Create app.py

app.py

@@ -0,0 +1,103 @@
+import gradio as gr
+import os
+from dotenv import load_dotenv
+from langchain_community.vectorstores import FAISS
+from langchain_community.document_loaders import TextLoader
+from langchain.text_splitter import CharacterTextSplitter
+from langchain_community.embeddings import SentenceTransformerEmbeddings
+from langchain.prompts import PromptTemplate
+from langchain_openai import ChatOpenAI
+# 若使用本地模型，替换为以下导入
+# from langchain_community.llms import HuggingFacePipeline
+# import transformers
+
+# 加载环境变量（HF Spaces需在Secrets中配置OPENAI_API_KEY）
+load_dotenv()
+openai_api_key = os.getenv("OPENAI_API_KEY")
+
+# 配置路径
+BASE_DIR = os.path.dirname(os.path.abspath(__file__))
+KNOWLEDGE_BASE_DIR = os.path.join(BASE_DIR, "knowledge_base")
+FAMILY_TREE_PATH = os.path.join(KNOWLEDGE_BASE_DIR, "family_tree.txt")
+PROMPT_TEMPLATE_PATH = os.path.join(KNOWLEDGE_BASE_DIR, "prompt_template.txt")
+GENERATION_VERSES_PATH = os.path.join(KNOWLEDGE_BASE_DIR, "generation_verses.txt")
+
+# 加载并处理知识库
+def load_knowledge_base():
+    # 加载家谱数据
+    loader = TextLoader(FAMILY_TREE_PATH, encoding="utf-8")
+    documents = loader.load()
+    # 文本分割
+    text_splitter = CharacterTextSplitter(chunk_size=500, chunk_overlap=50)
+    docs = text_splitter.split_documents(documents)
+    # 初始化嵌入模型
+    embeddings = SentenceTransformerEmbeddings(model_name="all-MiniLM-L6-v2")
+    # 创建向量库
+    db = FAISS.from_documents(docs, embeddings)
+    return db
+
+# 加载提示词模板
+def load_prompt_template():
+    with open(PROMPT_TEMPLATE_PATH, "r", encoding="utf-8") as f:
+        prompt_template = f.read()
+    return PromptTemplate.from_template(prompt_template)
+
+# 初始化RAG组件
+db = load_knowledge_base()
+prompt_template = load_prompt_template()
+
+# 初始化大语言模型
+llm = ChatOpenAI(
+    model_name="gpt-3.5-turbo",
+    temperature=0,  # 0为精准回答，不生成冗余内容
+    api_key=openai_api_key
+)
+
+# 若使用本地模型（示例：Llama-2-7B），替换以下代码
+# def load_local_llm():
+#     model_name = "meta-llama/Llama-2-7b-chat-hf"
+#     pipeline = transformers.pipeline(
+#         "text-generation",
+#         model=model_name,
+#         device_map="auto",
+#         model_kwargs={"torch_dtype": "auto"},
+#         max_new_tokens=500,
+#         temperature=0,
+#     )
+#     llm = HuggingFacePipeline(pipeline=pipeline)
+#     return llm
+# llm = load_local_llm()
+
+# 核心问答逻辑
+def answer_question(question):
+    # 检索相关文档
+    retriever = db.as_retriever(search_kwargs={"k": 3})  # 取top3相关结果
+    relevant_docs = retriever.get_relevant_documents(question)
+    # 拼接检索结果
+    context = "\n".join([doc.page_content for doc in relevant_docs])
+    # 加载字辈表（补充上下文）
+    with open(GENERATION_VERSES_PATH, "r", encoding="utf-8") as f:
+        generation_verses = f.read()
+    context += "\n" + generation_verses
+    # 构建提示词
+    prompt = prompt_template.format(context=context, question=question)
+    # 调用LLM生成回答
+    response = llm.invoke(prompt)
+    return response.content
+
+# 创建Gradio界面
+with gr.Blocks(title="冯氏家谱RAG机器人") as demo:
+    gr.Markdown("# 冯氏家谱RAG机器人")
+    gr.Markdown("### 请输入关于冯氏家谱的问题，支持亲属关系、生卒信息、字辈世代查询")
+    gr.Markdown("#### 示例问题：\n1. 冯乔福的子女有谁？\n2. 冯仁杰的出生地是什么？\n3. 宗字辈的上一辈是什么字辈？")
+    
+    with gr.Row():
+        question = gr.Textbox(label="输入问题", placeholder="请输入你的问题...", lines=2)
+        answer = gr.Textbox(label="回答结果", lines=5, interactive=False)
+    
+    submit_btn = gr.Button("提交查询")
+    submit_btn.click(fn=answer_question, inputs=question, outputs=answer)
+
+# 运行应用
+if __name__ == "__main__":
+    demo.launch()