fix: 修复智能体调用卡死问题

- 添加--json参数到openclaw agent命令
- 正确解析JSON输出结构: result.payloads[0].text
- 使用Popen替代subprocess.run，支持进程组杀死
- 超时时间从5分钟改为3分钟
- 添加os.setsid创建新进程组，确保超时时能杀死所有子进程
- 增强异常处理和日志记录

services/process_monitor.py

@@ -341,6 +341,9 @@ class ProcessMonitor:
         except Exception as e:
             logger.error(f"处理会话异常: {session_id} - {e}")
             db.update_session_status(session_id, 'failed')
+            # 确保清理
+            if session_id in self.active_sessions:
+                del self.active_sessions[session_id]
             return {'success': False, 'message': str(e)}
     
     def _start_step(self, session_id, product_name, step_num, step_name):
@@ -466,36 +469,65 @@ class ProcessMonitor:
     
     def _call_agent(self, task_text):
         """调用智能体执行任务"""
+        import signal
+        
         try:
             cmd = [
                 'openclaw', 'agent',
                 '--agent', 'hz4th_editor',
-                '--message', task_text
+                '--message', task_text,
+                '--json'  # 输出JSON格式以便解析
             ]
             
-            logger.info(f"调用智能体命令: openclaw agent --agent hz4th_editor --message '[任务文本 {len(task_text)} 字符]'")
+            logger.info(f"调用智能体命令: openclaw agent --agent hz4th_editor --message '[任务文本 {len(task_text)} 字符]' --json")
             
-            result = subprocess.run(
+            # 使用Popen以便更好地控制超时和进程杀死
+            proc = subprocess.Popen(
                 cmd,
-                capture_output=True,
-                text=True,
-                timeout=300  # 5分钟超时
+                stdout=subprocess.PIPE,
+                stderr=subprocess.PIPE,
+                preexec_fn=os.setsid  # 创建新进程组，方便杀死所有子进程
             )
             
-            if result.returncode == 0:
-                output = result.stdout.strip()
-                logger.info(f"智能体返回: {output[:500]}...")
-                return {'success': True, 'output': output}
-            else:
-                error = result.stderr.strip() or result.stdout.strip()
-                logger.error(f"智能体调用失败: {error}")
-                return {'success': False, 'error': error}
+            try:
+                stdout, stderr = proc.communicate(timeout=180)  # 3分钟超时
+                raw_output = stdout.decode('utf-8', errors='replace').strip()
+                
+                if proc.returncode == 0:
+                    # 解析JSON输出
+                    try:
+                        data = json.loads(raw_output)
+                        # 提取实际回复文本: result.payloads[0].text
+                        payloads = data.get('result', {}).get('payloads', [])
+                        if payloads and isinstance(payloads[0], dict):
+                            output = payloads[0].get('text', '')
+                        else:
+                            output = raw_output
+                        
+                        logger.info(f"智能体返回: {output[:500]}...")
+                        return {'success': True, 'output': output}
+                    except json.JSONDecodeError as e:
+                        logger.warning(f"JSON解析失败，使用原始输出: {e}")
+                        return {'success': True, 'output': raw_output}
+                else:
+                    error = stderr.decode('utf-8', errors='replace').strip() or raw_output
+                    logger.error(f"智能体调用失败(returncode={proc.returncode}): {error}")
+                    return {'success': False, 'error': error}
+            
+            except subprocess.TimeoutExpired:
+                # 超时，杀死整个进程组
+                logger.error(f"智能体执行超时(>3分钟)，杀死进程组")
+                try:
+                    os.killpg(os.getpgid(proc.pid), signal.SIGKILL)
+                except Exception:
+                    proc.kill()
+                proc.wait()
+                return {'success': False, 'error': '智能体执行超时(>3分钟)'}
         
-        except subprocess.TimeoutExpired:
-            return {'success': False, 'error': '智能体执行超时(>5分钟)'}
         except FileNotFoundError:
             return {'success': False, 'error': 'openclaw命令未找到'}
         except Exception as e:
+            logger.error(f"智能体调用异常: {e}")
             return {'success': False, 'error': str(e)}
     
     def _parse_agent_response(self, output):