chatCLI - server

---

Project Topic

• Develop servers & commands that enable use of chatgpt in CLI environments (Terminal/Linux environments)

Project Purpose

• The API request format provided by open API in cli environment is too complicated.
• The response format to the request sent to open ai api is hard to see at a glance the answer to the question I want.
• Let's make an easy-to-use command considering user convenience. (Let's abstract the use of chatgpt)

Expectation effect

• You do not need to know the complex request format of open ai api.
  • Header, Body format etc..
• You can use open ai api conveniently in the cli environment.
• It is easy to see the complex response format of open aiap at a glance.

How to run it

git clone https://github.com/YGwan/spring-chatgpt-communication.git
cd spring-chatgpt-communication/shell
source chatgptCommand.sh 

• User's guide
  • English Version
  • Korean Version

• Developer's guide
  • English Version
  • Korean Version

Project Structure

1. CLI ENV
   • It is an environment in which the user executes the desired commands.
   • Send a request in question format and show the response to the question.
   • UI Roles

2. Spring Server
   • ake the questions requested by the user and parse them into request packets in the format desired by the OpenAi Server.
     • OpenAiRequestEntity class
   • Parse the response to an OpenAi Server request sent by the server into a user-friendly response packet.
     • OpenAiResponseEntity class

3. OpenAi Server
   • It is actually a server operated by Open Ai.(Server with chatGPT service)
   • The server provides api for several services.
   • More api can be found at the corresponding address

technology of use

Springboot, ResTemplate, Shell Script, HTTPS

Core Technology

1. ObjectMapper
   • It is a technique used to serialize responses and deserialize requests when using the JSON format.
   • That is, it provides a function of performing conversion between JSON data and Java objects.
   • Class of Jackson Library.
   • When converting JSON data to Java objects, use the readValue() method of the ObjectMapper class.
   • When converting Java objects into JSON data, use the writeValue() method of the ObjectMapper class.

2. ResTemplate
   • Technology used to send HTTP requests for RESTful web services and receive responses.
   • In other words, it is a technology that is used when an external API is used in the spring and a request is sent to the server and a response to the request is required.
   • RestTemplate provides a variety of methods for creating and processing HTTP requests in an easy way.(GET, POST, PUT, DELETE etc..)
   • Headers, parameters, body, etc. required for the request can be set.
   • When receiving a response, the status code, header, and body of the response can be checked.
   • In the future, we will also add technology using WebClient instead of that technology. (Spring 5: RestTemplate Web Client)

3. Chatgpt Question Session Maintenance Method

   • When you ask a question on chatgpt, there is a problem in the CLI environment that the content of the previous question is not connected to the next question.
   • Method 1: Connect the questions. -> This also does not work normally because the answer to the question is not connected.
   • Method 2: Connect questions + answers to those questions. -> It works normally.
   
   

    public ResponseEntity<String> chat(@RequestBody String request) throws JsonProcessingException {
       chatRequest = chatRequest + request; // 이전 질문(chatRequest)을 현재 질문 request에 연결한다.
       HttpEntity<String> openAiRequest = openAiRequestEntity.chatParsed(chatRequest); // open ai api 요청 패킷을 만든다.
       ResponseEntity<String> openAiResponseEntity = this.openAiResponseEntity.chatParsed(openAiRequest); // 위에서 만든 패킷으로 open ai api에 요청을 보낸 후 결과값을 받아온다.(답만 파싱해서 받는다.)
       chatRequest = chatRequest + openAiResponseEntity.getBody(); // 답 또한 질문과 연결한다.
       return openAiResponseEntity;
    }

4. Open ai api server communication(CORS Policy)
   
   
   • By default, OpenAI servers are open to HTTPS communication.
   • So if you want to communicate using HTTP, you cannot communicate with the openAI server.
   • However, since the OpenAI server allowed LOCALHOST communication, it is possible to communicate using LOCALHOST.(LOCALHOST uses HTTP by default)

Supported Commands

Command	Command Format	Description
Help	chatAsk-help	Show user manual.
Ask questions	chatAsk -m "Content of questions"	chatgpt and show the answers to the questions.
End of session	chatAsk-clear	End of question session (provides refresh).)
Grammar Check	chatAsk -gc "Question Content"	Grammar Check for English Sentences. Correct the sentence if the grammar is wrong or show the sentence as it is.
Voice conversion	chatAsk -at voice file.m4a	When a voice file is sent, the contents of the voice file are displayed in text format.
Emotional Analysis	chatAsk -md "Question Content"	Emotional Analysis for Sentences. Emotional Analysis is classified into three categories: negative, positive, and netal.
Sentence Translation	chatAsk -t "Question Contents"	English sentences are translated into Korean and shown.
Sentence Summary	chatAsk -md "Question Content"	Shows a short summary of the long sentence.

Supported ENV(OS)

• MacOS
• Ubuntu

If you only modify the shell script to the correct format for that OS, it works fine for other OS.

version

• Spring Boot: 2.7.3
• java : 11