feat: Support Azure image tts stt model

apps/setting/models_provider/impl/azure_model_provider/azure_model_provider.py

@@ -12,13 +12,25 @@ from common.util.file_util import get_file_content
 from setting.models_provider.base_model_provider import IModelProvider, ModelProvideInfo, ModelInfo, \
     ModelTypeConst, ModelInfoManage
 from setting.models_provider.impl.azure_model_provider.credential.embedding import AzureOpenAIEmbeddingCredential
+from setting.models_provider.impl.azure_model_provider.credential.image import AzureOpenAIImageModelCredential
 from setting.models_provider.impl.azure_model_provider.credential.llm import AzureLLMModelCredential
+from setting.models_provider.impl.azure_model_provider.credential.stt import AzureOpenAISTTModelCredential
+from setting.models_provider.impl.azure_model_provider.credential.tti import AzureOpenAITextToImageModelCredential
+from setting.models_provider.impl.azure_model_provider.credential.tts import AzureOpenAITTSModelCredential
 from setting.models_provider.impl.azure_model_provider.model.azure_chat_model import AzureChatModel
 from setting.models_provider.impl.azure_model_provider.model.embedding import AzureOpenAIEmbeddingModel
+from setting.models_provider.impl.azure_model_provider.model.image import AzureOpenAIImage
+from setting.models_provider.impl.azure_model_provider.model.stt import AzureOpenAISpeechToText
+from setting.models_provider.impl.azure_model_provider.model.tti import AzureOpenAITextToImage
+from setting.models_provider.impl.azure_model_provider.model.tts import AzureOpenAITextToSpeech
 from smartdoc.conf import PROJECT_DIR
 
 base_azure_llm_model_credential = AzureLLMModelCredential()
 base_azure_embedding_model_credential = AzureOpenAIEmbeddingCredential()
+base_azure_image_model_credential = AzureOpenAIImageModelCredential()
+base_azure_tti_model_credential = AzureOpenAITextToImageModelCredential()
+base_azure_tts_model_credential = AzureOpenAITTSModelCredential()
+base_azure_stt_model_credential = AzureOpenAISTTModelCredential()
 
 default_model_info = ModelInfo('Azure OpenAI', '具体的基础模型由部署名决定', ModelTypeConst.LLM,
                                base_azure_llm_model_credential, AzureChatModel, api_version='2024-02-15-preview'
@@ -36,12 +48,47 @@ embedding_model_info = [
               ),
 ]
 
+image_model_info = [
+    ModelInfo('gpt-4o', '', ModelTypeConst.IMAGE,
+              base_azure_image_model_credential, AzureOpenAIImage, api_version='2023-05-15'
+              ),
+    ModelInfo('gpt-4o-mini', '', ModelTypeConst.IMAGE,
+              base_azure_image_model_credential, AzureOpenAIImage, api_version='2023-05-15'
+              ),
+]
+
+tti_model_info = [
+    ModelInfo('dall-e-3', '', ModelTypeConst.TTI,
+              base_azure_tti_model_credential, AzureOpenAITextToImage, api_version='2023-05-15'
+              ),
+]
+
+tts_model_info = [
+    ModelInfo('tts', '', ModelTypeConst.TTS,
+              base_azure_tts_model_credential, AzureOpenAITextToSpeech, api_version='2023-05-15'
+              ),
+]
+
+stt_model_info = [
+    ModelInfo('whisper', '', ModelTypeConst.STT,
+              base_azure_stt_model_credential, AzureOpenAISpeechToText, api_version='2023-05-15'
+              ),
+]
+
 model_info_manage = (
     ModelInfoManage.builder()
     .append_default_model_info(default_model_info)
     .append_model_info(default_model_info)
     .append_model_info_list(embedding_model_info)
     .append_default_model_info(embedding_model_info[0])
+    .append_model_info_list(image_model_info)
+    .append_default_model_info(image_model_info[0])
+    .append_model_info_list(stt_model_info)
+    .append_default_model_info(stt_model_info[0])
+    .append_model_info_list(tts_model_info)
+    .append_default_model_info(tts_model_info[0])
+    .append_model_info_list(tti_model_info)
+    .append_default_model_info(tti_model_info[0])
     .build()
 )
 

apps/setting/models_provider/impl/azure_model_provider/credential/image.py

@@ -0,0 +1,66 @@
+# coding=utf-8
+import base64
+import os
+from typing import Dict
+
+from langchain_core.messages import HumanMessage
+
+from common import forms
+from common.exception.app_exception import AppApiException
+from common.forms import BaseForm, TooltipLabel
+from setting.models_provider.base_model_provider import BaseModelCredential, ValidCode
+
+class AzureOpenAIImageModelParams(BaseForm):
+    temperature = forms.SliderField(TooltipLabel('温度', '较高的数值会使输出更加随机，而较低的数值会使其更加集中和确定'),
+                                    required=True, default_value=0.7,
+                                    _min=0.1,
+                                    _max=1.0,
+                                    _step=0.01,
+                                    precision=2)
+
+    max_tokens = forms.SliderField(
+        TooltipLabel('输出最大Tokens', '指定模型可生成的最大token个数'),
+        required=True, default_value=800,
+        _min=1,
+        _max=100000,
+        _step=1,
+        precision=0)
+
+
+
+class AzureOpenAIImageModelCredential(BaseForm, BaseModelCredential):
+    api_version = forms.TextInputField("API 版本 (api_version)", required=True)
+    api_base = forms.TextInputField('API 域名 (azure_endpoint)', required=True)
+    api_key = forms.PasswordInputField("API Key (api_key)", required=True)
+
+    def is_valid(self, model_type: str, model_name, model_credential: Dict[str, object], provider,
+                 raise_exception=False):
+        model_type_list = provider.get_model_type_list()
+        if not any(list(filter(lambda mt: mt.get('value') == model_type, model_type_list))):
+            raise AppApiException(ValidCode.valid_error.value, f'{model_type} 模型类型不支持')
+
+        for key in ['api_base', 'api_key', 'api_version']:
+            if key not in model_credential:
+                if raise_exception:
+                    raise AppApiException(ValidCode.valid_error.value, f'{key} 字段为必填字段')
+                else:
+                    return False
+        try:
+            model = provider.get_model(model_type, model_name, model_credential)
+            res = model.stream([HumanMessage(content=[{"type": "text", "text": "你好"}])])
+            for chunk in res:
+                print(chunk)
+        except Exception as e:
+            if isinstance(e, AppApiException):
+                raise e
+            if raise_exception:
+                raise AppApiException(ValidCode.valid_error.value, f'校验失败,请检查参数是否正确: {str(e)}')
+            else:
+                return False
+        return True
+
+    def encryption_dict(self, model: Dict[str, object]):
+        return {**model, 'api_key': super().encryption(model.get('api_key', ''))}
+
+    def get_model_params_setting_form(self, model_name):
+        return AzureOpenAIImageModelParams()

apps/setting/models_provider/impl/azure_model_provider/credential/stt.py

@@ -0,0 +1,43 @@
+# coding=utf-8
+from typing import Dict
+
+from common import forms
+from common.exception.app_exception import AppApiException
+from common.forms import BaseForm
+from setting.models_provider.base_model_provider import BaseModelCredential, ValidCode
+
+
+class AzureOpenAISTTModelCredential(BaseForm, BaseModelCredential):
+    api_version = forms.TextInputField("API 版本 (api_version)", required=True)
+    api_base = forms.TextInputField('API 域名 (azure_endpoint)', required=True)
+    api_key = forms.PasswordInputField("API Key (api_key)", required=True)
+
+    def is_valid(self, model_type: str, model_name, model_credential: Dict[str, object], provider,
+                 raise_exception=False):
+        model_type_list = provider.get_model_type_list()
+        if not any(list(filter(lambda mt: mt.get('value') == model_type, model_type_list))):
+            raise AppApiException(ValidCode.valid_error.value, f'{model_type} 模型类型不支持')
+
+        for key in ['api_base', 'api_key', 'api_version']:
+            if key not in model_credential:
+                if raise_exception:
+                    raise AppApiException(ValidCode.valid_error.value, f'{key} 字段为必填字段')
+                else:
+                    return False
+        try:
+            model = provider.get_model(model_type, model_name, model_credential)
+            model.check_auth()
+        except Exception as e:
+            if isinstance(e, AppApiException):
+                raise e
+            if raise_exception:
+                raise AppApiException(ValidCode.valid_error.value, f'校验失败,请检查参数是否正确: {str(e)}')
+            else:
+                return False
+        return True
+
+    def encryption_dict(self, model: Dict[str, object]):
+        return {**model, 'api_key': super().encryption(model.get('api_key', ''))}
+
+    def get_model_params_setting_form(self, model_name):
+        pass

apps/setting/models_provider/impl/azure_model_provider/credential/tti.py

@@ -0,0 +1,83 @@
+# coding=utf-8
+import base64
+import os
+from typing import Dict
+
+from langchain_core.messages import HumanMessage
+
+from common import forms
+from common.exception.app_exception import AppApiException
+from common.forms import BaseForm, TooltipLabel
+from setting.models_provider.base_model_provider import BaseModelCredential, ValidCode
+
+
+class AzureOpenAITTIModelParams(BaseForm):
+    size = forms.SingleSelect(
+        TooltipLabel('图片尺寸', '指定生成图片的尺寸, 如: 1024x1024'),
+        required=True,
+        default_value='1024x1024',
+        option_list=[
+            {'value': '1024x1024', 'label': '1024x1024'},
+            {'value': '1024x1792', 'label': '1024x1792'},
+            {'value': '1792x1024', 'label': '1792x1024'},
+        ],
+        text_field='label',
+        value_field='value'
+    )
+
+    quality = forms.SingleSelect(
+        TooltipLabel('图片质量', ''),
+        required=True,
+        default_value='standard',
+        option_list=[
+            {'value': 'standard', 'label': 'standard'},
+            {'value': 'hd', 'label': 'hd'},
+        ],
+        text_field='label',
+        value_field='value'
+    )
+
+    n = forms.SliderField(
+        TooltipLabel('图片数量', '指定生成图片的数量'),
+        required=True, default_value=1,
+        _min=1,
+        _max=10,
+        _step=1,
+        precision=0)
+
+
+class AzureOpenAITextToImageModelCredential(BaseForm, BaseModelCredential):
+    api_version = forms.TextInputField("API 版本 (api_version)", required=True)
+    api_base = forms.TextInputField('API 域名 (azure_endpoint)', required=True)
+    api_key = forms.PasswordInputField("API Key (api_key)", required=True)
+
+    def is_valid(self, model_type: str, model_name, model_credential: Dict[str, object], provider,
+                 raise_exception=False):
+        model_type_list = provider.get_model_type_list()
+        if not any(list(filter(lambda mt: mt.get('value') == model_type, model_type_list))):
+            raise AppApiException(ValidCode.valid_error.value, f'{model_type} 模型类型不支持')
+
+        for key in ['api_base', 'api_key', 'api_version']:
+            if key not in model_credential:
+                if raise_exception:
+                    raise AppApiException(ValidCode.valid_error.value, f'{key} 字段为必填字段')
+                else:
+                    return False
+        try:
+            model = provider.get_model(model_type, model_name, model_credential)
+            res = model.check_auth()
+            print(res)
+        except Exception as e:
+            if isinstance(e, AppApiException):
+                raise e
+            if raise_exception:
+                raise AppApiException(ValidCode.valid_error.value, f'校验失败,请检查参数是否正确: {str(e)}')
+            else:
+                return False
+        return True
+
+    def encryption_dict(self, model: Dict[str, object]):
+        return {**model, 'api_key': super().encryption(model.get('api_key', ''))}
+
+    def get_model_params_setting_form(self, model_name):
+        return AzureOpenAITTIModelParams()

apps/setting/models_provider/impl/azure_model_provider/credential/tts.py

@@ -0,0 +1,59 @@
+# coding=utf-8
+from typing import Dict
+
+from common import forms
+from common.exception.app_exception import AppApiException
+from common.forms import BaseForm, TooltipLabel
+from setting.models_provider.base_model_provider import BaseModelCredential, ValidCode
+
+class AzureOpenAITTSModelGeneralParams(BaseForm):
+    # alloy, echo, fable, onyx, nova, shimmer
+    voice = forms.SingleSelect(
+        TooltipLabel('Voice', '尝试不同的声音（合金、回声、寓言、缟玛瑙、新星和闪光），找到一种适合您所需的音调和听众的声音。当前的语音针对英语进行了优化。'),
+        required=True, default_value='alloy',
+        text_field='value',
+        value_field='value',
+        option_list=[
+            {'text': 'alloy', 'value': 'alloy'},
+            {'text': 'echo', 'value': 'echo'},
+            {'text': 'fable', 'value': 'fable'},
+            {'text': 'onyx', 'value': 'onyx'},
+            {'text': 'nova', 'value': 'nova'},
+            {'text': 'shimmer', 'value': 'shimmer'},
+        ])
+
+
+class AzureOpenAITTSModelCredential(BaseForm, BaseModelCredential):
+    api_version = forms.TextInputField("API 版本 (api_version)", required=True)
+    api_base = forms.TextInputField('API 域名 (azure_endpoint)', required=True)
+    api_key = forms.PasswordInputField("API Key (api_key)", required=True)
+
+    def is_valid(self, model_type: str, model_name, model_credential: Dict[str, object], provider,
+                 raise_exception=False):
+        model_type_list = provider.get_model_type_list()
+        if not any(list(filter(lambda mt: mt.get('value') == model_type, model_type_list))):
+            raise AppApiException(ValidCode.valid_error.value, f'{model_type} 模型类型不支持')
+
+        for key in ['api_base', 'api_key', 'api_version']:
+            if key not in model_credential:
+                if raise_exception:
+                    raise AppApiException(ValidCode.valid_error.value, f'{key} 字段为必填字段')
+                else:
+                    return False
+        try:
+            model = provider.get_model(model_type, model_name, model_credential)
+            model.check_auth()
+        except Exception as e:
+            if isinstance(e, AppApiException):
+                raise e
+            if raise_exception:
+                raise AppApiException(ValidCode.valid_error.value, f'校验失败,请检查参数是否正确: {str(e)}')
+            else:
+                return False
+        return True
+
+    def encryption_dict(self, model: Dict[str, object]):
+        return {**model, 'api_key': super().encryption(model.get('api_key', ''))}
+
+    def get_model_params_setting_form(self, model_name):
+        return AzureOpenAITTSModelGeneralParams()

apps/setting/models_provider/impl/azure_model_provider/model/image.py

@@ -0,0 +1,28 @@
+from typing import Dict
+
+from langchain_openai import AzureChatOpenAI
+from langchain_openai.chat_models import ChatOpenAI
+
+from common.config.tokenizer_manage_config import TokenizerManage
+from setting.models_provider.base_model_provider import MaxKBBaseModel
+
+
+def custom_get_token_ids(text: str):
+    tokenizer = TokenizerManage.get_tokenizer()
+    return tokenizer.encode(text)
+
+
+class AzureOpenAIImage(MaxKBBaseModel, AzureChatOpenAI):
+
+    @staticmethod
+    def new_instance(model_type, model_name, model_credential: Dict[str, object], **model_kwargs):
+        optional_params = MaxKBBaseModel.filter_optional_params(model_kwargs)
+        return AzureOpenAIImage(
+            model_name=model_name,
+            openai_api_key=model_credential.get('api_key'),
+            azure_endpoint=model_credential.get('api_base'),
+            openai_api_version=model_credential.get('api_version'),
+            openai_api_type="azure",
+            streaming=True,
+            **optional_params,
+        )

apps/setting/models_provider/impl/azure_model_provider/model/stt.py

@@ -0,0 +1,64 @@
+import asyncio
+import io
+from typing import Dict
+
+from openai import OpenAI, AzureOpenAI
+
+from common.config.tokenizer_manage_config import TokenizerManage
+from setting.models_provider.base_model_provider import MaxKBBaseModel
+from setting.models_provider.impl.base_stt import BaseSpeechToText
+
+
+def custom_get_token_ids(text: str):
+    tokenizer = TokenizerManage.get_tokenizer()
+    return tokenizer.encode(text)
+
+
+class AzureOpenAISpeechToText(MaxKBBaseModel, BaseSpeechToText):
+    api_base: str
+    api_key: str
+    api_version: str
+    model: str
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+        self.api_key = kwargs.get('api_key')
+        self.api_base = kwargs.get('api_base')
+        self.api_version = kwargs.get('api_version')
+
+    @staticmethod
+    def new_instance(model_type, model_name, model_credential: Dict[str, object], **model_kwargs):
+        optional_params = {}
+        if 'max_tokens' in model_kwargs and model_kwargs['max_tokens'] is not None:
+            optional_params['max_tokens'] = model_kwargs['max_tokens']
+        if 'temperature' in model_kwargs and model_kwargs['temperature'] is not None:
+            optional_params['temperature'] = model_kwargs['temperature']
+        return AzureOpenAISpeechToText(
+            model=model_name,
+            api_base=model_credential.get('api_base'),
+            api_key=model_credential.get('api_key'),
+            api_version=model_credential.get('api_version'),
+            **optional_params,
+        )
+
+    def check_auth(self):
+        client = AzureOpenAI(
+            azure_endpoint=self.api_base,
+            api_key=self.api_key,
+            api_version=self.api_version
+        )
+        response_list = client.models.with_raw_response.list()
+        # print(response_list)
+
+    def speech_to_text(self, audio_file):
+        client = AzureOpenAI(
+            azure_endpoint=self.api_base,
+            api_key=self.api_key,
+            api_version=self.api_version
+        )
+        audio_data = audio_file.read()
+        buffer = io.BytesIO(audio_data)
+        buffer.name = "file.mp3"  # this is the important line
+        res = client.audio.transcriptions.create(model=self.model, language="zh", file=buffer)
+        return res.text
+

apps/setting/models_provider/impl/azure_model_provider/model/tti.py

@@ -0,0 +1,61 @@
+from typing import Dict
+
+from openai import OpenAI, AzureOpenAI
+
+from common.config.tokenizer_manage_config import TokenizerManage
+from setting.models_provider.base_model_provider import MaxKBBaseModel
+from setting.models_provider.impl.base_tti import BaseTextToImage
+
+
+def custom_get_token_ids(text: str):
+    tokenizer = TokenizerManage.get_tokenizer()
+    return tokenizer.encode(text)
+
+
+class AzureOpenAITextToImage(MaxKBBaseModel, BaseTextToImage):
+    api_base: str
+    api_key: str
+    api_version: str
+    model: str
+    params: dict
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+        self.api_key = kwargs.get('api_key')
+        self.api_base = kwargs.get('api_base')
+        self.api_version = kwargs.get('api_version')
+        self.model = kwargs.get('model')
+        self.params = kwargs.get('params')
+
+    @staticmethod
+    def new_instance(model_type, model_name, model_credential: Dict[str, object], **model_kwargs):
+        optional_params = {'params': {'size': '1024x1024', 'quality': 'standard', 'n': 1}}
+        for key, value in model_kwargs.items():
+            if key not in ['model_id', 'use_local', 'streaming']:
+                optional_params['params'][key] = value
+        return AzureOpenAITextToImage(
+            model=model_name,
+            api_base=model_credential.get('api_base'),
+            api_key=model_credential.get('api_key'),
+            api_version=model_credential.get('api_version'),
+            **optional_params,
+        )
+
+    def is_cache_model(self):
+        return False
+
+    def check_auth(self):
+        chat = AzureOpenAI(api_key=self.api_key, azure_endpoint=self.api_base, api_version=self.api_version)
+        response_list = chat.models.with_raw_response.list()
+
+        # self.generate_image('生成一个小猫图片')
+
+    def generate_image(self, prompt: str, negative_prompt: str = None):
+        chat = AzureOpenAI(api_key=self.api_key, azure_endpoint=self.api_base, api_version=self.api_version)
+        res = chat.images.generate(model=self.model, prompt=prompt, **self.params)
+        file_urls = []
+        for content in res.data:
+            url = content.url
+            file_urls.append(url)
+
+        return file_urls

apps/setting/models_provider/impl/azure_model_provider/model/tts.py

@@ -0,0 +1,69 @@
+from typing import Dict
+
+from dashscope import api_key
+from dashscope.common.env import api_version
+from openai import OpenAI, AzureOpenAI
+
+from common.config.tokenizer_manage_config import TokenizerManage
+from setting.models_provider.base_model_provider import MaxKBBaseModel
+from setting.models_provider.impl.base_tts import BaseTextToSpeech
+
+
+def custom_get_token_ids(text: str):
+    tokenizer = TokenizerManage.get_tokenizer()
+    return tokenizer.encode(text)
+
+
+class AzureOpenAITextToSpeech(MaxKBBaseModel, BaseTextToSpeech):
+    api_base: str
+    api_key: str
+    api_version: str
+    model: str
+    params: dict
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+        self.api_key = kwargs.get('api_key')
+        self.api_base = kwargs.get('api_base')
+        self.api_version = kwargs.get('api_version')
+        self.model = kwargs.get('model')
+        self.params = kwargs.get('params')
+
+    @staticmethod
+    def new_instance(model_type, model_name, model_credential: Dict[str, object], **model_kwargs):
+        optional_params = {'params': {'voice': 'alloy'}}
+        for key, value in model_kwargs.items():
+            if key not in ['model_id', 'use_local', 'streaming']:
+                optional_params['params'][key] = value
+        return AzureOpenAITextToSpeech(
+            model=model_name,
+            api_base=model_credential.get('api_base'),
+            api_key=model_credential.get('api_key'),
+            api_version=model_credential.get('api_version'),
+            **optional_params,
+        )
+
+    def check_auth(self):
+        client = AzureOpenAI(
+            azure_endpoint=self.api_base,
+            api_key=self.api_key,
+            api_version=self.api_version
+        )
+        response_list = client.models.with_raw_response.list()
+        # print(response_list)
+
+    def text_to_speech(self, text):
+        client = AzureOpenAI(
+            azure_endpoint=self.api_base,
+            api_key=self.api_key,
+            api_version=self.api_version
+        )
+        with client.audio.speech.with_streaming_response.create(
+                model=self.model,
+                input=text,
+                **self.params
+        ) as response:
+            return response.read()
+
+    def is_cache_model(self):
+        return False