Added the comparison function for different models and the comparison function for different versions

Author: 2h0u4n

backend/agents/create_agent_info.py

@@ -365,14 +365,14 @@ async def create_agent_run_info(
     language: str = "zh",
     allow_memory_search: bool = True,
     is_debug: bool = False,
-    version_no_override: int | None = None,
+    override_version_no: int | None = None,
     override_model_id: int | None = None,
 ):
     # Determine which version_no to use based on is_debug flag
     # If is_debug=false, use the current published version (current_version_no)
     # If is_debug=true, use version 0 (draft/editing state)
-    if version_no_override is not None:
-        version_no = version_no_override
+    if override_version_no is not None:
+        version_no = override_version_no
     elif is_debug:
         version_no = 0
     else:

backend/apps/agent_app.py

@@ -70,11 +70,7 @@ async def agent_stop_api(conversation_id: int, authorization: Optional[str] = He
     stop agent run and preprocess tasks for specified conversation_id
     """
     user_id, _ = get_current_user_id(authorization)
-    if stop_agent_tasks(conversation_id, user_id).get("status") == "success":
-        return {"status": "success", "message": "agent run and preprocess tasks stopped successfully"}
-    else:
-        raise HTTPException(status_code=HTTPStatus.BAD_REQUEST,
-                            detail=f"no running agent or preprocess tasks found for conversation_id {conversation_id}")
+    return stop_agent_tasks(conversation_id, user_id)
 
 
 @agent_config_router.post("/search_info")

backend/services/agent_service.py

@@ -1582,7 +1582,7 @@ async def prepare_agent_run(
         language=language,
         allow_memory_search=allow_memory_search,
         is_debug=agent_request.is_debug,
-        version_no_override=agent_request.version_no,
+        override_version_no=agent_request.version_no,
         override_model_id=agent_request.model_id,
     )
     agent_run_manager.register_agent_run(
@@ -1937,8 +1937,8 @@ def stop_agent_tasks(conversation_id: int, user_id: str):
         return {"status": "success", "message": message}
     else:
         message = f"no running agent or preprocess tasks found for user_id {user_id}, conversation_id {conversation_id}"
-        logging.error(message)
-        return {"status": "error", "message": message}
+        logging.info(message)
+        return {"status": "success", "message": message, "already_stopped": True}
 
 
 async def get_agent_id_by_name(agent_name: str, tenant_id: str) -> int:

docker/generate_env.sh

@@ -109,25 +109,25 @@ update_env_file() {
   # Main Services
   # CONFIG_SERVICE_URL
   if grep -q "^CONFIG_SERVICE_URL=" ../.env; then
-    sed -i.bak "s~^CONFIG_SERVICE_URL=.*~CONFIG_SERVICE_URL=http://127.0.0.1:5010~" ../.env
+    sed -i.bak "s~^CONFIG_SERVICE_URL=.*~CONFIG_SERVICE_URL=http://localhost:5010~" ../.env
   else
     echo "" >> ../.env
     echo "# Main Services" >> ../.env
-    echo "CONFIG_SERVICE_URL=http://127.0.0.1:5010" >> ../.env
+    echo "CONFIG_SERVICE_URL=http://localhost:5010" >> ../.env
   fi
 
   # RUNTIME_SERVICE_URL
   if grep -q "^RUNTIME_SERVICE_URL=" ../.env; then
-    sed -i.bak "s~^RUNTIME_SERVICE_URL=.*~RUNTIME_SERVICE_URL=http://127.0.0.1:5014~" ../.env
+    sed -i.bak "s~^RUNTIME_SERVICE_URL=.*~RUNTIME_SERVICE_URL=http://localhost:5014~" ../.env
   else
-    echo "RUNTIME_SERVICE_URL=http://127.0.0.1:5014" >> ../.env
+    echo "RUNTIME_SERVICE_URL=http://localhost:5014" >> ../.env
   fi
 
   # ELASTICSEARCH_SERVICE
   if grep -q "^ELASTICSEARCH_SERVICE=" ../.env; then
-    sed -i.bak "s~^ELASTICSEARCH_SERVICE=.*~ELASTICSEARCH_SERVICE=http://127.0.0.1:5010/api~" ../.env
+    sed -i.bak "s~^ELASTICSEARCH_SERVICE=.*~ELASTICSEARCH_SERVICE=http://localhost:5010/api~" ../.env
   else
-    echo "ELASTICSEARCH_SERVICE=http://127.0.0.1:5010/api" >> ../.env
+    echo "ELASTICSEARCH_SERVICE=http://localhost:5010/api" >> ../.env
   fi
 
   # NEXENT_MCP_SERVER

frontend/server.js

@@ -22,10 +22,10 @@ const app = next({
 const handle = app.getRequestHandler();
 
 // Backend addresses
-const HTTP_BACKEND = process.env.HTTP_BACKEND || "http://127.0.0.1:5010"; // config
-const WS_BACKEND = process.env.WS_BACKEND || "ws://127.0.0.1:5014"; // runtime
+const HTTP_BACKEND = process.env.HTTP_BACKEND || "http://localhost:5010"; // config
+const WS_BACKEND = process.env.WS_BACKEND || "ws://localhost:5014"; // runtime
 const RUNTIME_HTTP_BACKEND =
-  process.env.RUNTIME_HTTP_BACKEND || "http://127.0.0.1:5014"; // runtime
+  process.env.RUNTIME_HTTP_BACKEND || "http://localhost:5014"; // runtime
 const MINIO_BACKEND = process.env.MINIO_ENDPOINT || "http://localhost:9010";
 const MARKET_BACKEND =
   process.env.MARKET_BACKEND || "https://market.nexent.tech"; // market

sdk/nexent/core/models/openai_llm.py

@@ -106,13 +106,9 @@ def __call__(self, messages: List[Dict[str, Any]], stop_sequences: Optional[List
 
         try:
             for chunk in current_request:
-                if not chunk.choices:
-                    chunk_list.append(chunk)
-                    continue
-
-                delta = chunk.choices[0].delta
-                new_token = getattr(delta, "content", None)
-                reasoning_content = getattr(delta, "reasoning_content", None)
+                new_token = chunk.choices[0].delta.content
+                reasoning_content = getattr(
+                    chunk.choices[0].delta, 'reasoning_content', None)
 
                 # Handle reasoning_content if it exists and is not null
                 if reasoning_content is not None:
@@ -134,7 +130,7 @@ def __call__(self, messages: List[Dict[str, Any]], stop_sequences: Optional[List
 
                     self.observer.add_model_new_token(new_token)
                     token_join.append(new_token)
-                    role = getattr(delta, "role", role)
+                    role = chunk.choices[0].delta.role
 
                 chunk_list.append(chunk)
                 if self.stop_event.is_set():
@@ -144,19 +140,15 @@ def __call__(self, messages: List[Dict[str, Any]], stop_sequences: Optional[List
                     raise RuntimeError(
                         "Model is interrupted by stop event")
 
-            if not chunk_list:
-                raise RuntimeError("Empty completion stream")
-
             # Send end marker
             self.observer.flush_remaining_tokens()
             model_output = "".join(token_join)
 
             # Extract token usage
             input_tokens = 0
             output_tokens = 0
-            usage_chunk = next((c for c in reversed(chunk_list) if getattr(c, "usage", None) is not None), None)
-            if usage_chunk is not None:
-                usage = usage_chunk.usage
+            if chunk_list and chunk_list[-1].usage is not None:
+                usage = chunk_list[-1].usage
                 input_tokens = usage.prompt_tokens
                 output_tokens = usage.completion_tokens if hasattr(
                     usage, 'completion_tokens') else usage.total_tokens

test/backend/app/test_agent_app.py

@@ -228,20 +228,18 @@ def test_agent_stop_api_not_found(mocker, mock_conversation_id):
     mock_get_user_id.return_value = ("test_user_id", "test_tenant_id")
 
     mock_stop_tasks = mocker.patch("apps.agent_app.stop_agent_tasks")
-    mock_stop_tasks.return_value = {"status": "error"}  # Simulate not found
+    mock_stop_tasks.return_value = {"status": "success", "message": "already stopped"}  # Simulate not found
 
     response = runtime_client.get(
         f"/agent/stop/{mock_conversation_id}",
         headers={"Authorization": "Bearer test_token"}
     )
 
-    # The app should raise HTTPException for non-success status
-    assert response.status_code == 400
+    assert response.status_code == 200
     mock_get_user_id.assert_called_once_with("Bearer test_token")
     mock_stop_tasks.assert_called_once_with(
         mock_conversation_id, "test_user_id")
-    assert "no running agent or preprocess tasks found" in response.json()[
-        "detail"]
+    assert response.json()["status"] == "success"
 
 
 def test_search_agent_info_api_success(mocker, mock_auth_header):
@@ -1953,4 +1951,4 @@ def test_list_published_agents_api_exception(mocker, mock_auth_header):
     )
 
     assert response.status_code == 500
-    assert "Published agents list error" in response.json()["detail"]
+    assert "Published agents list error" in response.json()["detail"]

test/backend/services/test_agent_service.py

@@ -3588,8 +3588,9 @@ def test_stop_agent_tasks(mock_preprocess_manager, mock_agent_run_manager):
     mock_agent_run_manager.stop_agent_run.return_value = False
     mock_preprocess_manager.stop_preprocess_tasks.return_value = False
     result = stop_agent_tasks(123, "test_user")
-    assert result["status"] == "error"
+    assert result["status"] == "success"
     assert "no running agent or preprocess tasks found" in result["message"]
+    assert result.get("already_stopped") is True
 
 
 @patch('backend.services.agent_service.search_agent_id_by_agent_name')