Add NCN to identifiers on ingest

ds-caselaw-ingester/lambda_function.py

@@ -2,9 +2,10 @@
 import os
 import tarfile
 import xml.etree.ElementTree as ET
-from typing import Dict, List, Tuple
 from urllib.parse import unquote_plus
 from xml.sax.saxutils import escape
+from caselawclient.models.identifiers.neutral_citation import NeutralCitationNumber
+from caselawclient.models.documents import DocumentURIString
 
 import boto3
 import rollbar
@@ -18,6 +19,11 @@
 from caselawclient.client_helpers import VersionAnnotation, VersionType
 from dotenv import load_dotenv
 from notifications_python_client.notifications import NotificationsAPIClient
+import logging
+from caselawclient.models.documents import Document
+
+logger = logging.getLogger("ingester")
+logger.setLevel(logging.DEBUG)
 
 load_dotenv()
 rollbar.init(os.getenv("ROLLBAR_TOKEN"), environment=os.getenv("ROLLBAR_ENV"))
@@ -437,6 +443,24 @@ def insert_document_xml(self) -> bool:
         api_client.insert_document_xml(self.uri, self.xml, annotation)
         return True
 
+    def set_document_identifiers(self) -> None:
+        doc = api_client.get_document_by_uri(DocumentURIString(self.uri))
+        if doc.identifiers:
+            msg = f"Ingesting, but identifiers already present for {self.uri}!"
+            logger.warning(msg)
+
+        try:
+            ncn = doc.neutral_citation
+        except AttributeError:
+            ncn = None
+
+        if ncn:
+            doc.identifiers.add(NeutralCitationNumber(ncn))
+            doc.identifiers.save(doc)
+            logger.info(f"Ingested document had NCN {ncn}")
+        else:
+            logger.info(f"Ingested document had NCN (NOT FOUND)")
+
     def send_updated_judgment_notification(self) -> None:
         personalisation = personalise_email(self.uri, self.metadata)
         if os.getenv("ROLLBAR_ENV") != "prod":
@@ -597,6 +621,7 @@ def upload_xml(self) -> None:
             raise DocumentInsertionError(
                 f"Judgment {self.uri} failed to insert into Marklogic. Consignment Ref: {self.consignment_reference}"
             )
+        self.set_document_identifiers()
 
     @property
     def upload_state(self) -> str:

ds-caselaw-ingester/tests.py

@@ -140,6 +140,8 @@ def test_handler_messages_v2(
         capsys,
     ):
         boto_session.return_value.client.return_value.download_file = create_fake_tdr_file
+        doc = apiclient.get_document_by_uri.return_value
+        doc.neutral_citation = None
 
         message = v2_message_raw
         event = {"Records": [{"Sns": {"Message": message}}, {"Sns": {"Message": message}}]}
@@ -161,6 +163,8 @@ def test_handler_messages_v2(
             payload=ANY,
         )
         assert annotation.call_count == 2
+        doc.identifiers.add.assert_not_called()
+        doc.identifiers.save.assert_not_called()
 
     @patch("lambda_function.api_client", autospec=True)
     @patch("lambda_function.boto3.session.Session")
@@ -180,6 +184,8 @@ def test_handler_messages_s3(
     ):
         """Test that, with appropriate stubs, an S3 message passes through the parsing process"""
         boto_session.return_value.client.return_value.download_file = create_fake_bulk_file
+        doc = apiclient.get_document_by_uri.return_value
+        doc.neutral_citation = "[2012] UKUT 82 (IAC)"
 
         message = s3_message_raw
         event = {"Records": [{"Sns": {"Message": message}}, {"Sns": {"Message": message}}]}
@@ -200,13 +206,16 @@ def test_handler_messages_s3(
         notify_new.assert_not_called()
         notify_updated.assert_not_called()
         modify_filename.assert_not_called()
+
         annotation.assert_called_with(
             ANY,
             automated=True,
             message="Updated document uploaded by Find Case Law",
             payload=ANY,
         )
         assert annotation.call_count == 2
+        assert doc.identifiers.add.call_args_list[0].args[0].value == "[2012] UKUT 82 (IAC)"
+        doc.identifiers.save.assert_called()
 
 
 class TestLambda: