删除了add_file_content

src/llm_models/model_client/base_client.py

@@ -117,13 +117,14 @@ class BaseClient:
     async def get_audio_transcriptions(
         self,
         model_info: ModelInfo,
-        message_list: list[Message],
+        audio_base64: str,
         extra_params: dict[str, Any] | None = None,
     ) -> APIResponse:
         """
         获取音频转录
         :param model_info: 模型信息
-        :param message_list: 消息列表，包含音频内容
+        :param audio_base64: base64编码的音频数据
+        :extra_params: 附加的请求参数
         :return: 音频转录响应
         """
         raise RuntimeError("This method should be overridden in subclasses")

src/llm_models/model_client/openai_client.py

@@ -2,6 +2,7 @@ import asyncio
 import io
 import json
 import re
+import base64
 from collections.abc import Iterable
 from typing import Callable, Any, Coroutine, Optional
 from json_repair import repair_json
@@ -536,19 +537,20 @@ class OpenaiClient(BaseClient):
     async def get_audio_transcriptions(
         self,
         model_info: ModelInfo,
-        message_list: list[Message],
+        audio_base64: str,
         extra_params: dict[str, Any] | None = None,
     ) -> APIResponse:
         """
         获取音频转录
         :param model_info: 模型信息
-        :param message_list: 消息列表，包含音频内容
+        :param audio_base64: base64编码的音频数据
-        :return: 转录响应
+        :extra_params: 附加的请求参数
+        :return: 音频转录响应
         """
         try:
             raw_response = await self.client.audio.transcriptions.create(
                 model=model_info.model_identifier,
-                file=message_list[0].content[0],
+                file=("audio.wav", io.BytesIO(base64.b64decode(audio_base64))),
                 extra_body=extra_params
             )
         except APIConnectionError as e:

src/llm_models/payload_content/message.py

@@ -1,6 +1,5 @@
-import base64
 from enum import Enum
-from io import BytesIO
+
 
 # 设计这系列类的目的是为未来可能的扩展做准备
 
@@ -34,7 +33,7 @@ class Message:
 class MessageBuilder:
     def __init__(self):
         self.__role: RoleType = RoleType.User
-        self.__content: list[tuple[str, str] | str | tuple[str, BytesIO]] = []
+        self.__content: list[tuple[str, str] | str] = []
         self.__tool_call_id: str | None = None
 
     def set_role(self, role: RoleType = RoleType.User) -> "MessageBuilder":
@@ -54,20 +53,6 @@ class MessageBuilder:
         """
         self.__content.append(text)
         return self
-
-    def add_file_content(
-        self, file_name: str, file_base64: str
-    ) -> "MessageBuilder":
-        """
-        添加文件内容
-        :param file_name: 文件名(包含类型后缀)
-        :param file_base64: 文件的base64编码
-        :return: MessageBuilder对象
-        """
-        if not file_name or not file_base64:
-            raise ValueError("文件名和base64编码不能为空")
-        self.__content.append((file_name, BytesIO(base64.b64decode(file_base64))))
-        return self
     
     def add_image_content(
         self, image_format: str, image_base64: str

src/llm_models/utils_model.py

@@ -114,11 +114,6 @@ class LLMRequest:
         Returns:
             (Optional[str]): 生成的文本描述或None
         """
-        # 请求体构建
-        message_builder = MessageBuilder()
-        message_builder.add_file_content(file_name="audio.wav", file_base64=voice_base64)
-        messages = [message_builder.build()]
-
         # 模型选择
         model_info, api_provider, client = self._select_model()
 
@@ -128,7 +123,7 @@ class LLMRequest:
             client=client,
             request_type=RequestType.AUDIO,
             model_info=model_info,
-            message_list=messages,
+            audio_base64=voice_base64,
         )
         return response.content or None
 
@@ -249,6 +244,7 @@ class LLMRequest:
         temperature: Optional[float] = None,
         max_tokens: Optional[int] = None,
         embedding_input: str = "",
+        audio_base64: str = ""
     ) -> APIResponse:
         """
         实际执行请求的方法
@@ -283,7 +279,7 @@ class LLMRequest:
                     assert message_list is not None, "message_list cannot be None for audio requests"
                     return await client.get_audio_transcriptions(
                         model_info=model_info,
-                        message_list=message_list,
+                        audio_base64=audio_base64,
                         extra_params=model_info.extra_params,
                     )
             except Exception as e: