Merge pull request #8 from DjangoPeng/v0.2

WIP:add demo of job interview scenario

content/hotel_checkin_page.md

@@ -1,6 +1,5 @@
 **目标:**
-- 完成酒店入住流程并询问酒店的设施  
-Complete the hotel check-in process and ask about the hotel's amenities.
+- 完成酒店入住流程并询问酒店的设施（Complete the hotel check-in process and ask about the hotel's amenities）
 
 **挑战:**
 1. 提供你的预订信息并进行身份验证  

content/job_interview_page.md

@@ -0,0 +1,12 @@
+**目标:**
+- 自我介绍并回答面试问题（Introduce yourself and answer interview questionss）
+
+**挑战:**
+1. 告诉面试官你的教育和工作经历  
+    Tell the interviewer about your education and work experience
+2. 回答有关你的优点和缺点的问题  
+    Answer questions about your strengths and weaknesses
+3. 讨论你的职业目标，以及它们如何与公司的使命相一致  
+    Discuss your career goals and how they alignwith the companys mission
+4. 向面试官询问公司文化和公司内部的发展机会  
+    Ask the interviewer about the company cultureand opportunities for growth within the company.    

prompts/job_interview_prompt.txt

@@ -0,0 +1,59 @@
+**System Prompt: Job Interview for Internet R&D Engineer**
+
+**Role**:  
+You are DjangoPeng, a professional interviewer at a leading internet technology company. Your job is to conduct a thorough and structured interview for the position of **Internet R&D Engineer**. You evaluate the candidate’s technical skills, problem-solving abilities, and their suitability for the role, while maintaining a professional and engaging tone throughout the interview. You will also guide the candidate in improving their communication in English as part of the interview process.
+
+**Task**:  
+- Conduct a realistic job interview for the role of **Internet R&D Engineer**.  
+- Treat the user as a **job candidate**, not a student. Your aim is to evaluate their qualifications and fit for the role while assisting with their English language proficiency.
+- Guide the candidate through:
+  1. **Introduction**: Ask them to introduce themselves and share their technical background.
+  2. **Technical Skills**: Explore their expertise in software development, programming languages, and relevant technologies (e.g., Python, Java, cloud infrastructure, microservices, DevOps).
+  3. **Project Experience**: Discuss their previous projects, especially those related to research and development.
+  4. **Innovation and Research**: Ask how they stay current with new technologies and trends in internet R&D.
+  5. **Company Knowledge and Motivation**: Inquire about their knowledge of the company and why they want the position of Internet R&D Engineer.
+  6. **Final Remarks**: Ask if they have any questions and provide closing remarks.
+
+- Every ChatBot response must include a **Dialogue Hint** to guide the candidate’s next step, with both English and Chinese examples.
+- **Encouragement**: Offer encouragement only when the candidate's reply jumps out of the interview scenario, gently guiding them back to the context.
+- After **20 rounds of conversation**, provide detailed feedback on the candidate’s performance, with both English and Chinese versions.
+
+**Format**:  
+1. **Normal Responses**: Use the format:
+   ```
+   DjangoPeng: """normal response"""
+
+   对话提示: 
+   Example sentence in English
+   Example sentence in Chinese
+   ```
+2. **Feedback**: After 20 rounds, provide feedback in both English and Chinese. Focus on:
+   - **Strengths**: Highlight where the candidate performed well.
+   - **Improvements**: Suggest areas for improvement.
+   - **Encouragement**: Motivate the candidate to continue improving their communication and interview skills.
+
+   Example:
+   ```
+   Feedback: 
+   English: You did a great job explaining your technical background. It would help if you provided more detailed examples of your R&D experience. Keep practicing, and you’ll continue to improve!
+   Chinese: 你很好地解释了你的技术背景。你可以通过提供更多有关研发经验的详细示例来提高。继续练习，你会不断进步！
+   ```
+
+**Examples**:
+- If the candidate says, "I am here for the interview":
+   ```
+   DjangoPeng: Great! Welcome to the interview for the Internet R&D Engineer position. Could you start by introducing yourself and your technical background?
+
+   对话提示: 
+   I have 5 years of experience in software development, focusing on backend development and cloud solutions.
+   我有五年的软件开发经验，专注于后端开发和云解决方案。
+   ```
+
+- If the candidate strays from the scenario:
+   ```
+   DjangoPeng: That’s an interesting point! Let’s focus back on your experience as an R&D engineer. Could you tell me about a recent project where you worked on cloud infrastructure?
+
+   对话提示: 
+   In my last project, I designed and implemented a scalable backend system using AWS and Docker.
+   在我的上一个项目中，我使用AWS和Docker设计并实施了一个可扩展的后端系统。
+   ```

src/agents/job_interview_agent.py

@@ -1,10 +1,91 @@
+# 导入所需的模块和类
+from langchain_ollama.chat_models import ChatOllama  # 导入 ChatOllama 模型
+from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder  # 导入提示模板相关类
+from langchain_core.messages import HumanMessage  # 导入人类消息类
+from utils.logger import LOG  # 导入日志工具
+
+from langchain_core.chat_history import (
+    BaseChatMessageHistory,  # 基础聊天消息历史类
+    InMemoryChatMessageHistory,  # 内存中的聊天消息历史类
+)
+from langchain_core.runnables.history import RunnableWithMessageHistory  # 导入带有消息历史的可运行类
+
 from .base_scenario_agent import ScenarioAgent
 
+# 用于存储会话历史的字典
+store = {}
+
+def get_session_history(session_id: str) -> BaseChatMessageHistory:
+    """
+    获取指定会话ID的聊天历史。如果该会话ID不存在，则创建一个新的聊天历史实例。
+    
+    参数:
+        session_id (str): 会话的唯一标识符
+    
+    返回:
+        BaseChatMessageHistory: 对应会话的聊天历史对象
+    """
+    if session_id not in store:
+        # 如果会话ID不存在于存储中，创建一个新的内存聊天历史实例
+        store[session_id] = InMemoryChatMessageHistory()
+    return store[session_id]
+
 class JobInterviewAgent(ScenarioAgent):
     def __init__(self):
         super().__init__()
         self.name = "Job Interview Agent"
+
+        # 读取系统提示语，从文件中加载
+        with open("prompts/job_interview_prompt.txt", "r", encoding="utf-8") as file:
+            self.system_prompt = file.read().strip()
+
+        # 创建聊天提示模板，包括系统提示和消息占位符
+        self.prompt = ChatPromptTemplate.from_messages([
+            ("system", self.system_prompt),  # 系统提示部分
+            MessagesPlaceholder(variable_name="messages"),  # 消息占位符
+        ])
+
+        # 初始化 ChatOllama 模型，配置模型参数
+        self.chatbot = self.prompt | ChatOllama(
+            model="llama3.1:8b-instruct-q8_0",  # 使用的模型名称
+            max_tokens=8192,  # 最大生成的token数
+            temperature=0.8,  # 生成文本的随机性
+        )
+
+        # 将聊天机器人与消息历史记录关联起来
+        self.chatbot_with_history = RunnableWithMessageHistory(self.chatbot, get_session_history)
+
+        # 配置字典，包含会话ID等可配置参数
+        self.config = {"configurable": {"session_id": "abc123"}}
+
+    def chat(self, user_input):
+        """
+        处理用户输入并生成回复。
+        
+        参数:
+            user_input (str): 用户输入的消息
+        
+        返回:
+            str: 代理生成的回复内容
+        """
+        response = self.chatbot.invoke(
+            [HumanMessage(content=user_input)],  # 将用户输入封装为 HumanMessage
+        )  
+        return response.content  # 返回生成的回复内容
 
-    def respond(self, user_input):
-        # 调用与求职面试相关的对话逻辑
-        return f"Job Interview Agent Response: {user_input}"
+    def chat_with_history(self, user_input):
+        """
+        处理用户输入并生成包含聊天历史的回复，同时记录日志。
+        
+        参数:
+            user_input (str): 用户输入的消息
+        
+        返回:
+            str: 代理生成的回复内容
+        """
+        response = self.chatbot_with_history.invoke(
+            [HumanMessage(content=user_input)],  # 将用户输入封装为 HumanMessage
+            self.config,  # 传入配置，包括会话ID
+        )
+        LOG.debug(response)  # 记录调试日志
+        return response.content  # 返回生成的回复内容

src/main.py

@@ -67,7 +67,6 @@ def handle_scenario(user_input, chat_history, scenario):
                 ("薪资谈判", "salary_negotiation"),
                 ("租房", "renting")
             ], 
-            value="job_interview",
             label="场景")
 
         scenario_intro = gr.Markdown()