dashscope 1.23.3__py3-none-any.whl → 1.23.5__py3-none-any.whl

dashscope/aigc/image_synthesis.py

@@ -171,7 +171,7 @@ class ImageSynthesis(BaseAsyncApi):
                 model, mask_image_url, api_key)
             if is_upload:
                 has_upload = True
-            input['mask_image_url'] = mask_image_url
+            input['mask_image_url'] = res_mask_image_url
 
         if base_image_url is not None and base_image_url:
             is_upload, res_base_image_url = check_and_upload_local(

dashscope/audio/asr/vocabulary.py

@@ -12,14 +12,15 @@ from dashscope.common.logging import logger
 
 
 class VocabularyServiceException(Exception):
-    def __init__(self, status_code: int, code: str,
+    def __init__(self, request_id: str, status_code: int, code: str,
                  error_message: str) -> None:
+        self._request_id = request_id
         self._status_code = status_code
         self._code = code
         self._error_message = error_message
 
     def __str__(self):
-        return f'Status Code: {self._status_code}, Code: {self._code}, Error Message: {self._error_message}'
+        return f'Request: {self._request_id}, Status Code: {self._status_code}, Code: {self._code}, Error Message: {self._error_message}'
 
 
 class VocabularyService(BaseApi):
@@ -86,7 +87,7 @@ class VocabularyService(BaseApi):
             self._last_request_id = response.request_id
             return response.output['vocabulary_id']
         else:
-            raise VocabularyServiceException(response.status_code,
+            raise VocabularyServiceException(response.request_id, response.status_code,
                                              response.code, response.message)
 
     def list_vocabularies(self,
@@ -117,7 +118,7 @@ class VocabularyService(BaseApi):
             self._last_request_id = response.request_id
             return response.output['vocabulary_list']
         else:
-            raise VocabularyServiceException(response.status_code,
+            raise VocabularyServiceException(response.request_id, response.status_code,
                                              response.code, response.message)
 
     def query_vocabulary(self, vocabulary_id: str) -> List[dict]:
@@ -134,7 +135,7 @@ class VocabularyService(BaseApi):
             self._last_request_id = response.request_id
             return response.output
         else:
-            raise VocabularyServiceException(response.status_code,
+            raise VocabularyServiceException(response.request_id, response.status_code,
                                              response.code, response.message)
 
     def update_vocabulary(self, vocabulary_id: str,
@@ -153,7 +154,7 @@ class VocabularyService(BaseApi):
             self._last_request_id = response.request_id
             return
         else:
-            raise VocabularyServiceException(response.status_code,
+            raise VocabularyServiceException(response.request_id, response.status_code,
                                              response.code, response.message)
 
     def delete_vocabulary(self, vocabulary_id: str) -> None:
@@ -169,7 +170,7 @@ class VocabularyService(BaseApi):
             self._last_request_id = response.request_id
             return
         else:
-            raise VocabularyServiceException(response.status_code,
+            raise VocabularyServiceException(response.request_id, response.status_code,
                                              response.code, response.message)
 
     def get_last_request_id(self):

dashscope/audio/tts_v2/enrollment.py

@@ -12,14 +12,15 @@ from dashscope.common.logging import logger
 
 
 class VoiceEnrollmentException(Exception):
-    def __init__(self, status_code: int, code: str,
+    def __init__(self, request_id: str, status_code: int, code: str,
                  error_message: str) -> None:
+        self._request_id = request_id
         self._status_code = status_code
         self._code = code
         self._error_message = error_message
 
     def __str__(self):
-        return f'Status Code: {self._status_code}, Code: {self._code}, Error Message: {self._error_message}'
+        return f'Request: {self._request_id}, Status Code: {self._status_code}, Code: {self._code}, Error Message: {self._error_message}'
 
 
 class VoiceEnrollmentService(BaseApi):
@@ -81,11 +82,11 @@ class VoiceEnrollmentService(BaseApi):
             'prefix': prefix,
             'url': url,
         }, )
+        self._last_request_id = response.request_id
         if response.status_code == 200:
-            self._last_request_id = response.request_id
             return response.output['voice_id']
         else:
-            raise VoiceEnrollmentException(response.status_code, response.code,
+            raise VoiceEnrollmentException(response.request_id, response.status_code, response.code,
                                            response.message)
 
     def list_voices(self,
@@ -111,11 +112,11 @@ class VoiceEnrollmentService(BaseApi):
                 'page_index': page_index,
                 'page_size': page_size,
             }, )
+        self._last_request_id = response.request_id
         if response.status_code == 200:
-            self._last_request_id = response.request_id
             return response.output['voice_list']
         else:
-            raise VoiceEnrollmentException(response.status_code, response.code,
+            raise VoiceEnrollmentException(response.request_id, response.status_code, response.code,
                                            response.message)
 
     def query_voice(self, voice_id: str) -> List[str]:
@@ -128,11 +129,11 @@ class VoiceEnrollmentService(BaseApi):
             'action': 'query_voice',
             'voice_id': voice_id,
         }, )
+        self._last_request_id = response.request_id
         if response.status_code == 200:
-            self._last_request_id = response.request_id
             return response.output
         else:
-            raise VoiceEnrollmentException(response.status_code, response.code,
+            raise VoiceEnrollmentException(response.request_id, response.status_code, response.code,
                                            response.message)
 
     def update_voice(self, voice_id: str, url: str) -> None:
@@ -146,11 +147,11 @@ class VoiceEnrollmentService(BaseApi):
             'voice_id': voice_id,
             'url': url,
         }, )
+        self._last_request_id = response.request_id
         if response.status_code == 200:
-            self._last_request_id = response.request_id
             return
         else:
-            raise VoiceEnrollmentException(response.status_code, response.code,
+            raise VoiceEnrollmentException(response.request_id, response.status_code, response.code,
                                            response.message)
 
     def delete_voice(self, voice_id: str) -> None:
@@ -162,11 +163,11 @@ class VoiceEnrollmentService(BaseApi):
             'action': 'delete_voice',
             'voice_id': voice_id,
         }, )
+        self._last_request_id = response.request_id
         if response.status_code == 200:
-            self._last_request_id = response.request_id
             return
         else:
-            raise VoiceEnrollmentException(response.status_code, response.code,
+            raise VoiceEnrollmentException(response.request_id, response.status_code, response.code,
                                            response.message)
 
     def get_last_request_id(self):

dashscope/audio/tts_v2/speech_synthesizer.py

@@ -134,9 +134,7 @@ class Request:
                 'task_group': 'audio',
                 'task': 'tts',
                 'function': 'SpeechSynthesizer',
-                'input': {
-                    'text': ''
-                },
+                'input': {},
                 'parameters': {
                     'voice': self.voice,
                     'volume': self.volume,
@@ -179,9 +177,7 @@ class Request:
                 'streaming': WebsocketStreamingMode.DUPLEX,
             },
             'payload': {
-                'input': {
-                    'text': ''
-                },
+                'input': {},
             },
         }
         return json.dumps(cmd)
@@ -514,6 +510,10 @@ class SpeechSynthesizer:
             otherwise, it will wait indefinitely.
         """
         # print('还不支持非流式语音合成sdk调用大模型，使用流式模拟')
+        if self.additional_params is None:
+            self.additional_params = {"enable_ssml":True}
+        else:
+            self.additional_params["enable_ssml"] = True
         if not self.callback:
             self.callback = ResultCallback()
         self.__start_stream()

dashscope/multimodal/dialog_state.py

@@ -0,0 +1,56 @@
+# dialog_state.py
+
+from enum import Enum
+
+
+class DialogState(Enum):
+    """
+    对话状态枚举类，定义了对话机器人可能处于的不同状态。
+
+    Attributes:
+        IDLE (str): 表示机器人处于空闲状态。
+        LISTENING (str): 表示机器人正在监听用户输入。
+        THINKING (str): 表示机器人正在思考。
+        RESPONDING (str): 表示机器人正在生成或回复中。
+    """
+    IDLE = 'Idle'
+    LISTENING = 'Listening'
+    THINKING = 'Thinking'
+    RESPONDING = 'Responding'
+
+
+class StateMachine:
+    """
+    状态机类，用于管理机器人的状态转换。
+
+    Attributes:
+        current_state (DialogState): 当前状态。
+    """
+
+    def __init__(self):
+        # 初始化状态机时设置初始状态为IDLE
+        self.current_state = DialogState.IDLE
+
+    def change_state(self, new_state: str) -> None:
+        """
+        更改当前状态到指定的新状态。
+
+        Args:
+            new_state (str): 要切换到的新状态。
+
+        Raises:
+            ValueError: 如果尝试切换到一个无效的状态，则抛出此异常。
+        """
+        if new_state in [state.value for state in DialogState]:
+            self.current_state = DialogState(new_state)
+        else:
+            raise ValueError("无效的状态类型")
+
+    def get_current_state(self) -> DialogState:
+        """
+        获取当前状态。
+
+        Returns:
+            DialogState: 当前状态。
+        """
+        return self.current_state

dashscope/multimodal/multimodal_constants.py

@@ -0,0 +1,27 @@
+# Copyright (c) Alibaba, Inc. and its affiliates.
+# -*- coding: utf-8 -*-
+
+# multimodal conversation request directive
+
+class RequestToRespondType:
+    TRANSCRIPT = 'transcript'
+    PROMPT = 'prompt'
+
+
+# multimodal conversation response directive
+RESPONSE_NAME_TASK_STARTED = "task-started"
+RESPONSE_NAME_RESULT_GENERATED = "result-generated"
+RESPONSE_NAME_TASK_FINISHED = "task-finished"
+
+RESPONSE_NAME_TASK_FAILED = "TaskFailed"
+RESPONSE_NAME_STARTED = "Started"
+RESPONSE_NAME_STOPPED = "Stopped"
+RESPONSE_NAME_STATE_CHANGED = "DialogStateChanged"
+RESPONSE_NAME_REQUEST_ACCEPTED = "RequestAccepted"
+RESPONSE_NAME_SPEECH_STARTED = "SpeechStarted"
+RESPONSE_NAME_SPEECH_ENDED = "SpeechEnded"  # 服务端检测到asr语音尾点时下发此事件,可选事件
+RESPONSE_NAME_RESPONDING_STARTED = "RespondingStarted"  # AI语音应答开始，sdk要准备接收服务端下发的语音数据
+RESPONSE_NAME_RESPONDING_ENDED = "RespondingEnded"  # AI语音应答结束
+RESPONSE_NAME_SPEECH_CONTENT = "SpeechContent"  # 用户语音识别出的文本，流式全量输出
+RESPONSE_NAME_RESPONDING_CONTENT = "RespondingContent"  # 统对外输出的文本，流式全量输出
+RESPONSE_NAME_ERROR = "Error"  # 服务端对话中报错

dashscope/multimodal/multimodal_dialog.py

@@ -0,0 +1,608 @@
+import json
+import platform
+import time
+import threading
+from abc import abstractmethod
+
+import websocket
+
+import dashscope
+from dashscope.common.logging import logger
+from dashscope.common.error import InputRequired
+from dashscope.multimodal import dialog_state
+from dashscope.multimodal.multimodal_constants import *
+from dashscope.multimodal.multimodal_request_params import RequestParameters, get_random_uuid, DashHeader, \
+    RequestBodyInput, DashPayload, RequestToRespondParameters, RequestToRespondBodyInput
+from dashscope.protocol.websocket import ActionType
+
+
+class MultiModalCallback:
+    """
+    语音聊天回调类，用于处理语音聊天过程中的各种事件。
+    """
+
+    def on_started(self, dialog_id: str) -> None:
+        """
+        通知对话开始
+
+        :param dialog_id: 回调对话ID
+        """
+        pass
+
+    def on_stopped(self) -> None:
+        """
+        通知对话停止
+        """
+        pass
+
+    def on_state_changed(self, state: 'dialog_state.DialogState') -> None:
+        """
+        对话状态改变
+
+        :param state: 新的对话状态
+        """
+        pass
+
+    def on_speech_audio_data(self, data: bytes) -> None:
+        """
+        合成音频数据回调
+
+        :param data: 音频数据
+        """
+        pass
+
+    def on_error(self, error) -> None:
+        """
+        发生错误时调用此方法。
+
+        :param error: 错误信息
+        """
+        pass
+
+    def on_connected(self) -> None:
+        """
+        成功连接到服务器后调用此方法。
+        """
+        pass
+
+    def on_responding_started(self):
+        """
+        回复开始回调
+        """
+        pass
+
+    def on_responding_ended(self, payload):
+        """
+        回复结束
+        """
+        pass
+
+    def on_speech_started(self):
+        """
+        检测到语音输入结束
+        """
+        pass
+
+    def on_speech_ended(self):
+        """
+        检测到语音输入结束
+        """
+        pass
+
+    def on_speech_content(self, payload):
+        """
+        语音识别文本
+
+        :param payload: text
+        """
+        pass
+
+    def on_responding_content(self, payload):
+        """
+        大模型回复文本。
+
+        :param payload: text
+        """
+        pass
+
+    def on_request_accepted(self):
+        """
+        打断请求被接受。
+        """
+        pass
+
+    def on_close(self, close_status_code, close_msg):
+        """
+        连接关闭时调用此方法。
+
+        :param close_status_code: 关闭状态码
+        :param close_msg: 关闭消息
+        """
+        pass
+
+
+class MultiModalDialog:
+    """
+    用于管理WebSocket连接以进行语音聊天的服务类。
+    """
+
+    def __init__(self,
+                 workspace_id: str,
+                 app_id: str,
+                 request_params: RequestParameters,
+                 multimodal_callback: MultiModalCallback,
+                 url: str = None,
+                 api_key: str = None,
+                 dialog_id: str = None,
+                 model: str = None
+                 ):
+        """
+                创建一个语音对话会话。
+
+                此方法用于初始化一个新的voice_chat会话，设置必要的参数以准备开始与模型的交互。
+                ：param workspace_id: 客户的workspace_id
+                :param app_id: 客户在管控台创建的应用id，可以根据值规律确定使用哪个对话系统
+                :param request_params: 请求参数集合
+                :param url: (str) API的URL地址。
+                :param multimodal_callback: (MultimodalCallback) 回调对象，用于处理来自服务器的消息。
+                :param api_key: (str) 应用程序接入的唯一key
+                :param dialog_id:对话id，如果传入表示承接上下文继续聊
+                :param model: 模型
+        """
+        if request_params is None:
+            raise InputRequired('request_params is required!')
+        if url is None:
+            url = dashscope.base_websocket_api_url
+        if api_key is None:
+            api_key = dashscope.api_key
+
+        self.request_params = request_params
+        self.model = model
+        self._voice_detection = None
+        self.thread = None
+        self.ws = None
+        self.request = _Request()
+        self._callback = multimodal_callback
+        self.url = url
+        self.api_key = api_key
+        self.workspace_id = workspace_id
+        self.app_id = app_id
+        self.dialog_id = dialog_id
+        self.dialog_state = dialog_state.StateMachine()
+        self.response = _Response(self.dialog_state, self._callback, self.close)  # 传递 self.close 作为回调
+
+    def _on_message(self, ws, message):
+        logger.debug(f"<<<<<<< Received message: {message}")
+        if isinstance(message, str):
+            self.response.handle_text_response(message)
+        elif isinstance(message, (bytes, bytearray)):
+            self.response.handle_binary_response(message)
+
+    def _on_error(self, ws, error):
+        logger.error(f"Error: {error}")
+        if self._callback:
+            self._callback.on_error(error)
+
+    def _on_close(self, ws, close_status_code, close_msg):
+        try:
+            logger.debug("WebSocket connection closed with status {} and message {}".format(close_status_code, close_msg))
+            if close_status_code is None:
+                close_status_code = 1000
+            if close_msg is None:
+                close_msg = "websocket is closed"
+            self._callback.on_close(close_status_code, close_msg)
+        except Exception as e:
+            logger.error(f"Error: {e}")
+
+    def _on_open(self, ws):
+        self._callback.on_connected()
+
+    # def _on_pong(self, _):
+    #     _log.debug("on pong")
+
+    def start(self, dialog_id, enable_voice_detection=False):
+        """
+        初始化WebSocket连接并发送启动请求
+        :param dialog_id: 上下位继承标志位。新对话无需设置。
+               如果继承之前的对话历史，则需要记录之前的dialog_id并传入
+        :param enable_voice_detection: 是否开启语音检测,可选参数 默认False
+        """
+        self._voice_detection = enable_voice_detection
+        self._connect(self.api_key)
+        logger.debug("connected with server.")
+        self._send_start_request(dialog_id, self.request_params)
+
+    def start_speech(self):
+        """开始上传语音数据"""
+        _send_speech_json = self.request.generate_common_direction_request("SendSpeech", self.dialog_id)
+        self._send_text_frame(_send_speech_json)
+
+    def send_audio_data(self, speech_data: bytes):
+        """发送语音数据"""
+        self.__send_binary_frame(speech_data)
+
+    def stop_speech(self):
+        """停止上传语音数据"""
+        _send_speech_json = self.request.generate_common_direction_request("StopSpeech", self.dialog_id)
+        self._send_text_frame(_send_speech_json)
+
+    def interrupt(self):
+        """请求服务端开始说话"""
+        _send_speech_json = self.request.generate_common_direction_request("RequestToSpeak", self.dialog_id)
+        self._send_text_frame(_send_speech_json)
+
+    def request_to_respond(self,
+                           request_type: str,
+                           text: str,
+                           parameters: RequestToRespondParameters = None):
+        """请求服务端直接文本合成语音"""
+        _send_speech_json = self.request.generate_request_to_response_json(direction_name="RequestToRespond",
+                                                                           dialog_id=self.dialog_id,
+                                                                           request_type=request_type, text=text,
+                                                                           parameters=parameters)
+        self._send_text_frame(_send_speech_json)
+
+    @abstractmethod
+    def request_to_respond_prompt(self, text):
+        """请求服务端通过文本请求回复文本答复"""
+        return
+
+    def local_responding_started(self):
+        """本地tts播放开始"""
+        _send_speech_json = self.request.generate_common_direction_request("LocalRespondingStarted", self.dialog_id)
+        self._send_text_frame(_send_speech_json)
+
+    def local_responding_ended(self):
+        """本地tts播放结束"""
+        _send_speech_json = self.request.generate_common_direction_request("LocalRespondingEnded", self.dialog_id)
+        self._send_text_frame(_send_speech_json)
+
+    def stop(self):
+        if self.ws is None or not self.ws.sock or not self.ws.sock.connected:
+            self._callback.on_close(1001, "websocket is not connected")
+            return
+        _send_speech_json = self.request.generate_stop_request("Stop", self.dialog_id)
+        self._send_text_frame(_send_speech_json)
+
+    def get_dialog_state(self) -> dialog_state.DialogState:
+        return self.dialog_state.get_current_state()
+
+    def get_conversation_mode(self) -> str:
+        """get mode of conversation: support tap2talk/push2talk/duplex"""
+        return self.request_params.upstream.mode
+
+    """内部方法"""
+
+    def _send_start_request(self, dialog_id: str, request_params: RequestParameters):
+        """发送'Start'请求"""
+        _start_json = self.request.generate_start_request(
+            workspace_id=self.workspace_id,
+            direction_name="Start",
+            dialog_id=dialog_id,
+            app_id=self.app_id,
+            request_params=request_params,
+            model=self.model
+        )
+        # send start request
+        self._send_text_frame(_start_json)
+
+    def _run_forever(self):
+        self.ws.run_forever(ping_interval=5, ping_timeout=4)
+
+    def _connect(self, api_key: str):
+        """初始化WebSocket连接并发送启动请求。"""
+        self.ws = websocket.WebSocketApp(self.url, header=self.request.get_websocket_header(api_key),
+                                         on_open=self._on_open,
+                                         on_message=self._on_message,
+                                         on_error=self._on_error,
+                                         on_close=self._on_close)
+        self.thread = threading.Thread(target=self._run_forever)
+        self.ws.ping_interval = 3
+        self.thread.daemon = True
+        self.thread.start()
+
+        self._wait_for_connection()
+
+    def close(self):
+        if self.ws is None or not self.ws.sock or not self.ws.sock.connected:
+            return
+        self.ws.close()
+
+    def _wait_for_connection(self):
+        """等待WebSocket连接建立"""
+        timeout = 5
+        start_time = time.time()
+        while not (self.ws.sock and self.ws.sock.connected) and (time.time() - start_time) < timeout:
+            time.sleep(0.1)  # 短暂休眠，避免密集轮询
+
+    def _send_text_frame(self, text: str):
+        logger.info('>>>>>> send text frame : %s' % text)
+        self.ws.send(text, websocket.ABNF.OPCODE_TEXT)
+
+    def __send_binary_frame(self, binary: bytes):
+        # _log.info('send binary frame length: %d' % len(binary))
+        self.ws.send(binary, websocket.ABNF.OPCODE_BINARY)
+
+    def __del__(self):
+        self.cleanup()
+
+    def cleanup(self):
+        """清理所有资源"""
+        try:
+            if self.ws:
+                self.ws.close()
+            if self.thread and self.thread.is_alive():
+                # 设置标志位通知线程退出
+                self.thread.join(timeout=2)
+            # 清除引用
+            self.ws = None
+            self.thread = None
+            self._callback = None
+            self.response = None
+        except Exception as e:
+            logger.error(f"Error in cleanup: {e}")
+
+
+class _Request:
+    def __init__(self):
+        # websocket header
+        self.ws_headers = None
+        # request body for voice chat
+        self.header = None
+        self.payload = None
+        # params
+        self.task_id = None
+        self.app_id = None
+        self.workspace_id = None
+
+    def get_websocket_header(self, api_key):
+        ua = 'dashscope/%s; python/%s; platform/%s; processor/%s' % (
+            '1.18.0',  # dashscope version
+            platform.python_version(),
+            platform.platform(),
+            platform.processor(),
+        )
+        self.ws_headers = {
+            "User-Agent": ua,
+            "Authorization": f"bearer {api_key}",
+            "Accept": "application/json"
+        }
+        logger.info('websocket header: {}'.format(self.ws_headers))
+        return self.ws_headers
+
+    def generate_start_request(self, direction_name: str,
+                               dialog_id: str,
+                               workspace_id: str,
+                               app_id: str,
+                               request_params: RequestParameters,
+                               model: str = None
+                               ) -> str:
+        """
+        构建语音聊天服务的启动请求数据.
+        :param app_id: 管控台应用id
+        :param request_params: start请求body中的parameters
+        :param direction_name:
+        :param dialog_id: 对话ID.
+        :param workspace_id: 管控台工作区id
+        :param model: 模型
+        :return: 启动请求字典.
+        """
+        self._get_dash_request_header(ActionType.START)
+        self._get_dash_request_payload(direction_name, dialog_id, app_id, workspace_id=workspace_id,
+                                       request_params=request_params, model=model)
+
+        cmd = {
+            "header": self.header,
+            "payload": self.payload
+        }
+        return json.dumps(cmd)
+
+    def generate_common_direction_request(self, direction_name: str, dialog_id: str) -> str:
+        """
+        构建语音聊天服务的命令请求数据.
+        :param direction_name: 命令.
+        :param dialog_id: 对话ID.
+        :return: 命令请求json.
+        """
+        self._get_dash_request_header(ActionType.CONTINUE)
+        self._get_dash_request_payload(direction_name, dialog_id, self.app_id)
+        cmd = {
+            "header": self.header,
+            "payload": self.payload
+        }
+        return json.dumps(cmd)
+
+    def generate_stop_request(self, direction_name: str, dialog_id: str) -> str:
+        """
+        构建语音聊天服务的启动请求数据.
+        :param direction_name:指令名称
+        :param dialog_id: 对话ID.
+        :return: 启动请求json.
+        """
+        self._get_dash_request_header(ActionType.FINISHED)
+        self._get_dash_request_payload(direction_name, dialog_id, self.app_id)
+
+        cmd = {
+            "header": self.header,
+            "payload": self.payload
+        }
+        return json.dumps(cmd)
+
+    @abstractmethod
+    def generate_request_to_response_json(self, direction_name: str, dialog_id: str, request_type: str, text: str,
+                                          parameters: RequestToRespondParameters = None) -> str:
+        """
+        构建语音聊天服务的命令请求数据.
+        :param direction_name: 命令.
+        :param dialog_id: 对话ID.
+        :param request_type: 服务应该采取的交互类型，transcript 表示直接把文本转语音，prompt 表示把文本送大模型回答
+        :param text: 文本.
+        :param parameters: 命令请求body中的parameters
+        :return: 命令请求字典.
+        """
+        self._get_dash_request_header(ActionType.CONTINUE)
+
+        custom_input = RequestToRespondBodyInput(
+            app_id=self.app_id,
+            directive=direction_name,
+            dialog_id=dialog_id,
+            type_=request_type,
+            text=text
+        )
+
+        self._get_dash_request_payload(direction_name, dialog_id, self.app_id, request_params=parameters,
+                                       custom_input=custom_input)
+        cmd = {
+            "header": self.header,
+            "payload": self.payload
+        }
+        return json.dumps(cmd)
+
+    def _get_dash_request_header(self, action: str):
+        """
+        构建多模对话请求的请求协议Header
+        :param action: ActionType 百炼协议action 支持：run-task, continue-task, finish-task
+        """
+        if self.task_id is None:
+            self.task_id = get_random_uuid()
+        self.header = DashHeader(action=action, task_id=self.task_id).to_dict()
+
+    def _get_dash_request_payload(self, direction_name: str,
+                                  dialog_id: str, app_id: str, workspace_id: str = None,
+                                  request_params: RequestParameters = None, custom_input=None, model: str = None):
+        """
+        构建多模对话请求的请求协议payload
+        :param direction_name: 对话协议内部的指令名称
+        :param dialog_id: 对话ID.
+        :param app_id: 管控台应用id
+        :param request_params: start请求body中的parameters
+        :param custom_input: 自定义输入
+        :param model: 模型
+        """
+        if custom_input is not None:
+            input = custom_input
+        else:
+            input = RequestBodyInput(
+                workspace_id=workspace_id,
+                app_id=app_id,
+                directive=direction_name,
+                dialog_id=dialog_id
+            )
+
+        self.payload = DashPayload(
+            model=model,
+            input=input,
+            parameters=request_params
+        ).to_dict()
+
+
+class _Response:
+    def __init__(self, state: dialog_state.StateMachine, callback: MultiModalCallback, close_callback=None):
+        super().__init__()
+        self.dialog_id = None  # 对话ID.
+        self.dialog_state = state
+        self._callback = callback
+        self._close_callback = close_callback  # 保存关闭回调函数
+
+    def handle_text_response(self, response_json: str):
+        """
+        处理语音聊天服务的响应数据.
+        :param response_json: 从服务接收到的原始JSON字符串响应。
+        """
+        logger.info("<<<<<< server response: %s" % response_json)
+        try:
+            # 尝试将消息解析为JSON
+            json_data = json.loads(response_json)
+            if "status_code" in json_data["header"] and json_data["header"]["status_code"] != 200:
+                logger.error("Server returned invalid message: %s" % response_json)
+                if self._callback:
+                    self._callback.on_error(response_json)
+                return
+            if "event" in json_data["header"] and json_data["header"]["event"] == "task-failed":
+                logger.error("Server returned invalid message: %s" % response_json)
+                if self._callback:
+                    self._callback.on_error(response_json)
+                return None
+
+            payload = json_data["payload"]
+            if "output" in payload and payload["output"] is not None:
+                response_event = payload["output"]["event"]
+                logger.info("Server response event: %s" % response_event)
+                self._handle_text_response_in_conversation(response_event=response_event, response_json=json_data)
+            del json_data
+
+        except json.JSONDecodeError:
+            logger.error("Failed to parse message as JSON.")
+
+    def _handle_text_response_in_conversation(self, response_event: str, response_json: dict):
+        payload = response_json["payload"]
+        try:
+            if response_event == RESPONSE_NAME_STARTED:
+                self._handle_started(payload["output"])
+            elif response_event == RESPONSE_NAME_STOPPED:
+                self._handle_stopped()
+            elif response_event == RESPONSE_NAME_STATE_CHANGED:
+                self._handle_state_changed(payload["output"]["state"])
+                logger.debug("service response change state: %s" % payload["output"]["state"])
+            elif response_event == RESPONSE_NAME_REQUEST_ACCEPTED:
+                self._handle_request_accepted()
+            elif response_event == RESPONSE_NAME_SPEECH_STARTED:
+                self._handle_speech_started()
+            elif response_event == RESPONSE_NAME_SPEECH_ENDED:
+                self._handle_speech_ended()
+            elif response_event == RESPONSE_NAME_RESPONDING_STARTED:
+                self._handle_responding_started()
+            elif response_event == RESPONSE_NAME_RESPONDING_ENDED:
+                self._handle_responding_ended(payload)
+            elif response_event == RESPONSE_NAME_SPEECH_CONTENT:
+                self._handle_speech_content(payload)
+            elif response_event == RESPONSE_NAME_RESPONDING_CONTENT:
+                self._handle_responding_content(payload)
+            elif response_event == RESPONSE_NAME_ERROR:
+                self._callback.on_error(json.dumps(response_json))
+            else:
+                logger.error("Unknown response name: {}", response_event)
+        except json.JSONDecodeError:
+            logger.error("Failed to parse message as JSON.")
+
+    def handle_binary_response(self, message: bytes):
+        # logger.debug('<<<recv binary {}'.format(len(message)))
+        self._callback.on_speech_audio_data(message)
+
+    def _handle_request_accepted(self):
+        self._callback.on_request_accepted()
+
+    def _handle_started(self, payload: dict):
+        self.dialog_id = payload["dialog_id"]
+        self._callback.on_started(self.dialog_id)
+
+    def _handle_stopped(self):
+        self._callback.on_stopped()
+        if self._close_callback is not None:
+            self._close_callback()
+
+    def _handle_state_changed(self, state: str):
+        """
+        处理语音聊天状态流转.
+        :param state: 状态.
+        """
+        self.dialog_state.change_state(state)
+        self._callback.on_state_changed(self.dialog_state.get_current_state())
+
+    def _handle_speech_started(self):
+        self._callback.on_speech_started()
+
+    def _handle_speech_ended(self):
+        self._callback.on_speech_ended()
+
+    def _handle_responding_started(self):
+        self._callback.on_responding_started()
+
+    def _handle_responding_ended(self, payload: dict):
+        self._callback.on_responding_ended(payload)
+
+    def _handle_speech_content(self, payload: dict):
+        self._callback.on_speech_content(payload)
+
+    def _handle_responding_content(self, payload: dict):
+        self._callback.on_responding_content(payload)

dashscope/multimodal/multimodal_request_params.py

@@ -0,0 +1,269 @@
+from dataclasses import dataclass, field, asdict
+import uuid
+
+
+def get_random_uuid() -> str:
+    """生成并返回32位UUID字符串"""
+    return uuid.uuid4().hex
+
+
+@dataclass
+class DashHeader:
+    action: str
+    task_id: str = field(default=get_random_uuid())
+    streaming: str = field(default="duplex")  # 默认为 duplex
+
+    def to_dict(self):
+        return {
+            "action": self.action,
+            "task_id": self.task_id,
+            "request_id": self.task_id,
+            "streaming": self.streaming
+        }
+
+
+class DashPayloadParameters:
+    def to_dict(self):
+        pass
+
+
+class DashPayloadInput:
+    def to_dict(self):
+        pass
+
+
+@dataclass
+class DashPayload:
+    task_group: str = field(default="aigc")
+    function: str = field(default="generation")
+    model: str = field(default="")
+    task: str = field(default="multimodal-generation")
+    parameters: DashPayloadParameters = field(default=None)
+    input: DashPayloadInput = field(default=None)
+
+    def to_dict(self):
+        payload = {
+            "task_group": self.task_group,
+            "function": self.function,
+            "model": self.model,
+            "task": self.task,
+        }
+
+        if self.parameters is not None:
+            payload["parameters"] = self.parameters.to_dict()
+
+        if self.input is not None:
+            payload["input"] = self.input.to_dict()
+
+        return payload
+
+
+@dataclass
+class RequestBodyInput(DashPayloadInput):
+    workspace_id: str
+    app_id: str
+    directive: str
+    dialog_id: str
+
+    def to_dict(self):
+        return {
+            "workspace_id": self.workspace_id,
+            "app_id": self.app_id,
+            "directive": self.directive,
+            "dialog_id": self.dialog_id
+        }
+
+
+@dataclass
+class Upstream:
+    """struct for upstream"""
+    audio_format: str = field(default="pcm")  # 上行语音格式，默认pcm.支持pcm/opus
+    type: str = field(default="AudioOnly")  # 上行类型：AudioOnly 仅语音通话; AudioAndVideo 上传视频
+    mode: str = field(default="tap2talk")  # 客户端交互模式 push2talk/tap2talk/duplex
+    # sample_rate: int  # 合成音频采样率
+
+    def to_dict(self):
+        return {
+            "type": self.type,
+            "mode": self.mode,
+            "audio_format": self.audio_format,
+            # "sample_rate": self.sample_rate
+        }
+
+
+@dataclass
+class Downstream:
+    # transcript  返回用户语音识别结果
+    # dialog 返回对话系统回答中间结果
+    # 可以设置多种，以逗号分割，默认为transcript
+    voice: str = field(default="")  # 语音音色
+    sample_rate: int = field(default=0)  # 语音音色 # 合成音频采样率
+    intermediate_text: str = field(default="transcript")  # 控制返回给用户那些中间文本：
+    debug: bool = field(default=False)  # 控制是否返回debug信息
+    # type_: str = field(default="Audio", metadata={"alias": "type"})  # 下行类型：Text：不需要下发语音;Audio：输出语音，默认值
+    audio_format: str = field(default="pcm")  # 下行语音格式，默认pcm 。支持pcm/mp3
+    volume: int = field(default=50)  # 语音音量 0-100
+    pitch_rate: int = field(default=100)  # 语音语调 50-200
+    speech_rate: int = field(default=100)  # 语音语速 50-200
+
+    def to_dict(self):
+        stream: dict = {
+            "intermediate_text": self.intermediate_text,
+            "debug": self.debug,
+            # "type": self.type_,
+            "audio_format": self.audio_format,
+            "volume": self.volume,
+            "pitch_rate": self.pitch_rate,
+            "speech_rate": self.speech_rate
+        }
+        if self.voice != "":
+            stream["voice"] = self.voice
+        if self.sample_rate != 0:
+            stream["sample_rate"] = self.sample_rate
+        return stream
+
+
+@dataclass
+class DialogAttributes:
+    agent_id: str = field(default=None)
+    prompt: str = field(default=None)
+    vocabulary_id: str = field(default=None)
+
+    def to_dict(self):
+        return {
+            "agent_id": self.agent_id,
+            "prompt": self.prompt,
+            "vocabulary_id": self.vocabulary_id
+        }
+
+
+@dataclass
+class Locations:
+    city_name: str = field(default=None)
+    latitude: str = field(default=None)
+    longitude: str = field(default=None)
+
+    def to_dict(self):
+        return {
+            "city_name": self.city_name,
+            "latitude": self.latitude,
+            "longitude": self.longitude
+        }
+
+
+@dataclass
+class Network:
+    ip: str = field(default=None)
+
+    def to_dict(self):
+        return {
+            "ip": self.ip
+        }
+
+
+@dataclass
+class Device:
+    uuid: str = field(default=None)
+
+    def to_dict(self):
+        return {
+            "uuid": self.uuid
+        }
+
+
+@dataclass
+class ClientInfo:
+    user_id: str
+    device: Device = field(default=None)
+    network: Network = field(default=None)
+    location: Locations = field(default=None)
+
+    def to_dict(self):
+        info = {
+            "user_id": self.user_id
+        }
+        if self.device is not None:
+            info["device"] = self.device.to_dict()
+        if self.network is not None:
+            info["network"] = self.network.to_dict()
+        if self.location is not None:
+            info["location"] = self.location.to_dict()
+        return info
+
+
+@dataclass
+class BizParams:
+    user_defined_params: dict = field(default=None)
+    user_defined_tokens: dict = field(default=None)
+    tool_prompts: dict = field(default=None)
+    user_prompt_params: dict = field(default=None)
+    user_query_params: dict = field(default=None)
+
+    def to_dict(self):
+        params = {}
+        if self.user_defined_params is not None:
+            params["user_defined_params"] = self.user_defined_params
+        if self.user_defined_tokens is not None:
+            params["user_defined_tokens"] = self.user_defined_tokens
+        if self.tool_prompts is not None:
+            params["tool_prompts"] = self.tool_prompts
+        if self.user_prompt_params is not None:
+            params["user_prompt_params"] = self.user_prompt_params
+        if self.user_query_params is not None:
+            params["user_query_params"] = self.user_query_params
+        return params
+
+
+@dataclass
+class RequestParameters(DashPayloadParameters):
+    upstream: Upstream
+    downstream: Downstream
+    client_info: ClientInfo
+    dialog_attributes: DialogAttributes = field(default=None)
+    biz_params: BizParams = field(default=None)
+
+    def to_dict(self):
+        params = {
+            "upstream": self.upstream.to_dict(),
+            "downstream": self.downstream.to_dict(),
+            "client_info": self.client_info.to_dict(),
+        }
+
+        if self.dialog_attributes is not None:
+            params["dialog_attributes"] = self.dialog_attributes.to_dict()
+        if self.biz_params is not None:
+            params["biz_params"] = self.biz_params.to_dict()
+        return params
+
+
+@dataclass
+class RequestToRespondParameters(DashPayloadParameters):
+    images: list = field(default=None)
+    biz_params: BizParams = field(default=None)
+
+    def to_dict(self):
+        params = {
+        }
+        if self.images is not None:
+            params["images"] = self.images
+        if self.biz_params is not None:
+            params["biz_params"] = self.biz_params.to_dict()
+        return params
+
+
+@dataclass
+class RequestToRespondBodyInput(DashPayloadInput):
+    app_id: str
+    directive: str
+    dialog_id: str
+    type_: str = field(metadata={"alias": "type"})
+    text: str = field(default="")
+
+    def to_dict(self):
+        return {
+            "app_id": self.app_id,
+            "directive": self.directive,
+            "dialog_id": self.dialog_id,
+            "type": self.type_,
+            "text": self.text
+        }

dashscope/version.py

@@ -1,3 +1,3 @@
 # Copyright (c) Alibaba, Inc. and its affiliates.
 
-__version__ = '1.23.3'
+__version__ = '1.23.5'

{dashscope-1.23.3.dist-info → dashscope-1.23.5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: dashscope
-Version: 1.23.3
+Version: 1.23.5
 Summary: dashscope client sdk library
 Home-page: https://dashscope.aliyun.com/
 Author: Alibaba Cloud

{dashscope-1.23.3.dist-info → dashscope-1.23.5.dist-info}/RECORD

@@ -3,13 +3,13 @@ dashscope/cli.py,sha256=amegoTkGOs6TlHMdoo4JVOqBePo3lGs745rc7leEyrE,24020
 dashscope/files.py,sha256=vRDQygm3lOqBZR73o7KNHs1iTBVuvLncuwJNxIYjzAU,3981
 dashscope/model.py,sha256=B5v_BtYLPqj6raClejBgdKg6WTGwhH_f-20pvsQqmsk,1491
 dashscope/models.py,sha256=dE4mzXkl85G343qVylSGpURPRdA5pZSqXlx6PcxqC_Q,1275
-dashscope/version.py,sha256=KoL2HDS3KUg-imtukZreW80cuyVG_B6Uy4FbPH7nZe4,74
+dashscope/version.py,sha256=AEebMkhuRmXDos0SQUqvNS4ui-KBHqJSwnTbSjynlLU,74
 dashscope/aigc/__init__.py,sha256=AuRhu_vA1K0tbs_C6DgcZYhTvxMuzDgpwHJNHzEPIHg,442
 dashscope/aigc/chat_completion.py,sha256=ONlyyssIbfaKKcFo7cEKhHx5OCF2XX810HFzIExW1ho,14813
 dashscope/aigc/code_generation.py,sha256=p_mxDKJLQMW0IjFD46JRlZuEZCRESSVKEfLlAevBtqw,10936
 dashscope/aigc/conversation.py,sha256=95xEEY4ThZJysj5zy3aMw7ql9KLJVfD_1iHv9QZ17Ew,14282
 dashscope/aigc/generation.py,sha256=xMcMu16rICTdjZiD_sPqYV_Ltdp4ewGzzfC7JD9VApY,17948
-dashscope/aigc/image_synthesis.py,sha256=RGyjhSl1-5YzaYBtAMaoemh6Ij_Xw0l_z9IGKD5qgWI,13359
+dashscope/aigc/image_synthesis.py,sha256=Jgqmyv4jRxikgX7J18QrKKQ4OZAMxs6Q6YXObae3DhI,13363
 dashscope/aigc/multimodal_conversation.py,sha256=1rZZRk_1lCdbVs7Rx1kJ5LvwWE1put5p_dQKdCX0ysY,5574
 dashscope/aigc/video_synthesis.py,sha256=XQ3-NKYFmj5cIbUbLTbI0-FyC_fQp8eds6QmD1ZHj_0,13015
 dashscope/api_entities/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -34,14 +34,14 @@ dashscope/audio/asr/asr_phrase_manager.py,sha256=vHOLExaKCtjedkihIu7gyfQyarR9rN5
 dashscope/audio/asr/recognition.py,sha256=b_aAPvOKjpWdSiYhM_hp30sZ06QdmNBSDJwhiv78kHM,20932
 dashscope/audio/asr/transcription.py,sha256=lYzPjh7jJQwjMoxx8-AY0YCMBKNKO0bi7xd5tZGSHPc,9094
 dashscope/audio/asr/translation_recognizer.py,sha256=JgBmhkIl_kqH8uVwop6Fba5KlXccftKFrhaygN9PKjU,39680
-dashscope/audio/asr/vocabulary.py,sha256=TEmN4JqXFpg9c1o3BUyj-rA7L58T85iURukJIupEiv0,6528
+dashscope/audio/asr/vocabulary.py,sha256=N0pMS2x1lDxqJ14FgTGKctfuVkR2_hlEsCNWFcgYpTY,6717
 dashscope/audio/qwen_tts/__init__.py,sha256=JS3axY1grqO0aTIJufZ3KS1JsU6yf6y4K2CQlNvUK9I,132
 dashscope/audio/qwen_tts/speech_synthesizer.py,sha256=7LHR-PXhn-VE1cCOp_82Jq0zE9rMc3xy3dszUeyLLNs,2927
 dashscope/audio/tts/__init__.py,sha256=xYpMFseUZGgqgj_70zcX2VsLv-L7qxJ3d-bbdj_hO0I,245
 dashscope/audio/tts/speech_synthesizer.py,sha256=vD1xQV-rew8qAsIaAGH5amsNtB0SqdtNhVHhJHGQ-xk,7622
 dashscope/audio/tts_v2/__init__.py,sha256=me9a3_7KsHQxcJ8hx4SeKlY1e_ThHVvGMw7Yn0uoscM,333
-dashscope/audio/tts_v2/enrollment.py,sha256=DlkV4P1unbtgg9vQNfDfnBqA61yYtIwRHgUdwPh4AF4,6192
-dashscope/audio/tts_v2/speech_synthesizer.py,sha256=aL_VNKMYwiw5pKv3SvuAzWtZF0bpyvntc3Kh_iJroNM,19980
+dashscope/audio/tts_v2/enrollment.py,sha256=-nrlywYSOP73Bm9ETTSxNnlp-B8ezJcUmd59mVvyvgk,6361
+dashscope/audio/tts_v2/speech_synthesizer.py,sha256=qUoLga8HpvNVdbN5n_orxrgZ28yD6Lhwuwqeoi1T7yA,20056
 dashscope/client/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 dashscope/client/base_api.py,sha256=aWNy_xm02GXuLKVgWnYJht2nI4ZHSGfYIcr52SML15A,41239
 dashscope/common/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -64,6 +64,11 @@ dashscope/embeddings/multimodal_embedding.py,sha256=NwjQsdkKgUz51ozGjqFDzVlLcZjY
 dashscope/embeddings/text_embedding.py,sha256=4W1V-Ggj0WJhY5MdP2xoUTteXlWk8TUtI6y2gRUVCUk,2060
 dashscope/io/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 dashscope/io/input_output.py,sha256=0aXrRJFo1ZqYm_AJWR_w88O4-Btn9np2zUhrrUdBdfw,3992
+dashscope/multimodal/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+dashscope/multimodal/dialog_state.py,sha256=CtOdfGWhq0ePG3bc8-7inhespETtPD4QDli1513hd1A,1522
+dashscope/multimodal/multimodal_constants.py,sha256=MlKJkOIGoEzp3c3jXzeEE4zw0oNHv4HkEQeJGx7kLaM,1228
+dashscope/multimodal/multimodal_dialog.py,sha256=21yIQyrRzRcfDSbo0IMcEu-KYMji70cEH1p3NLsSR-Q,22372
+dashscope/multimodal/multimodal_request_params.py,sha256=chFDxPvbKcEyLEkVdsBFWUChsHaKnBAvPcG-IOjVRq4,7775
 dashscope/nlp/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 dashscope/nlp/understanding.py,sha256=00ado-ibYEzBRT0DgKGd3bohQDNW73xnFhJ_1aa87lw,2880
 dashscope/protocol/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -87,9 +92,9 @@ dashscope/tokenizers/tokenizer.py,sha256=3FQVDvMNkCW9ccYeJdjrd_PIMMD3Xv7aNZkaYOE
 dashscope/tokenizers/tokenizer_base.py,sha256=5EJIFuizMWESEmLmbd38yJnfeHmPnzZPwsO4aOGjpl4,707
 dashscope/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 dashscope/utils/oss_utils.py,sha256=L5LN3lN8etVxSL_jkZydstvEKpnTG9CY0zcvPGQ5LBo,7383
-dashscope-1.23.3.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-dashscope-1.23.3.dist-info/METADATA,sha256=lqzVJ_GvC8JbdUYLTDjC2OL0bdp1Vzk5RaMD9aFIkYE,6798
-dashscope-1.23.3.dist-info/WHEEL,sha256=G16H4A3IeoQmnOrYV4ueZGKSjhipXx8zc8nu9FGlvMA,92
-dashscope-1.23.3.dist-info/entry_points.txt,sha256=raEp5dOuj8whJ7yqZlDM8WQ5p2RfnGrGNo0QLQEnatY,50
-dashscope-1.23.3.dist-info/top_level.txt,sha256=woqavFJK9zas5xTqynmALqOtlafghjsk63Xk86powTU,10
-dashscope-1.23.3.dist-info/RECORD,,
+dashscope-1.23.5.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+dashscope-1.23.5.dist-info/METADATA,sha256=tERMxFGxdS3EOMhTiNl3jaQCAf-k40aUNJXZsMERc5s,6798
+dashscope-1.23.5.dist-info/WHEEL,sha256=G16H4A3IeoQmnOrYV4ueZGKSjhipXx8zc8nu9FGlvMA,92
+dashscope-1.23.5.dist-info/entry_points.txt,sha256=raEp5dOuj8whJ7yqZlDM8WQ5p2RfnGrGNo0QLQEnatY,50
+dashscope-1.23.5.dist-info/top_level.txt,sha256=woqavFJK9zas5xTqynmALqOtlafghjsk63Xk86powTU,10
+dashscope-1.23.5.dist-info/RECORD,,