Home

Awesome

it’s an “unofficial" or "community-maintained” library. 这是一个非官方的社区维护的库。有bug欢迎朋友们指出，互相学习。

原创发布转载注明出处！

文档地址：https://chatgpt-java.unfbx.com/

星火大模型Java SDK: https://github.com/Grt1228/SparkDesk-Java

To English Doc -> English Doc

💬 联系我，ChatGPT相关学习交流咨询全部免费。

一起探讨chatgpt-java，SDK问题咨询<br/>项目产品开发交流群失效关注公众号恢复：chatgpt-java
<img src="https://user-images.githubusercontent.com/27008803/225246389-7b452214-f3fe-4a70-bd3e-832a0ed34288.jpg" width="210" height="300" alt="二维码" /><img src="https://g-photo.oss-cn-shanghai.aliyuncs.com/hd15.jpg" width="210" height="210" alt="二维码" />
<!--<img src="https://user-images.githubusercontent.com/27008803/225246581-15e90f78-5438-4637-8e7d-14c68ca13b59.jpg" width="210" height="300" alt="二维码" />-->

📖 项目简介

ChatGPT的Java客户端 OpenAI官方Api的Java SDK，可以快速接入项目使用。支持OpenAI官方全部接口，同时支持Tokens计算。参考文档：Tokens_README.md。

AssistantDall-e-3FineTuneJobTTSChatCompletionWithPicture
自定义知识库、AI助理AI生成图片微调job文本转语音附加图片的ChatCompletion
TikTokenChatCompletionsImagesSpeech To Text余额查询
Token计算GPT-3.5、4.0对话模型GPT-3.0对话图片模型语音转文字，语音翻译余额查询
EmbeddingsFilesModerationsFine-tuneModels
嵌入自定义训练模型文本审核，敏感词鉴别微调模型检索相关

支持流式输出：

流式输出实现方式小程序安卓iosH5
SSE参考：OpenAISSEEventSourceListener不支持支持支持支持
WebSocket参考：OpenAIWebSocketEventSourceListener支持支持支持支持

基于本SDK开发的Demo项目，整合Spring Boot 实现CahtGPT对话模式，思路可以参考： https://github.com/Grt1228/chatgpt-steam-output

🚩 特性

📑 更新日志

🚀 快速开始

本项目支持默认输出和流式输出。完整SDK测试案例参考：

SDK测试案例Tokens计算测试案例完整Function-Call调用案例
OpenAiClientTest 和OpenAiStreamClientTestTokens计算参考：TikTokensTestOpenAiClientFunctionTest

方式一

1、导入pom依赖

<dependency>
    <groupId>com.unfbx</groupId>
    <artifactId>chatgpt-java</artifactId>
    <version>1.0.14-beta1</version>
</dependency>

2、流式客户端使用示例：

更多SDK示例参考：OpenAiStreamClientTest

默认OkHttpClient

public class Test {
    public static void main(String[] args) {
        OpenAiStreamClient client = OpenAiStreamClient.builder()
                .apiKey(Arrays.asList("sk-********","sk-********"))
                //自定义key的获取策略：默认KeyRandomStrategy
                //.keyStrategy(new KeyRandomStrategy())
                .keyStrategy(new FirstKeyStrategy())
                //自己做了代理就传代理地址，没有可不不传
//                .apiHost("https://自己代理的服务器地址/")
                .build();
        //聊天模型：gpt-3.5
        ConsoleEventSourceListener eventSourceListener = new ConsoleEventSourceListener();
        Message message = Message.builder().role(Message.Role.USER).content("你好啊我的伙伴！").build();
        ChatCompletion chatCompletion = ChatCompletion.builder().messages(Arrays.asList(message)).build();
        client.streamChatCompletion(chatCompletion, eventSourceListener);
        CountDownLatch countDownLatch = new CountDownLatch(1);
        try {
            countDownLatch.await();
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
    }
}

自定义OkHttpClient客户端使用示例：

public class Test {
    public static void main(String[] args) {
        //国内访问需要做代理，国外服务器不需要
        Proxy proxy = new Proxy(Proxy.Type.HTTP, new InetSocketAddress("127.0.0.1", 7890));
        HttpLoggingInterceptor httpLoggingInterceptor = new HttpLoggingInterceptor(new OpenAILogger());
        //！！！！千万别再生产或者测试环境打开BODY级别日志！！！！
        //！！！生产或者测试环境建议设置为这三种级别：NONE,BASIC,HEADERS,！！！
        httpLoggingInterceptor.setLevel(HttpLoggingInterceptor.Level.HEADERS);
        OkHttpClient okHttpClient = new OkHttpClient
                .Builder()
                .proxy(proxy)//自定义代理
                .addInterceptor(httpLoggingInterceptor)//自定义日志
                .connectTimeout(30, TimeUnit.SECONDS)//自定义超时时间
                .writeTimeout(30, TimeUnit.SECONDS)//自定义超时时间
                .readTimeout(30, TimeUnit.SECONDS)//自定义超时时间
                .build();
        OpenAiStreamClient client = OpenAiStreamClient.builder()
                .apiKey(Arrays.asList("sk-********","sk-********"))
                //自定义key的获取策略：默认KeyRandomStrategy
                //.keyStrategy(new KeyRandomStrategy())
                .keyStrategy(new FirstKeyStrategy())
                .okHttpClient(okHttpClient)
                //自己做了代理就传代理地址，没有可不不传
//                .apiHost("https://自己代理的服务器地址/")
                .build();
    }
}

输出日志（text是持续输出的）：

23:03:59.158 [省略无效信息] INFO com.unfbx.chatgpt.sse.ConsoleEventSourceListener - OpenAI建立sse连接...
23:03:59.160 [省略无效信息] INFO com.unfbx.chatgpt.sse.ConsoleEventSourceListener - OpenAI返回数据：{"id": "cmpl-6pIHnOOJiiUEVMesXwxzzcSQFoZHj", "object": "text_completion", "created": 1677683039, "choices": [{"text": "\n", "index": 0, "logprobs": null, "finish_reason": null}], "model": "text-davinci-003"}
23:03:59.172 [省略无效信息] INFO com.unfbx.chatgpt.sse.ConsoleEventSourceListener - OpenAI返回数据：{"id": "cmpl-6pIHnOOJiiUEVMesXwxzzcSQFoZHj", "object": "text_completion", "created": 1677683039, "choices": [{"text": "\n", "index": 0, "logprobs": null, "finish_reason": null}], "model": "text-davinci-003"}
23:03:59.251 [省略无效信息] INFO com.unfbx.chatgpt.sse.ConsoleEventSourceListener - OpenAI返回数据：{"id": "cmpl-6pIHnOOJiiUEVMesXwxzzcSQFoZHj", "object": "text_completion", "created": 1677683039, "choices": [{"text": "\u5fc3", "index": 0, "logprobs": null, "finish_reason": null}], "model": "text-davinci-003"}
23:03:59.313 [省略无效信息] INFO com.unfbx.chatgpt.sse.ConsoleEventSourceListener - OpenAI返回数据：{"id": "cmpl-6pIHnOOJiiUEVMesXwxzzcSQFoZHj", "object": "text_completion", "created": 1677683039, "choices": [{"text": "\u60c5", "index": 0, "logprobs": null, "finish_reason": null}], "model": "text-davinci-003"}
23:03:59.380 [省略无效信息] INFO com.unfbx.chatgpt.sse.ConsoleEventSourceListener - OpenAI返回数据：{"id": "cmpl-6pIHnOOJiiUEVMesXwxzzcSQFoZHj", "object": "text_completion", "created": 1677683039, "choices": [{"text": "\u8212", "index": 0, "logprobs": null, "finish_reason": null}], "model": "text-davinci-003"}
23:03:59.439 [省略无效信息] INFO com.unfbx.chatgpt.sse.ConsoleEventSourceListener - OpenAI返回数据：{"id": "cmpl-6pIHnOOJiiUEVMesXwxzzcSQFoZHj", "object": "text_completion", "created": 1677683039, "choices": [{"text": "\u7545", "index": 0, "logprobs": null, "finish_reason": null}], "model": "text-davinci-003"}
23:03:59.532 [省略无效信息] INFO com.unfbx.chatgpt.sse.ConsoleEventSourceListener - OpenAI返回数据：{"id": "cmpl-6pIHnOOJiiUEVMesXwxzzcSQFoZHj", "object": "text_completion", "created": 1677683039, "choices": [{"text": "\uff0c", "index": 0, "logprobs": null, "finish_reason": null}], "model": "text-davinci-003"}
23:03:59.579 [省略无效信息] INFO com.unfbx.chatgpt.sse.ConsoleEventSourceListener - OpenAI返回数据：{"id": "cmpl-6pIHnOOJiiUEVMesXwxzzcSQFoZHj", "object": "text_completion", "created": 1677683039, "choices": [{"text": "\u5fc3", "index": 0, "logprobs": null, "finish_reason": null}], "model": "text-davinci-003"}
23:03:59.641 [省略无效信息] INFO com.unfbx.chatgpt.sse.ConsoleEventSourceListener - OpenAI返回数据：{"id": "cmpl-6pIHnOOJiiUEVMesXwxzzcSQFoZHj", "object": "text_completion", "created": 1677683039, "choices": [{"text": "\u65f7", "index": 0, "logprobs": null, "finish_reason": null}], "model": "text-davinci-003"}
23:03:59.673 [省略无效信息] INFO com.unfbx.chatgpt.sse.ConsoleEventSourceListener - OpenAI返回数据：{"id": "cmpl-6pIHnOOJiiUEVMesXwxzzcSQFoZHj", "object": "text_completion", "created": 1677683039, "choices": [{"text": "\u795e", "index": 0, "logprobs": null, "finish_reason": null}], "model": "text-davinci-003"}
23:03:59.751 [省略无效信息] INFO com.unfbx.chatgpt.sse.ConsoleEventSourceListener - OpenAI返回数据：{"id": "cmpl-6pIHnOOJiiUEVMesXwxzzcSQFoZHj", "object": "text_completion", "created": 1677683039, "choices": [{"text": "\u6021", "index": 0, "logprobs": null, "finish_reason": null}], "model": "text-davinci-003"}
23:03:59.782 [省略无效信息] INFO com.unfbx.chatgpt.sse.ConsoleEventSourceListener - OpenAI返回数据：{"id": "cmpl-6pIHnOOJiiUEVMesXwxzzcSQFoZHj", "object": "text_completion", "created": 1677683039, "choices": [{"text": "\u3002", "index": 0, "logprobs": null, "finish_reason": null}], "model": "text-davinci-003"}
23:03:59.815 [省略无效信息] INFO com.unfbx.chatgpt.sse.ConsoleEventSourceListener - OpenAI返回数据：[DONE]
23:03:59.815 [省略无效信息] INFO com.unfbx.chatgpt.sse.ConsoleEventSourceListener - OpenAI返回数据结束了
23:03:59.815 [省略无效信息] INFO com.unfbx.chatgpt.sse.ConsoleEventSourceListener - OpenAI关闭sse连接...

3、默认客户端使用示例（支持全部API）：

更多SDK示例参考：OpenAiClientTest

默认OkHttpClient

public class Test {
    public static void main(String[] args) {
        OpenAiClient openAiClient = OpenAiClient.builder()
                .apiKey(Arrays.asList("sk-********","sk-********"))
                //自定义key的获取策略：默认KeyRandomStrategy
                //.keyStrategy(new KeyRandomStrategy())
                .keyStrategy(new FirstKeyStrategy())
                //自己做了代理就传代理地址，没有可不不传
//                .apiHost("https://自己代理的服务器地址/")
                .build();
                //聊天模型：gpt-3.5
        Message message = Message.builder().role(Message.Role.USER).content("你好啊我的伙伴！").build();
        ChatCompletion chatCompletion = ChatCompletion.builder().messages(Arrays.asList(message)).build();
        ChatCompletionResponse chatCompletionResponse = openAiClient.chatCompletion(chatCompletion);
        chatCompletionResponse.getChoices().forEach(e -> {
            System.out.println(e.getMessage());
        });
    }
}

自定义OkHttpClient客户端使用示例：

public class Test {
    public static void main(String[] args) {
        //国内访问需要做代理，国外服务器不需要
        Proxy proxy = new Proxy(Proxy.Type.HTTP, new InetSocketAddress("127.0.0.1", 7890));
        HttpLoggingInterceptor httpLoggingInterceptor = new HttpLoggingInterceptor(new OpenAILogger());
        //！！！！千万别再生产或者测试环境打开BODY级别日志！！！！
        //！！！生产或者测试环境建议设置为这三种级别：NONE,BASIC,HEADERS,！！！
        httpLoggingInterceptor.setLevel(HttpLoggingInterceptor.Level.HEADERS);
        OkHttpClient okHttpClient = new OkHttpClient
                .Builder()
                .proxy(proxy)//自定义代理
                .addInterceptor(httpLoggingInterceptor)//自定义日志输出
                .addInterceptor(new OpenAiResponseInterceptor())//自定义返回值拦截
                .connectTimeout(10, TimeUnit.SECONDS)//自定义超时时间
                .writeTimeout(30, TimeUnit.SECONDS)//自定义超时时间
                .readTimeout(30, TimeUnit.SECONDS)//自定义超时时间
                .build();
        //构建客户端
        OpenAiClient openAiClient = OpenAiClient.builder()
                .apiKey(Arrays.asList("sk-********","sk-********"))
                //自定义key的获取策略：默认KeyRandomStrategy
                //.keyStrategy(new KeyRandomStrategy())
                .keyStrategy(new FirstKeyStrategy())
                .okHttpClient(okHttpClient)
                //自己做了代理就传代理地址，没有可不不传
//                .apiHost("https://自己代理的服务器地址/")
                .build();
                //聊天模型：gpt-3.5
        Message message = Message.builder().role(Message.Role.USER).content("你好啊我的伙伴！").build();
        ChatCompletion chatCompletion = ChatCompletion.builder().messages(Arrays.asList(message)).build();
        ChatCompletionResponse chatCompletionResponse = openAiClient.chatCompletion(chatCompletion);
        chatCompletionResponse.getChoices().forEach(e -> {
            System.out.println(e.getMessage());
        });
    }
}

方式二（下载源码直接运行）

下载源码打包

❔ QA

QA
如何实现连续对话？issues：https://github.com/Grt1228/chatgpt-java/issues/8
如何实现流式输出？升级1.0.2版本，参考源码：OpenAiStreamClientTest
如何整合SpringBoot实现流式输出的Api接口？参考另外一个项目：chatgpt-steam-output
最新版GPT-3.5-TURBO是否支持？升级1.0.3 已经支持ChatCompletion, 参考测试案例：OpenAiStreamClientTest 和OpenAiStreamClientTest
最新版语言转文字和语言翻译是否支持？升级1.0.3 已经支持whisper参考测试案例：OpenAiStreamClientTest 和OpenAiStreamClientTest

📌 国内访问解决方案

国内访问可以看下这个解决方案：https://github.com/noobnooc/noobnooc/discussions/9

📋 开发案例收集

基于本SDK开发案例收集：chatgpt-java SDK案例征集

🌟 Star History

Star History Chart

🙏 鸣谢

站在巨人的肩膀：

☕ 打赏

如果项目对你有帮助，可以选择请我喝杯奶茶

<img width="180" alt="微信截图_20230405222411" src="https://user-images.githubusercontent.com/27008803/230111508-3179cf30-e128-4b2e-9645-157266c491ce.png"> <img width="164" alt="微信截图_20230405222357" src="https://user-images.githubusercontent.com/27008803/230111525-322f5036-d06d-46bb-94d1-db8ce9ed2adf.png">

开源协议：LICENSE