规划一个聊天机器人
智能化完全于依托于GPT, 而产品化是我们需要考虑的事情 比如，如何去构建一个聊天机器人 聊天机器人它的处理逻辑其实非常的清晰 我们输入问题调用 GPT 然后，GPT 给我们生成回答就可以了 需要注意的是，聊天机器人不同于调用API进行一个简单的测试 我们和聊天机器人的对话，可能是多轮的一个对话 在这时候，我们去调用API的时候，就需要将我们多轮的问答都传递给GPT才行新增一些实现类，结构如下
以下Java版代码来源于网络，可基于此逻辑，改造成其他编程语言
src
main java com.xxx.gpt.client util ChatContextHolder.java ChatBotClient.java … test java com.xxx.gpt.client.test FunctionCallTest.java …ChatContextHolder.java
package com.xxx.gpt.client.util;
import com.xxx.gpt.client.entity.Message;
import java.util.ArrayList;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
public class ChatContextHolder {
private static Map<String, List<Message>> context = new HashMap<>();
public static List<Message> get(String id) {
// TODO 限制轮数，或者限制token数量
List<Message> messages = context.get(id);
if (messages == null) {
messages = new ArrayList<>();
context.put(id, messages);
}
return messages;
}
public static void add(String id, String msg) {
Message message = Message.builder().content(msg).build();
add(id, message);
}
public static void add(String id, Message message) {
List<Message> messages = context.get(id);
if (messages == null) {
messages = new ArrayList<>();
context.put(id, messages);
}
messages.add(message);
}
public static void remove(String id) {
context.remove(id);
}
}
ChatBotClient.java
package com.xxx.gpt.client;
import com.xxx.gpt.client.entity.Message;
import com.xxx.gpt.client.listener.ConsoleStreamListener;
import com.xxx.gpt.client.util.ChatContextHolder;
import com.xxx.gpt.client.util.Proxys;
import lombok.SneakyThrows;
import lombok.extern.slf4j.Slf4j;
import java.io.BufferedReader;
import java.io.IOException;
import java.io.InputStreamReader;
import java.net.Proxy;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.List;
import java.util.UUID;
import java.util.concurrent.CountDownLatch;
import java.util.stream.Collectors;
@Slf4j
public class ChatBotClient {
public static Proxy proxy = Proxy.NO_PROXY;
public static void main(String[] args) {
System.out.println("ChatGPT - Java command-line interface");
System.out.println("Press enter twice to submit your question.");
System.out.println();
System.out.println("按两次回车以提交您的问题！！！");
String chatUuid = UUID.randomUUID().toString();
String key = "sk-adfas";
proxy = Proxys.http("127.0.0.1", 7890);
while (true) {
String prompt = getInput("\nYou:\n");
ChatGPTStreamClient chatGPT = ChatGPTStreamClient.builder()
.apiKey(key)
.proxy(proxy)
.build()
.init();
System.out.println("AI: ");
// 卡住
CountDownLatch countDownLatch = new CountDownLatch(1);
Message message = Message.of(prompt);
ChatContextHolder.add(chatUuid, message);
ConsoleStreamListener listener = new ConsoleStreamListener() {
@Override
public void onError(Throwable throwable, String response) {
throwable.printStackTrace();
countDownLatch.countDown();
}
};
listener.setOnComplate(msg -> {
ChatContextHolder.add(chatUuid, Message.ofAssistant(msg));
countDownLatch.countDown();
});
chatGPT.streamChatCompletion(ChatContextHolder.get(chatUuid), listener);
try {
countDownLatch.await();
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
@SneakyThrows
public static String getInput(String prompt) {
System.out.print(prompt);
BufferedReader reader = new BufferedReader(new InputStreamReader(System.in));
List<String> lines = new ArrayList<>();
String line;
try {
while ((line = reader.readLine()) != null && !line.isEmpty()) {
lines.add(line);
}
} catch (IOException e) {
e.printStackTrace();
}
return lines.stream().collect(Collectors.joining("\n"));
}
}