diff --git a/main/xiaozhi-server/core/connection.py b/main/xiaozhi-server/core/connection.py index 20b55a91..850f6f7a 100644 --- a/main/xiaozhi-server/core/connection.py +++ b/main/xiaozhi-server/core/connection.py @@ -116,6 +116,8 @@ class ConnectionHandler: self.memory = _memory self.intent = _intent + self.is_exiting = False # 标记是否正在执行退出流程 + # 为每个连接单独管理声纹识别 self.voiceprint_provider = None @@ -289,6 +291,10 @@ class ConnectionHandler: async def _route_message(self, message): """消息路由""" + # 退出状态丢弃所有消息 + if self.is_exiting: + return + # 检查是否已经获取到真实的绑定状态 if not self.bind_completed_event.is_set(): # 还没有获取到真实状态，等待直到获取到真实状态或超时 @@ -991,11 +997,23 @@ class ConnectionHandler: ) futures_with_data.append((future, tool_call_data)) + TOOL_CALL_TIMEOUT = 30 # 等待协程结束（实际等待时长为最慢的那个） tool_results = [] for future, tool_call_data in futures_with_data: - result = future.result() - tool_results.append((result, tool_call_data)) + try: + result = future.result(timeout=TOOL_CALL_TIMEOUT) + tool_results.append((result, tool_call_data)) + except Exception as e: + self.logger.bind(tag=TAG).error( + f"工具调用超时或异常: {tool_call_data['name']}, 错误: {e}" + ) + # 超时时返回错误响应，避免整个流程卡死 + from plugins_func.register import Action, ActionResponse + tool_results.append(( + ActionResponse(action=Action.ERROR, result="哎呀，网络遇到点问题，请稍后再试下！"), + tool_call_data + )) # 统一处理所有工具调用结果 if tool_results: diff --git a/main/xiaozhi-server/core/handle/abortHandle.py b/main/xiaozhi-server/core/handle/abortHandle.py index b7bf1611..b74b2893 100644 --- a/main/xiaozhi-server/core/handle/abortHandle.py +++ b/main/xiaozhi-server/core/handle/abortHandle.py @@ -7,6 +7,10 @@ TAG = __name__ async def handleAbortMessage(conn: "ConnectionHandler"): + if conn.close_after_chat or conn.is_exiting: + conn.logger.bind(tag=TAG).info("退出流程中被打断，直接关闭连接") + return + conn.logger.bind(tag=TAG).info("Abort message received") # 设置成打断状态，会自动打断llm、tts任务 conn.client_abort = True diff --git a/main/xiaozhi-server/core/handle/intentHandler.py b/main/xiaozhi-server/core/handle/intentHandler.py index a9e3ee15..35ade945 100644 --- a/main/xiaozhi-server/core/handle/intentHandler.py +++ b/main/xiaozhi-server/core/handle/intentHandler.py @@ -32,6 +32,10 @@ async def handle_user_intent(conn: "ConnectionHandler", text): if await check_direct_exit(conn, filtered_text): return True + # 明确再见不被打断 + if conn.is_exiting: + return True + # 检查是否是唤醒词 if await checkWakeupWords(conn, filtered_text): return True @@ -57,6 +61,7 @@ async def check_direct_exit(conn: "ConnectionHandler", text): if text == cmd: conn.logger.bind(tag=TAG).info(f"识别到明确的退出命令: {text}") await send_stt_message(conn, text) + conn.is_exiting = True await conn.close() return True return False @@ -144,7 +149,9 @@ async def process_intent_result( # 使用executor执行函数调用和结果处理 def process_function_call(): conn.dialogue.put(Message(role="user", content=original_text)) - + + # 使用统一工具处理器处理所有工具调用，添加超时保护 + TOOL_CALL_TIMEOUT = 30 # 使用统一工具处理器处理所有工具调用 try: result = asyncio.run_coroutine_threadsafe( @@ -152,11 +159,11 @@ async def process_intent_result( conn, function_call_data ), conn.loop, - ).result() + ).result(timeout=TOOL_CALL_TIMEOUT) except Exception as e: conn.logger.bind(tag=TAG).error(f"工具调用失败: {e}") result = ActionResponse( - action=Action.ERROR, result=str(e), response=str(e) + action=Action.ERROR, result="工具调用超时，请一会再试下哈", response="工具调用超时，请一会再试下哈" ) if result: diff --git a/main/xiaozhi-server/core/handle/receiveAudioHandle.py b/main/xiaozhi-server/core/handle/receiveAudioHandle.py index 4f69aa99..280d72cf 100644 --- a/main/xiaozhi-server/core/handle/receiveAudioHandle.py +++ b/main/xiaozhi-server/core/handle/receiveAudioHandle.py @@ -15,6 +15,8 @@ TAG = __name__ async def handleAudioMessage(conn: "ConnectionHandler", audio): + if conn.is_exiting: + return # 当前片段是否有人说话 have_voice = conn.vad.is_vad(conn, audio) # 如果设备刚刚被唤醒，短暂忽略VAD检测 diff --git a/main/xiaozhi-server/core/providers/llm/openai/openai.py b/main/xiaozhi-server/core/providers/llm/openai/openai.py index 58ec725c..e6be8ddb 100644 --- a/main/xiaozhi-server/core/providers/llm/openai/openai.py +++ b/main/xiaozhi-server/core/providers/llm/openai/openai.py @@ -17,8 +17,12 @@ class LLMProvider(LLMProviderBase): self.base_url = config.get("base_url") else: self.base_url = config.get("url") - timeout = config.get("timeout", 300) - self.timeout = int(timeout) if timeout else 300 + custom_timeout = httpx.Timeout( + pool=2.0, + connect=3.0, + write=5.0, + read=10.0 + ) param_defaults = { "max_tokens": int, @@ -45,7 +49,7 @@ class LLMProvider(LLMProviderBase): model_key_msg = check_model_key("LLM", self.api_key) if model_key_msg: logger.bind(tag=TAG).error(model_key_msg) - self.client = openai.OpenAI(api_key=self.api_key, base_url=self.base_url, timeout=httpx.Timeout(self.timeout)) + self.client = openai.OpenAI(api_key=self.api_key, base_url=self.base_url, timeout=custom_timeout) @staticmethod def normalize_dialogue(dialogue): diff --git a/main/xiaozhi-server/core/providers/tts/base.py b/main/xiaozhi-server/core/providers/tts/base.py index 4fef87dd..bc99dcec 100644 --- a/main/xiaozhi-server/core/providers/tts/base.py +++ b/main/xiaozhi-server/core/providers/tts/base.py @@ -357,7 +357,7 @@ class TTSProviderBase(ABC): sendAudioMessage(self.conn, sentence_type, audio_datas, text), self.conn.loop, ) - future.result() + future.result(timeout=15) # 记录输出和报告 if self.conn.max_output_size > 0 and text: diff --git a/main/xiaozhi-server/core/providers/tts/huoshan_double_stream.py b/main/xiaozhi-server/core/providers/tts/huoshan_double_stream.py index 152aa33f..cdca9ed8 100644 --- a/main/xiaozhi-server/core/providers/tts/huoshan_double_stream.py +++ b/main/xiaozhi-server/core/providers/tts/huoshan_double_stream.py @@ -149,6 +149,7 @@ class TTSProvider(TTSProviderBase): self.cluster = config.get("cluster") self.resource_id = config.get("resource_id") self.activate_session = False + self.timeout = 30 if config.get("private_voice"): self.voice = config.get("private_voice") else: @@ -307,7 +308,7 @@ class TTSProvider(TTSProviderBase): self.start_session(self.conn.sentence_id), loop=self.conn.loop, ) - future.result() + future.result(self.timeout) self.before_stop_play_files.clear() logger.bind(tag=TAG).debug("TTS会话启动成功") except Exception as e: @@ -324,7 +325,7 @@ class TTSProvider(TTSProviderBase): self.text_to_speak(message.content_detail, None), loop=self.conn.loop, ) - future.result() + future.result(self.timeout) logger.bind(tag=TAG).debug("TTS文本发送成功") except Exception as e: logger.bind(tag=TAG).error(f"发送TTS文本失败: {str(e)}") @@ -344,7 +345,7 @@ class TTSProvider(TTSProviderBase): self.finish_session(self.conn.sentence_id), loop=self.conn.loop, ) - future.result() + future.result(self.timeout) except Exception as e: logger.bind(tag=TAG).error(f"结束TTS会话失败: {str(e)}") continue diff --git a/main/xiaozhi-server/plugins_func/functions/handle_exit_intent.py b/main/xiaozhi-server/plugins_func/functions/handle_exit_intent.py index bd24f372..154c4f89 100644 --- a/main/xiaozhi-server/plugins_func/functions/handle_exit_intent.py +++ b/main/xiaozhi-server/plugins_func/functions/handle_exit_intent.py @@ -31,6 +31,7 @@ handle_exit_intent_function_desc = { "handle_exit_intent", handle_exit_intent_function_desc, ToolType.SYSTEM_CTL ) def handle_exit_intent(conn: "ConnectionHandler", say_goodbye: str | None = None): + conn.is_exiting = True # 处理退出意图 try: if say_goodbye is None: