From 47bdd9d5a589c232b520bd94ae85e014502cc009 Mon Sep 17 00:00:00 2001
From: jeremygan2021 <jeremygan2021@gmail.com>
Date: Wed, 5 Nov 2025 00:29:54 +0800
Subject: [PATCH 1/2] =?UTF-8?q?model=20=E6=9B=B4=E6=8D=A2?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 fastapi_server/server.py | 19 +++++++++++--------
 1 file changed, 11 insertions(+), 8 deletions(-)

diff --git a/fastapi_server/server.py b/fastapi_server/server.py
index 394d404..aa9328c 100644
--- a/fastapi_server/server.py
+++ b/fastapi_server/server.py
@@ -26,6 +26,8 @@ class ChatCompletionRequest(BaseModel):
     max_tokens: Optional[int] = Field(default=500, description="最大生成token数")
     stream: Optional[bool] = Field(default=False, description="是否流式返回")
     thread_id: Optional[int] = Field(default=3, description="线程ID，用于多轮对话")
+    llm_provider: Optional[str] = Field(default="openai", description="LLM提供商")
+    base_url: Optional[str] = Field(default="https://dashscope.aliyuncs.com/compatible-mode/v1", description="LLM API基础URL")
 
 class ChatCompletionResponseChoice(BaseModel):
     index: int
@@ -73,14 +75,6 @@ app.add_middleware(
     allow_headers=["*"],
 )
 
-# 初始化Pipeline
-pipeline_config = PipelineConfig()
-pipeline_config.llm_name = "qwen-flash"
-pipeline_config.llm_provider = "openai"
-pipeline_config.base_url = "https://dashscope.aliyuncs.com/compatible-mode/v1"
-
-pipeline = Pipeline(pipeline_config)
-
 @app.post("/v1/chat/completions", response_model=ChatCompletionResponse)
 async def chat_completions(
     request: ChatCompletionRequest, 
@@ -103,6 +97,15 @@ async def chat_completions(
         if not user_message:
             raise HTTPException(status_code=400, detail="缺少用户消息")
         
+        # 动态创建PipelineConfig
+        pipeline_config = PipelineConfig()
+        pipeline_config.llm_name = request.model
+        pipeline_config.llm_provider = request.llm_provider
+        pipeline_config.base_url = request.base_url
+        
+        # 创建新的Pipeline实例
+        pipeline = Pipeline(pipeline_config)
+        
         # 调用pipeline的chat方法
         response_content = pipeline.chat(
             inp=user_message,

From 4e83426d16460e122187eb88181abfc49c0a07ef Mon Sep 17 00:00:00 2001
From: jeremygan2021 <jeremygan2021@gmail.com>
Date: Wed, 5 Nov 2025 00:37:09 +0800
Subject: [PATCH 2/2] =?UTF-8?q?=E5=B0=81=E8=A3=85fastAPI=20openAI=E6=8E=A5?=
 =?UTF-8?q?=E5=8F=A3=E8=A7=84=E8=8C=83?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 fastapi_server/server.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fastapi_server/server.py b/fastapi_server/server.py
index aa9328c..834d88a 100644
--- a/fastapi_server/server.py
+++ b/fastapi_server/server.py
@@ -91,7 +91,7 @@ async def chat_completions(
         for message in request.messages:
             if message.role == "user":
                 user_message = message.content
-            elif message.role == "system":
+            elif message.role == "system" or message.role == "assistant":
                 system_message = message.content
         
         if not user_message: