Get ack files only from day before the ack file is received.

Take care of upper and lower case of file names and contents
Add a test for s3 get_list_of_files_by_suffix

app/aws/s3.py

@@ -109,7 +109,7 @@ def upload_letters_pdf(reference, crown, filedata):
         upload_file_name, current_app.config['LETTERS_PDF_BUCKET_NAME']))
 
 
-def get_list_of_files_by_suffix(bucket_name, subfolder='', suffix=''):
+def get_list_of_files_by_suffix(bucket_name, subfolder='', suffix='', last_modified=None):
     s3_client = client('s3', current_app.config['AWS_REGION'])
     paginator = s3_client.get_paginator('list_objects_v2')
 
@@ -120,6 +120,7 @@ def get_list_of_files_by_suffix(bucket_name, subfolder='', suffix=''):
 
     for page in page_iterator:
         for obj in page['Contents']:
-            key = obj['Key']
+            key = obj['Key'].lower()
-            if key.endswith(suffix):
+            if key.endswith(suffix.lower()):
+                if not last_modified or obj['LastModified'] >= last_modified:
                     yield key

app/celery/scheduled_tasks.py

@@ -1,3 +1,4 @@
+import pytz
 from datetime import (
     date,
     datetime,
@@ -452,29 +453,39 @@ def daily_stats_template_usage_by_month():
 @notify_celery.task(name='raise-alert-if-no-letter-ack-file')
 @statsd(namespace="tasks")
 def letter_raise_alert_if_no_ack_file_for_zip():
-    # get a list of today's zip files
+    # get a list of zip files since yesterday
     zip_file_list = []
+
     for key in s3.get_list_of_files_by_suffix(bucket_name=current_app.config['LETTERS_PDF_BUCKET_NAME'],
-                                              subfolder=datetime.utcnow().strftime('%Y-%m-%d'), suffix='.ZIP'):
+                                              subfolder=datetime.utcnow().strftime('%Y-%m-%d'),
+                                              suffix='.zip'):
         zip_file_list.append(key)
 
     # get acknowledgement file
     ack_file_list = []
+    yesterday = datetime.now(tz=pytz.utc) - timedelta(days=1)
     for key in s3.get_list_of_files_by_suffix(bucket_name=current_app.config['DVLA_RESPONSE_BUCKET_NAME'],
-                                              subfolder='root/dispatch', suffix='.ACK.txt'):
+                                              subfolder='root/dispatch', suffix='.ACK.txt', lastModified=yesterday):
         ack_file_list.append(key)
 
-    todaystr = datetime.utcnow().strftime('%Y%m%d')
+    today_str = datetime.utcnow().strftime('%Y%m%d')
+    zip_not_today = []
 
     for key in ack_file_list:
-        if todaystr in key:
+        if today_str in key:
             content = s3.get_s3_file(current_app.config['DVLA_RESPONSE_BUCKET_NAME'], key)
-
             for zip_file in content.split('\n'):    # each line
                 s = zip_file.split('|')
                 for zf in zip_file_list:
-                    if s[0] in zf:
+                    if s[0].lower() in zf.lower():
                         zip_file_list.remove(zf)
+                    else:
+                        zip_not_today.append(s[0])
 
     if zip_file_list:
         raise NoAckFileReceived(message=zip_file_list)
+
+    if zip_not_today:
+        current_app.logger.info(
+            "letter ack contains zip that is not for today {} ".format(zip_not_today)
+        )

tests/app/aws/test_s3.py

@@ -1,7 +1,7 @@
 from unittest.mock import call
 from datetime import datetime, timedelta
 import pytest
-
+import pytz
 from flask import current_app
 
 from freezegun import freeze_time
@@ -11,7 +11,8 @@ from app.aws.s3 import (
     get_s3_file,
     filter_s3_bucket_objects_within_date_range,
     remove_transformed_dvla_file,
-    upload_letters_pdf
+    upload_letters_pdf,
+    get_list_of_files_by_suffix,
 )
 from tests.app.conftest import datetime_in_past
 
@@ -173,3 +174,37 @@ def test_upload_letters_pdf_puts_in_tomorrows_bucket_after_half_five(notify_api,
         # in tomorrow's folder, but still has this evening's timestamp
         file_location='2017-12-05/NOTIFY.FOO.D.2.C.C.20171204173100.PDF'
     )
+
+
+@freeze_time("2018-01-11 00:00:00")
+@pytest.mark.parametrize('suffix_str, days_before, returned_no', [
+    ('.ACK.txt', None, 1),
+    ('.ack.txt', None, 1),
+    ('.ACK.TXT', None, 1),
+    ('', None, 2),
+    ('', 1, 1),
+])
+def test_get_list_of_files_by_suffix(notify_api, mocker, suffix_str, days_before, returned_no):
+    paginator_mock = mocker.patch('app.aws.s3.client')
+    multiple_pages_s3_object = [
+        {
+            "Contents": [
+                single_s3_object_stub('bar/foo.ACK.txt', datetime_in_past(1, 0)),
+            ]
+        },
+        {
+            "Contents": [
+                single_s3_object_stub('bar/foo1.rs.txt', datetime_in_past(2, 0)),
+            ]
+        }
+    ]
+    paginator_mock.return_value.get_paginator.return_value.paginate.return_value = multiple_pages_s3_object
+    if (days_before):
+        key = get_list_of_files_by_suffix('foo-bucket', subfolder='bar', suffix=suffix_str,
+                                          last_modified=datetime.now(tz=pytz.utc) - timedelta(days=days_before))
+    else:
+        key = get_list_of_files_by_suffix('foo-bucket', subfolder='bar', suffix=suffix_str)
+
+    assert sum(1 for x in key) == returned_no
+    for k in key:
+        assert k == 'bar/foo.ACK.txt'

tests/app/celery/test_scheduled_tasks.py

@@ -1029,20 +1029,18 @@ def test_dao_fetch_monthly_historical_stats_by_template_null_template_id_not_cou
     assert len(result) == 1
 
 
-def mock_s3_get_list_match(bucket_name, subfolder='', suffix=''):
+def mock_s3_get_list_match(bucket_name, subfolder='', suffix='', lastModified=None):
 
     if subfolder == '2018-01-11':
         return ['NOTIFY.20180111175007.ZIP', 'NOTIFY.20180111175008.ZIP']
-    print(suffix)
     if subfolder == 'root/dispatch':
         return ['root/dispatch/NOTIFY.20180111175733.ACK.txt']
 
 
-def mock_s3_get_list_diff(bucket_name, subfolder='', suffix=''):
+def mock_s3_get_list_diff(bucket_name, subfolder='', suffix='', lastModified=None):
     if subfolder == '2018-01-11':
         return ['NOTIFY.20180111175007.ZIP', 'NOTIFY.20180111175008.ZIP', 'NOTIFY.20180111175009.ZIP',
                 'NOTIFY.20180111175010.ZIP']
-    print(suffix)
     if subfolder == 'root/dispatch':
         return ['root/dispatch/NOTIFY.20180111175733.ACK.txt']