dashscope 1.23.7__py3-none-any.whl → 1.23.9__py3-none-any.whl

dashscope/assistants/assistants.py

@@ -23,8 +23,8 @@ class Assistants(CreateMixin, CancelMixin, DeleteMixin, ListObjectMixin,
         name: str = None,
         description: str = None,
         instructions: str = None,
-        tools: Optional[str] = [],
-        file_ids: Optional[str] = [],
+        tools: Optional[List[Dict]] = None,
+        file_ids: Optional[List[str]] = [],
         metadata: Dict = {},
     ):
         obj = {}
@@ -36,7 +36,7 @@ class Assistants(CreateMixin, CancelMixin, DeleteMixin, ListObjectMixin,
             obj['description'] = description
         if instructions:
             obj['instructions'] = instructions
-        if tools:
+        if tools is not None:
             obj['tools'] = tools
         obj['file_ids'] = file_ids
         obj['metadata'] = metadata
@@ -50,7 +50,7 @@ class Assistants(CreateMixin, CancelMixin, DeleteMixin, ListObjectMixin,
              name: str = None,
              description: str = None,
              instructions: str = None,
-             tools: Optional[List[Dict]] = [],
+             tools: Optional[List[Dict]] = None,
              file_ids: Optional[List[str]] = [],
              metadata: Dict = None,
              workspace: str = None,
@@ -93,7 +93,7 @@ class Assistants(CreateMixin, CancelMixin, DeleteMixin, ListObjectMixin,
                name: str = None,
                description: str = None,
                instructions: str = None,
-               tools: Optional[List[Dict]] = [],
+               tools: Optional[List[Dict]] = None,
                file_ids: Optional[List[str]] = [],
                metadata: Dict = None,
                workspace: str = None,
@@ -219,8 +219,8 @@ class Assistants(CreateMixin, CancelMixin, DeleteMixin, ListObjectMixin,
                name: str = None,
                description: str = None,
                instructions: str = None,
-               tools: Optional[str] = [],
-               file_ids: Optional[str] = [],
+               tools: Optional[List[Dict]] = None,
+               file_ids: Optional[List[str]] = [],
                metadata: Dict = None,
                workspace: str = None,
                api_key: str = None,

dashscope/audio/__init__.py

@@ -1,5 +1,5 @@
 # Copyright (c) Alibaba, Inc. and its affiliates.
 
-from . import asr, tts, tts_v2, qwen_tts
+from . import asr, tts, tts_v2, qwen_tts, qwen_tts_realtime, qwen_omni
 
-__all__ = [asr, tts, tts_v2, qwen_tts]
+__all__ = [asr, tts, tts_v2, qwen_tts, qwen_tts_realtime, qwen_omni]

dashscope/audio/qwen_omni/__init__.py

@@ -0,0 +1,11 @@
+# Copyright (c) Alibaba, Inc. and its affiliates.
+
+from .omni_realtime import (AudioFormat, MultiModality, OmniRealtimeCallback,
+                            OmniRealtimeConversation)
+
+__all__ = [
+    'OmniRealtimeCallback',
+    'AudioFormat',
+    'MultiModality',
+    'OmniRealtimeConversation',
+]

dashscope/audio/qwen_omni/omni_realtime.py

@@ -0,0 +1,415 @@
+# Copyright (c) Alibaba, Inc. and its affiliates.
+
+import json
+import platform
+import threading
+import time
+from typing import List
+import uuid
+from enum import Enum, unique
+
+import dashscope
+import websocket
+from dashscope.common.error import InputRequired, ModelRequired
+from dashscope.common.logging import logger
+
+
+class OmniRealtimeCallback:
+    """
+    An interface that defines callback methods for getting omni-realtime results. # noqa E501
+    Derive from this class and implement its function to provide your own data.
+    """
+    def on_open(self) -> None:
+        pass
+
+    def on_close(self, close_status_code, close_msg) -> None:
+        pass
+
+    def on_event(self, message: str) -> None:
+        pass
+
+
+@unique
+class AudioFormat(Enum):
+    # format, sample_rate, channels, bit_rate, name
+    PCM_16000HZ_MONO_16BIT = ('pcm', 16000, 'mono', '16bit', 'pcm16')
+    PCM_24000HZ_MONO_16BIT = ('pcm', 24000, 'mono', '16bit', 'pcm16')
+
+    def __init__(self, format, sample_rate, channels, bit_rate, format_str):
+        self.format = format
+        self.sample_rate = sample_rate
+        self.channels = channels
+        self.bit_rate = bit_rate
+        self.format_str = format_str
+
+    def __repr__(self):
+        return self.format_str
+
+    def __str__(self):
+        return f'{self.format.upper()} with {self.sample_rate}Hz sample rate, {self.channels} channel, {self.bit_rate} bit rate:  {self.format_str}'
+
+
+class MultiModality(Enum):
+    """
+    MultiModality
+    """
+    TEXT = 'text'
+    AUDIO = 'audio'
+
+    def __str__(self):
+        return self.name
+
+
+class OmniRealtimeConversation:
+    def __init__(
+        self,
+        model,
+        callback: OmniRealtimeCallback,
+        headers=None,
+        workspace=None,
+        url=None,
+        additional_params=None,
+    ):
+        """
+        Qwen Omni Realtime SDK
+        Parameters:
+        -----------
+        model: str
+            Model name.
+        headers: Dict
+            User-defined headers.
+        callback: OmniRealtimeCallback
+            Callback to receive real-time omni results.
+        workspace: str
+            Dashscope workspace ID.
+        url: str
+            Dashscope WebSocket URL.
+        additional_params: Dict
+            Additional parameters for the Dashscope API.
+        """
+
+        if model is None:
+            raise ModelRequired('Model is required!')
+        if callback is None:
+            raise ModelRequired('Callback is required!')
+        if url is None:
+            url = f'wss://dashscope.aliyuncs.com/api-ws/v1/realtime?model={model}'
+        else:
+            url = f'{url}?model={model}'
+        self.url = url
+        self.apikey = dashscope.api_key
+        self.user_headers = headers
+        self.user_workspace = workspace
+        self.model = model
+        self.config = {}
+        self.callback = callback
+        self.ws = None
+        self.session_id = None
+        self.last_message = None
+        self.last_response_id = None
+        self.last_response_create_time = None
+        self.last_first_text_delay = None
+        self.last_first_audio_delay = None
+        self.metrics = []
+
+    def _generate_event_id(self):
+        '''
+        generate random event id: event_xxxx
+        '''
+        return 'event_' + uuid.uuid4().hex
+
+    def _get_websocket_header(self, ):
+        ua = 'dashscope/%s; python/%s; platform/%s; processor/%s' % (
+            '1.18.0',  # dashscope version
+            platform.python_version(),
+            platform.platform(),
+            platform.processor(),
+        )
+        headers = {
+            'user-agent': ua,
+            'Authorization': 'bearer ' + self.apikey,
+        }
+        if self.user_headers:
+            headers = {**self.user_headers, **headers}
+        if self.user_workspace:
+            headers = {
+                **headers,
+                'X-DashScope-WorkSpace': self.user_workspace,
+            }
+        return headers
+
+    def connect(self) -> None:
+        '''
+        connect to server, create session and return default session configuration
+        '''
+        self.ws = websocket.WebSocketApp(
+            self.url,
+            header=self._get_websocket_header(),
+            on_message=self.on_message,
+            on_error=self.on_error,
+            on_close=self.on_close,
+        )
+        self.thread = threading.Thread(target=self.ws.run_forever)
+        self.thread.daemon = True
+        self.thread.start()
+        timeout = 5  # 最长等待时间（秒）
+        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
+        if not (self.ws.sock and self.ws.sock.connected):
+            raise TimeoutError(
+                'websocket connection could not established within 5s. '
+                'Please check your network connection, firewall settings, or server status.'
+            )
+        self.callback.on_open()
+
+    def __send_str(self, data: str, enable_log: bool = True):
+        if enable_log:
+            logger.debug('[omni realtime] send string: {}'.format(data))
+        self.ws.send(data)
+
+    def update_session(self,
+                       output_modalities: List[MultiModality],
+                       voice: str,
+                       input_audio_format: AudioFormat = AudioFormat.
+                       PCM_16000HZ_MONO_16BIT,
+                       output_audio_format: AudioFormat = AudioFormat.
+                       PCM_24000HZ_MONO_16BIT,
+                       enable_input_audio_transcription: bool = True,
+                       input_audio_transcription_model: str = None,
+                       enable_turn_detection: bool = True,
+                       turn_detection_type: str = 'server_vad',
+                       prefix_padding_ms: int = 300,
+                       turn_detection_threshold: float = 0.2,
+                       turn_detection_silence_duration_ms: int = 800,
+                       turn_detection_param: dict = None,
+                       **kwargs) -> None:
+        '''
+        update session configuration, should be used before create response
+
+        Parameters
+        ----------
+        output_modalities: list[MultiModality]
+            omni output modalities to be used in session
+        voice: str
+            voice to be used in session
+        input_audio_format: AudioFormat
+            input audio format
+        output_audio_format: AudioFormat
+            output audio format
+        enable_turn_detection: bool
+            enable turn detection
+        turn_detection_threshold: float
+            turn detection threshold, range [-1, 1]
+            In a noisy environment, it may be necessary to increase the threshold to reduce false detections
+            In a quiet environment, it may be necessary to decrease the threshold to improve sensitivity
+        turn_detection_silence_duration_ms: int
+            duration of silence in milliseconds to detect turn, range [200, 6000]
+        '''
+        self.config = {
+            'modalities': [m.value for m in output_modalities],
+            'voice': voice,
+            'input_audio_format': input_audio_format.format_str,
+            'output_audio_format': output_audio_format.format_str,
+        }
+        if enable_input_audio_transcription:
+            self.config['input_audio_transcription'] = {
+                'model': input_audio_transcription_model,
+            }
+        else:
+            self.config['input_audio_transcription'] = None
+        if enable_turn_detection:
+            self.config['turn_detection'] = {
+                'type': turn_detection_type,
+                'threshold': turn_detection_threshold,
+                'prefix_padding_ms': prefix_padding_ms,
+                'silence_duration_ms': turn_detection_silence_duration_ms,
+            }
+            if turn_detection_param is not None:
+                self.config['turn_detection'].update(turn_detection_param)
+        else:
+            self.config['turn_detection'] = None
+        self.config.update(kwargs)
+        self.__send_str(
+            json.dumps({
+                'event_id': self._generate_event_id(),
+                'type': 'session.update',
+                'session': self.config
+            }))
+
+    def append_audio(self, audio_b64: str) -> None:
+        '''
+        send audio in base64 format
+
+        Parameters
+        ----------
+        audio_b64: str
+            base64 audio string
+        '''
+        logger.debug('[omni realtime] append audio: {}'.format(len(audio_b64)))
+        self.__send_str(
+            json.dumps({
+                'event_id': self._generate_event_id(),
+                'type': 'input_audio_buffer.append',
+                'audio': audio_b64
+            }), False)
+
+    def append_video(self, video_b64: str) -> None:
+        '''
+        send one image frame in video in base64 format
+
+        Parameters
+        ----------
+        video_b64: str
+            base64 image string
+        '''
+        logger.debug('[omni realtime] append video: {}'.format(len(video_b64)))
+        self.__send_str(
+            json.dumps({
+                'event_id': self._generate_event_id(),
+                'type': 'input_image_buffer.append',
+                'image': video_b64
+            }), False)
+
+    def commit(self, ) -> None:
+        '''
+        Commit the audio and video sent before.
+        When in Server VAD mode, the client does not need to use this method,
+        the server will commit the audio automatically after detecting vad end.
+        '''
+        self.__send_str(
+            json.dumps({
+                'event_id': self._generate_event_id(),
+                'type': 'input_audio_buffer.commit'
+            }))
+
+    def clear_appended_audio(self, ) -> None:
+        '''
+        clear the audio sent to server before.
+        '''
+        self.__send_str(
+            json.dumps({
+                'event_id': self._generate_event_id(),
+                'type': 'input_audio_buffer.clear'
+            }))
+
+    def create_response(self,
+                        instructions: str = None,
+                        output_modalities: List[MultiModality] = None) -> None:
+        '''
+        create response, use audio and video commited before to request llm.
+        When in Server VAD mode, the client does not need to use this method,
+        the server will create response automatically after detecting vad
+        and sending commit.
+
+        Parameters
+        ----------
+        instructions: str
+            instructions to llm
+        output_modalities: list[MultiModality]
+            omni output modalities to be used in session
+        '''
+        request = {
+            'event_id': self._generate_event_id(),
+            'type': 'response.create',
+            'response': {}
+        }
+        request['response']['instructions'] = instructions
+        if output_modalities:
+            request['response']['modalities'] = [
+                m.value for m in output_modalities
+            ]
+        self.__send_str(json.dumps(request))
+
+    def cancel_response(self, ) -> None:
+        '''
+        cancel the current response
+        '''
+        self.__send_str(
+            json.dumps({
+                'event_id': self._generate_event_id(),
+                'type': 'response.cancel'
+            }))
+
+    def send_raw(self, raw_data: str) -> None:
+        '''
+        send raw data to server
+        '''
+        self.__send_str(raw_data)
+
+    def close(self, ) -> None:
+        '''
+        close the connection to server
+        '''
+        self.ws.close()
+
+    # 监听消息的回调函数
+    def on_message(self, ws, message):
+        if isinstance(message, str):
+            logger.debug('[omni realtime] receive string {}'.format(
+                message[:1024]))
+            try:
+                # 尝试将消息解析为JSON
+                json_data = json.loads(message)
+                self.last_message = json_data
+                self.callback.on_event(json_data)
+                if 'type' in message:
+                    if 'session.created' == json_data['type']:
+                        self.session_id = json_data['session']['id']
+                    if 'response.created' == json_data['type']:
+                        self.last_response_id = json_data['response']['id']
+                        self.last_response_create_time = time.time() * 1000
+                        self.last_first_audio_delay = None
+                        self.last_first_text_delay = None
+                    elif 'response.audio_transcript.delta' == json_data[
+                            'type']:
+                        if self.last_response_create_time and self.last_first_text_delay is None:
+                            self.last_first_text_delay = time.time(
+                            ) * 1000 - self.last_response_create_time
+                    elif 'response.audio.delta' == json_data['type']:
+                        if self.last_response_create_time and self.last_first_audio_delay is None:
+                            self.last_first_audio_delay = time.time(
+                            ) * 1000 - self.last_response_create_time
+                    elif 'response.done' == json_data['type']:
+                        logger.info(
+                            '[Metric] response: {}, first text delay: {}, first audio delay: {}'
+                            .format(self.last_response_id,
+                                    self.last_first_text_delay,
+                                    self.last_first_audio_delay))
+            except json.JSONDecodeError:
+                logger.error('Failed to parse message as JSON.')
+                raise Exception('Failed to parse message as JSON.')
+        elif isinstance(message, (bytes, bytearray)):
+            # 如果失败，认为是二进制消息
+            logger.error(
+                'should not receive binary message in omni realtime api')
+            logger.debug('[omni realtime] receive binary {} bytes'.format(
+                len(message)))
+
+    def on_close(self, ws, close_status_code, close_msg):
+        self.callback.on_close(close_status_code, close_msg)
+
+    # WebSocket发生错误的回调函数
+    def on_error(self, ws, error):
+        print(f'websocket closed due to {error}')
+        raise Exception(f'websocket closed due to {error}')
+
+    # 获取上一个任务的taskId
+    def get_session_id(self) -> str:
+        return self.session_id
+
+    def get_last_message(self) -> str:
+        return self.last_message
+
+    def get_last_message(self) -> str:
+        return self.last_message
+
+    def get_last_response_id(self) -> str:
+        return self.last_response_id
+
+    def get_last_first_text_delay(self):
+        return self.last_first_text_delay
+
+    def get_last_first_audio_delay(self):
+        return self.last_first_audio_delay

dashscope/audio/qwen_tts_realtime/__init__.py

@@ -0,0 +1,10 @@
+# Copyright (c) Alibaba, Inc. and its affiliates.
+
+from .qwen_tts_realtime import (AudioFormat, QwenTtsRealtimeCallback,
+                            QwenTtsRealtime)
+
+__all__ = [
+    'AudioFormat',
+    'QwenTtsRealtimeCallback',
+    'QwenTtsRealtime',
+]

dashscope/audio/qwen_tts_realtime/qwen_tts_realtime.py

@@ -0,0 +1,314 @@
+# Copyright (c) Alibaba, Inc. and its affiliates.
+
+import json
+import platform
+import threading
+import time
+import uuid
+from enum import Enum, unique
+
+import dashscope
+import websocket
+from dashscope.common.error import InputRequired, ModelRequired
+from dashscope.common.logging import logger
+
+
+class QwenTtsRealtimeCallback:
+    """
+    An interface that defines callback methods for getting omni-realtime results. # noqa E501
+    Derive from this class and implement its function to provide your own data.
+    """
+    def on_open(self) -> None:
+        pass
+
+    def on_close(self, close_status_code, close_msg) -> None:
+        pass
+
+    def on_event(self, message: str) -> None:
+        pass
+
+
+@unique
+class AudioFormat(Enum):
+    # format, sample_rate, channels, bit_rate, name
+    PCM_24000HZ_MONO_16BIT = ('pcm', 24000, 'mono', '16bit', 'pcm16')
+
+    def __init__(self, format, sample_rate, channels, bit_rate, format_str):
+        self.format = format
+        self.sample_rate = sample_rate
+        self.channels = channels
+        self.bit_rate = bit_rate
+        self.format_str = format_str
+
+    def __repr__(self):
+        return self.format_str
+
+    def __str__(self):
+        return f'{self.format.upper()} with {self.sample_rate}Hz sample rate, {self.channels} channel, {self.bit_rate} bit rate:  {self.format_str}'
+
+
+class QwenTtsRealtime:
+    def __init__(
+        self,
+        model,
+        headers=None,
+        callback: QwenTtsRealtimeCallback = None,
+        workspace=None,
+        url=None,
+        additional_params=None,
+    ):
+        """
+        Qwen Tts Realtime SDK
+        Parameters:
+        -----------
+        model: str
+            Model name.
+        headers: Dict
+            User-defined headers.
+        callback: OmniRealtimeCallback
+            Callback to receive real-time omni results.
+        workspace: str
+            Dashscope workspace ID.
+        url: str
+            Dashscope WebSocket URL.
+        additional_params: Dict
+            Additional parameters for the Dashscope API.
+        """
+
+        if model is None:
+            raise ModelRequired('Model is required!')
+        if url is None:
+            url = f'wss://dashscope.aliyuncs.com/api-ws/v1/realtime?model={model}'
+        else:
+            url = f'{url}?model={model}'
+        self.url = url
+        self.apikey = dashscope.api_key
+        self.user_headers = headers
+        self.user_workspace = workspace
+        self.model = model
+        self.config = {}
+        self.callback = callback
+        self.ws = None
+        self.session_id = None
+        self.last_message = None
+        self.last_response_id = None
+        self.last_first_text_time = None
+        self.last_first_audio_delay = None
+        self.metrics = []
+
+    def _generate_event_id(self):
+        '''
+        generate random event id: event_xxxx
+        '''
+        return 'event_' + uuid.uuid4().hex
+
+    def _get_websocket_header(self, ):
+        ua = 'dashscope/%s; python/%s; platform/%s; processor/%s' % (
+            '1.18.0',  # dashscope version
+            platform.python_version(),
+            platform.platform(),
+            platform.processor(),
+        )
+        headers = {
+            'user-agent': ua,
+            'Authorization': 'bearer ' + self.apikey,
+        }
+        if self.user_headers:
+            headers = {**self.user_headers, **headers}
+        if self.user_workspace:
+            headers = {
+                **headers,
+                'X-DashScope-WorkSpace': self.user_workspace,
+            }
+        return headers
+
+    def connect(self) -> None:
+        '''
+        connect to server, create session and return default session configuration
+        '''
+        self.ws = websocket.WebSocketApp(
+            self.url,
+            header=self._get_websocket_header(),
+            on_message=self.on_message,
+            on_error=self.on_error,
+            on_close=self.on_close,
+        )
+        self.thread = threading.Thread(target=self.ws.run_forever)
+        self.thread.daemon = True
+        self.thread.start()
+        timeout = 5  # 最长等待时间（秒）
+        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
+        if not (self.ws.sock and self.ws.sock.connected):
+            raise TimeoutError(
+                'websocket connection could not established within 5s. '
+                'Please check your network connection, firewall settings, or server status.'
+            )
+        self.callback.on_open()
+
+    def __send_str(self, data: str, enable_log: bool = True):
+        if enable_log:
+            logger.debug('[qwen tts realtime] send string: {}'.format(data))
+        self.ws.send(data)
+
+    def update_session(self,
+                       voice: str,
+                       response_format: AudioFormat = AudioFormat.
+                       PCM_24000HZ_MONO_16BIT,
+                       mode: str = 'server_commit',
+                       **kwargs) -> None:
+        '''
+        update session configuration, should be used before create response
+
+        Parameters
+        ----------
+        voice: str
+            voice to be used in session
+        response_format: AudioFormat
+            output audio format
+        mode: str
+            response mode, server_commit or commit
+        '''
+        self.config = {
+            'voice': voice,
+            'mode': mode,
+            'response_format': response_format.format,
+            'sample_rate': response_format.sample_rate,
+        }
+        self.config.update(kwargs)
+        self.__send_str(
+            json.dumps({
+                'event_id': self._generate_event_id(),
+                'type': 'session.update',
+                'session': self.config
+            }))
+
+    def append_text(self, text: str) -> None:
+        '''
+        send text
+
+        Parameters
+        ----------
+        text: str
+            text to send
+        '''
+        self.__send_str(
+            json.dumps({
+                'event_id': self._generate_event_id(),
+                'type': 'input_text_buffer.append',
+                'text': text
+            }))
+        if self.last_first_text_time is None:
+            self.last_first_text_time = time.time() * 1000
+
+    def commit(self, ) -> None:
+        '''
+        commit the text sent before, create response and start synthesis audio.
+        '''
+        self.__send_str(
+            json.dumps({
+                'event_id': self._generate_event_id(),
+                'type': 'input_text_buffer.commit'
+            }))
+
+    def clear_appended_text(self, ) -> None:
+        '''
+        clear the text sent to server before.
+        '''
+        self.__send_str(
+            json.dumps({
+                'event_id': self._generate_event_id(),
+                'type': 'input_text_buffer.clear'
+            }))
+
+    def cancel_response(self, ) -> None:
+        '''
+        cancel the current response
+        '''
+        self.__send_str(
+            json.dumps({
+                'event_id': self._generate_event_id(),
+                'type': 'response.cancel'
+            }))
+
+    def send_raw(self, raw_data: str) -> None:
+        '''
+        send raw data to server
+        '''
+        self.__send_str(raw_data)
+
+    def finish(self, ) -> None:
+        '''
+        finish input text stream, server will synthesis all text in buffer and close the connection
+        '''
+        self.__send_str(
+            json.dumps({
+                'event_id': self._generate_event_id(),
+                'type': 'session.finish'
+            }))
+
+    def close(self, ) -> None:
+        '''
+        close the connection to server
+        '''
+        self.ws.close()
+
+    # 监听消息的回调函数
+    def on_message(self, ws, message):
+        if isinstance(message, str):
+            logger.debug('[omni realtime] receive string {}'.format(
+                message[:1024]))
+            try:
+                # 尝试将消息解析为JSON
+                json_data = json.loads(message)
+                self.last_message = json_data
+                self.callback.on_event(json_data)
+                if 'type' in message:
+                    if 'session.created' == json_data['type']:
+                        self.session_id = json_data['session']['id']
+                    if 'response.created' == json_data['type']:
+                        self.last_response_id = json_data['response']['id']
+                    elif 'response.audio.delta' == json_data['type']:
+                        if self.last_first_text_time and self.last_first_audio_delay is None:
+                            self.last_first_audio_delay = time.time(
+                            ) * 1000 - self.last_first_text_time
+                    elif 'response.done' == json_data['type']:
+                        logger.debug(
+                            '[Metric] response: {}, first audio delay: {}'
+                            .format(self.last_response_id,
+                                    self.last_first_audio_delay))
+            except json.JSONDecodeError:
+                logger.error('Failed to parse message as JSON.')
+                raise Exception('Failed to parse message as JSON.')
+        elif isinstance(message, (bytes, bytearray)):
+            # 如果失败，认为是二进制消息
+            logger.error(
+                'should not receive binary message in omni realtime api')
+            logger.debug('[omni realtime] receive binary {} bytes'.format(
+                len(message)))
+
+    def on_close(self, ws, close_status_code, close_msg):
+        logger.debug(
+            '[omni realtime] connection closed with code {} and message {}'.format(
+                close_status_code, close_msg))
+        self.callback.on_close(close_status_code, close_msg)
+
+    # WebSocket发生错误的回调函数
+    def on_error(self, ws, error):
+        print(f'websocket closed due to {error}')
+        raise Exception(f'websocket closed due to {error}')
+
+    # 获取上一个任务的taskId
+    def get_session_id(self):
+        return self.session_id
+
+    def get_last_message(self):
+        return self.last_message
+
+    def get_last_response_id(self):
+        return self.last_response_id
+
+    def get_first_audio_delay(self):
+        return self.last_first_audio_delay

dashscope/embeddings/text_embedding.py

@@ -15,6 +15,7 @@ class TextEmbedding(BaseApi):
         text_embedding_v1 = 'text-embedding-v1'
         text_embedding_v2 = 'text-embedding-v2'
         text_embedding_v3 = 'text-embedding-v3'
+        text_embedding_v4 = 'text-embedding-v4'
 
     @classmethod
     def call(cls,

dashscope/multimodal/multimodal_request_params.py

@@ -83,12 +83,15 @@ class Upstream:
     # sample_rate: int  # 合成音频采样率
 
     def to_dict(self):
-        return {
+        upstream: dict = {
             "type": self.type,
             "mode": self.mode,
             "audio_format": self.audio_format,
             # "sample_rate": self.sample_rate
         }
+        if self.pass_through_params is not None:
+            upstream.update(self.pass_through_params)
+        return upstream
 
 
 @dataclass
@@ -105,6 +108,7 @@ class Downstream:
     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
+    pass_through_params: dict = field(default=None)
 
     def to_dict(self):
         stream: dict = {
@@ -120,6 +124,8 @@ class Downstream:
             stream["voice"] = self.voice
         if self.sample_rate != 0:
             stream["sample_rate"] = self.sample_rate
+        if self.pass_through_params is not None:
+            stream.update(self.pass_through_params)
         return stream
 
 
@@ -199,6 +205,7 @@ class BizParams:
     user_prompt_params: dict = field(default=None)
     user_query_params: dict = field(default=None)
     videos: list = field(default=None)
+    pass_through_params: dict = field(default=None)
 
     def to_dict(self):
         params = {}
@@ -214,6 +221,8 @@ class BizParams:
             params["user_query_params"] = self.user_query_params
         if self.videos is not None:
             params["videos"] = self.videos
+        if self.pass_through_params is not None:
+            params.update(self.pass_through_params)
         return params
 
 

dashscope/version.py

@@ -1,3 +1,3 @@
 # Copyright (c) Alibaba, Inc. and its affiliates.
 
-__version__ = '1.23.7'
+__version__ = '1.23.9'

{dashscope-1.23.7.dist-info → dashscope-1.23.9.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dashscope
-Version: 1.23.7
+Version: 1.23.9
 Summary: dashscope client sdk library
 Home-page: https://dashscope.aliyun.com/
 Author: Alibaba Cloud

{dashscope-1.23.7.dist-info → dashscope-1.23.9.dist-info}/RECORD

@@ -3,7 +3,7 @@ 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=HwgMDxa-K4tdaSJnt84_qDc7qTyV4eqR_QjdjT4ulxc,74
+dashscope/version.py,sha256=ACHUqhomdzZ7aih_Y368NUmXBBfj2GY5GqSf2NoTta4,74
 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
@@ -27,17 +27,21 @@ dashscope/app/application.py,sha256=Whf_ij4RHOaY12_xdS8uj8HVNCwkTp_MRdrFTryF1Kg,
 dashscope/app/application_response.py,sha256=z9BKcb9QuV_TzHkqtwaCwap2GQ1mP48uDD7gIc2il98,7038
 dashscope/assistants/__init__.py,sha256=hjCTuv13yFaXyUqlexAU-RaO0Ahq3P7VK9_LkSbkGVU,434
 dashscope/assistants/assistant_types.py,sha256=DQ_lOust10wjiV38Nlsu3HaGYNbGlJoaLjEza82Wotk,4262
-dashscope/assistants/assistants.py,sha256=_ADX4s8Mlfhfq7kb8xMLztNX_09NAW7DouvbVLY0jpw,10884
+dashscope/assistants/assistants.py,sha256=t8BGu9K2YbNpBenUyFFDMx5WpZ5x5OtbssJkj0xfVo0,10930
 dashscope/assistants/files.py,sha256=Ol2h7L2vNV8kgWqum2B-3B9vtLEHB_2KWt0K7e96Bmg,6750
-dashscope/audio/__init__.py,sha256=RKw-A1PLA5F4jiXI3AnuRSYeHFzQl2OW2ZGW6xyHS1Q,132
+dashscope/audio/__init__.py,sha256=7e3ejVsDJxEbMHN-9E0nEDfU-CnnQ4JgtgUxqNs0IG4,192
 dashscope/audio/asr/__init__.py,sha256=JoCenJAUVOQXPmAn1toKeFYCfc8BqNn0NKpqjuJvNJc,1055
 dashscope/audio/asr/asr_phrase_manager.py,sha256=vHOLExaKCtjedkihIu7gyfQyarR9rN5JZn79LvlCpco,7693
 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=N0pMS2x1lDxqJ14FgTGKctfuVkR2_hlEsCNWFcgYpTY,6717
+dashscope/audio/qwen_omni/__init__.py,sha256=MEFxmyxr5H6bW22l_R9073Pl6Ka6knvhrATGT-4UBjI,298
+dashscope/audio/qwen_omni/omni_realtime.py,sha256=eBmoOxuKcfzMHuXsQWCrIIKmso9iEzYylOeYZ5upv-w,14869
 dashscope/audio/qwen_tts/__init__.py,sha256=JS3axY1grqO0aTIJufZ3KS1JsU6yf6y4K2CQlNvUK9I,132
 dashscope/audio/qwen_tts/speech_synthesizer.py,sha256=7LHR-PXhn-VE1cCOp_82Jq0zE9rMc3xy3dszUeyLLNs,2927
+dashscope/audio/qwen_tts_realtime/__init__.py,sha256=vVkmeJr_mEAn_O0Rh5AU3ICg6qIZqppUryJ5lY8VYPo,254
+dashscope/audio/qwen_tts_realtime/qwen_tts_realtime.py,sha256=8bOAMcDasTHwSLb9xAGJoj9eUPpQHh2aWvonV6Kf1U4,10367
 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
@@ -62,14 +66,14 @@ dashscope/embeddings/__init__.py,sha256=XQ7vKr8oZM2CmdOduE53BWy6_Qpn9xUPkma64yw8
 dashscope/embeddings/batch_text_embedding.py,sha256=lVhvTS8McYfXuqt_8CmmhA6bPqD0nrGv965kjYG_j0E,8842
 dashscope/embeddings/batch_text_embedding_response.py,sha256=ZfkJMUq8GRsFA6XUTsiAsIySqGJH-VPi2P9Ba1KTU-s,2056
 dashscope/embeddings/multimodal_embedding.py,sha256=NwjQsdkKgUz51ozGjqFDzVlLcZjY0m1JNdH1EyAY0a4,4109
-dashscope/embeddings/text_embedding.py,sha256=4W1V-Ggj0WJhY5MdP2xoUTteXlWk8TUtI6y2gRUVCUk,2060
+dashscope/embeddings/text_embedding.py,sha256=2MPEyMB99xueDbvFg9kKAe8bgHMDEaFLaFa6GzDWDHg,2108
 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=z_QVq01E43FAqKQnDu9vdf89d1zuYlWyANewWTEXVJM,1282
 dashscope/multimodal/multimodal_dialog.py,sha256=HymlaQYp7SgJdoKbT27SNiviyRRoM91zklNBwTHmm1Q,23939
-dashscope/multimodal/multimodal_request_params.py,sha256=7A4UhsbYjcX7aAJwWI1xZEt0e1bSgPcu5pJAinaZyx0,7907
+dashscope/multimodal/multimodal_request_params.py,sha256=9Dlvyy0u67K5FtMfWkCRLHKsevTM8jvT2V-OljZP5sM,8350
 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
@@ -93,9 +97,9 @@ 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-1.23.7.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-dashscope-1.23.7.dist-info/METADATA,sha256=JxFpTi5zU9f1O1uLoTH1OoV6iV3K3aAqshhD1MTDtFY,7123
-dashscope-1.23.7.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-dashscope-1.23.7.dist-info/entry_points.txt,sha256=e9C3sOf9zDYL0O5ROEGX6FT8w-QK_kaGRWmPZDHAFys,49
-dashscope-1.23.7.dist-info/top_level.txt,sha256=woqavFJK9zas5xTqynmALqOtlafghjsk63Xk86powTU,10
-dashscope-1.23.7.dist-info/RECORD,,
+dashscope-1.23.9.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+dashscope-1.23.9.dist-info/METADATA,sha256=tQQhkm5TxNwI2TCmWzT11dwWd9UGFb7cI1cojbo9Dpw,7123
+dashscope-1.23.9.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+dashscope-1.23.9.dist-info/entry_points.txt,sha256=e9C3sOf9zDYL0O5ROEGX6FT8w-QK_kaGRWmPZDHAFys,49
+dashscope-1.23.9.dist-info/top_level.txt,sha256=woqavFJK9zas5xTqynmALqOtlafghjsk63Xk86powTU,10
+dashscope-1.23.9.dist-info/RECORD,,