Compare two identifiers, only add if not present

src/caselawclient/models/identifiers/__init__.py

@@ -101,6 +101,10 @@ def as_xml_tree(self) -> etree._Element:
     def url_slug(self) -> str:
         return self.schema.compile_identifier_url_slug(self.value)
 
+    def similar(self, other: "Identifier") -> bool:
+        "Is this the same as another identifier (in value and schema)?"
+        return self.value == other.value and self.schema == other.schema
+
 
 class Identifiers(dict[str, Identifier]):
     def validate(self) -> None:
@@ -109,8 +113,13 @@ def validate(self) -> None:
                 msg = "Key of {identifier} in Identifiers is {uuid} not {identifier.uuid}"
                 raise UUIDMismatchError(msg)
 
+    def contains_similar(self, other_identifier: Identifier) -> bool:
+        "Do the identifier's value and namespace already exist in this group?"
+        return any(other_identifier.similar(identifier) for identifier in self.values())
+
     def add(self, identifier: Identifier) -> None:
-        self[identifier.uuid] = identifier
+        if not self.contains_similar(identifier):
+            self[identifier.uuid] = identifier
 
     def __delitem__(self, key: Union[Identifier, str]) -> None:
         if isinstance(key, Identifier):

tests/models/identifiers/test_identifiers.py

@@ -2,18 +2,19 @@
 from lxml import etree
 
 from caselawclient.models.identifiers import Identifier, Identifiers, IdentifierSchema
+from caselawclient.models.identifiers.neutral_citation import NeutralCitationNumber
 
 
 @pytest.fixture
 def identifiers():
     return Identifiers(
-        {"id-1": Identifier(uuid="id-1", value="TEST-111"), "id-2": Identifier(uuid="id-2", value="TEST-222")}
+        {"id-1": TestIdentifier(uuid="id-1", value="TEST-111"), "id-2": TestIdentifier(uuid="id-2", value="TEST-222")}
     )
 
 
 @pytest.fixture
 def id3():
-    return Identifier(uuid="id-3", value="TEST-333")
+    return TestIdentifier(uuid="id-3", value="TEST-333")
 
 
 class TestIdentifierSchema(IdentifierSchema):
@@ -55,6 +56,21 @@ def test_xml_representation(self):
             etree.fromstring(expected_xml), strip_text=True
         )
 
+    def test_similar_different_types(self):
+        id_a = TestIdentifier("X")
+        id_b = NeutralCitationNumber("X")
+        assert not id_a.similar(id_b)
+
+    def test_similar_same_type_and_value(self):
+        id_a = NeutralCitationNumber("X")
+        id_b = NeutralCitationNumber("X")
+        assert id_a.similar(id_b)
+
+    def test_similar_different_values(self):
+        id_a = NeutralCitationNumber("Y")
+        id_b = NeutralCitationNumber("X")
+        assert not id_a.similar(id_b)
+
 
 class TestIdentifiersCRUD:
     def test_delete(self, identifiers):
@@ -72,3 +88,8 @@ def test_add_identifier(self, identifiers, id3):
         identifiers.add(id3)
         assert identifiers["id-3"] == id3
         assert len(identifiers) == 3
+
+    def test_contains(self, identifiers):
+        assert identifiers.contains_similar(TestIdentifier(value="TEST-111"))
+        assert not identifiers.contains_similar(TestIdentifier(value="TEST-333"))
+        assert not identifiers.contains_similar(NeutralCitationNumber(value="TEST-111"))