fix: eagerly validate conversation on service-api and explore chat endpoints (#38801)

Signed-off-by: Yufeng He <40085740+he-yufeng@users.noreply.github.com>
Co-authored-by: autofix-ci[bot] <114827586+autofix-ci[bot]@users.noreply.github.com>

api/controllers/console/explore/completion.py

@@ -36,6 +36,7 @@ from models import Account
 from models.model import AppMode, InstalledApp
 from services.app_generate_service import AppGenerateService
 from services.app_task_service import AppTaskService
+from services.conversation_service import ConversationService
 from services.errors.llm import InvokeRateLimitError
 
 from .. import console_ns
@@ -188,6 +189,15 @@ class ChatApi(InstalledAppResource):
         db.session.commit()
 
         try:
+            # Eagerly validate conversation to avoid hanging on invalid conversation_id
+            if payload.conversation_id:
+                ConversationService.get_conversation(
+                    app_model=app_model,
+                    conversation_id=payload.conversation_id,
+                    user=current_user,
+                    session=db.session(),
+                )
+
             response = AppGenerateService.generate(
                 session=session,
                 app_model=app_model,

api/controllers/service_api/app/completion.py

@@ -40,12 +40,14 @@ from core.errors.error import (
     QuotaExceededError,
 )
 from core.helper.trace_id_helper import get_external_trace_id, get_trace_session_id, omit_trace_session_id_from_payload
+from extensions.ext_database import db
 from graphon.model_runtime.errors.invoke import InvokeError
 from libs import helper
 from libs.helper import UUIDStrOrEmpty
 from models.model import App, AppMode, EndUser
 from services.app_generate_service import AppGenerateService
 from services.app_task_service import AppTaskService
+from services.conversation_service import ConversationService
 from services.errors.app import IsDraftWorkflowError, WorkflowIdFormatError, WorkflowNotFoundError
 from services.errors.llm import InvokeRateLimitError
 
@@ -377,6 +379,15 @@ class ChatApi(Resource):
         streaming = _resolve_agent_app_streaming(app_mode=app_mode, response_mode=payload.response_mode)
 
         try:
+            # Eagerly validate conversation to avoid hanging on invalid conversation_id
+            if payload.conversation_id:
+                ConversationService.get_conversation(
+                    app_model=app_model,
+                    conversation_id=payload.conversation_id,
+                    user=end_user,
+                    session=db.session(),
+                )
+
             response = AppGenerateService.generate(
                 session=session,
                 app_model=app_model,

api/tests/unit_tests/controllers/console/explore/test_completion.py

@@ -1,3 +1,4 @@
+import uuid
 from inspect import unwrap
 from unittest.mock import MagicMock, PropertyMock, patch
 
@@ -310,6 +311,37 @@ class TestChatApi:
             with pytest.raises(completion_module.NotFound):
                 method(api, MagicMock(), user, chat_app)
 
+    def test_invalid_conversation_id_fails_fast_as_not_found(self, app: Flask, chat_app, user) -> None:
+        # A nonexistent conversation_id must fail fast as 404, before the streaming
+        # generator is created. Previously the lookup only ran inside the generator,
+        # so an invalid id surfaced as a hang instead of a clean error.
+        payload_patch = patch.object(
+            type(completion_module.console_ns),
+            "payload",
+            new_callable=PropertyMock,
+            return_value={"inputs": {}, "query": "hi", "conversation_id": str(uuid.uuid4())},
+        )
+        generate_mock = MagicMock(return_value={"ok": True})
+
+        api = completion_module.ChatApi()
+        method = unwrap(api.post)
+
+        with (
+            app.test_request_context("/", json={}),
+            payload_patch,
+            patch.object(
+                completion_module.ConversationService,
+                "get_conversation",
+                side_effect=completion_module.services.errors.conversation.ConversationNotExistsError(),
+            ),
+            patch.object(completion_module.AppGenerateService, "generate", generate_mock),
+        ):
+            with pytest.raises(completion_module.NotFound):
+                method(api, MagicMock(), user, chat_app)
+
+        # The lookup must run before generation, so the generator is never started.
+        generate_mock.assert_not_called()
+
     def test_app_unavailable_chat(self, app: Flask, chat_app, user, payload_patch):
         api = completion_module.ChatApi()
         method = unwrap(api.post)

api/tests/unit_tests/controllers/service_api/app/test_completion.py

@@ -42,6 +42,7 @@ from graphon.model_runtime.errors.invoke import InvokeError
 from models.model import App, AppMode, EndUser
 from services.app_generate_service import AppGenerateService
 from services.app_task_service import AppTaskService
+from services.conversation_service import ConversationService
 from services.errors.app import IsDraftWorkflowError, WorkflowIdFormatError, WorkflowNotFoundError
 from services.errors.conversation import ConversationNotExistsError
 from services.errors.llm import InvokeRateLimitError
@@ -554,6 +555,35 @@ class TestChatApiController:
             with pytest.raises(AgentNotPublishedError):
                 handler(api, session=Mock(), app_model=app_model, end_user=end_user)
 
+    def test_invalid_conversation_id_fails_fast_as_not_found(self, app: Flask, monkeypatch: pytest.MonkeyPatch) -> None:
+        # A well-formed but nonexistent conversation_id must fail fast as 404, before the
+        # streaming generator is created. Previously the lookup only ran inside the generator,
+        # so an invalid id surfaced as a hang instead of a clean error.
+        monkeypatch.setattr(
+            ConversationService,
+            "get_conversation",
+            lambda *_args, **_kwargs: (_ for _ in ()).throw(ConversationNotExistsError()),
+        )
+
+        generate_mock = Mock(return_value={"text": "unused"})
+        monkeypatch.setattr(AppGenerateService, "generate", generate_mock)
+
+        api = ChatApi()
+        handler = unwrap(api.post)
+        app_model = SimpleNamespace(mode=AppMode.CHAT.value, id="app-1")
+        end_user = SimpleNamespace()
+
+        with app.test_request_context(
+            "/chat-messages",
+            method="POST",
+            json={"inputs": {}, "query": "hi", "conversation_id": str(uuid.uuid4())},
+        ):
+            with pytest.raises(NotFound):
+                handler(api, session=Mock(), app_model=app_model, end_user=end_user)
+
+        # The lookup must run before generation, so the generator is never started.
+        generate_mock.assert_not_called()
+
 
 class TestChatStopApiController:
     def test_wrong_mode(self, app: Flask) -> None: