fix: 修复了严重bug

* feat: Clean up dashboard UI, shift workflow WS to SSE, and add file upload support

- Removed Monitoring view and associated `/ws/state` cluster websocket route.
- Modified workflow tracing from WebSocket (`/api/v1/workflow/ws/{trace_id}`) to Server-Sent Events (`/api/v1/workflow/sse/{trace_id}`) for unidirectional pushes, introducing a new `/api/v1/workflow/reply/{trace_id}` POST route to handle incoming client replies.
- Cleaned up dummy data and unneeded links in the chat layout (LeftPanel, ChatPanel).
- Implemented file upload functionality: added a `/api/v1/adapter/client/upload` endpoint to the backend which saves files to a local `uploads` directory, and added an integrated file input triggered via the `+` button in the frontend chat interface to facilitate uploading with an automated chat message confirmation.

Co-authored-by: zhaoxi826 <198742034+zhaoxi826@users.noreply.github.com>

* fix: prevent global_state_machine actor from being garbage collected

Added `lifetime="detached"` and kept a local reference to the `GlobalStateMachine`
actor in `main.py` so that it doesn't get cleaned up by Ray due to going out
of scope, which was causing `ray.get_actor('global_state_machine')` calls to fail
in API route handlers (resulting in 500 errors).

Co-authored-by: zhaoxi826 <198742034+zhaoxi826@users.noreply.github.com>

* fix: resolve named actor addressing failure across Ray processes via explicit namespace

The `ray.get_actor` calls in API routes executing within a Ray Serve worker were failing to
resolve the actors created by the main process because the implicit random namespace of
`ray.init()` did not match the namespace of the Ray Serve application scope.

Instead of overriding garbage collection via `lifetime="detached"` (which can lead to actor
leakage), this assigns an explicit `namespace="pretor"` when initializing Ray in the main process,
and uses the identical namespace in `ray_hook.py` when looking up named actors. Also retains the
local variable assignments in `main.py` to prevent them from being eliminated as unused variables.

Co-authored-by: zhaoxi826 <198742034+zhaoxi826@users.noreply.github.com>

* fix: defer actor lookup in WorkflowRunningEngine to avoid startup race conditions

The `WorkflowRunningEngine` was trying to fetch the `global_state_machine` actor
during its `__init__` method via `ray_actor_hook()`. Since actor creation requests are
dispatched asynchronously, the `global_state_machine` might not be completely
registered and discoverable via `ray.get_actor()` by the time the `WorkflowRunningEngine`'s
`__init__` is evaluated.

Moved the actor lookup to the async `run()` method, which gets executed after the engine
itself is fully up, allowing time for other components to become available in the global
Ray namespace.

Co-authored-by: zhaoxi826 <198742034+zhaoxi826@users.noreply.github.com>

---------

Co-authored-by: google-labs-jules[bot] <161369871+google-labs-jules[bot]@users.noreply.github.com>
Co-authored-by: zhaoxi826 <198742034+zhaoxi826@users.noreply.github.com>

frontend/src/App.tsx

@@ -1,6 +1,5 @@
 import { useState, useEffect } from 'react';
 import { Sidebar } from './components/Layout/Sidebar';
-import { MonitoringLayout } from './components/Monitoring/MonitoringLayout';
 import { SettingsLayout } from './components/Settings/SettingsLayout';
 import { AgentLayout } from './components/Agent/AgentLayout';
 import { ResourceLayout } from './components/Resource/ResourceLayout';
@@ -12,7 +11,7 @@ import { AuthPage } from './components/Auth/AuthPage';
 function App() {
   const [isAuthenticated, setIsAuthenticated] = useState(false);
   const [activeTab, setActiveTab] = useState('chats'); // For LeftPanel
-  const [currentView, setCurrentView] = useState('dashboard'); // 'dashboard', 'settings', 'monitoring', 'agent', 'resource'
+  const [currentView, setCurrentView] = useState('dashboard'); // 'dashboard', 'settings', 'agent', 'resource'
   const [settingsTab, setSettingsTab] = useState('users'); // For SettingsLayout
   const [agentTab, setAgentTab] = useState('worker'); // For AgentLayout
   const [resourceTab, setResourceTab] = useState('skill'); // For ResourceLayout
@@ -38,9 +37,7 @@ function App() {
       <Sidebar currentView={currentView} setCurrentView={setCurrentView} />
 
       {/* Main Content Area depending on view */}
-      {currentView === 'monitoring' ? (
-        <MonitoringLayout />
-      ) : currentView === 'agent' ? (
+      {currentView === 'agent' ? (
         <AgentLayout agentTab={agentTab} setAgentTab={setAgentTab} />
       ) : currentView === 'resource' ? (
         <ResourceLayout resourceTab={resourceTab} setResourceTab={setResourceTab} />

frontend/src/components/Chat/ChatPanel.tsx

@@ -1,4 +1,4 @@
-import { useState } from 'react';
+import React, { useState } from 'react';
 import { MessageSquare, Activity, Terminal, ChevronRight, Plus } from 'lucide-react';
 import apiClient from '../../api/client';
 
@@ -21,6 +21,45 @@ export function ChatPanel() {
   ]);
   const [input, setInput] = useState('');
   const [loading, setLoading] = useState(false);
+  const fileInputRef = React.useRef<HTMLInputElement>(null);
+
+  const handleFileUpload = async (e: React.ChangeEvent<HTMLInputElement>) => {
+    const file = e.target.files?.[0];
+    if (!file) return;
+
+    const formData = new FormData();
+    formData.append('file', file);
+
+    setLoading(true);
+    try {
+      const response = await apiClient.post('/api/v1/adapter/client/upload', formData, {
+        headers: {
+          'Content-Type': 'multipart/form-data'
+        }
+      });
+      const aiMessage: ChatMessage = {
+        id: Date.now().toString(),
+        sender: 'ai',
+        text: `已上传文件: ${response.data.filename}`,
+        timestamp: new Date()
+      };
+      setMessages(prev => [...prev, aiMessage]);
+    } catch (error) {
+      console.error("Error uploading file", error);
+      const errorMessage: ChatMessage = {
+        id: Date.now().toString(),
+        sender: 'ai',
+        text: "文件上传失败。",
+        timestamp: new Date()
+      };
+      setMessages(prev => [...prev, errorMessage]);
+    } finally {
+      setLoading(false);
+      if (fileInputRef.current) {
+        fileInputRef.current.value = '';
+      }
+    }
+  };
 
   const handleSendMessage = async () => {
     if (!input.trim()) return;
@@ -124,7 +163,14 @@ export function ChatPanel() {
       {/* Chat Input */}
       <div className="p-4 bg-white border-t border-slate-200">
         <div className="relative flex items-center">
+          <input
+            type="file"
+            ref={fileInputRef}
+            onChange={handleFileUpload}
+            className="hidden"
+          />
           <button
+            onClick={() => fileInputRef.current?.click()}
             className="absolute left-2 p-1.5 text-slate-400 hover:text-blue-600 hover:bg-blue-50 rounded-lg transition-colors z-10 cursor-pointer"
             title="Add attachment"
           >
@@ -146,10 +192,6 @@ export function ChatPanel() {
             <ChevronRight size={18} />
           </button>
         </div>
-        <div className="flex mt-2 space-x-3 px-2">
-          <span className="text-[10px] text-slate-400 cursor-pointer hover:text-blue-500">Run diagnostics</span>
-          <span className="text-[10px] text-slate-400 cursor-pointer hover:text-blue-500">View recent errors</span>
-        </div>
       </div>
     </div>
   );

frontend/src/components/Chat/LeftPanel.tsx

@@ -154,14 +154,6 @@ export function LeftPanel({ activeTab, setActiveTab, selectedWorkflow, setSelect
           )}
           {activeTab === 'chats' && (
             <div className="space-y-2">
-              <div className="p-3 rounded-lg border border-slate-100 hover:border-blue-200 hover:bg-slate-50 cursor-pointer transition-all">
-                <div className="font-medium text-sm text-slate-700 mb-1">System Architecture</div>
-                <p className="text-xs text-slate-400 line-clamp-1">Can you explain the MoE model...</p>
-              </div>
-              <div className="p-3 rounded-lg border border-slate-100 hover:border-blue-200 hover:bg-slate-50 cursor-pointer transition-all">
-                <div className="font-medium text-sm text-slate-700 mb-1">Log Analysis Helper</div>
-                <p className="text-xs text-slate-400 line-clamp-1">Show me the errors from yesterday.</p>
-              </div>
             </div>
           )}
         </div>

frontend/src/components/Chat/RightPanel.tsx

@@ -16,56 +16,43 @@ export function RightPanel({ selectedWorkflow }: RightPanelProps) {
       return;
     }
 
-    let ws: WebSocket | null = null;
-    let reconnectTimeout: ReturnType<typeof setTimeout>;
-    let retryCount = 0;
-    const maxRetryCount = 10;
-    const baseDelay = 1000;
+    let eventSource: EventSource | null = null;
 
     const connect = () => {
-      const protocol = window.location.protocol === 'https:' ? 'wss:' : 'ws:';
+      const protocol = window.location.protocol;
       const host = window.location.host;
 
-      const wsBase = import.meta.env.VITE_API_BASE_URL
-        ? import.meta.env.VITE_API_BASE_URL.replace(/^http/, 'ws')
-        : `${protocol}//${host}`;
+      const apiBase = import.meta.env.VITE_API_BASE_URL || `${protocol}//${host}`;
 
-      // Using the workflow router WS endpoint
-      ws = new WebSocket(`${wsBase}/api/v1/workflow/ws/${selectedWorkflow}`);
+      // Using the workflow router SSE endpoint
+      eventSource = new EventSource(`${apiBase}/api/v1/workflow/sse/${selectedWorkflow}`);
 
-      ws.onopen = () => {
+      eventSource.onopen = () => {
         setIsConnected(true);
-        retryCount = 0; // reset on successful connection
         setMessages([]); // clear previous traces
       };
 
-      ws.onmessage = (event) => {
+      eventSource.onmessage = (event) => {
         try {
           setMessages(prev => [...prev, event.data]);
         } catch (e) {
-          console.error("Error receiving workflow websocket message", e);
+          console.error("Error receiving workflow SSE message", e);
         }
       };
 
-      ws.onclose = () => {
+      eventSource.onerror = (error) => {
+        console.error("EventSource failed.", error);
         setIsConnected(false);
-        if (retryCount < maxRetryCount) {
-          const delay = baseDelay * Math.pow(2, retryCount);
-          retryCount++;
-          console.log(`WebSocket closed. Reconnecting in ${delay}ms... (Attempt ${retryCount})`);
-          reconnectTimeout = setTimeout(connect, delay);
-        } else {
-          console.error("Max WebSocket reconnect attempts reached.");
-        }
+        // EventSource automatically attempts to reconnect, so we can just let it be,
+        // or we could close it if we wanted to handle retries manually.
       };
     };
 
     connect();
 
     return () => {
-      clearTimeout(reconnectTimeout);
-      if (ws) {
-        ws.close();
+      if (eventSource) {
+        eventSource.close();
       }
     };
   }, [selectedWorkflow]);

frontend/src/components/Layout/Sidebar.tsx

@@ -1,5 +1,5 @@
 
-import { Activity, MessageSquare, MonitorPlay, Settings, Bot, Box } from 'lucide-react';
+import { Activity, MessageSquare, Settings, Bot, Box } from 'lucide-react';
 
 interface SidebarProps {
   currentView: string;
@@ -24,13 +24,6 @@ export function Sidebar({ currentView, setCurrentView }: SidebarProps) {
         >
           <MessageSquare size={18} />
         </button>
-        <button
-          onClick={() => setCurrentView('monitoring')}
-          className={`p-1.5 rounded-lg transition-colors ${currentView === 'monitoring' ? 'text-blue-600 bg-blue-50' : 'text-slate-400 hover:text-blue-500 hover:bg-blue-50'}`}
-          title="Monitoring"
-        >
-          <MonitorPlay size={18} />
-        </button>
         <button
           onClick={() => setCurrentView('agent')}
           className={`p-1.5 rounded-lg transition-colors ${currentView === 'agent' ? 'text-blue-600 bg-blue-50' : 'text-slate-400 hover:text-blue-500 hover:bg-blue-50'}`}

frontend/src/components/Monitoring/MonitoringDashboard.tsx

@@ -1,154 +0,0 @@
-import { Server, Cpu, HardDrive, Box } from 'lucide-react';
-import type { ClusterNode } from '../../types';
-import { useClusterState } from '../../hooks/useClusterState';
-
-export function MonitoringDashboard() {
-  const { nodes, isConnected } = useClusterState();
-
-  const totalNodes = nodes.length;
-  let totalCpu = 0;
-  let totalMemory = 0;
-  let totalGpu = 0;
-
-  nodes.forEach((node: ClusterNode) => {
-    totalCpu += node.resources?.CPU || 0;
-    totalMemory += node.resources?.memory || 0;
-    totalGpu += node.resources?.GPU || 0;
-  });
-
-  return (
-    <div className="flex-1 flex flex-col bg-slate-50 overflow-hidden">
-      <div className="p-6 border-b border-slate-200 bg-white shadow-sm z-10 flex items-center justify-between">
-        <div>
-          <h2 className="text-xl font-semibold text-slate-800">Ray Cluster Monitoring</h2>
-          <p className="text-sm text-slate-500 mt-1">Real-time resource utilization across all nodes.</p>
-        </div>
-        <div className="flex items-center space-x-2">
-           <span className={`flex h-2.5 w-2.5 rounded-full ${isConnected ? 'bg-green-500 animate-pulse' : 'bg-red-500'}`}></span>
-           <span className="text-sm font-medium text-slate-600">{isConnected ? 'Connected' : 'Disconnected'}</span>
-        </div>
-      </div>
-
-      <div className="flex-1 overflow-y-auto p-6 space-y-6">
-        {/* Cluster Global Metrics */}
-        <div className="grid grid-cols-4 gap-4">
-          <div className="bg-white p-4 rounded-xl border border-slate-200 shadow-sm flex items-center">
-             <div className="w-12 h-12 rounded-lg bg-blue-50 flex items-center justify-center mr-4">
-               <Server size={24} className="text-blue-600" />
-             </div>
-             <div>
-               <p className="text-xs text-slate-500 font-medium">TOTAL NODES</p>
-               <p className="text-2xl font-bold text-slate-800">{totalNodes}</p>
-             </div>
-          </div>
-          <div className="bg-white p-4 rounded-xl border border-slate-200 shadow-sm flex items-center">
-             <div className="w-12 h-12 rounded-lg bg-indigo-50 flex items-center justify-center mr-4">
-               <Cpu size={24} className="text-indigo-600" />
-             </div>
-             <div>
-               <p className="text-xs text-slate-500 font-medium">TOTAL CPU CORES</p>
-               <p className="text-2xl font-bold text-slate-800">{totalCpu}</p>
-             </div>
-          </div>
-          <div className="bg-white p-4 rounded-xl border border-slate-200 shadow-sm flex items-center">
-             <div className="w-12 h-12 rounded-lg bg-green-50 flex items-center justify-center mr-4">
-               <HardDrive size={24} className="text-green-600" />
-             </div>
-             <div>
-               <p className="text-xs text-slate-500 font-medium">TOTAL RAM</p>
-               <p className="text-2xl font-bold text-slate-800">
-                  {totalMemory > 0 ? `${(totalMemory / (1024 * 1024 * 1024)).toFixed(2)} GB` : '0 GB'}
-               </p>
-             </div>
-          </div>
-          <div className="bg-white p-4 rounded-xl border border-slate-200 shadow-sm flex items-center">
-             <div className="w-12 h-12 rounded-lg bg-purple-50 flex items-center justify-center mr-4">
-               <Box size={24} className="text-purple-600" />
-             </div>
-             <div>
-               <p className="text-xs text-slate-500 font-medium">TOTAL GPUS</p>
-               <p className="text-2xl font-bold text-slate-800">{totalGpu}</p>
-             </div>
-          </div>
-        </div>
-
-        {/* Node List */}
-        <div className="bg-white border border-slate-200 rounded-xl shadow-sm overflow-hidden">
-           <table className="w-full text-left text-sm">
-            <thead className="bg-slate-50 border-b border-slate-200 text-slate-500">
-              <tr>
-                <th className="px-6 py-4 font-medium">Node ID / Name</th>
-                <th className="px-6 py-4 font-medium">Status</th>
-                <th className="px-6 py-4 font-medium">CPU (Used / Total)</th>
-                <th className="px-6 py-4 font-medium">RAM (Used / Total)</th>
-                <th className="px-6 py-4 font-medium">GPU (Used / Total)</th>
-              </tr>
-            </thead>
-            <tbody className="divide-y divide-slate-100">
-              {nodes.map((node: ClusterNode, i: number) => {
-                const totalCpu = node.resources?.CPU || 0;
-                const remainingCpu = node.remaining?.CPU || 0;
-                const usedCpu = totalCpu - remainingCpu;
-                const cpuPercent = totalCpu > 0 ? (usedCpu / totalCpu) * 100 : 0;
-
-                const totalRam = node.resources?.memory || 0;
-                const remainingRam = node.remaining?.memory || 0;
-                const usedRam = totalRam - remainingRam;
-                const ramPercent = totalRam > 0 ? (usedRam / totalRam) * 100 : 0;
-
-                const totalGpu = node.resources?.GPU || 0;
-                const remainingGpu = node.remaining?.GPU || 0;
-                const usedGpu = totalGpu - remainingGpu;
-                const gpuPercent = totalGpu > 0 ? (usedGpu / totalGpu) * 100 : 0;
-
-                return (
-                  <tr key={i} className="hover:bg-slate-50 transition-colors">
-                    <td className="px-6 py-4 font-medium text-slate-800 flex flex-col">
-                      <span>{node.node_name || 'Unknown'}</span>
-                      <span className="text-xs text-slate-400 font-mono">{node.node_id}</span>
-                    </td>
-                    <td className="px-6 py-4">
-                      <span className={`flex items-center text-xs font-medium ${node.alive ? 'text-green-600' : 'text-red-600'}`}>
-                        <span className={`w-2 h-2 rounded-full mr-2 ${node.alive ? 'bg-green-500' : 'bg-red-500'}`}></span>
-                        {node.alive ? 'Alive' : 'Dead'}
-                      </span>
-                    </td>
-                    <td className="px-6 py-4">
-                      <div className="flex items-center">
-                        <span className="w-16 text-right mr-2 text-xs">{usedCpu.toFixed(1)} / {totalCpu}</span>
-                        <div className="w-16 bg-slate-100 rounded-full h-1.5"><div className={`h-1.5 rounded-full ${cpuPercent > 80 ? 'bg-red-500' : 'bg-indigo-500'}`} style={{ width: `${cpuPercent}%` }}></div></div>
-                      </div>
-                    </td>
-                    <td className="px-6 py-4 text-slate-600 text-xs">
-                       <div className="flex items-center">
-                        <span className="w-24 text-right mr-2 text-xs">{(usedRam / (1024**3)).toFixed(1)}G / {(totalRam / (1024**3)).toFixed(1)}G</span>
-                        <div className="w-16 bg-slate-100 rounded-full h-1.5"><div className={`h-1.5 rounded-full ${ramPercent > 80 ? 'bg-red-500' : 'bg-green-500'}`} style={{ width: `${ramPercent}%` }}></div></div>
-                      </div>
-                    </td>
-                    <td className="px-6 py-4">
-                      {totalGpu > 0 ? (
-                        <div className="flex items-center">
-                          <span className="w-16 text-right mr-2 text-xs">{usedGpu} / {totalGpu}</span>
-                          <div className="w-16 bg-slate-100 rounded-full h-1.5"><div className={`h-1.5 rounded-full ${gpuPercent > 80 ? 'bg-red-500' : 'bg-purple-500'}`} style={{ width: `${gpuPercent}%` }}></div></div>
-                        </div>
-                      ) : (
-                        <span className="text-slate-400 italic text-xs">No GPU</span>
-                      )}
-                    </td>
-                  </tr>
-                );
-              })}
-              {nodes.length === 0 && (
-                 <tr>
-                    <td colSpan={5} className="px-6 py-8 text-center text-slate-500 text-sm">
-                       No node data available. {isConnected ? 'Waiting for cluster state...' : 'Check connection.'}
-                    </td>
-                 </tr>
-              )}
-            </tbody>
-          </table>
-        </div>
-      </div>
-    </div>
-  );
-}

frontend/src/components/Monitoring/MonitoringLayout.tsx

@@ -1,33 +0,0 @@
-import { useState } from 'react';
-import { Server } from 'lucide-react';
-import { MonitoringDashboard } from './MonitoringDashboard';
-
-export function MonitoringLayout() {
-  const [activeTab, setActiveTab] = useState('cluster');
-
-  return (
-    <div className="flex-1 flex bg-slate-50 overflow-hidden">
-       {/* Monitoring Inner Sidebar */}
-       <div className="w-64 bg-white border-r border-slate-200 flex flex-col z-0">
-          <div className="p-6 border-b border-slate-100">
-             <h2 className="text-lg font-semibold text-slate-800">Monitoring</h2>
-          </div>
-          <div className="flex-1 p-4 space-y-2 overflow-y-auto">
-             <button
-               onClick={() => setActiveTab('cluster')}
-               className={`w-full flex items-center px-4 py-3 text-sm font-medium rounded-xl transition-all ${activeTab === 'cluster' ? 'bg-blue-50 text-blue-600' : 'text-slate-600 hover:bg-slate-50 hover:text-slate-900'}`}
-             >
-               <Server size={18} className="mr-3" />
-               Cluster Monitor
-             </button>
-             {/* Future monitoring tabs (e.g., Application Logs, Agent Metrics) can go here */}
-          </div>
-       </div>
-
-       {/* Monitoring Main Content */}
-       <div className="flex-1 overflow-y-auto">
-          {activeTab === 'cluster' && <MonitoringDashboard />}
-       </div>
-    </div>
-  );
-}

main.py

@@ -25,6 +25,7 @@ async def start_system():
     }
 
     ray.init(ignore_reinit_error=True,
+             namespace="pretor",
              dashboard_host="0.0.0.0",
              dashboard_port=8265,
              runtime_env={"env_vars": env_vars})
@@ -34,8 +35,21 @@ async def start_system():
     postgres_database = PostgresDatabase.options(name='postgres_database').remote()
     await postgres_database.init_db.remote()
 
-    # 3. 启动全局状态机
-    global_state_machine = GlobalStateMachine.options(name='global_state_machine').remote(postgres_database)
+    global_state_machine = GlobalStateMachine.options(
+        name='global_state_machine',
+        namespace='pretor',
+        lifetime='detached'
+    ).remote(postgres_database)
+
+    print("正在等待 GlobalStateMachine 初始化并加载注册表...")
+    try:
+        # 强制执行初始化方法并阻塞等待结果。
+        # 如果 __init__ 或 init_state_machine 中有任何报错，会立刻在这里抛出！
+        await global_state_machine.init_state_machine.remote()
+        print("GlobalStateMachine 初始化成功！")
+    except Exception as e:
+        print(f"\n[致命错误] GlobalStateMachine 启动失败！真实报错如下：\n{e}\n")
+        return
 
     # 4. 启动核心节点
     supervisory_node = SupervisoryNode.options(name='supervisory_node').remote()

pretor/api/cluster.py

@@ -12,34 +12,8 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 
-from fastapi import APIRouter, WebSocket, WebSocketDisconnect
-import ray
-import asyncio
+from fastapi import APIRouter
 
 cluster_router = APIRouter(prefix="/api/v1/cluster", tags=["cluster"])
 
-@cluster_router.websocket("/ws/state")
-async def update_cluster_state(websocket: WebSocket):
-    await websocket.accept()
-    try:
-        while True:
-            nodes = ray.nodes()
-            payload = [
-                {
-                    "node_id": node.get("NodeID"),
-                    "node_name": node.get("NodeName"),
-                    "alive": node.get("Alive"),
-                    "resources": node.get("Resources", {}),
-                    "remaining": node.get("RemainingResources", {})
-                }
-                for node in nodes
-            ]
-            await websocket.send_json(payload)
-            await asyncio.sleep(10)
-    except WebSocketDisconnect:
-        pass
-    except RuntimeError as e:
-        if "closed" not in str(e) and "GeneratorExit" not in str(e):
-            raise
-    except Exception:
-        pass
+# Monitor websocket API temporarily removed

pretor/api/platform/frontend.py

@@ -12,11 +12,13 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 
-from fastapi import APIRouter, Depends, HTTPException, status
+from fastapi import APIRouter, Depends, HTTPException, status, UploadFile, File
 from pydantic import BaseModel
 from pretor.utils.access import Accessor, TokenData
 from pretor.api.platform.event import PretorEvent
 from pretor.utils.ray_hook import ray_actor_hook
+import os
+import shutil
 
 from pretor.utils.logger import get_logger
 logger = get_logger('frontend')
@@ -45,3 +47,17 @@ async def create_message(message: Message,
     else:
         return {"message": message}
 
+@client_router.post("/upload")
+async def upload_file(file: UploadFile = File(...),
+                      token_data: TokenData = Depends(Accessor.get_current_user)):
+    try:
+        upload_dir = "uploads"
+        os.makedirs(upload_dir, exist_ok=True)
+        file_path = os.path.join(upload_dir, file.filename)
+        with open(file_path, "wb") as buffer:
+            shutil.copyfileobj(file.file, buffer)
+        logger.info(f"用户 {token_data.username} 上传了文件: {file.filename}")
+        return {"filename": file.filename, "message": f"File {file.filename} uploaded successfully"}
+    except Exception as e:
+        logger.error(f"文件上传失败: {e}")
+        raise HTTPException(status_code=500, detail="文件上传失败")

pretor/api/workflow.py

@@ -14,26 +14,37 @@
 
 
 from pretor.utils.ray_hook import ray_actor_hook
-from fastapi import APIRouter, WebSocket, WebSocketDisconnect
-
+from fastapi import APIRouter, Request
+from fastapi.responses import StreamingResponse
+import asyncio
 
 workflow_router = APIRouter(prefix="/api/v1/workflow", tags=["workflow"])
 
-@workflow_router.websocket("/ws/{trace_id}")
-async def get_workflow(websocket: WebSocket, trace_id: str):
-    await websocket.accept()
-    global_state_machine = ray_actor_hook("global_state_machine")
-    try:
-        while True:
-            await websocket.send(await global_state_machine.get_workflow.remote(trace_id))
-            await websocket.send_text(await global_state_machine.get_pending.remote(trace_id))
-            response = await websocket.receive_text()
-            await global_state_machine.put_received(trace_id, response)
-    except WebSocketDisconnect:
-        pass
-    except RuntimeError as e:
-        if "closed" not in str(e) and "GeneratorExit" not in str(e):
-            raise
-    except Exception:
-        pass
+@workflow_router.get("/sse/{trace_id}")
+async def get_workflow_sse(trace_id: str, request: Request):
+    global_state_machine = ray_actor_hook("global_state_machine").global_state_machine
+
+    async def event_generator():
+        try:
+            while True:
+                if await request.is_disconnected():
+                    break
+
+                # You might also want to send the workflow state periodically or when updated
+                # Here we just wait for pending messages and send them
+                message = await global_state_machine.get_pending.remote(trace_id)
+                # Ensure the message is formatted as SSE
+                yield f"data: {message}\n\n"
+        except asyncio.CancelledError:
+            pass
+
+    return StreamingResponse(event_generator(), media_type="text/event-stream")
+
+@workflow_router.post("/reply/{trace_id}")
+async def post_workflow_reply(trace_id: str, request: Request):
+    data = await request.json()
+    reply_msg = data.get("message", "")
+    global_state_machine = ray_actor_hook("global_state_machine").global_state_machine
+    await global_state_machine.put_received.remote(trace_id, reply_msg)
+    return {"status": "ok"}
 

pretor/core/global_state_machine/global_state_machine.py

@@ -28,16 +28,22 @@ from pretor.core.global_state_machine.individual_manager import GlobalIndividual
 @ray.remote
 class GlobalStateMachine:
     def __init__(self, postgres_database: PostgresDatabase):
-
+        import sys
+        print("GSM __init__ START", file=sys.stderr, flush=True)
         self.event_dict: Dict[str, PretorEvent] = {}
-
+        print("  event_dict done", file=sys.stderr, flush=True)
         self._global_provider_manager = ProviderManager(postgres_database)
+        print("  provider_manager done", file=sys.stderr, flush=True)
         self._global_tool_manager = GlobalToolManager()
+        print("  tool_manager done", file=sys.stderr, flush=True)
         self._global_workflow_template_manager = WorkflowManager()
+        print("  workflow_template_manager done", file=sys.stderr, flush=True)
         self._global_skill_manager = GlobalSkillManager()
+        print("  skill_manager done", file=sys.stderr, flush=True)
         self._global_individual_manager = GlobalIndividualManager()
-
+        print("  individual_manager done", file=sys.stderr, flush=True)
         self.postgres_database = postgres_database
+        print("GSM __init__ DONE", file=sys.stderr, flush=True)
 
     async def init_state_machine(self):
         await self._global_provider_manager.init_provider_register(self.postgres_database)

pretor/core/workflow/workflow_runner.py

@@ -268,9 +268,11 @@ class WorkflowRunningEngine:
         self.consciousness_node = consciousness_node
         self.control_node = control_node
         self.supervisory_node = supervisory_node
-        self.global_state_machine = ray_actor_hook("global_state_machine").global_state_machine
+        self.global_state_machine = None
 
     async def run(self):
+        # Move actor hook to async start so we don't race during __init__ across cluster
+        self.global_state_machine = ray_actor_hook("global_state_machine").global_state_machine
         self.workflow_queue = asyncio.Queue()
         self.runner_engine = {
             f"runner_{i}": asyncio.create_task(self.runner(i))

pretor/utils/ray_hook.py

@@ -35,7 +35,7 @@ class ActorList:
 @lru_cache(maxsize=128)
 def _get_cached_actor_handle(actor_name: str):
     """缓存接口"""
-    return ray.get_actor(actor_name)
+    return ray.get_actor(actor_name, namespace="pretor")
 
 def clear_actor_cache():
     """清理接口"""

tests/core/database/table/table_provider_test.py

@@ -1,4 +1,3 @@
-import pytest
 from pretor.core.database.table.provider import Provider
 
 def test_provider_table():

tests/core/database/table/table_user_test.py

@@ -1,4 +1,3 @@
-import pytest
 from pretor.core.database.table.user import User
 
 def test_user_table():

tests/core/global_state_machine/model_provider/base_provider_test.py

@@ -1,4 +1,3 @@
-import pytest
 from pretor.core.global_state_machine.model_provider.base_provider import Provider, ProviderArgs, ProviderStatus
 
 def test_provider_status():

tests/core/global_state_machine/provider_manager_test.py

@@ -5,7 +5,6 @@ from pretor.core.global_state_machine.provider_manager import ProviderManager
 
 @pytest.mark.asyncio
 async def test_provider_manager_init():
-    from pretor.core.global_state_machine.provider_manager import ProviderManager
     mock_postgres = MagicMock()
     mock_provider1 = MagicMock()
     mock_provider1.provider_title = "title1"

tests/core/workflow/workflow_template_generator/workflow_template_generator_test.py

@@ -1,4 +1,3 @@
-import pytest
 from unittest.mock import patch, MagicMock
 from pretor.core.workflow.workflow_template_generator.workflow_template_generator import WorkflowTemplateGenerator
 

tests/core/workflow/workflow_template_manager_test.py

@@ -1,4 +1,3 @@
-import pytest
 import json
 from unittest.mock import MagicMock, patch, mock_open
 from pathlib import Path

tests/utils/access_test.py

@@ -83,7 +83,6 @@ def test_decode_token_validation_error():
     token = "valid.jwt.invalid.payload"
     payload = {"wrong": "payload"}
 
-    import pydantic
     from fastapi import HTTPException
 
     with patch("jwt.decode", return_value=payload):