GPT-PR

GPT-PR is an open-source command-line tool designed to streamline your GitHub workflow for opening PRs. Leveraging OpenAI's ChatGPT API, it automatically opens a GitHub Pull Request with a predefined description and title directly from your current project directory.

Table of Contents

• Features
• Prerequisites
• Installation
• Configuration
• Usage
• How to Contribute
• Roadmap

Features

• Analyzes the diff changes of the current branch against the main branch.
• Provides an option to exclude certain file changes from PR generation (for instance, you can ignore a package.lock file with 5k lines changed).
• Incorporates commit messages into the process.

Prerequisites

Before getting started, make sure you have the following installed:

• Python 3.7 or higher
• Pipenv

Installation

You can install and use GPT-PR in one of two ways. Choose the option that best suits your needs.

Option 1: Using pip install (Recommended)

1. Install the package:

pip install -U gpt-pr

Note: Use this command to update gpt-pr package to the latest version.

2. Setup API keys for GitHub and OpenAI, take a look at Configuration.

3. Inside the Git repository you are working on, ensure you have pushed your branch to origin, then run:

gpt-pr --help

Option 2: Cloning the code

1. Clone the repository:

git clone https://github.com/alissonperez/gpt-pr.git

2. Navigate to the project directory and install dependencies:

cd gpt-pr
pipenv install

After setting up API keys (Configuration), you can use GPT-PR within any git project directory. Suppose you've cloned this project to ~/workplace/gpt-pr, here's how you can use it:

PYTHONPATH=~/workplace/gpt-pr/gpt-pr \
PIPENV_PIPFILE=~/workplace/gpt-pr/Pipfile \
pipenv run python ~/workplace/gpt-pr/gptpr/main.py --help

Configuration

Setting up GitHub Token (GH_TOKEN)

GPT-PR tool will look for a GH_TOKEN in current shell env var OR in gpt-pr config file (at ~/.gpt-pr.ini).

To authenticate with GitHub, generate and export a GitHub Personal Access Token:

1. Navigate to GitHub's Personal Access Token page.
2. Click "Generate new token."
3. Provide a description and select the required permissions repo for the token.
4. Click "Generate token" at the bottom of the page.
5. Copy the generated token.
6. Set gh_token config running (supposing your gh token is ghp_4Mb1QEr9gY5e8Lk3tN1KjPzX7W9z2V4HtJ2b):

gpt-pr-config set gh_token ghp_4Mb1QEr9gY5e8Lk3tN1KjPzX7W9z2V4HtJ2b

Or just export it as an environment variable in your shell initializer:

export GH_TOKEN=your_generated_token_here

Setting up OpenAI API Key (OPENAI_API_KEY)

GPT-PR tool will look for a OPENAI_API_KEY env var in current shell OR in gpt-pr config file (at ~/.gpt-pr.ini).

This project needs to interact with the ChatGPT API to generate the pull request description. So, you need to generate and export an OpenAI API Key:

1. Navigate to OpenAI's API Key page.
2. If you don't have an account, sign up and log in.
3. Go to the API Keys section and click "Create new key."
4. Provide a description and click "Create."
5. Copy the generated API key.
6. Set openai_api_key config running (supposing your openai_api_key is QEr9gY5e8Lk3tN1KjPzX7W9z2V4Ht):

gpt-pr-config set openai_api_key QEr9gY5e8Lk3tN1KjPzX7W9z2V4Ht

Or just export it as an environment variable in your shell initializer:

export OPENAI_API_KEY=your_generated_api_key_here

Change OpenAI model

To change OpenAI model, just run:

gpt-pr-config set openai_model gpt-3.5-turbo

To see a full list of available models, access OpenAI Models Documentation

See all configs available

To print all default configs and what is being used, just run:

gpt-pr-config print

Reset config

To reset any config to default value, just run:

gpt-pr-config reset config-name

Example:

gpt-pr-config reset openai_model

Usage

Generating Github Pull Requests

To create a Pull request from your current branch commits to merge with main branch, just run:

gpt-pr

If you would like to compare with other base branch that is not main, just use -b param:

gpt-pr -b my-other-branch

Usage help

To show help commands:

gpt-pr -h

Output:

How to contribute?

Please follow our CONTRIBUTING guide.

Roadmap

• Improve execution method, possibly through a shell script or at least an alias in bash rc files.
  • Change to use with pip installation and console_scripts entry point.
• Fetch GitHub PR templates from the current project.
• Add configuration to set which LLM and model should be used (OpenAI GPT, Mistral, etc...)
• Add unit tests.