https://github.com/skywind3000/vim-gpt-commit
:rocket: Generate git commit message using ChatGPT and Ollama !!
https://github.com/skywind3000/vim-gpt-commit
chatgpt chatgpt-app neovim neovim-plugin neovim-plugins vim vim-plugin vim-plugins
Last synced: 7 months ago
JSON representation
:rocket: Generate git commit message using ChatGPT and Ollama !!
- Host: GitHub
- URL: https://github.com/skywind3000/vim-gpt-commit
- Owner: skywind3000
- License: mit
- Created: 2024-02-11T12:34:32.000Z (almost 2 years ago)
- Default Branch: master
- Last Pushed: 2025-02-07T16:25:21.000Z (11 months ago)
- Last Synced: 2025-06-10T00:09:05.906Z (7 months ago)
- Topics: chatgpt, chatgpt-app, neovim, neovim-plugin, neovim-plugins, vim, vim-plugin, vim-plugins
- Language: Python
- Homepage:
- Size: 49.8 KB
- Stars: 48
- Watchers: 2
- Forks: 6
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Preface
Generate proper git commit message using ChatGPT (or [Ollama](https://github.com/ollama/ollama)) in Vim:
This plugin provides a `:GptCommit` command to generate commit message and append in the current buffer.
Usually, it can be used when editing the `COMMIT_EDITMSG` file, or in a fugitive commit window.
## Quick start
For vim-plug:
```vimL
Plug 'skywind3000/vim-gpt-commit'
let g:gpt_commit_key = 'Your openai apikey'
" uncomment this line below to enable proxy
" let g:gpt_commit_proxy = 'socks5://127.0.0.1:1080'
" uncomment the following lines if you want to use Ollama:
" let g:gpt_commit_engine = 'ollama'
" let g:gpt_commit_ollama_url = 'http://127.0.0.1:11434/api/chat'
" let g:gpt_commit_ollama_model = 'llama2'
```
For lazy:
```lua
{
'skywind3000/vim-gpt-commit',
config = function ()
-- if you don't want to set your api key directly, add to your .zshrc:
-- export OPENAI_API_KEY='sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx'
vim.g.gpt_commit_key = os.getenv("OPENAI_API_KEY")
-- uncomment this line below to enable proxy
-- vim.g.gpt_commit_proxy = 'socks5://127.0.0.1:1080'
-- uncomment the following lines if you want to use Ollama:
-- vim.g.gpt_commit_engine = 'ollama'
-- vim.g.gpt_commit_ollama_url = 'http://127.0.0.1:11434/api/chat'
-- vim.g.gpt_commit_ollama_model = 'llama2'
end,
},
```
Requirements:
- Python: A Python 3 executable binary needs to be located in your `$PATH`, but vim's `+python3` feature is not required.
## Command
This plugin provides only one command:
```VimL
:GptCommit[!] [repo_path]
```
It will generate git commit message and append to the current cursor position. If the `repo_path` is omitted, the directory name of the current buffer will be used.
If the bang (`!`) is included, `:GptCommit` will not append any text to the current buffer, instead, it will only store the messages in the unnamed register (`*`). You can use them any time manually by `"*p` or display the content by:
```VimL
:echo @*
```
You will see the result in the command line.
## Options
| Name | Required | Default | Description |
|-|-|-|-|
| g:gpt_commit_key | Yes | `''` | Your openai apikey |
| g:gpt_commit_proxy | - | `''` | proxy |
| g:gpt_commit_concise | - | `0` | set to 1 to generate less verbose message |
| g:gpt_commit_lang | - | `''` | output language |
| g:gpt_commit_model | - | `'gpt-3.5-turbo'` | ChatGPT model |
| g:gpt_commit_staged | - | `1` | set to 1 to use staged diff |
| g:gpt_commit_max_line | - | `160` | max diff lines to reference |
| g:gpt_gpt_commit_max_logs | - | `0` | max log history (git log) use as context |
| g:gpt_commit_url | - | `''` | alternative request URL, see #1 |
| g:gpt_commit_python | - | `''` | specify the Python executable file explicitly |
| g:gpt_commit_engine | - | `'chatgpt'` | change to 'ollama' to use Ollama |
| g:gpt_commit_ollama_model | Yes | `''` | ollama model |
| g:gpt_commit_ollama_url | - | `''` | explicitly setting ollama URL|
Note #1: the default URL is "https://api.openai.com", can be changed by setting `g:gpt_commit_url`.
## Script
For command line usage, you can execute `gptcommit.py` in the `bin` directory:
```
usage: python3 gptcommit.py repo_path
available options:
--key=xxx required, your openai apikey
--staged optional, use staged diff if present
--proxy=xxx optional, proxy support
--maxline=num optional, max diff lines to feed ChatGPT, default ot 160
--model=xxx optional, can be gpt-3.5-turbo (default) or something
--lang=xxx optional, output language
--concise optional, generate concise message if present
--utf8 optional, output utf-8 encoded text if present
--url=xxx optional, alternative openai request url
--maxlogs=num optional, use git log entries to enrich the context
```
It can be integrated into other editors.
## Credit
If you find this plugin amusing, you might also be interested in my other vim plugins, such as:
- [vim-auto-popmenu](https://github.com/skywind3000/vim-auto-popmenu): Display the completion menu automantically.
- [vim-color-export](https://github.com/skywind3000/vim-color-export): A tool to backport NeoVim colorschemes to Vim.
- [vim-navigator](https://github.com/skywind3000/vim-navigator): Navigate your commands easily.
- [vim-rt-format](https://github.com/skywind3000/vim-rt-format): Prettify current line when you press Enter.
- [vim-quickui](https://github.com/skywind3000/vim-quickui): The missing UI extensions for Vim 8.2 (and NeoVim 0.4).
- [asynctasks.vim](https://github.com/skywind3000/asynctasks.vim): Modern task system for project building/testing/deploying.