feat: Deploy final version with latest google-genai SDK

app.py

@@ -5,13 +5,14 @@ import asyncio
 import os
 import random
 from contextlib import asynccontextmanager
+from typing import List
 
 import uvicorn
 from fastapi import FastAPI, Request, HTTPException
 from fastapi.responses import StreamingResponse, JSONResponse
 from fastapi.middleware.cors import CORSMiddleware
-from google import genai
-from google.genai import types
+# 正确的导入方式
+import google.generativeai as genai
 
 # 配置日志
 logging.basicConfig(
@@ -21,6 +22,7 @@ logging.basicConfig(
 )
 logger = logging.getLogger(__name__)
 
+
 # !!! 重要：管理您的 API 密钥 !!!
 # 为了安全，强烈建议您通过 Hugging Face Space 的 "Settings" -> "Secrets"
 # 来设置您的 API 密钥。例如，创建一个名为 "GOOGLE_API_KEYS" 的 Secret，
@@ -31,8 +33,8 @@ logger = logging.getLogger(__name__)
 
 # 为了方便首次测试，您可以先临时使用下面的列表：
 API_KEYS = [
-    "YOUR_GOOGLE_API_KEY_1",  # 请替换成您的第一个有效密钥
-    "YOUR_GOOGLE_API_KEY_2",  # 请替换成您的第二个有效密钥
+    "AIzaSyCJGYHjn3m41mYpft9j3G9-RXWkDAcAsPs", # 请替换成您的第一个有效密钥
+    "AIzaSyCoDA9F9bsCXJx9p1CQqdKpeSO4n31DPt0", # 请替换成您的第二个有效密钥
 ]
 
 
@@ -47,7 +49,7 @@ def get_model_name(requested_model: str) -> str:
         return requested_model
     return "gemini-1.5-flash-latest"
 
-def convert_messages_to_gemini_format(messages):
+def convert_messages_to_gemini_format(messages: List[dict]):
     gemini_messages = []
     system_instruction = None
     for message in messages:
@@ -64,29 +66,34 @@ def convert_messages_to_gemini_format(messages):
 
 def handle_error_response(error):
     error_str = str(error).lower()
+    if "api_key_invalid" in error_str:
+        return "提供的 API 密钥无效。", "invalid_request_error"
     if "prompt_feedback" in error_str:
         return "请求因安全或内容策略被拒绝。", "content_filter"
-    elif "api_key_invalid" in error_str:
-        return "提供的 API 密钥无效。", "invalid_request_error"
     return f"发生未知错误: {error}", "stop"
 
 def get_api_key():
-    if not API_KEYS or all(key.startswith("YOUR_")):
+    if not API_KEYS or all("YOUR_GOOGLE_API_KEY" in key for key in API_KEYS):
         raise ValueError("API 密钥列表为空或未配置。请在 app.py 中设置它们或使用 Space Secrets。")
     return random.choice(API_KEYS)
 
-
 @asynccontextmanager
 async def lifespan(app: FastAPI):
     logger.info("应用启动...")
-    if not API_KEYS or all(key.startswith("YOUR_")):
-         logger.warning("警告：未检测到有效的 Google API 密钥。应用可能会在处理请求时失败。")
-    else:
-        logger.info(f"已加载 {len(API_KEYS)} 个 API 密钥。")
+    try:
+        if not API_KEYS or all("YOUR_GOOGLE_API_KEY" in key for key in API_KEYS):
+            logger.warning("警告：未检测到有效的 Google API 密钥。应用可能会在处理请求时失败。")
+        else:
+            # 启动时用一个密钥测试配置是否正确
+            genai.configure(api_key=API_KEYS[0])
+            logger.info(f"已加载 {len(API_KEYS)} 个 API 密钥，并成功配置。")
+    except Exception as e:
+        logger.error(f"Gemini 配置失败: {e}")
     yield
     logger.info("应用关闭。")
 
-app = FastAPI(lifespan=lifespan, title="Gemini API Proxy", version="1.4.0")
+
+app = FastAPI(lifespan=lifespan, title="Gemini API Proxy", version="1.5.0")
 
 app.add_middleware(
     CORSMiddleware,
@@ -96,13 +103,14 @@ app.add_middleware(
     allow_headers=["*"],
 )
 
+# 使用 genai.types 来访问类型定义
 SAFETY_SETTINGS = [
-    types.SafetySetting(category=cat, threshold=types.HarmBlockThreshold.BLOCK_NONE)
+    genai.types.SafetySetting(category=cat, threshold=genai.types.HarmBlockThreshold.BLOCK_NONE)
     for cat in [
-        types.HarmCategory.HARM_CATEGORY_HARASSMENT,
-        types.HarmCategory.HARM_CATEGORY_HATE_SPEECH,
-        types.HarmCategory.HARM_CATEGORY_SEXUALLY_EXPLICIT,
-        types.HarmCategory.HARM_CATEGORY_DANGEROUS_CONTENT,
+        genai.types.HarmCategory.HARM_CATEGORY_HARASSMENT,
+        genai.types.HarmCategory.HARM_CATEGORY_HATE_SPEECH,
+        genai.types.HarmCategory.HARM_CATEGORY_SEXUALLY_EXPLICIT,
+        genai.types.HarmCategory.HARM_CATEGORY_DANGEROUS_CONTENT,
     ]
 ]
 
@@ -118,10 +126,8 @@ async def stream_generator(model_name: str, gemini_messages, generation_config):
         ):
             if chunk.text:
                 data = {
-                    "id": f"chatcmpl-{int(time.time())}",
-                    "object": "chat.completion.chunk",
-                    "created": int(time.time()),
-                    "model": model_name,
+                    "id": f"chatcmpl-{int(time.time())}", "object": "chat.completion.chunk",
+                    "created": int(time.time()), "model": model_name,
                     "choices": [{"delta": {"content": chunk.text}}],
                 }
                 yield f"data: {json.dumps(data, ensure_ascii=False)}\n\n"
@@ -130,20 +136,16 @@ async def stream_generator(model_name: str, gemini_messages, generation_config):
         logger.error(f"流式响应生成时出错: {e}")
         error_message, finish_reason = handle_error_response(e)
         error_data = {
-            "id": f"chatcmpl-{int(time.time())}",
-            "object": "chat.completion.chunk",
-            "created": int(time.time()),
-            "model": model_name,
+            "id": f"chatcmpl-{int(time.time())}", "object": "chat.completion.chunk",
+            "created": int(time.time()), "model": model_name,
             "choices": [{"delta": {"content": error_message}, "finish_reason": finish_reason}],
         }
         yield f"data: {json.dumps(error_data, ensure_ascii=False)}\n\n"
     
     finally:
         final_data = {
-            "id": f"chatcmpl-{int(time.time())}",
-            "object": "chat.completion.chunk",
-            "created": int(time.time()),
-            "model": model_name,
+            "id": f"chatcmpl-{int(time.time())}", "object": "chat.completion.chunk",
+            "created": int(time.time()), "model": model_name,
             "choices": [{"delta": {}, "finish_reason": "stop"}],
         }
         yield f"data: {json.dumps(final_data, ensure_ascii=False)}\n\n"
@@ -157,10 +159,10 @@ async def chat_completions(request: Request):
     messages = body.get("messages", [])
     gemini_messages, system_instruction = convert_messages_to_gemini_format(messages)
     
-    generation_config = types.GenerationConfig(
+    generation_config = genai.types.GenerationConfig(
         temperature=body.get("temperature", 0.7),
         top_p=body.get("top_p", 1.0),
-        max_output_tokens=body.get("max_tokens", 2048),
+        max_output_tokens=body.get("max_tokens", 8192),
     )
 
     if system_instruction:
@@ -181,14 +183,11 @@ async def chat_completions(request: Request):
                 contents=gemini_messages,
                 generation_config=generation_config
             )
-            
             return {
-                "id": f"chatcmpl-{int(time.time())}",
-                "object": "chat.completion",
-                "created": int(time.time()),
-                "model": model_name,
+                "id": f"chatcmpl-{int(time.time())}", "object": "chat.completion",
+                "created": int(time.time()), "model": model_name,
                 "choices": [{"message": {"role": "assistant", "content": response.text}, "finish_reason": "stop"}],
-                "usage": {"prompt_tokens": 0, "completion_tokens": 0, "total_tokens": 0} # Usage info not easily available
+                "usage": {"prompt_tokens": 0, "completion_tokens": 0, "total_tokens": 0}
             }
         except Exception as e:
             logger.error(f"非流式响应生成时出错: {e}")

requirements.txt

@@ -1,5 +1,5 @@
 fastapi
 uvicorn
-google-generativeai
+google-genai==1.20.0
 loguru
 httpx