task: add real estate story

llm/llm_thinking_engine.py

@@ -0,0 +1,144 @@
+from typing import Optional, Dict
+from dataclasses import dataclass
+import os
+from config.settings import settings
+from openai import OpenAI
+from utils import logger
+
+
+@dataclass
+class LLMConfig:
+    """LLM配置类"""
+    api_key: Optional[str] = None
+    base_url: str = "https://dashscope.aliyuncs.com/compatible-mode/v1"
+    model: str = "deepseek-v3.2"
+    enable_thinking: bool = True
+    temperature: float = 0.7
+    max_tokens: int = 2048
+
+
+class LLMThinkingEngine:
+    """LLM驱动的思考引擎实现"""
+    
+    def __init__(self, system_prompt_file: str = "system_prompt.txt", config: Optional[LLMConfig] = None):
+        """
+        初始化LLMThinkingEngine
+        
+        Args:
+            config: LLM配置对象，如果为None则使用默认配置
+        """
+        self.system_prompt_file = system_prompt_file
+        self.config = config or LLMConfig()
+        self._init_client()
+    
+    def _init_client(self):
+        """初始化OpenAI客户端"""
+        api_key = self.config.api_key or settings.LLM_API_KEY
+        self.client = OpenAI(
+            api_key=api_key,
+            base_url=self.config.base_url,
+        )
+    
+    def think(self, user_input: str) -> str:
+        """
+        基于LLM进行思考，返回下一步的行动
+        
+        Args:
+            user_input: 用户输入内容
+            
+        Returns:
+            Thought: 思考结果，包含行动类型和内容
+        """
+        # 构建适用于LLM的消息
+        messages = self._build_messages(user_input)
+        logger.info(f"LLM构建的消息: {messages}")
+        # 调用LLM进行思考
+        thinking_content, response_content = self._call_llm(messages)
+        
+        # logger.info(f"LLM思考结果: thinking_content={thinking_content}, response_content={response_content}")
+        return response_content
+    
+    def _build_messages(self, user_input: str) -> list[Dict[str, str]]:
+        """
+        构建发送给LLM的消息
+
+        Args:
+            user_input: 用户输入内容
+
+        Returns:
+            消息列表，包含系统提示、历史和当前输入
+        """
+        messages = []
+        
+        # 系统提示
+        system_prompt = self._get_system_prompt()
+        messages.append({"role": "system", "content": system_prompt})
+    
+        # 用户输入
+        messages.append({
+            "role": "user",
+            "content": user_input
+        })
+        
+        return messages
+    
+    def _get_system_prompt(self) -> str:
+        """
+        获取系统提示词
+        
+        Returns:
+            系统提示词
+        """
+        prompt_path = os.path.join(
+            os.path.dirname(__file__), 
+            "prompts", 
+            self.system_prompt_file
+        )
+        with open(prompt_path, "r", encoding="utf-8") as f:
+            return f.read()
+    
+    def _call_llm(self, messages: list[Dict[str, str]]) -> tuple[str, str]:
+        """
+        调用LLM API
+        
+        Args:
+            messages: 消息列表
+            
+        Returns:
+            (thinking_content, response_content): 思考内容和响应内容
+        """
+        thinking_content = ""
+        response_content = ""
+        
+        try:
+            completion = self.client.chat.completions.create(
+                model=self.config.model,
+                messages=messages,
+                temperature=self.config.temperature,
+                max_tokens=self.config.max_tokens,
+                extra_body={"enable_thinking": self.config.enable_thinking},
+                stream=True
+            )
+            
+            # 流式处理响应
+            for chunk in completion:
+                delta = chunk.choices[0].delta
+                
+                # 收集思考内容
+                if hasattr(delta, "reasoning_content") and delta.reasoning_content:
+                    thinking_content += delta.reasoning_content
+                
+                # 收集响应内容
+                if hasattr(delta, "content") and delta.content:
+                    response_content += delta.content
+            
+        except Exception as e:
+            # 错误处理
+            response_content = f"调用LLM时出错：{str(e)}"
+        
+        return thinking_content, response_content
+    
+    def set_config(self, config: LLMConfig):
+        """更新LLM配置"""
+        self.config = config
+        self._init_client()