Merge pull request #1924 from alphagov/replay-letters-in-error

Added a method to reply letters that are in the error folder on S3.

app/celery/letters_pdf_tasks.py

@@ -26,7 +26,9 @@ from app.letters.utils import (
     get_reference_from_filename,
     move_scanned_pdf_to_test_or_live_pdf_bucket,
     upload_letter_pdf,
-    move_failed_pdf, ScanErrorType)
+    move_failed_pdf, ScanErrorType, move_error_pdf_to_scan_bucket,
+    get_file_names_from_error_bucket
+)
 from app.models import (
     KEY_TYPE_TEST,
     NOTIFICATION_CREATED,
@@ -192,7 +194,9 @@ def process_virus_scan_failed(filename):
             )
         )
 
-    raise VirusScanError('notification id {} Virus scan failed: {}'.format(notification.id, filename))
+    error = VirusScanError('notification id {} Virus scan failed: {}'.format(notification.id, filename))
+    current_app.logger.exception(error)
+    raise error
 
 
 @notify_celery.task(name='process-virus-scan-error')
@@ -208,8 +212,9 @@ def process_virus_scan_error(filename):
                 updated_count
             )
         )
-
-    raise VirusScanError('notification id {} Virus scan error: {}'.format(notification.id, filename))
+    error = VirusScanError('notification id {} Virus scan error: {}'.format(notification.id, filename))
+    current_app.logger.exception(error)
+    raise error
 
 
 def update_letter_pdf_status(reference, status):
@@ -219,3 +224,29 @@ def update_letter_pdf_status(reference, status):
             'status': status,
             'updated_at': datetime.utcnow()
         })
+
+
+def replay_letters_in_error(filename=None):
+    # This method can be used to replay letters that end up in the ERROR directory.
+    # We had an incident where clamAV was not processing the virus scan.
+    if filename:
+        move_error_pdf_to_scan_bucket(filename)
+        # call task to add the filename to anti virus queue
+        current_app.logger.info("Calling scan_file for: {}".format(filename))
+        notify_celery.send_task(
+            name=TaskNames.SCAN_FILE,
+            kwargs={'filename': filename},
+            queue=QueueNames.ANTIVIRUS,
+        )
+    else:
+        error_files = get_file_names_from_error_bucket()
+        for item in error_files:
+            moved_file_name = item.key.split('/')[1]
+            current_app.logger.info("Calling scan_file for: {}".format(moved_file_name))
+            move_error_pdf_to_scan_bucket(moved_file_name)
+            # call task to add the filename to anti virus queue
+            notify_celery.send_task(
+                name=TaskNames.SCAN_FILE,
+                kwargs={'filename': moved_file_name},
+                queue=QueueNames.ANTIVIRUS,
+            )

app/letters/utils.py

@@ -110,6 +110,21 @@ def move_failed_pdf(source_filename, scan_error_type):
     _move_s3_object(scan_bucket, source_filename, scan_bucket, target_filename)
 
 
+def move_error_pdf_to_scan_bucket(source_filename):
+    scan_bucket = current_app.config['LETTERS_SCAN_BUCKET_NAME']
+    error_file = 'ERROR/' + source_filename
+
+    _move_s3_object(scan_bucket, error_file, scan_bucket, source_filename)
+
+
+def get_file_names_from_error_bucket():
+    s3 = boto3.resource('s3')
+    scan_bucket = current_app.config['LETTERS_SCAN_BUCKET_NAME']
+    bucket = s3.Bucket(scan_bucket)
+
+    return bucket.objects.filter(Prefix="ERROR")
+
+
 def get_letter_pdf(notification):
     is_test_letter = notification.key_type == KEY_TYPE_TEST and notification.template.is_precompiled_letter
     if is_test_letter:

tests/app/celery/test_letters_pdf_tasks.py

@@ -20,7 +20,8 @@ from app.celery.letters_pdf_tasks import (
     letter_in_created_state,
     process_virus_scan_passed,
     process_virus_scan_failed,
-    process_virus_scan_error)
+    process_virus_scan_error, replay_letters_in_error
+)
 from app.letters.utils import get_letter_pdf_filename, ScanErrorType
 from app.models import (
     KEY_TYPE_NORMAL,
@@ -363,3 +364,25 @@ def test_process_letter_task_check_virus_scan_error(sample_letter_notification,
     assert "Virus scan error:" in str(e)
     mock_move_failed_pdf.assert_called_once_with(filename, ScanErrorType.ERROR)
     assert sample_letter_notification.status == NOTIFICATION_TECHNICAL_FAILURE
+
+
+def test_replay_letters_in_error_for_all_letters_in_error_bucket(notify_api, mocker):
+    import boto3
+    mockObject = boto3.resource('s3').Object('ERROR', 'ERROR/file_name')
+    mocker.patch("app.celery.letters_pdf_tasks.get_file_names_from_error_bucket", return_value=[mockObject])
+    mock_move = mocker.patch("app.celery.letters_pdf_tasks.move_error_pdf_to_scan_bucket")
+    mock_celery = mocker.patch("app.celery.letters_pdf_tasks.notify_celery.send_task")
+    replay_letters_in_error()
+    mock_move.assert_called_once_with('file_name')
+    mock_celery.assert_called_once_with(name='scan-file', kwargs={'filename': 'file_name'}, queue='antivirus-tasks')
+
+
+def test_replay_letters_in_error_for_one_file(notify_api, mocker):
+    import boto3
+    mockObject = boto3.resource('s3').Object('ERROR', 'ERROR/file_name')
+    mocker.patch("app.celery.letters_pdf_tasks.get_file_names_from_error_bucket", return_value=[mockObject])
+    mock_move = mocker.patch("app.celery.letters_pdf_tasks.move_error_pdf_to_scan_bucket")
+    mock_celery = mocker.patch("app.celery.letters_pdf_tasks.notify_celery.send_task")
+    replay_letters_in_error("file_name")
+    mock_move.assert_called_once_with('file_name')
+    mock_celery.assert_called_once_with(name='scan-file', kwargs={'filename': 'file_name'}, queue='antivirus-tasks')