Merge pull request #19 from 00make/main

update: 添加Gemini LLM支持并更新配置

config.yaml

@@ -82,7 +82,7 @@ LLM:
     # 可在这里找到你的api key https://bigmodel.cn/usercenter/proj-mgmt/apikeys
     model_name: glm-4-flash
     url: https://open.bigmodel.cn/api/paas/v4/
-    api_key: 你的ChatGLMLLM api key
+    api_key: 47ab8569b25f4c3d8ea25830a33a64ca.4cF037cViBB1df5o
   OllamaLLM:
     # 定义LLM API类型
     type: ollama
@@ -95,6 +95,10 @@ LLM:
     # 如果使用DifyLLM，配置文件里prompt(提示词)是无效的，需要在dify控制台设置提示词
     base_url: https://api.dify.cn/v1
     api_key: 你的DifyLLM api key
+  GeminiLLM:
+    type: gemini
+    api_key: 你的gemini api key
+    model_name: "gemini-1.5-pro"  #  gemini-1.5-pro 是免费的
 TTS:
   # 当前支持的type为edge、doubao，可自行适配
   EdgeTTS:

core/providers/llm/gemini/gemini.py

@@ -0,0 +1,80 @@
+import logging
+import google.generativeai as genai
+from core.providers.llm.base import LLMProviderBase
+
+logger = logging.getLogger(__name__)
+
+class LLMProvider(LLMProviderBase):
+    def __init__(self, config):
+        """初始化Gemini LLM Provider"""
+        self.model_name = config.get("model_name", "gemini-1.5-pro") 
+        self.api_key = config.get("api_key")
+        
+        if not self.api_key or "你" in self.api_key:
+            logger.error("你还没配置Gemini LLM的密钥，请在配置文件中配置密钥，否则无法正常工作")
+            return
+            
+        try:
+            # 初始化Gemini客户端
+            genai.configure(api_key=self.api_key)
+            self.model = genai.GenerativeModel(self.model_name)
+            
+            # 设置生成参数
+            self.generation_config = {
+                "temperature": 0.7,
+                "top_p": 0.9,
+                "top_k": 40,
+                "max_output_tokens": 2048,
+            }
+            self.chat = None
+        except Exception as e:
+            logger.error(f"Gemini初始化失败: {e}")
+            self.model = None
+
+    def response(self, session_id, dialogue):
+        """生成Gemini对话响应"""
+        if not self.model:
+            yield "【Gemini服务未正确初始化】"
+            return
+
+        try:
+            # 处理对话历史
+            chat_history = []
+            for msg in dialogue[:-1]:  # 历史对话
+                role = "model" if msg["role"] == "assistant" else "user"
+                content = msg["content"].strip()
+                if content:
+                    chat_history.append({
+                        "role": role,
+                        "parts": [content]
+                    })
+
+            # 获取当前消息
+            current_msg = dialogue[-1]["content"]
+
+            # 创建新的聊天会话
+            chat = self.model.start_chat(history=chat_history)
+            
+            # 发送消息并获取流式响应
+            response = chat.send_message(
+                current_msg,
+                stream=True,
+                generation_config=self.generation_config
+            )
+
+            # 处理流式响应
+            for chunk in response:
+                if hasattr(chunk, 'text') and chunk.text:
+                    yield chunk.text
+
+        except Exception as e:
+            error_msg = str(e)
+            logger.error(f"Gemini响应生成错误: {error_msg}")
+            
+            # 针对不同错误返回友好提示
+            if "Rate limit" in error_msg:
+                yield "【Gemini服务请求太频繁,请稍后再试】"
+            elif "Invalid API key" in error_msg:
+                yield "【Gemini API key无效】"
+            else:
+                yield f"【Gemini服务响应异常: {error_msg}】"

requirements.txt

@@ -8,5 +8,6 @@ pydub==0.25.1
 funasr==1.2.3
 torchaudio==2.2.2
 openai==1.61.0
+google-generativeai==0.8.4
 edge_tts==7.0.0
 httpx==0.27.2