diff --git a/api/services/trigger/webhook_service.py b/api/services/trigger/webhook_service.py index 4b3e1330fd..5c4607d400 100644 --- a/api/services/trigger/webhook_service.py +++ b/api/services/trigger/webhook_service.py @@ -33,6 +33,11 @@ from services.errors.app import QuotaExceededError from services.trigger.app_trigger_service import AppTriggerService from services.workflow.entities import WebhookTriggerData +try: + import magic +except ImportError: + magic = None # type: ignore[assignment] + logger = logging.getLogger(__name__) @@ -317,7 +322,8 @@ class WebhookService: try: file_content = request.get_data() if file_content: - file_obj = cls._create_file_from_binary(file_content, "application/octet-stream", webhook_trigger) + mimetype = cls._detect_binary_mimetype(file_content) + file_obj = cls._create_file_from_binary(file_content, mimetype, webhook_trigger) return {"raw": file_obj.to_dict()}, {} else: return {"raw": None}, {} @@ -341,6 +347,18 @@ class WebhookService: body = {"raw": ""} return body, {} + @staticmethod + def _detect_binary_mimetype(file_content: bytes) -> str: + """Guess MIME type for binary payloads using python-magic when available.""" + if magic is not None: + try: + detected = magic.from_buffer(file_content[:1024], mime=True) + if detected: + return detected + except Exception: + logger.debug("python-magic detection failed for octet-stream payload") + return "application/octet-stream" + @classmethod def _process_file_uploads( cls, files: Mapping[str, FileStorage], webhook_trigger: WorkflowWebhookTrigger diff --git a/api/tests/unit_tests/services/test_webhook_service.py b/api/tests/unit_tests/services/test_webhook_service.py index 920b1e91b6..d788657589 100644 --- a/api/tests/unit_tests/services/test_webhook_service.py +++ b/api/tests/unit_tests/services/test_webhook_service.py @@ -110,6 +110,70 @@ class TestWebhookServiceUnit: assert webhook_data["method"] == "POST" assert webhook_data["body"]["raw"] == "raw text content" + def test_extract_octet_stream_body_uses_detected_mime(self): + """Octet-stream uploads should rely on detected MIME type.""" + app = Flask(__name__) + binary_content = b"plain text data" + + with app.test_request_context( + "/webhook", method="POST", headers={"Content-Type": "application/octet-stream"}, data=binary_content + ): + webhook_trigger = MagicMock() + mock_file = MagicMock() + mock_file.to_dict.return_value = {"file": "data"} + + with ( + patch.object(WebhookService, "_detect_binary_mimetype", return_value="text/plain") as mock_detect, + patch.object(WebhookService, "_create_file_from_binary") as mock_create, + ): + mock_create.return_value = mock_file + body, files = WebhookService._extract_octet_stream_body(webhook_trigger) + + assert body["raw"] == {"file": "data"} + assert files == {} + mock_detect.assert_called_once_with(binary_content) + mock_create.assert_called_once() + args = mock_create.call_args[0] + assert args[0] == binary_content + assert args[1] == "text/plain" + assert args[2] is webhook_trigger + + def test_detect_binary_mimetype_uses_magic(self, monkeypatch): + """python-magic output should be used when available.""" + fake_magic = MagicMock() + fake_magic.from_buffer.return_value = "image/png" + monkeypatch.setattr("services.trigger.webhook_service.magic", fake_magic) + + result = WebhookService._detect_binary_mimetype(b"binary data") + + assert result == "image/png" + fake_magic.from_buffer.assert_called_once() + + def test_detect_binary_mimetype_fallback_without_magic(self, monkeypatch): + """Fallback MIME type should be used when python-magic is unavailable.""" + monkeypatch.setattr("services.trigger.webhook_service.magic", None) + + result = WebhookService._detect_binary_mimetype(b"binary data") + + assert result == "application/octet-stream" + + def test_detect_binary_mimetype_handles_magic_exception(self, monkeypatch): + """Fallback MIME type should be used when python-magic raises an exception.""" + try: + import magic as real_magic + except ImportError: + pytest.skip("python-magic is not installed") + + fake_magic = MagicMock() + fake_magic.from_buffer.side_effect = real_magic.MagicException("magic error") + monkeypatch.setattr("services.trigger.webhook_service.magic", fake_magic) + + with patch("services.trigger.webhook_service.logger") as mock_logger: + result = WebhookService._detect_binary_mimetype(b"binary data") + + assert result == "application/octet-stream" + mock_logger.debug.assert_called_once() + def test_extract_webhook_data_invalid_json(self): """Test webhook data extraction with invalid JSON.""" app = Flask(__name__)