[NRL-1509] Add patient number to invalid pointer report

Author: mattdean3-nhs

reports/find_malformed_urls.py

@@ -21,14 +21,14 @@ def _validate_url(document: str):
     """
     docref = DocumentReference.model_validate_json(document)
 
-    validator = DocumentReferenceValidator()
-    result = validator.validate(data=docref)
-    for i, content in enumerate(result.content):
+    for content in docref.content:
         if content.attachment.contentType == "application/pdf":
             url = content.attachment.url
             if url[-1] == "/":
                 raise RuntimeError("Malformed URL found: " + str(url))
 
+    validator = DocumentReferenceValidator()
+    result = validator.validate(data=docref)
     if not result.is_valid:
         raise RuntimeError("Failed to validate document: " + str(result.issues))
 
@@ -51,22 +51,25 @@ def _find_malformed_urls(
         "PaginationConfig": {"PageSize": 50},
     }
 
-    malformed_pointers = []
+    malformed_pointers: list[dict[str, Any]] = []
     total_scanned_count = 0
 
     start_time = datetime.now(tz=timezone.utc)
 
     for page in paginator.paginate(**params):
         for item in page["Items"]:
-            pointer_id = item.get("id", {}).get("S")
+            pointer_id = item.get("id", {}).get("S", "")
             custodian, _ = pointer_id.split("-", 1)
             # only need to check pointers created by the specified custodians
             if custodian in custodian_ods_codes:
+                patient_number = item.get("nhs_number", {}).get("S", "")
                 document = item.get("document", {}).get("S", "")
                 try:
                     _validate_url(document)
                 except Exception as exc:
-                    malformed_pointers.append((pointer_id, exc))
+                    malformed_pointers.append(
+                        {"id": pointer_id, "patient_number": patient_number, "err": exc}
+                    )
 
         total_scanned_count += page["ScannedCount"]
 
@@ -84,8 +87,8 @@ def _find_malformed_urls(
 
     print("Writing malformed_pointers to file ./malformed_pointers.txt ...")  # noqa
     with open("malformed_pointers.txt", "w") as f:
-        for _id, err in malformed_pointers:
-            f.write(f"{_id}: {err}\n")
+        for malformed_pointer_info in malformed_pointers:
+            f.write(f"{malformed_pointer_info}\n")
 
     return {
         "malformed_pointers": len(malformed_pointers),