From 5dcd7aa862c9becff215690b96b06d2c01562ee8 Mon Sep 17 00:00:00 2001
From: Richard Smith <43441481+thisusernameisnowtaken@users.noreply.github.com>
Date: Mon, 27 Nov 2023 12:39:19 +0000
Subject: [PATCH] Prmdr 520 accept apostrophes in patient names for bulk upload
 (#165)

* Add apostrophe to LG validation regex
* Update unit tests
---
 .../handlers/back_channel_logout_handler.py   |  9 +++++---
 .../test_back_channel_logout_handler.py       |  1 -
 .../unit/utils/test_lloyd_george_validator.py | 23 +++++++++++++++++++
 lambdas/utils/lloyd_george_validator.py       |  4 ++--
 4 files changed, 31 insertions(+), 6 deletions(-)

diff --git a/lambdas/handlers/back_channel_logout_handler.py b/lambdas/handlers/back_channel_logout_handler.py
index 55996f62e..ea81cac35 100644
--- a/lambdas/handlers/back_channel_logout_handler.py
+++ b/lambdas/handlers/back_channel_logout_handler.py
@@ -72,7 +72,7 @@ def logout_handler(token):
 def remove_session_from_dynamo_db(sid):
     dynamodb_name = os.environ["AUTH_DYNAMODB_NAME"]
     dynamodb_service = DynamoDBService()
-    
+
     filter_sid = Attr("sid").eq(sid)
     db_response = dynamodb_service.scan_table(
         dynamodb_name, filter_expression=filter_sid
@@ -83,7 +83,10 @@ def remove_session_from_dynamo_db(sid):
         ndr_session_id = items[0]["NDRSessionId"]
 
         dynamodb_service.delete_item(
-            key={"NDRSessionId": ndr_session_id }, table_name=dynamodb_name
+            key={"NDRSessionId": ndr_session_id}, table_name=dynamodb_name
         )
 
-    logger.info(f"Session removed for sid: {sid} and NDRSessionId {ndr_session_id }", {"Result": "Successful logout"})
+    logger.info(
+        f"Session removed for sid: {sid} and NDRSessionId {ndr_session_id }",
+        {"Result": "Successful logout"},
+    )
diff --git a/lambdas/tests/unit/handlers/test_back_channel_logout_handler.py b/lambdas/tests/unit/handlers/test_back_channel_logout_handler.py
index 444996701..be1e43da7 100644
--- a/lambdas/tests/unit/handlers/test_back_channel_logout_handler.py
+++ b/lambdas/tests/unit/handlers/test_back_channel_logout_handler.py
@@ -1,4 +1,3 @@
-
 import pytest
 from botocore.exceptions import ClientError
 from handlers.back_channel_logout_handler import lambda_handler
diff --git a/lambdas/tests/unit/utils/test_lloyd_george_validator.py b/lambdas/tests/unit/utils/test_lloyd_george_validator.py
index c965cd687..85a0ef3d9 100644
--- a/lambdas/tests/unit/utils/test_lloyd_george_validator.py
+++ b/lambdas/tests/unit/utils/test_lloyd_george_validator.py
@@ -51,6 +51,16 @@ def test_valid_file_name_special_characters():
         assert False, "One or more of the files do not match naming convention"
 
 
+def test_valid_file_name_with_apostrophe():
+    try:
+        file_name = (
+            "1of1_Lloyd_George_Record_[Joé Blöggê's-Glüë]_[1111111111]_[25-12-2019].pdf"
+        )
+        validate_file_name(file_name)
+    except LGInvalidFilesException:
+        assert False, "One or more of the files do not match naming convention"
+
+
 def test_files_with_duplication():
     with pytest.raises(LGInvalidFilesException):
         lg_file_list = [
@@ -105,6 +115,19 @@ def test_files_list_with_invalid_name():
     assert str(e.value) == "One or more of the files do not match naming convention"
 
 
+def test_file_name_with_apostrophe_as_name():
+    """
+    This is an edge case which currently passes.
+    As part of prmdr-520 it was decided that it was acceptable to have an apostrophe accepted as a name
+    This is because patient names will only ever come from PDS
+    """
+    try:
+        file_name = "1of1_Lloyd_George_Record_[']_[1111111111]_[25-12-2019].pdf"
+        validate_file_name(file_name)
+    except LGInvalidFilesException:
+        assert False, "One or more of the files do not match naming convention"
+
+
 def test_files_without_missing_files():
     try:
         lg_file_list = [
diff --git a/lambdas/utils/lloyd_george_validator.py b/lambdas/utils/lloyd_george_validator.py
index 8099dad0c..d0462349c 100644
--- a/lambdas/utils/lloyd_george_validator.py
+++ b/lambdas/utils/lloyd_george_validator.py
@@ -28,7 +28,7 @@ def validate_lg_file_type(file_type: str):
 
 
 def validate_file_name(name: str):
-    lg_regex = r"[0-9]+of[0-9]+_Lloyd_George_Record_\[[A-Za-z À-ÿ\-]+]_\[[0-9]{10}]_\[\d\d-\d\d-\d\d\d\d].pdf"
+    lg_regex = r"[0-9]+of[0-9]+_Lloyd_George_Record_\[[A-Za-z À-ÿ\-']+]_\[[0-9]{10}]_\[\d\d-\d\d-\d\d\d\d].pdf"
     if not re.fullmatch(lg_regex, name):
         raise LGInvalidFilesException(
             "One or more of the files do not match naming convention"
@@ -85,7 +85,7 @@ def validate_lg_file_names(file_name_list: list[str], nhs_number: Optional[str]
 def extract_info_from_filename(filename: str) -> dict:
     page_number = r"(?P<page_no>[1-9][0-9]*)"
     total_page_number = r"(?P<total_page_no>[1-9][0-9]*)"
-    patient_name = r"(?P<patient_name>[A-Za-z À-ÿ\-]+)"
+    patient_name = r"(?P<patient_name>[A-Za-z À-ÿ\-']+)"
     nhs_number = r"(?P<nhs_number>\d{10})"
     date_of_birth = r"(?P<date_of_birth>\d\d-\d\d-\d\d\d\d)"