TST: Add an integration test for #83

integrationtests/test_issues.py

@@ -114,7 +114,7 @@ def test_issue_23(clear_database):
     assert rec_crawled.get_property("identifying_prop").value == "identifier"
     assert rec_crawled.get_property("prop_b") is not None
     assert rec_crawled.get_property("prop_b").value == "something_else"
-    # no interaction with the database yet, so the rrecord shouldn't have a prop_a yet
+    # no interaction with the database yet, so the record shouldn't have a prop_a yet
     assert rec_crawled.get_property("prop_a") is None
 
     # synchronize with database and update the record
@@ -133,3 +133,78 @@ def test_issue_23(clear_database):
         "identifying_prop").value == rec_crawled.get_property("identifying_prop").value
     assert rec_retrieved.get_property(
         "prop_b").value == rec_crawled.get_property("prop_b").value
+
+
+def test_issue_83(clear_database):
+    """https://gitlab.com/linkahead/linkahead-crawler/-/issues/83. Test that
+    names don't need to be unique for referenced entities if they are not part
+    of the identifiable.
+
+    """
+
+    # Very simple data model
+    identifying_prop = db.Property(name="IdentifyingProp", datatype=db.INTEGER).insert()
+    referenced_type = db.RecordType(name="ReferencedType").add_property(
+        name=identifying_prop.name, importance=db.OBLIGATORY).insert()
+    referencing_type = db.RecordType(name="ReferencingType").add_property(
+        name=referenced_type.name, datatype=db.LIST(referenced_type.name)).insert()
+
+    # Define identifiables. ReferencingType by name, ReferencedType by
+    # IdentifyingProp and not by name.
+    ident = CaosDBIdentifiableAdapter()
+    ident.register_identifiable(referenced_type.name, db.RecordType().add_parent(
+        name=referenced_type.name).add_property(name=identifying_prop.name))
+    ident.register_identifiable(referencing_type.name, db.RecordType().add_parent(
+        name=referencing_type.name).add_property(name="name"))
+
+    crawler = Crawler(identifiableAdapter=ident)
+
+    ref_target1 = db.Record(name="RefTarget").add_parent(
+        name=referenced_type.name).add_property(name=identifying_prop.name, value=1)
+    ref_target2 = db.Record(name="RefTarget").add_parent(
+        name=referenced_type.name).add_property(name=identifying_prop.name, value=2)
+
+    referencing1 = db.Record(name="Referencing1").add_parent(
+        name=referencing_type.name).add_property(name=referenced_type.name, value=[ref_target1])
+    referencing2 = db.Record(name="Referencing2").add_parent(
+        name=referencing_type.name).add_property(name=referenced_type.name, value=[ref_target2])
+    referencing3 = db.Record(name="Referencing3").add_parent(name=referencing_type.name).add_property(
+        name=referenced_type.name, value=[ref_target1, ref_target2])
+
+    records = db.Container().extend(
+        [ref_target1, ref_target2, referencing1, referencing2, referencing3])
+
+    ins, ups = crawler.synchronize(crawled_data=records, unique_names=False)
+    assert len(ins) == len(records)
+    assert len(ups) == 0
+
+    retrieved_target1 = db.execute_query(
+        f"FIND {referenced_type.name} WITH {identifying_prop.name}=1", unique=True)
+    retrieved_target2 = db.execute_query(
+        f"FIND {referenced_type.name} WITH {identifying_prop.name}=2", unique=True)
+    assert retrieved_target2.name == retrieved_target1.name
+    assert retrieved_target1.name == ref_target1.name
+    assert retrieved_target1.id != retrieved_target2.id
+
+    retrieved_referencing1 = db.execute_query(
+        f"FIND {referencing_type.name} WITH name={referencing1.name}", unique=True)
+    assert retrieved_referencing1.get_property(referenced_type.name) is not None
+    assert retrieved_referencing1.get_property(referenced_type.name).value == [
+        retrieved_target1.id]
+    assert retrieved_referencing1.get_property(referenced_type.name).value != [
+        retrieved_target2.id]
+
+    retrieved_referencing2 = db.execute_query(
+        f"FIND {referencing_type.name} WITH name={referencing2.name}", unique=True)
+    assert retrieved_referencing2.get_property(referenced_type.name) is not None
+    assert retrieved_referencing2.get_property(referenced_type.name).value == [
+        retrieved_target2.id]
+    assert retrieved_referencing2.get_property(referenced_type.name).value != [
+        retrieved_target1.id]
+
+    retrieved_referencing3 = db.execute_query(
+        f"FIND {referencing_type.name} WITH name={referencing3.name}", unique=True)
+    assert retrieved_referencing3.get_property(referenced_type.name) is not None
+    assert len(retrieved_referencing3.get_property(referenced_type.name).value) == 2
+    assert retrieved_target1.id in retrieved_referencing3.get_property(referenced_type.name).value
+    assert retrieved_target2.id in retrieved_referencing3.get_property(referenced_type.name).value