3. LangChain4j-RAG，实现简单的text-sql功能

1. 简介

前两章我们讲了如何使用LangChain4J进行AI交互, 其中包括

1. 使用ChatLanguageModel、ChatMessage、ChatMemory等底层组件进行灵活/自由的与AI交互 传送门
2. 使用AI Services高级对象, 只关注业务逻辑, 使用简单的Api即可进行AI交互 传送门

RAG（Retrieval-Augmented Generation，检索增强生成）: 是一种利用向量数据库召回专业知识并融入Prompt的大模型应用方案.

RAG其实分为两个阶段: 构建索引 和 检索

1.1 构建索引

构建索引的过程其实是构建知识库的过程 如上图上半部分, 执行步骤如下:

1. 数据准备 (数据形式不限 如 文件, 字符串, 数据表内容 甚至是多媒体类型都可以)
2. 读取数据内容并分割成文本块 (数据规模很可能非常庞大，整体存储具有难度，并且在查询的时候可能仅仅和其中的一个或几个段落有关系，所以需要分块处理 将解析后的文档内容切分为适当的片段)
3. 将分割的文本块向量化 (转换成机器能识别的内容)
4. 将向量化的内容存储进向量数据库

1.2 检索

检索过程其实就是当我们进行提问的时候, 先使用之前准备好的知识库把提问内容相关的知识检索出来,然后一并将问题和检索结果交给大模型,辅助大模型生成回答 如上图下半部分 执行步骤如下:

1. 用户提问
2. 将用户提问的内容向量化
3. 使用问题向量化的结果到向量数据库进行检索,并返回相关的知识
4. 将问题和相关的知识组织成提示词交给大模型
5. 大模型根据提示词思考并生成回答
6. 响应用户

1.3 解决了什么问题

RAG核心是为了弥补传统通用大模型的两大短板：

1. 知识时效性不足：传统大模型训练数据固定，难以及时获取最新信息（如 近期的新闻、政策），而 RAG 通过实时检索外部知识库，让 AI 能调用最新数据生成回答。
2. 专业领域知识受限 ：通用模型对垂直领域（如医疗、法律）的深度知识掌握有限，RAG 可连接行业专属数据库，提升回答的专业性和准确性。
   简言之，RAG 让 AI "边查边答"，解决 "知识旧" 和 "不够专" 的问题，让回答更实时、更精准。

2. 实践案例

本章我们使用LangChain4J来实现一个简单的RAG需求: 整理一份学生成绩系统的数据库表ddl信息, 通过RAG实现让大模型帮我们生成sql语句, 比如向大模型提问: 张铁牛上学期语文考试成绩是多少? 希望AI给我们响应查询sql语句.

2.1 环境信息

使用SDK版本信息如下：

Java: 21
SpringBoot: 3.4.5
LangChain4j: 1.0.1
LLM: (使用在线的百炼(阿里)平台)
	embedding模型: text-embedding-v3  
	chat模型: qwen-plus
PGVector(postgresql版本的向量数据库): 0.8.0-pg17 

2.1.1 部署PGVector

推荐使用docker-compose部署, yml文件如下:

version: '3'
services:
  pgvector:  
    container_name: pgvector
    restart: always
    image: pgvector/pgvector:0.8.0-pg17
    privileged: true
    ports:
      - 5431:5432
    environment:
      POSTGRES_USER: root
      POSTGRES_PASSWORD: 123456
      PGDATA: /var/lib/postgresql/data/
    volumes:
      - ./data:/var/lib/postgresql/data/

2.2 Maven

<?xml version="1.0" encoding="UTF-8"?>
<project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
         xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 https://maven.apache.org/xsd/maven-4.0.0.xsd">
    <modelVersion>4.0.0</modelVersion>
    <parent>
        <groupId>org.springframework.boot</groupId>
        <artifactId>spring-boot-starter-parent</artifactId>
        <version>3.4.5</version>
        <relativePath/> <!-- lookup parent from repository -->
    </parent>
    <groupId>com.genuai</groupId>
    <artifactId>langchain-test</artifactId>
    <version>0.0.1-SNAPSHOT</version>
    <name>langchain-test</name>
    <description>langchain-test</description>

    <properties>
        <java.version>21</java.version>
        <guava.version>33.0.0-jre</guava.version>
    </properties>
    <dependencyManagement>
        <dependencies>
            <dependency>
                <groupId>dev.langchain4j</groupId>
                <artifactId>langchain4j-bom</artifactId>
                <version>1.0.1</version>
                <type>pom</type>
                <scope>import</scope>
            </dependency>
        </dependencies>
    </dependencyManagement>

    <dependencies>
        <dependency>
            <groupId>org.springframework.boot</groupId>
            <artifactId>spring-boot-starter-web</artifactId>
        </dependency>

        <dependency>
            <groupId>dev.langchain4j</groupId>
            <artifactId>langchain4j-open-ai-spring-boot-starter</artifactId>
        </dependency>

        <dependency>
            <groupId>dev.langchain4j</groupId>
            <artifactId>langchain4j-spring-boot-starter</artifactId>
        </dependency>

        <dependency>
            <groupId>dev.langchain4j</groupId>
            <artifactId>langchain4j-open-ai</artifactId>
        </dependency>

        <dependency>
            <groupId>com.google.guava</groupId>
            <artifactId>guava</artifactId>
            <version>${guava.version}</version>
        </dependency>

        <dependency>
            <groupId>dev.langchain4j</groupId>
            <artifactId>langchain4j-reactor</artifactId>
        </dependency>

        <dependency>
            <groupId>dev.langchain4j</groupId>
            <artifactId>langchain4j-pgvector</artifactId>
        </dependency>

        <dependency>
            <groupId>org.projectlombok</groupId>
            <artifactId>lombok</artifactId>
            <optional>true</optional>
        </dependency>
      
        <dependency>
            <groupId>org.springframework.boot</groupId>
            <artifactId>spring-boot-starter-test</artifactId>
            <scope>test</scope>
        </dependency>
    </dependencies>

    <build>
        <plugins>
            <plugin>
                <groupId>org.apache.maven.plugins</groupId>
                <artifactId>maven-compiler-plugin</artifactId>
            </plugin>

            <plugin>
                <groupId>org.springframework.boot</groupId>
                <artifactId>spring-boot-maven-plugin</artifactId>
                <configuration>
                    <excludes>
                        <exclude>
                            <groupId>org.projectlombok</groupId>
                            <artifactId>lombok</artifactId>
                        </exclude>
                    </excludes>
                </configuration>
            </plugin>
        </plugins>
    </build>
</project>

2.3 构建模型和存储对象

private static ChatModel chatModel;

private static EmbeddingModel embeddingModel;

private static EmbeddingStore<TextSegment> embeddingStore;

private static JdkHttpClientBuilder jdkHttpClientBuilder;

@BeforeAll
public static void init_embed_model() {
    jdkHttpClientBuilder = JdkHttpClient
            .builder()
            .httpClientBuilder(HttpClient.newBuilder());
    chatModel = OpenAiChatModel
            .builder()
            .httpClientBuilder(jdkHttpClientBuilder)
            .apiKey(System.getenv("LLM_API_KEY"))
            .modelName("qwen-plus")
            .baseUrl("https://dashscope.aliyuncs.com/compatible-mode/v1")
            // 打印请求日志
            //.logRequests(true)
            // 打印响应日志
            //.logResponses(true)
            .build();
    embeddingModel = OpenAiEmbeddingModel
            .builder()
            .httpClientBuilder(jdkHttpClientBuilder)
            .apiKey(System.getenv("LLM_API_KEY"))
            .modelName("text-embedding-v3")
            .baseUrl("https://dashscope.aliyuncs.com/compatible-mode/v1")
            .build();
    embeddingStore = PgVectorEmbeddingStore
            .builder()
            .host("localhost")                           // 必需：PostgresSQL 实例的主机
            .port(5431)                                  // 必需：PostgresSQL 实例的端口
            .database("postgres")                        // 必需：数据库名称
            .user("root")                                // 必需：数据库用户
            .password("123456")                          // 必需：数据库密码
            .table("my_sql_embeddings")                  // 必需：存储嵌入的表名
            .dimension(embeddingModel.dimension())       // 必需：嵌入的维度
            .metadataStorageConfig(DefaultMetadataStorageConfig.defaultConfig()) // 将元数据存储配置
            .build();
}

2.4 测试向量化

测试使用向量模型 将sql信息转为向量数据

@Test
public void should_print_embedding_content_when_embed_sql_resource() {
    // 资源信息在文章末尾
    final Document document = ClassPathDocumentLoader.loadDocument("student_ddl.sql");
    // 创建 SQL 内容的分割器
    DocumentSplitter splitter = new DocumentByRegexSplitter(
            ";",     // 分割方式(正则)
            ";",     // 连接方式(分割完之后每段内容末尾的分隔符)
            2000,    // 每段内容最大片段长度
            100      // 相邻片段之间的最大重叠字符数 (为了保证内容的完整性 分割时片段之间可能会重叠)
    );
    final List<TextSegment> textSegments = splitter.split(document);
    final Response<List<Embedding>> embedResult = embeddingModel.embedAll(textSegments);
    final List<Embedding> content = embedResult.content();
    log.info("embedding content: {}", content);
}

测试结果如下:

[main] INFO com.genuai.langchaintest.rag.AiRagTest -- embedding content: [Embedding { vector = [-0.05850703, -0.0043222117, -0.054852795, 0.0036689683, -0.07005912, -0.015255443, -0.0048452974, 0.060707428, 
...
-0.0153250545, -0.06653458, -0.0023748502, -0.011913025, 0.052188545, 0.02347709, -0.0043522767, -0.017399414, 0.028711455, 0.0027141145, 0.008481608, 0.002006506, -0.00725541, 0.011253882] }]

2.5 测试构建索引

@Test
public void should_store_vector_when_embed_sql_resource() {
  final Document document = ClassPathDocumentLoader.loadDocument("student_ddl.sql");
  DocumentSplitter splitter = new DocumentByRegexSplitter(";",";", 2000, 100);
  final List<TextSegment> textSegments = splitter.split(document);
  final Response<List<Embedding>> embedResult = embeddingModel.embedAll(textSegments);
  final List<Embedding> content = embedResult.content();
  // 将向量化结果和元数据一并存储到向量库
  embeddingStore.addAll(content, textSegments);
}

测试结果如下:

2.6 测试检索

@Test
public void should_return_sql_when_chat_embed_sql_resource() {
    ContentRetriever contentRetriever = EmbeddingStoreContentRetriever
            .builder()
            .embeddingStore(embeddingStore)
            .embeddingModel(embeddingModel)
            // 最大返回多少条相关的结果
            .maxResults(10)
            // 最小的得分(检索后每条数据都会有跟当前问题匹配度的得分, 得分越高越相近)
            .minScore(0.65)
            .build();
	// 使用ai services 进行ai call
    AiAssistantServiceWithMemoryTest assistant = AiServices
            .builder(AiAssistantServiceWithMemoryTest.class)
            .chatModel(chatModel)
            // 注入内容检索器
            .contentRetriever(contentRetriever)
            .chatMemoryProvider(memoryId -> MessageWindowChatMemory
                    .builder()
                    .id(memoryId)
                    .maxMessages(10)
                    .build())
            .build();
    String id = "zhangtieniu-01";
    final String q = "张铁牛上学期语文考试成绩是多少";
    final String a = assistant.chatSql(id, q);
    log.info("call ai q:{}\na: {}", q, a);
}

AI Services

import dev.langchain4j.service.MemoryId;
import dev.langchain4j.service.SystemMessage;
import dev.langchain4j.service.UserMessage;
import dev.langchain4j.service.spring.AiService;

/**
 * ai svc
 *
 * @author ludangxin
 * @date 2025/6/5
 */
@AiService
public interface AiAssistantServiceWithMemoryTest {
    String chat(@MemoryId String memoryId, @UserMessage String message);

    @SystemMessage("你是一名sql分析专家 我会将sql相关的ddl给你, 需要你根据ddl生成合理且可执行的sql语句并返回")
    String chatSql(@MemoryId String memoryId, @UserMessage String message);
}

测试结果如下(开启了请求相应日志) :

[main] INFO dev.langchain4j.http.client.log.LoggingHttpClient -- HTTP request:
- method: POST
- url: https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions
- headers: [Authorization: Beare...ff], [User-Agent: langchain4j-openai], [Content-Type: application/json]
- body: {
  "model" : "qwen-plus",
  "messages" : [ {
    "role" : "system",
    "content" : "你是一名sql分析专家 我会将sql相关的ddl给你, 需要你根据ddl生成合理且可执行的sql语句并返回"
  }, {
    "role" : "user",
    "content" : "张铁牛上学期语文考试成绩是多少\n\nAnswer using the following information:\nALTER TABLE courses ADD CONSTRAINT fk_courses_department_id FOREIGN KEY (department_id) REFERENCES departments(department_id);\nALTER TABLE courses ADD CONSTRAINT fk_courses_teacher_id FOREIGN KEY (teacher_id) REFERENCES teachers(teacher_id);\nALTER TABLE teachers ADD CONSTRAINT fk_teachers_department_id FOREIGN KEY (department_id) REFERENCES departments(department_id);\nALTER TABLE exam_arrangements ADD CONSTRAINT fk_exam_arrangements_course_id FOREIGN KEY (course_id) REFERENCES courses(course_id);\nALTER TABLE scores ADD CONSTRAINT fk_scores_student_id FOREIGN KEY (student_id) REFERENCES students(student_id);\nALTER TABLE scores ADD CONSTRAINT fk_scores_exam_id FOREIGN KEY (exam_id) REFERENCES exam_arrangements(exam_id)\n\n-- 创建成绩表\nCREATE TABLE scores (\n                        score_id INT PRIMARY KEY AUTO_INCREMENT,\n                        student_id INT NOT NULL COMMENT '学生ID',\n                        exam_id INT NOT NULL COMMENT '考试ID',\n                        score DECIMAL(5,2) COMMENT '成绩',\n                        score_type ENUM('原始分', '平时分', '卷面分', '最终分') NOT NULL DEFAULT '最终分' COMMENT '成绩类型',\n                        remark VARCHAR(200) COMMENT '备注',\n                        created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',\n                        updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间',\n                        UNIQUE KEY unique_student_exam (student_id, exam_id)\n);\n\n-- 创建索引\nCREATE INDEX idx_students_student_no ON students(student_no);\nCREATE INDEX idx_students_major_id ON students(major_id);\nCREATE INDEX idx_students_class_id ON students(class_id);\nCREATE INDEX idx_majors_department_id ON majors(department_id);\nCREATE INDEX idx_classes_major_id ON classes(major_id);\nCREATE INDEX idx_courses_department_id ON courses(department_id);\nCREATE INDEX idx_courses_teacher_id ON courses(teacher_id);\nCREATE INDEX idx_teachers_department_id ON teachers(department_id);\nCREATE INDEX idx_exam_arrangements_course_id ON exam_arrangements(course_id);\nCREATE INDEX idx_scores_student_id ON scores(student_id);\nCREATE INDEX idx_scores_exam_id ON scores(exam_id);\n\n-- 添加外键约束\nALTER TABLE students ADD CONSTRAINT fk_students_major_id FOREIGN KEY (major_id) REFERENCES majors(major_id);\nALTER TABLE students ADD CONSTRAINT fk_students_class_id FOREIGN KEY (class_id) REFERENCES classes(class_id);\nALTER TABLE majors ADD CONSTRAINT fk_majors_department_id FOREIGN KEY (department_id) REFERENCES departments(department_id);\nALTER TABLE classes ADD CONSTRAINT fk_classes_major_id FOREIGN KEY (major_id) REFERENCES majors(major_id)\n\n-- 创建教师表\nCREATE TABLE teachers (\n                          teacher_id INT PRIMARY KEY AUTO_INCREMENT,\n                          teacher_no VARCHAR(20) UNIQUE NOT NULL COMMENT '教师编号',\n                          name VARCHAR(50) NOT NULL COMMENT '姓名',\n                          gender ENUM('男', '女', '其他') COMMENT '性别',\n                          title VARCHAR(50) COMMENT '职称',\n                          department_id INT NOT NULL COMMENT '所属院系ID',\n                          email VARCHAR(100) UNIQUE COMMENT '邮箱',\n                          phone VARCHAR(20) COMMENT '电话',\n                          created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',\n                          updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间'\n);\n\n-- 创建考试安排表\nCREATE TABLE exam_arrangements (\n                                   exam_id INT PRIMARY KEY AUTO_INCREMENT,\n                                   course_id INT NOT NULL COMMENT '课程ID',\n                                   exam_date DATE NOT NULL COMMENT '考试日期',\n                                   start_time TIME NOT NULL COMMENT '开始时间',\n                                   end_time TIME NOT NULL COMMENT '结束时间',\n                                   exam_room VARCHAR(50) NOT NULL COMMENT '考场',\n                                   invigilator VARCHAR(100) COMMENT '监考老师',\n                                   exam_type ENUM('期中', '期末', '补考', '重修') NOT NULL COMMENT '考试类型',\n                                   academic_year VARCHAR(20) NOT NULL COMMENT '学年',\n                                   semester TINYINT NOT NULL COMMENT '学期',\n                                   created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',\n                                   updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间'\n)\n\n-- 创建学生信息表\nCREATE TABLE students (\n                          student_id INT PRIMARY KEY AUTO_INCREMENT,\n                          student_no VARCHAR(20) UNIQUE NOT NULL COMMENT '学号',\n                          name VARCHAR(50) NOT NULL COMMENT '姓名',\n                          gender ENUM('男', '女', '其他') COMMENT '性别',\n                          birth_date DATE COMMENT '出生日期',\n                          email VARCHAR(100) UNIQUE COMMENT '邮箱',\n                          phone VARCHAR(20) COMMENT '电话',\n                          address VARCHAR(200) COMMENT '地址',\n                          enrollment_date DATE NOT NULL COMMENT '入学日期',\n                          major_id INT NOT NULL COMMENT '专业ID',\n                          class_id INT NOT NULL COMMENT '班级ID',\n                          status ENUM('在读', '休学', '毕业', '退学') NOT NULL DEFAULT '在读' COMMENT '状态',\n                          created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',\n                          updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间'\n);\n\n-- 创建专业表\nCREATE TABLE majors (\n                        major_id INT PRIMARY KEY AUTO_INCREMENT,\n                        major_name VARCHAR(50) UNIQUE NOT NULL COMMENT '专业名称',\n                        department_id INT NOT NULL COMMENT '院系ID',\n                        created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',\n                        updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间'\n)\n\n-- 创建院系表\nCREATE TABLE departments (\n                             department_id INT PRIMARY KEY AUTO_INCREMENT,\n                             department_name VARCHAR(50) UNIQUE NOT NULL COMMENT '院系名称',\n                             dean VARCHAR(50) COMMENT '院长',\n                             phone VARCHAR(20) COMMENT '联系电话',\n                             email VARCHAR(100) COMMENT '邮箱',\n                             created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',\n                             updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间'\n);\n\n-- 创建班级表\nCREATE TABLE classes (\n                         class_id INT PRIMARY KEY AUTO_INCREMENT,\n                         class_name VARCHAR(50) NOT NULL COMMENT '班级名称',\n                         major_id INT NOT NULL COMMENT '专业ID',\n                         grade INT NOT NULL COMMENT '年级',\n                         head_teacher VARCHAR(50) COMMENT '班主任',\n                         created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',\n                         updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间'\n);\n\n-- 创建课程表\nCREATE TABLE courses (\n                         course_id INT PRIMARY KEY AUTO_INCREMENT,\n                         course_code VARCHAR(20) UNIQUE NOT NULL COMMENT '课程代码',\n                         course_name VARCHAR(100) NOT NULL COMMENT '课程名称',\n                         credit TINYINT NOT NULL COMMENT '学分',\n                         course_type ENUM('必修课', '选修课', '公共课') NOT NULL COMMENT '课程类型',\n                         department_id INT NOT NULL COMMENT '开课院系ID',\n                         teacher_id INT NOT NULL COMMENT '授课教师ID',\n                         created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',\n                         updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间'\n)"
  } ],
  "stream" : false
}

[main] INFO dev.langchain4j.http.client.log.LoggingHttpClient -- HTTP response:
- status code: 200
- headers: [:status: 200], [content-length: 2644], [content-type: application/json], [date: Wed, 11 Jun 2025 02:20:11 GMT], [req-arrive-time: 1749608380626], [req-cost-time: 31767], [resp-start-time: 1749608412394], [server: istio-envoy], [set-cookie: acw_tc=c98b1f3b-b6e6-9271-a000-d232726b3d2c47d53afe44ffd30275d4f0a5315ca361;path=/;HttpOnly;Max-Age=1800], [vary: Origin,Access-Control-Request-Method,Access-Control-Request-Headers, Accept-Encoding], [x-dashscope-call-gateway: true], [x-envoy-upstream-service-time: 31762], [x-request-id: c98b1f3b-b6e6-9271-a000-d232726b3d2c]
- body: {"choices":[{"message":{"role":"assistant","content":"根据提供的DDL和问题"张铁牛上学期语文考试成绩是多少"，我们需要从多个表中提取相关信息。以下是解决问题的步骤：\n\n1. **确定学生ID**：通过`students`表找到"张铁牛"的`student_id`。\n2. **确定课程ID**：通过`courses`表找到"语文"课程的`course_id`。\n3. **确定考试ID**：通过`exam_arrangements`表找到与"语文"课程相关的考试ID，并确保是"上学期"的考试（可以通过`semester`字段判断）。\n4. **查询成绩**：通过`scores`表找到该学生在该考试中的成绩。\n\n以下是完整的SQL语句：\n\n```sql\n-- Step 1: 获取张铁牛的学生ID\nSELECT student_id \nFROM students \nWHERE name = '张铁牛';\n\n-- Step 2: 获取语文课程的课程ID\nSELECT course_id \nFROM courses \nWHERE course_name = '语文';\n\n-- Step 3: 获取上学期语文考试的考试ID\nSELECT exam_id \nFROM exam_arrangements \nWHERE course_id = (SELECT course_id FROM courses WHERE course_name = '语文') \n  AND semester = (SELECT MAX(semester) FROM exam_arrangements WHERE academic_year = (SELECT academic_year FROM exam_arrangements ORDER BY exam_date DESC LIMIT 1));\n\n-- Step 4: 查询张铁牛上学期语文考试的成绩\nSELECT s.score \nFROM scores s\nJOIN students st ON s.student_id = st.student_id\nJOIN exam_arrangements ea ON s.exam_id = ea.exam_id\nJOIN courses c ON ea.course_id = c.course_id\nWHERE st.name = '张铁牛'\n  AND c.course_name = '语文'\n  AND ea.semester = (SELECT MAX(semester) FROM exam_arrangements WHERE academic_year = (SELECT academic_year FROM exam_arrangements ORDER BY exam_date DESC LIMIT 1));\n```\n\n### 解释：\n1. **Step 1**：通过`students`表查找"张铁牛"的`student_id`。\n2. **Step 2**：通过`courses`表查找"语文"课程的`course_id`。\n3. **Step 3**：通过`exam_arrangements`表查找与"语文"课程相关且属于"上学期"的考试ID。\n   - 使用`MAX(semester)`确保获取的是最近学年的上学期考试。\n4. **Step 4**：将以上结果联合查询，从`scores`表中提取张铁牛在上学期语文考试中的成绩。\n\n请确保数据库中已存在相关数据以供查询。如果需要进一步调整或补充信息，请告知！"},"finish_reason":"stop","index":0,"logprobs":null}],"object":"chat.completion","usage":{"prompt_tokens":1498,"completion_tokens":558,"total_tokens":2056,"prompt_tokens_details":{"cached_tokens":0}},"created":1749608412,"system_fingerprint":null,"model":"qwen-plus","id":"chatcmpl-c98b1f3b-b6e6-9271-a000-d232726b3d2c"}

[main] INFO com.genuai.langchaintest.rag.AiRagTest -- call ai q:张铁牛上学期语文考试成绩是多少
a: 根据提供的DDL和问题"张铁牛上学期语文考试成绩是多少"，我们需要从多个表中提取相关信息。以下是解决问题的步骤：

1. **确定学生ID**：通过`students`表找到"张铁牛"的`student_id`。
2. **确定课程ID**：通过`courses`表找到"语文"课程的`course_id`。
3. **确定考试ID**：通过`exam_arrangements`表找到与"语文"课程相关的考试ID，并确保是"上学期"的考试（可以通过`semester`字段判断）。
4. **查询成绩**：通过`scores`表找到该学生在该考试中的成绩。

以下是完整的SQL语句：

```sql
-- Step 1: 获取张铁牛的学生ID
SELECT student_id 
FROM students 
WHERE name = '张铁牛';

-- Step 2: 获取语文课程的课程ID
SELECT course_id 
FROM courses 
WHERE course_name = '语文';

-- Step 3: 获取上学期语文考试的考试ID
SELECT exam_id 
FROM exam_arrangements 
WHERE course_id = (SELECT course_id FROM courses WHERE course_name = '语文') 
  AND semester = (SELECT MAX(semester) FROM exam_arrangements WHERE academic_year = (SELECT academic_year FROM exam_arrangements ORDER BY exam_date DESC LIMIT 1));

-- Step 4: 查询张铁牛上学期语文考试的成绩
SELECT s.score 
FROM scores s
JOIN students st ON s.student_id = st.student_id
JOIN exam_arrangements ea ON s.exam_id = ea.exam_id
JOIN courses c ON ea.course_id = c.course_id
WHERE st.name = '张铁牛'
  AND c.course_name = '语文'
  AND ea.semester = (SELECT MAX(semester) FROM exam_arrangements WHERE academic_year = (SELECT academic_year FROM exam_arrangements ORDER BY exam_date DESC LIMIT 1));
```

### 解释：
1. **Step 1**：通过`students`表查找"张铁牛"的`student_id`。
2. **Step 2**：通过`courses`表查找"语文"课程的`course_id`。
3. **Step 3**：通过`exam_arrangements`表查找与"语文"课程相关且属于"上学期"的考试ID。
   - 使用`MAX(semester)`确保获取的是最近学年的上学期考试。
4. **Step 4**：将以上结果联合查询，从`scores`表中提取张铁牛在上学期语文考试中的成绩。

请确保数据库中已存在相关数据以供查询。如果需要进一步调整或补充信息，请告知！

小结

本章通过使用RAG特性实现一个简单的text-sql功能实现, 直观的感受了一下RAG的强大之处, 下一章我们将使用SpringBoot实现快速集成LangChain4J, 通过简单的配置即可实现AI的调用。

student_ddl.sql

-- 创建学生信息表
CREATE TABLE students (
      student_id INT PRIMARY KEY AUTO_INCREMENT,
      student_no VARCHAR(20) UNIQUE NOT NULL COMMENT '学号',
      name VARCHAR(50) NOT NULL COMMENT '姓名',
      gender ENUM('男', '女', '其他') COMMENT '性别',
      birth_date DATE COMMENT '出生日期',
      email VARCHAR(100) UNIQUE COMMENT '邮箱',
      phone VARCHAR(20) COMMENT '电话',
      address VARCHAR(200) COMMENT '地址',
      enrollment_date DATE NOT NULL COMMENT '入学日期',
      major_id INT NOT NULL COMMENT '专业ID',
      class_id INT NOT NULL COMMENT '班级ID',
      status ENUM('在读', '休学', '毕业', '退学') NOT NULL DEFAULT '在读' COMMENT '状态',
      created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',
      updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间'
);

-- 创建专业表
CREATE TABLE majors (
      major_id INT PRIMARY KEY AUTO_INCREMENT,
      major_name VARCHAR(50) UNIQUE NOT NULL COMMENT '专业名称',
      department_id INT NOT NULL COMMENT '院系ID',
      created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',
      updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间'
);

-- 创建院系表
CREATE TABLE departments (
      department_id INT PRIMARY KEY AUTO_INCREMENT,
      department_name VARCHAR(50) UNIQUE NOT NULL COMMENT '院系名称',
      dean VARCHAR(50) COMMENT '院长',
      phone VARCHAR(20) COMMENT '联系电话',
      email VARCHAR(100) COMMENT '邮箱',
      created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',
      updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间'
);

-- 创建班级表
CREATE TABLE classes (
      class_id INT PRIMARY KEY AUTO_INCREMENT,
      class_name VARCHAR(50) NOT NULL COMMENT '班级名称',
      major_id INT NOT NULL COMMENT '专业ID',
      grade INT NOT NULL COMMENT '年级',
      head_teacher VARCHAR(50) COMMENT '班主任',
      created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',
      updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间'
);

-- 创建课程表
CREATE TABLE courses (
      course_id INT PRIMARY KEY AUTO_INCREMENT,
      course_code VARCHAR(20) UNIQUE NOT NULL COMMENT '课程代码',
      course_name VARCHAR(100) NOT NULL COMMENT '课程名称',
      credit TINYINT NOT NULL COMMENT '学分',
      course_type ENUM('必修课', '选修课', '公共课') NOT NULL COMMENT '课程类型',
      department_id INT NOT NULL COMMENT '开课院系ID',
      teacher_id INT NOT NULL COMMENT '授课教师ID',
      created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',
      updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间'
);

-- 创建教师表
CREATE TABLE teachers (
      teacher_id INT PRIMARY KEY AUTO_INCREMENT,
      teacher_no VARCHAR(20) UNIQUE NOT NULL COMMENT '教师编号',
      name VARCHAR(50) NOT NULL COMMENT '姓名',
      gender ENUM('男', '女', '其他') COMMENT '性别',
      title VARCHAR(50) COMMENT '职称',
      department_id INT NOT NULL COMMENT '所属院系ID',
      email VARCHAR(100) UNIQUE COMMENT '邮箱',
      phone VARCHAR(20) COMMENT '电话',
      created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',
      updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间'
);

-- 创建考试安排表
CREATE TABLE exam_arrangements (
      exam_id INT PRIMARY KEY AUTO_INCREMENT,
      course_id INT NOT NULL COMMENT '课程ID',
      exam_date DATE NOT NULL COMMENT '考试日期',
      start_time TIME NOT NULL COMMENT '开始时间',
      end_time TIME NOT NULL COMMENT '结束时间',
      exam_room VARCHAR(50) NOT NULL COMMENT '考场',
      invigilator VARCHAR(100) COMMENT '监考老师',
      exam_type ENUM('期中', '期末', '补考', '重修') NOT NULL COMMENT '考试类型',
      academic_year VARCHAR(20) NOT NULL COMMENT '学年',
      semester TINYINT NOT NULL COMMENT '学期',
      created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',
      updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间'
);

-- 创建成绩表
CREATE TABLE scores (
      score_id INT PRIMARY KEY AUTO_INCREMENT,
      student_id INT NOT NULL COMMENT '学生ID',
      exam_id INT NOT NULL COMMENT '考试ID',
      score DECIMAL(5,2) COMMENT '成绩',
      score_type ENUM('原始分', '平时分', '卷面分', '最终分') NOT NULL DEFAULT '最终分' COMMENT '成绩类型',
      remark VARCHAR(200) COMMENT '备注',
      created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',
      updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间',
      UNIQUE KEY unique_student_exam (student_id, exam_id)
);

-- 创建索引
CREATE INDEX idx_students_student_no ON students(student_no);
CREATE INDEX idx_students_major_id ON students(major_id);
CREATE INDEX idx_students_class_id ON students(class_id);
CREATE INDEX idx_majors_department_id ON majors(department_id);
CREATE INDEX idx_classes_major_id ON classes(major_id);
CREATE INDEX idx_courses_department_id ON courses(department_id);
CREATE INDEX idx_courses_teacher_id ON courses(teacher_id);
CREATE INDEX idx_teachers_department_id ON teachers(department_id);
CREATE INDEX idx_exam_arrangements_course_id ON exam_arrangements(course_id);
CREATE INDEX idx_scores_student_id ON scores(student_id);
CREATE INDEX idx_scores_exam_id ON scores(exam_id);

-- 添加外键约束
ALTER TABLE students ADD CONSTRAINT fk_students_major_id FOREIGN KEY (major_id) REFERENCES majors(major_id);
ALTER TABLE students ADD CONSTRAINT fk_students_class_id FOREIGN KEY (class_id) REFERENCES classes(class_id);
ALTER TABLE majors ADD CONSTRAINT fk_majors_department_id FOREIGN KEY (department_id) REFERENCES departments(department_id);
ALTER TABLE classes ADD CONSTRAINT fk_classes_major_id FOREIGN KEY (major_id) REFERENCES majors(major_id);
ALTER TABLE courses ADD CONSTRAINT fk_courses_department_id FOREIGN KEY (department_id) REFERENCES departments(department_id);
ALTER TABLE courses ADD CONSTRAINT fk_courses_teacher_id FOREIGN KEY (teacher_id) REFERENCES teachers(teacher_id);
ALTER TABLE teachers ADD CONSTRAINT fk_teachers_department_id FOREIGN KEY (department_id) REFERENCES departments(department_id);
ALTER TABLE exam_arrangements ADD CONSTRAINT fk_exam_arrangements_course_id FOREIGN KEY (course_id) REFERENCES courses(course_id);
ALTER TABLE scores ADD CONSTRAINT fk_scores_student_id FOREIGN KEY (student_id) REFERENCES students(student_id);
ALTER TABLE scores ADD CONSTRAINT fk_scores_exam_id FOREIGN KEY (exam_id) REFERENCES exam_arrangements(exam_id);