CodeGPT.nvim

CodeGPT is a plugin for neovim that provides commands to interact with ChatGPT. The focus is around code related usages. So code completion, refactorings, generating docs, etc.

Installation

• Set environment variable OPENAI_API_KEY to your openai api key.
• The plugins 'plenary' and 'nui' are also required.
• OpenAI's tokenizer tiktoken is recommended for accurate token count estimate.

Installing with Lazy.

{
    "dpayne/CodeGPT.nvim",
    dependencies = {
      'nvim-lua/plenary.nvim',
      'MunifTanjim/nui.nvim',
    },
    config = function()
        require("codegpt.config")
    end
}

Installing with packer.

use({
   "dpayne/CodeGPT.nvim",
   requires = {
      "MunifTanjim/nui.nvim",
      "nvim-lua/plenary.nvim",
   },
   config = function()
      require("codegpt.config")
   end
})

Installing with plugged.

Plug("nvim-lua/plenary.nvim")
Plug("MunifTanjim/nui.nvim")
Plug("dpayne/CodeGPT.nvim")

Installing OpenAI's tokenizer

pip install tiktoken

Commands

The top-level command is :Chat. The behavior is different depending on whether text is selected and/or arguments are passed.

Completion

• :Chat with text selection will trigger the completion command, ChatGPT will try to complete the selected code snippet.

Code Edit

• :Chat some instructions with text selection and command args will invoke the code_edit command. This will treat the command args as instructions on what to do with the code snippet. In the below example, :Chat refactor to use iteration will apply the instruction refactor to use iteration to the selected code.

Code Edit

• :Chat <command> if there is only one argument and that argument matches a command, it will invoke that command with the given text selection. In the below example :Chat tests will attempt to write units for the selected code.

Chat

• :Chat hello world without any text selection will trigger the chat command. This will send the arguments hello world to ChatGPT and show the results in a popup.

A full list of predefined commands are below

command	input	Description
completion	text selection	Will ask ChatGPT to complete the selected code.
code_edit	text selection and command args	Will ask ChatGPT to apply the given instructions (the command args) to the selected code.
explain	text selection	Will ask ChatGPT to explain the selected code.
question	text selection	Will pass the commands args to ChatGPT and return the answer in a text popup.
debug	text selection	Will pass the code selectiont to ChatGPT analyze it for bugs, the results will be in a text popup.
doc	text selection	Will ask ChatGPT to document the selected code.
opt	text selection	Will ask ChatGPT to optimize the selected code.
tests	text selection	Will ask ChatGPT to write unit tests for the selected code.
chat	command args	Will pass the given command args to ChatGPT and return the response in a popup.

Overriding Command Configurations

The configuration option vim.g["codegpt_commands_defaults"] = {} can be used to override command configurations. This is a lua table with a list of commands and the options you want to override.

vim.g["codegpt_commands_defaults"] = {
  ["completion"] = {
      user_message_template = "This is a template of the message passed to chat gpt. Hello, the code snippet is {{text_selection}}."
}

The above, overrides the message template for the completion command.

A full list of overrides

name	default	description
model	"gpt-3.5-turbo"	The model to use.
max_tokens	4096	The maximum number of tokens to use including the prompt tokens.
temperature	0.6	0 -> 1, what sampling temperature to use.
system_message_template	""	Helps set the behavior of the assistant.
user_message_template	""	Instructs the assistant.
callback_type	"replace_lines"	Controls what the plugin does with the response
language_instructions	{}	A table of filetype => instructions. The current buffer's filetype is used in this lookup. This is useful trigger different instructions for different languages.

Templates

The system_message_template and the user_message_template can contain template macros. For example:

macro	description
{{filetype}}	The filetype of the current buffer.
{{text_selection}}	The selected text in the current buffer.
{{language}}	The name of the programming language in the current buffer.
{{command_args}}	Everything passed to the command as an argument, joined with spaces. See below.
{{language_instructions}}	The found value in the language_instructions map. See below.

Language Instructions

Some commands have templates that use the {{language_instructions}} macro to allow for additional instructions for specific filetypes.

vim.g["codegpt_commands_defaults"] = {
  ["completion"] = {
      language_instructions = {
          cpp = "Use trailing return type.",
      },
  }
}

The above adds a specific Use trailing return type. to the command completion for the filetype cpp.

Command Args

Commands are normally a single value, for example :Chat completion. Normally, a command such as :Chat completion value will be interpreted as a code_edit command, with the arguments "completion value", and not completion with "value". You can make commands accept additional arguments by using the {{command_args}} macro anywhere in either user_message_template or system_message_template. For example:

vim.g["codegpt_commands"] = {
  ["testwith"] = {
      user_message_template =
        "Write tests for the following code: ```{{filetype}}\n{{text_selection}}```\n{{command_args}} " ..
        "Only return the code snippet and nothing else."
  }
}

After defining this command, any :Chat command that has testwith as its first argument will be handled. For example, :Chat testwith some additional instructions will be interpreted as testwith with "some additional instructions".

Custom Commands

Custom commands can be added to the vim.g["codegpt_commands"] configuration option to extend the available commands.

vim.g["codegpt_commands"] = {
  ["modernize"] = {
      user_message_template = "I have the following {{language}} code: ```{{filetype}}\n{{text_selection}}```\nModernize the above code. Use current best practices. Only return the code snippet and comments. {{language_instructions}}",
      language_instructions = {
          cpp = "Refactor the code to use trailing return type, and the auto keyword where applicable.",
      },
  }
}

The above configuration adds the command :Chat modernize that attempts modernize the selected code snippet.

Command Defaults

The default command configuration is:

{
    model = "gpt-3.5-turbo",
    max_tokens = 4096,
    temperature = 0.6,
    number_of_choices = 1,
    system_message_template = "",
    user_message_template = "",
    callback_type = "replace_lines",
}

More Configuration Options

Custom status hooks

You can add custom hooks to update your status line or other ui elements, for example, this code updates the status line colour to yellow whilst the request is in progress.

vim.g["codegpt_hooks"] = {
    request_started = function()
        vim.cmd("hi StatusLine ctermbg=NONE ctermfg=yellow")
    end,
  request_finished = vim.schedule_wrap(function()
        vim.cmd("hi StatusLine ctermbg=NONE ctermfg=NONE")
    end)
}

Lualine Status Component

There is a convenience function get_status so that you can add a status component to lualine.

local CodeGPTModule = require("codegpt")

require('lualine').setup({
    sections = {
        -- ...
        lualine_x = { CodeGPTModule.get_status, "encoding", "fileformat" },
        -- ...
    }
})

Popup options

Popup commands

The default filetype of the text popup window is markdown. You can change this by setting the codegpt_popup_options variable.

vim.g["codegpt_text_popup_filetype"] = "markdown"

Popup commands

vim.g["codegpt_ui_commands"] = {
  -- some default commands, you can remap the keys
  quit = "q", -- key to quit the popup
  use_as_output = "<c-o>", -- key to use the popup content as output and replace the original lines
  use_as_input = "<c-i>", -- key to use the popup content as input for a new API request
}
vim.g["codegpt_ui_commands"] = {
  -- tables as defined by nui.nvim https://github.com/MunifTanjim/nui.nvim/tree/main/lua/nui/popup#popupmap
  {"n", "<c-l>", function() print("do something") end, {noremap = false, silent = false}}
}

Popup layouts

vim.g["codegpt_popup_options"] = {
  -- a table as defined by nui.nvim https://github.com/MunifTanjim/nui.nvim/tree/main/lua/nui/popup#popupupdate_layout
  relative = "editor",
  position = "50%",
  size = {
    width = "80%",
    height = "80%"
  }
}

Popup border

vim.g["codegpt_popup_border"] = {
  -- a table as defined by nui.nvim https://github.com/MunifTanjim/nui.nvim/tree/main/lua/nui/popup#border
  style = "rounded"
}

Popup window options

-- Enable text wrapping and line numbers
vim.g["codegpt_popup_window_options"] = {
  wrap = true,
  linebreak = true,
  relativenumber = true,
  number = true,
}

Move completion to popup window

For any command, you can override the callback type to move the completion to a popup window. An example below is for overriding the completion command.

require("codegpt.config")

vim.g["codegpt_commands"] = {
  ["completion"] = {
    callback_type = "code_popup",
  },
}

Horizontal or vertical split window

If you prefer a horizontal or vertical split window, you can change the popup type to horizontal or vertical.

-- options are "horizontal", "vertical", or "popup". Default is "popup"
vim.g["codegpt_popup_type"] = "horizontal"

To set the height of the horizontal window or the width of the vertical popup, you can use codegpt_horizontal_popup_size and codegpt_horizontal_popup_size variables.

vim.g["codegpt_horizontal_popup_size"] = "20%"
vim.g["codegpt_vertical_popup_size"] = "20%"

Miscellaneous Configuration Options

-- Open API key and api endpoint
vim.g["codegpt_openai_api_key"] = os.getenv("OPENAI_API_KEY")
vim.g["codegpt_chat_completions_url"] = "https://api.openai.com/v1/chat/completions"
vim.g["codegpt_openai_api_provider"] = "OpenAI" -- or Azure

-- clears visual selection after completion
vim.g["codegpt_clear_visual_selection"] = true

Callback Types

Callback types control what to do with the response

name	Description
replace_lines	replaces the current lines with the response. If no text is selected it will insert the response at the cursor.
text_popup	Will display the result in a text popup window.
code_popup	Will display the results in a popup window with the filetype set to the filetype of the current buffer

Template Variables

name	Description
language	Programming language of the current buffer.
filetype	filetype of the current buffer.
text_selection	Any selected text.
command_args	Command arguments.
filetype_instructions	filetype specific instructions.

Example Configuration

Note that CodeGPT should work without any configuration. This is an example configuration that shows some of the options available:

require("codegpt.config")

-- Override the default chat completions url, this is useful to override when testing custom commands
-- vim.g["codegpt_chat_completions_url"] = "http://127.0.0.1:800/test"

vim.g["codegpt_commands"] = {
  ["tests"] = {
    -- Language specific instructions for java filetype
    language_instructions = {
        java = "Use the TestNG framework.",
    },
  },
  ["doc"] = {
    -- Language specific instructions for python filetype
    language_instructions = {
        python = "Use the Google style docstrings."
    },

    -- Overrides the max tokens to be 1024
    max_tokens = 1024,
  },
  ["code_edit"] = {
    -- Overrides the system message template
    system_message_template = "You are {{language}} developer.",

    -- Overrides the user message template
    user_message_template = "I have the following {{language}} code: ```{{filetype}}\n{{text_selection}}```\nEdit the above code. {{language_instructions}}",

    -- Display the response in a popup window. The popup window filetype will be the filetype of the current buffer.
    callback_type = "code_popup",
  },
  -- Custom command
  ["modernize"] = {
    user_message_template = "I have the following {{language}} code: ```{{filetype}}\n{{text_selection}}```\nModernize the above code. Use current best practices. Only return the code snippet and comments. {{language_instructions}}",
    language_instructions = {
        cpp = "Use modern C++ syntax. Use auto where possible. Do not import std. Use trailing return type. Use the c++11, c++14, c++17, and c++20 standards where applicable.",
    },
  }
}

Goals

• Code related usages.
• Simple.
• Easy to add custom commands.