💡 How To? Chat GPT 리뷰어 도입하기

선택하기

GitHub MarketPlace를 탐색하면 많은 AI 코드 리뷰어를 찾을 수 있다.

1. ChatGPT CodeReviewer

ChatGPT CodeReviewer - GitHub Marketplace

설정이 간편하고, 한국어도 지원해준다.

아래 코드와 같이

• top_p
• temperature
• maxtoken

은 default 값이 있어 제외한다

https://github.com/anc95/ChatGPT-CodeReview/blob/main/src/chat.ts

import { ChatGPTAPI } from 'chatgpt';
export class Chat {
  private chatAPI: ChatGPTAPI;

  constructor(apikey: string) {
    this.chatAPI = new ChatGPTAPI({
      apiKey: apikey,
      apiBaseUrl:
        process.env.OPENAI_API_ENDPOINT || 'https://api.openai.com/v1',
      completionParams: {
        model: process.env.MODEL || 'gpt-3.5-turbo',
        temperature: +(process.env.temperature || 0) || 1,
        top_p: +(process.env.top_p || 0) || 1,
        max_tokens: process.env.max_tokens
          ? +process.env.max_tokens
          : undefined,
      },
    });
  }

  private generatePrompt = (patch: string) => {
    const answerLanguage = process.env.LANGUAGE
      ? `Answer me in ${process.env.LANGUAGE},`
      : '';

    const prompt =
      process.env.PROMPT ||
        'Below is a code patch, please help me do a brief code review on it. Any bug risks and/or improvement suggestions are welcome:';

    return `${prompt}, ${answerLanguage}:
    ${patch}
    `;
  };

  public codeReview = async (patch: string) => {
    if (!patch) {
      return '';
    }

    console.time('code-review cost');
    const prompt = this.generatePrompt(patch);

    const res = await this.chatAPI.sendMessage(prompt);

    console.timeEnd('code-review cost');
    return res.text;
  };
}

우리가 사용할 4o mini의 기본 최대 출력 토큰이 16,384 밖에 되지 않아 undefined 상태로 둘 예정

간혹 너무 많은 변경사항이 잡혀 리뷰로 넘어갈 것을 대비.

AI의 Context - Window가 128,000 tokens 로 넉넉하긴 하지만

MAX_PATCH_LENGTH 값만 10,000으로 제한해서 사용

프롬프트 개선과 진행상황을 정확히 판단하기 위해 debug mode 활성화

코드 변경을 의해 fork 하고 fork한 레포로 변경

name: Code Review

permissions:
  contents: read
  pull-requests: write

on:
  pull_request:
    types: [opened, reopened, synchronize]
    branches:
      - dev

jobs:
  test:
    runs-on: ubuntu-latest
    steps:
      - uses: LuizyHub/ChatGPT-CodeReview@main
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
          OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
          LANGUAGE: Korean
          OPENAI_API_ENDPOINT: https://api.openai.com/v1
          DEBUG: true
          MODEL: gpt-4o-mini
          MAX_PATCH_LENGTH: 10000

도입

https://github.com/woowa-techcamp-2024/Team5-Guys/pull/11

후기

https://github.com/woowa-techcamp-2024/Team5-Guys/pull/35

• AI가 전체 context를 가지고 리뷰를 진행하지 않는다
• 변화간 단일 파일만 보고 리뷰하기 하기때문에 전체 PR의 정보를 모르고 말을 많이한다
• Too Much Talk…

2. Code Review GPT

[Code Review GPT - GitHub Marketplace](https://github.com/marketplace/actions/code-review-gpt

기본적으로 Conventional Commit을 따르고 Star수를 보아 인기와 사용자가 꽤 있는것으로 보인다

다만, 설정에 용이하지 못하고 한국어 리뷰는 어려워 보인다

설정할것

• MODEL
• OPENAI_API_KEY
• GITHUB_TOKEN

Prompt

https://github.com/mattzcarey/code-review-gpt/blob/main/packages/code-review-gpt/src/review/prompt/prompts.ts

You are an expert {Language} developer, your task is to review a set of pull requests.
You are given a list of filenames and their partial contents, but note that you might not have the full context of the code.

Only review lines of code which have been changed (added or removed) in the pull request. The code looks similar to the output of a git diff command. Lines which have been removed are prefixed with a minus (-) and lines which have been added are prefixed with a plus (+). Other lines are added to provide context but should be ignored in the review.

Begin your review by evaluating the changed code using a risk score similar to a LOGAF score but measured from 1 to 5, where 1 is the lowest risk to the code base if the code is merged and 5 is the highest risk which would likely break something or be unsafe.

In your feedback, focus on highlighting potential bugs, improving readability if it is a problem, making code cleaner, and maximising the performance of the programming language. Flag any API keys or secrets present in the code in plain text immediately as highest risk. Rate the changes based on SOLID principles if applicable.

Do not comment on breaking functions down into smaller, more manageable functions unless it is a huge problem. Also be aware that there will be libraries and techniques used which you are not familiar with, so do not comment on those unless you are confident that there is a problem.

Use markdown formatting for the feedback details. Also do not include the filename or risk level in the feedback details.

Ensure the feedback details are brief, concise, accurate. If there are multiple similar issues, only comment on the most critical.

Include brief example code snippets in the feedback details for your suggested changes when you're confident your suggestions are improvements. Use the same programming language as the file under review.
If there are multiple improvements you suggest in the feedback details, use an ordered list to indicate the priority of the changes.

Format the response in a valid JSON format as a list of feedbacks, where the value is an object containing the filename ("fileName"),  risk score ("riskScore") and the feedback ("details"). The schema of the JSON feedback object must be:
{
  "fileName": {
    "type": "string"
  },
  "riskScore": {
    "type": "number"
  },
  "details": {
    "type": "string"
  }
}

The filenames and file contents to review are provided below as a list of JSON objects:

프롬프트와 동작을 살피면, 변화에 대해 직접 위험도와 분석을 받고, 위험도를 기준으로 정렬해서 하나의 리뷰를 남겨준다.

직접 한국어 지원은 안되지만, 프롬프트를 변경해서 한국어 답변을 만들 수 있을 것으로 보인다.

도입

• https://github.com/woowa-techcamp-2024/Team5-Guys/pull/49

후기

• https://github.com/woowa-techcamp-2024/Team5-Guys/pull/100#issuecomment-2299320711

파일에 대해 종합적인 리뷰를 한 점은 전에 비해 훨신 긍정적

하지만, 여전히 변경된 파일에 대한 context를 가지고 있기때문에 리뷰는 한정적이다.

코드 변경 이력에 대한 context가 있으면 좋겠지만, 현실적으로 많이 어려울 것 같다.