基于类定制本地Tool

1、pom.xml

<properties>
    <maven.compiler.source>17</maven.compiler.source>
    <maven.compiler.target>17</maven.compiler.target>
    <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
    <langchain4j.version>0.35.0</langchain4j.version>
</properties>

<parent>
    <groupId>org.springframework.boot</groupId>
    <artifactId>spring-boot-starter-parent</artifactId>
    <version>3.2.1</version>
</parent>

<dependencies>
    <dependency>
        <groupId>dev.langchain4j</groupId>
        <artifactId>langchain4j</artifactId>
        <version>${langchain4j.version}</version>
    </dependency>
    <dependency>
        <groupId>dev.langchain4j</groupId>
        <artifactId>langchain4j-open-ai</artifactId>
        <version>${langchain4j.version}</version>
    </dependency>

  <dependency>
    <groupId>dev.langchain4j</groupId>
    <artifactId>langchain4j-dashscope</artifactId>
    <version>${langchain4j.version}</version>
    <exclusions>
        <exclusion>
            <artifactId>slf4j-simple</artifactId>
            <groupId>org.slf4j</groupId>
        </exclusion>
    </exclusions>
  </dependency>

    <dependency>
        <groupId>org.springframework.boot</groupId>
        <artifactId>spring-boot-starter-web</artifactId>
    </dependency>
    <dependency>
        <groupId>dev.langchain4j</groupId>
        <artifactId>langchain4j-dashscope-spring-boot-starter</artifactId>
        <version>0.35.0</version>
    </dependency>
</dependencies>

2、代码

Weather

import dev.langchain4j.agent.tool.P;
import dev.langchain4j.agent.tool.Tool;

public class Weather {

    @Tool("获取某一个具体城市的天气")
    public String getWeather(@P("指定的城市")String city) {
        return "明天 " + city + " 天气晴朗";

    }
}

WeatherToolDemo

import dev.langchain4j.agent.tool.ToolExecutionRequest;
import dev.langchain4j.agent.tool.ToolSpecification;
import dev.langchain4j.agent.tool.ToolSpecifications;
import dev.langchain4j.data.message.AiMessage;
import dev.langchain4j.data.message.ChatMessage;
import dev.langchain4j.data.message.ToolExecutionResultMessage;
import dev.langchain4j.data.message.UserMessage;
import dev.langchain4j.model.chat.ChatLanguageModel;
import dev.langchain4j.model.dashscope.QwenChatModel;
import dev.langchain4j.service.tool.DefaultToolExecutor;
import java.util.ArrayList;
import java.util.List;
import java.util.UUID;

public class WeatherToolDemo {

    public static void main(String [] args) {
        // 构建模型
        ChatLanguageModel model = QwenChatModel.builder()
                .apiKey("sk-3bfe0dfa79b94aa9b2d2da0a9286b9b1")
                .build();
        // 指定ToolSpecification
        List<ToolSpecification> toolSpecifications = ToolSpecifications.toolSpecificationsFrom(Weather.class);
        // 构建对话消息
        List<ChatMessage> chatMessages = new ArrayList<>();
        UserMessage userMessage = UserMessage.from("北京的天气怎么样？");
        chatMessages.add(userMessage);
        // 调用大模型，生成工具调用请求
        AiMessage aiMessage = model.generate(chatMessages, toolSpecifications).content();
        List<ToolExecutionRequest> toolExecutionRequests = aiMessage.toolExecutionRequests();
        toolExecutionRequests.forEach(toolExecutionRequest -> {
            System.out.println("调用工具方法： " + toolExecutionRequest.name());
            System.out.println("调用参数： " + toolExecutionRequest.arguments());
        });
        chatMessages.add(aiMessage);
        // 把工具请求以及工具的执行结果一起加入到对话消息队列表中
        Weather weather = new Weather();
        toolExecutionRequests.forEach(toolExecutionRequest -> {
            DefaultToolExecutor executor = new DefaultToolExecutor(weather, toolExecutionRequest);
            String result = executor.execute(toolExecutionRequest, UUID.randomUUID().toString());
            System.out.println("工具执行结果" + result);
            ToolExecutionResultMessage resultMessage = ToolExecutionResultMessage.from(toolExecutionRequest, result);
            chatMessages.add(resultMessage);
        });
        // 调用大模型，生成最终结果
        AiMessage message = model.generate(chatMessages).content();
        System.out.println("最终结果：" + message.text());
    }
}

3、验证结果

调用工具方法： getWeather
调用参数： {"city": "北京"}
工具执行结果明天 北京 天气晴朗
最终结果：北京明天的天气将会是晴朗的。如果你有出行计划，可以放心安排，记得做好防晒措施哦！需要我为你提供更详细的天气信息吗？