feat: 兼容OpenAI格式的模型服务代理

app/config/ai_configs.py

@@ -2,11 +2,6 @@ from app.config.env import env
 
 ai_configs = {
   # /*---------------------------------------local-------------------------------------------*/
-  "local": {
-    "model": "deepseek-r1-distill-qwen-7b",
-    "url": "http://127.0.0.1:1234/v1/chat/completions",
-    "key": env.llm_key_local
-  },
   "local_glm": {
     "model": "glm-4-9b-0414",
     "url": "http://127.0.0.1:1234/v1/chat/completions",
@@ -17,38 +12,12 @@ ai_configs = {
     "url": "http://127.0.0.1:1234/v1/embeddings",
     "key": env.llm_key_local
   },
-  # /*---------------------------------------deepseek-------------------------------------------*/
-  'deepseek-v3': {
-    'model': 'deepseek-chat',
-    'url': 'https://api.deepseek.com/chat/completions',
-    'key': env.llm_key_deepseek,
-  },
-  'deepseek-r1': {
-    'model': 'deepseek-reasoner',
-    'url': 'https://api.deepseek.com/chat/completions',
-    'key': env.llm_key_deepseek,
-  },
   # /*---------------------------------------huoshan-------------------------------------------*/
-  'huoshan-deepseek-v3': {
-    'model': 'deepseek-v3-250324',
-    'url': 'https://ark.cn-beijing.volces.com/api/v3/chat/completions',
-    'key': env.llm_key_huoshan,
-  },
-  'huoshan-deepseek-r1': {
-    'model': 'deepseek-r1-distill-qwen-7b-250120',
-    'url': 'https://ark.cn-beijing.volces.com/api/v3/chat/completions',
-    'key': env.llm_key_huoshan,
-  },
   'huoshan-doubao': {
     'model': 'doubao-1-5-lite-32k-250115',
     'url': 'https://ark.cn-beijing.volces.com/api/v3/chat/completions',
     'key': env.llm_key_huoshan,
   },
-  'huoshan-think-pro': {
-    'model': 'doubao-1-5-thinking-pro-250415',
-    'url': 'https://ark.cn-beijing.volces.com/api/v3/chat/completions',
-    'key': env.llm_key_huoshan,
-  },
   'huoshan-doubao-seed': {
     'model': 'doubao-seed-1-6-250615',
     'url': 'https://ark.cn-beijing.volces.com/api/v3/chat/completions',
@@ -59,12 +28,6 @@ ai_configs = {
     'url': 'https://ark.cn-beijing.volces.com/api/v3/chat/completions',
     'key': env.llm_key_huoshan,
   },
-
-  'huoshan-embedding-240715': {
-    "model": 'doubao-embedding-text-240715',
-    "url": 'https://ark.cn-beijing.volces.com/api/v3/embeddings',
-    'key': env.llm_key_huoshan,
-  },
   # /*---------------------------------------bailian-------------------------------------------*/
   'bailian-qwen-turbo': {
     'model': 'qwen-turbo',
@@ -76,6 +39,11 @@ ai_configs = {
     'url': 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions',
     'key': env.llm_key_bailian,
   },
+  'bailian-qwen3.6-plus': {
+    'model': 'qwen3.6-plus',
+    'url': 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions',
+    'key': env.llm_key_bailian,
+  },
   'bailian-embedding': {
     'model': 'text-embedding-v4',
     'url': 'https://dashscope.aliyuncs.com/compatible-mode/v1/embeddings',

app/routes.py

@@ -6,6 +6,7 @@ from app.controller.add_graph_proxy_route import add_graph_proxy_route
 from app.controller.add_test_route import add_test_route
 from app.model.LlmDemoMdel import LlmDemoModel
 from app.model.ResumeTemplateModel import ResumeTemplateModel
+from app.utils.add_llm_routes import add_llm_routes
 from app.utils.add_model_routes import add_model_routes
 from app.utils.next_id import add_next_id_route
 
@@ -19,6 +20,10 @@ routes = [
   add_next_id_route,            # 生成ID接口
   add_file_route,               # 文件上传接口
 
+  lambda app: add_llm_routes(app, 'huoshan-doubao'),
+  lambda app: add_llm_routes(app, 'doubao-vision-lite'),
+  lambda app: add_llm_routes(app, 'bailian-qwen3.6-plus'),
+
   lambda app: add_model_routes(app,LlmDemoModel,'/llm_demo'),                     # LlmDemo 测试用户模块
   lambda app: add_model_routes(app,ResumeTemplateModel,'/llm_resume_template'),   # 简历模板
 ]

app/utils/add_llm_routes.py

@@ -0,0 +1,61 @@
+import httpx
+from fastapi import APIRouter, Request, HTTPException, FastAPI
+from fastapi.responses import StreamingResponse
+
+from app.config.ai_configs import ai_configs
+
+
+async def proxy_openai_request(request: Request, config_key: str):
+  """
+  通用代理函数：将请求转发至指定的上游模型服务
+  """
+  if config_key not in ai_configs:
+    raise HTTPException(status_code=404, detail=f"Config {config_key} not found")
+
+  config = ai_configs[config_key]
+  target_url = config["url"]
+  api_key = config["key"]
+  real_model_name = config["model"]
+
+  # 1. 获取原始请求体并修改模型名称
+  body = await request.json()
+  body["model"] = real_model_name  # 强制替换为上游真正识别的模型名
+
+  # 2. 准备请求头
+  headers = {
+    "Authorization": f"Bearer {api_key}",
+    "Content-Type": "application/json"
+  }
+
+  client = httpx.AsyncClient()
+
+  # 3. 处理流式响应 (OpenAI 协议的关键)
+  if body.get("stream", False):
+    async def stream_generator():
+      async with client.stream("POST", target_url, json=body, headers=headers, timeout=60.0) as response:
+        async for chunk in response.aiter_bytes():
+          yield chunk
+      await client.aclose()
+
+    return StreamingResponse(stream_generator(), media_type="text/event-stream")
+
+  # 4. 处理非流式响应
+  else:
+    response = await client.post(target_url, json=body, headers=headers, timeout=60.0)
+    await client.aclose()
+    return response.json()
+
+
+def add_llm_routes(app: FastAPI, config_name: str):
+  """
+  动态生成路由的辅助函数
+  """
+  # 自动根据配置判断是 chat 还是 embeddings 路径
+  is_embedding = "embedding" in config_name or "embeddings" in ai_configs[config_name]["url"]
+  path = "/v1/embeddings" if is_embedding else "/v1/chat/completions"
+
+  @app.post(f"/{config_name}{path}")
+  async def dynamic_proxy(request: Request):
+    return await proxy_openai_request(request, config_name)
+
+  print(f"🚀 Proxy LLM route: /{config_name}{path}")