Parse empty dates as None, warn on invalid non-empty dates

src/caselawclient/models/documents.py

@@ -1,4 +1,5 @@
 import datetime
+import warnings
 from functools import cached_property
 from typing import TYPE_CHECKING, Any, Dict, NewType, Optional
 
@@ -26,6 +27,11 @@
     uri_for_s3,
 )
 
+
+class UnparsableDate(Warning):
+    pass
+
+
 DOCUMENT_STATUS_HOLD = "On hold"
 """ This document has been placed on hold to actively prevent publication. """
 
@@ -171,10 +177,19 @@ def document_date_as_string(self) -> str:
         )
 
     @cached_property
-    def document_date_as_date(self) -> datetime.date:
-        return datetime.datetime.strptime(
-            self.document_date_as_string, "%Y-%m-%d"
-        ).date()
+    def document_date_as_date(self) -> Optional[datetime.date]:
+        if not self.document_date_as_string:
+            return None
+        try:
+            return datetime.datetime.strptime(
+                self.document_date_as_string, "%Y-%m-%d"
+            ).date()
+        except ValueError:
+            warnings.warn(
+                f"Unparsable date encountered: {self.document_date_as_string}",
+                UnparsableDate,
+            )
+            return None
 
     def get_manifestation_datetimes(self, name: str) -> list[datetime.datetime]:
         iso_datetimes = self._get_xpath_match_strings(

tests/models/test_documents.py

@@ -17,6 +17,7 @@
     CannotPublishUnpublishableDocument,
     Document,
     DocumentNotSafeForDeletion,
+    UnparsableDate,
 )
 
 
@@ -490,6 +491,54 @@ def test_date_as_string(self, opening_tag, closing_tag, mock_api_client):
             "test/1234", show_unpublished=True
         )
 
+    @pytest.mark.parametrize(
+        "opening_tag, closing_tag",
+        [
+            ("judgment", "judgment"),
+            ('doc name="pressSummary"', "doc"),
+        ],
+    )
+    def test_bad_date_as_string(self, opening_tag, closing_tag, mock_api_client):
+        mock_api_client.get_judgment_xml_bytestring.return_value = f"""
+            <akomaNtoso xmlns:uk="https://caselaw.nationalarchives.gov.uk/akn"
+                xmlns="http://docs.oasis-open.org/legaldocml/ns/akn/3.0">
+                <{opening_tag}>
+                    <meta>
+                        <identification>
+                            <FRBRWork>
+                                <FRBRdate date="kitten"/>
+                            </FRBRWork>
+                        </identification>
+                    </meta>
+                </{closing_tag}>
+            </akomaNtoso>
+        """.encode(
+            "utf-8"
+        )
+
+        document = Document("test/1234", mock_api_client)
+
+        assert document.document_date_as_string == "kitten"
+        with pytest.warns(UnparsableDate):
+            assert document.document_date_as_date is None
+        mock_api_client.get_judgment_xml_bytestring.assert_called_once_with(
+            "test/1234", show_unpublished=True
+        )
+
+    def test_absent_date_as_string(self, mock_api_client):
+        mock_api_client.get_judgment_xml_bytestring.return_value = """
+            <akomaNtoso xmlns:uk="https://caselaw.nationalarchives.gov.uk/akn"
+                xmlns="http://docs.oasis-open.org/legaldocml/ns/akn/3.0">
+            </akomaNtoso>
+        """.encode(
+            "utf-8"
+        )
+
+        document = Document("test/1234", mock_api_client)
+
+        assert document.document_date_as_string == ""
+        assert document.document_date_as_date is None
+
     def test_dates(self, mock_api_client):
         mock_api_client.get_judgment_xml_bytestring.return_value = """
             <akomaNtoso xmlns:uk="https://caselaw.nationalarchives.gov.uk/akn"