def redact_pii_documents_handler(event, context):
"""Redaction Lambda function handler."""
LOG.info('Received event with requestId: %s', event[REQUEST_ID])
LOG.debug(f'Raw event {event}')
InputEventValidator.validate(event)
invoke_args = json.loads(event[S3OL_CONFIGURATION][PAYLOAD]) if event[S3OL_CONFIGURATION][PAYLOAD] else {}
language_code = invoke_args.get(LANGUAGE_CODE, DEFAULT_LANGUAGE_CODE)
redaction_config = RedactionConfig(**invoke_args)
object_get_context = event[GET_OBJECT_CONTEXT]
s3ol_access_point = event[S3OL_CONFIGURATION][S3OL_ACCESS_POINT_ARN]
s3 = S3Client(s3ol_access_point)
cloud_watch = CloudWatchClient()
comprehend = ComprehendClient(s3ol_access_point=s3ol_access_point, session_id=event[REQUEST_ID], user_agent=DEFAULT_USER_AGENT,
endpoint_url=COMPREHEND_ENDPOINT_URL)
exception_handler = ExceptionHandler(s3)
LOG.debug("Pii Entity Types to be redacted:" + str(redaction_config.pii_entity_types))
processed_document = False
document = Document('')
try:
def time_bound_task():
nonlocal processed_document
nonlocal document
PartialObjectRequestValidator.validate(event)
pii_classification_segmenter = Segmenter(DOCUMENT_MAX_SIZE_CONTAINS_PII_ENTITIES)
pii_redaction_segmenter = Segmenter(DOCUMENT_MAX_SIZE_DETECT_PII_ENTITIES)
redactor = Redactor(redaction_config)
time1 = time.time()
text, http_headers, status_code = s3.download_file_from_presigned_url(object_get_context[INPUT_S3_URL],
event[USER_REQUEST][HEADERS])
time2 = time.time()
LOG.info(f"Downloaded the file in : {(time2 - time1)} seconds")
document = redact(text, pii_classification_segmenter, pii_redaction_segmenter, redactor,
comprehend, redaction_config, language_code)
processed_document = True
time1 = time.time()
LOG.info(f"Pii redaction completed within {(time1 - time2)} seconds. Returning back the response to S3")
redacted_text_bytes = document.redacted_text.encode('utf-8')
http_headers[CONTENT_LENGTH] = len(redacted_text_bytes)
s3.respond_back_with_data(redacted_text_bytes, http_headers, object_get_context[REQUEST_ROUTE],
object_get_context[REQUEST_TOKEN], status_code)
execute_task_with_timeout(context.get_remaining_time_in_millis() - RESERVED_TIME_FOR_CLEANUP, time_bound_task)
except Exception as generated_exception:
exception_handler.handle_exception(generated_exception, object_get_context[REQUEST_ROUTE], object_get_context[REQUEST_TOKEN])
finally:
if PUBLISH_CLOUD_WATCH_METRICS:
pii_entities = get_interested_pii(document, redaction_config)
publish_metrics(cloud_watch, s3, comprehend, processed_document, len(pii_entities) > 0, language_code,
s3ol_access_point, pii_entities)
LOG.info("Responded back to s3 successfully")
def test_default_exception_handler(self):
s3_client = MagicMock()
ExceptionHandler(s3_client).handle_exception(Exception(), "SomeRoute", "SomeToken")
s3_client.respond_back_with_error.assert_called_once_with(S3_STATUS_CODES.INTERNAL_SERVER_ERROR_500,
S3_ERROR_CODES.InternalError,
"An internal error occurred while processing the file",
"SomeRoute", "SomeToken")
def test_file_size_limit_exceeded_handler(self):
s3_client = MagicMock()
ExceptionHandler(s3_client).handle_exception(FileSizeLimitExceededException(), "SomeRoute", "SomeToken")
s3_client.respond_back_with_error.assert_called_once_with(S3_STATUS_CODES.BAD_REQUEST_400,
S3_ERROR_CODES.EntityTooLarge,
"Size of the requested object exceeds maximum file size supported",
"SomeRoute", "SomeToken")
def test_timeout_exception_handler(self):
s3_client = MagicMock()
ExceptionHandler(s3_client).handle_exception(TimeoutException(),
"SomeRoute", "SomeToken")
s3_client.respond_back_with_error.assert_called_once_with(S3_STATUS_CODES.BAD_REQUEST_400,
S3_ERROR_CODES.RequestTimeout,
"Failed to complete document processing within time limit",
"SomeRoute", "SomeToken")
def test_restricted_document_exception_handler(self):
s3_client = MagicMock()
ExceptionHandler(s3_client).handle_exception(RestrictedDocumentException(),
"SomeRoute", "SomeToken")
s3_client.respond_back_with_error.assert_called_once_with(S3_STATUS_CODES.FORBIDDEN_403,
S3_ERROR_CODES.AccessDenied,
"Document Contains PII",
"SomeRoute", "SomeToken")
def test_s3_download_exception_handler(self):
s3_client = MagicMock()
ExceptionHandler(s3_client).handle_exception(S3DownloadException("InternalError", "Internal Server Error"),
"SomeRoute", "SomeToken")
s3_client.respond_back_with_error.assert_called_once_with(S3_STATUS_CODES.INTERNAL_SERVER_ERROR_500,
S3_ERROR_CODES.InternalError,
"Internal Server Error",
"SomeRoute", "SomeToken")
def test_invalid_configuration_exception_handler(self):
s3_client = MagicMock()
ExceptionHandler(s3_client).handle_exception(InvalidConfigurationException("Missconfigured knob"),
"SomeRoute", "SomeToken")
s3_client.respond_back_with_error.assert_called_once_with(S3_STATUS_CODES.BAD_REQUEST_400,
S3_ERROR_CODES.InvalidRequest,
"Lambda function has been incorrectly setup",
"SomeRoute", "SomeToken")
def test_unsupported_file_exception_handling_return_error(self):
s3_client = MagicMock()
ExceptionHandler(s3_client). \
handle_exception(UnsupportedFileException(file_content="SomeContent", http_headers={'h1': 'v1'}), "SomeRoute", "SomeToken")
s3_client.respond_back_with_error.assert_called_once_with(S3_STATUS_CODES.BAD_REQUEST_400,
S3_ERROR_CODES.UnexpectedContent,
"Unsupported file encountered for determining Pii",
"SomeRoute", "SomeToken")
def test_unsupported_file_exception_handling_return_unknown_error(self):
s3_client = MagicMock()
self.assertRaises(Exception, ExceptionHandler(s3_client).handle_exception,
UnsupportedFileException(file_content="SomeContent", http_headers={'h1': 'v1'}), "SomeRoute", "SomeToken")
def test_unsupported_file_exception_handling_do_not_fail(self):
s3_client = MagicMock()
ExceptionHandler(s3_client). \
handle_exception(UnsupportedFileException(file_content="SomeContent", http_headers={'h1': 'v1'}), "SomeRoute", "SomeToken")
s3_client.respond_back_with_data.assert_called_once_with("SomeContent", {'h1': 'v1'}, "SomeRoute", "SomeToken")