fix(data-extraction): Handle malformed date_of_birth fields (#204)

fix(data-extraction): Handle malformed date_of_birth fields

src/aki_prj23_transparenzregister/models/company.py

@@ -124,7 +124,7 @@ class PersonToCompanyRelationship(CompanyRelationship):
     """Extension of CompanyRelationship with extras for Person."""
 
     name: PersonName
-    date_of_birth: str
+    date_of_birth: str | None
 
 
 @dataclass

src/aki_prj23_transparenzregister/utils/data_extraction/unternehmensregister/transform.py

@@ -52,6 +52,22 @@ def transform_xml_to_json(source_dir: str, target_dir: str) -> None:
                 json_file.write(json.dumps(data))
 
 
+def parse_date_of_birth(data: dict) -> str | None:
+    """Retreives the date of birth from a stakeholder entry if possible.
+
+    Args:
+        data (dict): Stakeholder data
+
+    Returns:
+        str | None: date of birth or None if not found
+    """
+    if "Geburt" in (base := data["Beteiligter"]["Natuerliche_Person"]):
+        base = base["Geburt"]["Geburtsdatum"]
+        if isinstance(base, str):
+            return base
+    return None
+
+
 def parse_stakeholder(data: dict) -> CompanyRelationship | None:
     """Extract the company stakeholder/relation from a single "Beteiligung".
 
@@ -92,7 +108,7 @@ def parse_stakeholder(data: dict) -> CompanyRelationship | None:
                 }
             )
         return PersonToCompanyRelationship(
-            **{
+            **{  # type: ignore
                 "name": PersonName(
                     **{
                         "firstname": data["Beteiligter"]["Natuerliche_Person"][
@@ -103,11 +119,7 @@ def parse_stakeholder(data: dict) -> CompanyRelationship | None:
                         ]["Nachname"],
                     }
                 ),
-                "date_of_birth": data["Beteiligter"]["Natuerliche_Person"]["Geburt"][
-                    "Geburtsdatum"
-                ]
-                if "Geburt" in data["Beteiligter"]["Natuerliche_Person"]
-                else None,
+                "date_of_birth": parse_date_of_birth(data),
                 "location": Location(
                     **{
                         "city": data["Beteiligter"]["Natuerliche_Person"]["Anschrift"][

src/aki_prj23_transparenzregister/utils/data_transfer.py

@@ -395,19 +395,13 @@ def add_person_relation(person: dict[str, Any], company_id: int, db: Session) ->
             db=db,
         )
     except DataInvalidError:
-        # logger.exception("Test except: ")
         if date_of_brith:
-            # print("Break")
             raise
-        # TODO enable the following line
-        # logger.warning(f"No date of birth birth for {name['firstname']} {name['lastname']}")
+        logger.debug(
+            f"No date of birth birth for {person['name']['lastname']}, {person['name']['firstname']}"
+        )
         db.rollback()
         return
-    except TypeError as error:
-        # TODO remove unhashable type catcher
-        if "unhashable type: 'dict'" in str(error):
-            return
-        raise
     except Exception:
         logger.exception("Test except")
         raise

tests/utils/data_extraction/unternehmensregister/transform_test.py

@@ -82,6 +82,27 @@ def test_parse_stakeholder_person() -> None:
     assert transform.parse_stakeholder(data) == expected_result
 
 
+def test_parse_stakeholder_person_missing_date_of_birth() -> None:
+    data = {
+        "Beteiligter": {
+            "Natuerliche_Person": {
+                "Voller_Name": {"Vorname": "Stephen", "Nachname": "King"},
+                "Anschrift": {"Ort": "Maine"},
+                "Geburt": {"Geburtsdatum": {"@xsi:nil": "true"}},
+            }
+        },
+        "Rolle": {"Rollenbezeichnung": {"content": "Geschäftsleiter(in)"}},
+    }
+    expected_result = PersonToCompanyRelationship(
+        role=RelationshipRoleEnum.GESCHAEFTSLEITER,  # type: ignore
+        date_of_birth=None,
+        name=PersonName(**{"firstname": "Stephen", "lastname": "King"}),
+        type=CompanyRelationshipEnum.PERSON,
+        location=Location(**{"city": "Maine"}),
+    )
+    assert transform.parse_stakeholder(data) == expected_result
+
+
 def test_parse_stakeholder_org() -> None:
     data = {
         "Beteiligter": {