Fix timestamp validation for EA and batch

Author: dlzhry2nhs

lambdas/filenameprocessor/src/constants.py

@@ -38,6 +38,10 @@
     Exception: 500,
 }
 
+# Filename timestamp constants
+VALID_TIMESTAMP_LENGTH = 17
+VALID_TIMEZONE_OFFSETS = {"00", "01"}
+
 
 class FileStatus(StrEnum):
     """File status constants"""

lambdas/filenameprocessor/src/file_name_processor.py

@@ -264,7 +264,6 @@ def handle_extended_attributes_file(
             FileStatus.PROCESSING,
         )
 
-        # TODO: agree the prefix with DPS
         dest_file_key = f"{DPS_DESTINATION_PREFIX}/{file_key}"
         copy_file_to_external_bucket(
             bucket_name,

lambdas/filenameprocessor/src/file_validation.py

@@ -3,7 +3,14 @@
 from datetime import datetime
 from re import match
 
-from constants import EXTENDED_ATTRIBUTES_FILE_PREFIX, EXTENDED_ATTRIBUTES_VACC_TYPE, VALID_EA_VERSIONS, VALID_VERSIONS
+from constants import (
+    EXTENDED_ATTRIBUTES_FILE_PREFIX,
+    EXTENDED_ATTRIBUTES_VACC_TYPE,
+    VALID_EA_VERSIONS,
+    VALID_TIMESTAMP_LENGTH,
+    VALID_TIMEZONE_OFFSETS,
+    VALID_VERSIONS,
+)
 from elasticache import (
     get_supplier_system_from_cache,
     get_valid_vaccine_types_from_cache,
@@ -20,20 +27,32 @@ def is_file_in_directory_root(file_key: str) -> bool:
 
 def is_valid_datetime(timestamp: str) -> bool:
     """
-    Returns a bool to indicate whether the timestamp is a valid datetime in the format 'YYYYmmddTHHMMSSzz'
-    where 'zz' is a two digit number indicating the timezone
+    Returns a bool to indicate whether the file timestamp matches the expected NHS file submission format:
+    - YYYY = Year
+    - MM = Month (01-12)
+    - DD = Date (01-31)
+    - T = fixed value of “T”
+    - hh = Hours (00-23)
+    - mm = Minutes (00-59)
+    - ss = Seconds (00-59)
+    - time zone offset = 00 for GMT, 01 for BST
+
+    e.g. 20220514T10081501
     """
     # Check that datetime (excluding timezone) is a valid datetime in the expected format.
-    if len(timestamp) < 15:
+    if len(timestamp) != VALID_TIMESTAMP_LENGTH:
         return False
 
-    # Note that any digits after the seconds (i.e. from the 16th character onwards, usually expected to represent
-    # timezone), do not need to be validated
     try:
         datetime.strptime(timestamp[:15], "%Y%m%dT%H%M%S")
     except ValueError:
         return False
 
+    time_zone_offset = timestamp[-2:]
+
+    if time_zone_offset not in VALID_TIMEZONE_OFFSETS:
+        return False
+
     return True
 
 

lambdas/filenameprocessor/tests/test_file_key_validation.py

@@ -45,12 +45,13 @@ def test_is_valid_datetime(self, _):
         """Tests that is_valid_datetime returns True for valid datetimes, and false otherwise"""
         # Test case tuples are structured as (date_time_string, expected_result)
         test_cases = [
-            ("20200101T12345600", True),  # Valid datetime string with timezone
-            ("20200101T123456", True),  # Valid datetime string without timezone
+            ("20200101T12345600", True),  # Valid datetime string with timezone offset "00" GMT
+            ("20200101T12345601", True),  # Valid datetime string with timezone offset "01" BST
+            ("20200101T123456", False),  # Invalid datetime string without timezone
             (
                 "20200101T123456extracharacters",
-                True,
-            ),  # Valid datetime string with additional characters
+                False,
+            ),  # Invalid datetime string with additional characters
             ("20201301T12345600", False),  # Invalid month
             ("20200100T12345600", False),  # Invalid day
             ("20200230T12345600", False),  # Invalid combination of month and day
@@ -124,7 +125,7 @@ def test_validate_extended_attributes_file_key(self, mock_get_redis_client):
             ),
             # Valid extended attributes file key with different organization code
             (
-                "Vaccination_Extended_Attributes_v1_5_YGM41_20221231T23595999.csv",
+                "Vaccination_Extended_Attributes_v1_5_YGM41_20221231T23595900.csv",
                 "YGM41",
             ),
         ]