosDBIdentifiableAdapter(IdentifiableAdapter):
         """Load identifiables defined in a yaml file"""
         with open(path, "r", encoding="utf-8") as yaml_f:
             identifiable_data = yaml.safe_load(yaml_f)
+        self.load_from_yaml_object(identifiable_data)
+
+    def load_from_yaml_object(self, identifiable_data):
+        """Load identifiables defined in a yaml object.
+        """
 
         for key, value in identifiable_data.items():
             rt = db.RecordType().add_parent(key)

unittests/test_sync_graph.py

@@ -651,3 +651,28 @@ def test_set_each_scalar_value():
     assert a.properties[0].value == 42
     _set_each_scalar_value(a, lambda x: x == 42, lambda x: None)
     assert a.properties[0].value is None
+
+
+def test_merge_referenced_by():
+    """Merging two entities that are referenced by a third entity.
+
+    See also https://gitlab.com/linkahead/linkahead-crawler/-/issues/95
+    """
+    ident = CaosDBIdentifiableAdapter()
+    ident.load_from_yaml_object({
+        "RT_A": ["name"],
+        "RT_B": [{"is_referenced_by": "RT_A"}, "my_id"]
+    })
+
+    crawled_data: list = []
+    references: list = []
+    for ii in [0, 1]:
+        rec = db.Record().add_parent("RT_B").add_property("my_id", value=ii)
+        references.append(rec)
+        crawled_data.append(rec)
+    rec_a = db.Record(name="A").add_parent("RT_A")
+    rec_a.add_property("my_ref", value=references)
+    crawled_data.append(rec_a)
+
+    sync_graph = SyncGraph(crawled_data, ident)
+    assert sync_graph is not None