talon-tree-sitter-service

This project aims to bring high level voice commands to Talon for efficient vocal programming in any text editor and should serve as an early prototype for the system described in my master thesis.

The current implementation is a HTTP server written in PureScript that uses TreeSitter to extract syntactic information from an Elm program and exposes it to Talon in order to increase accuracy when dictating Elm code, and enable structural navigation and editing in Vim.

Installation

Start by making sure you have the following programs installed:

• node@12 + npm (install using nvm)
• vim (>= 8.0) / nvim

If you are on Linux/Max, clone this repository into your Talon user directory (~/.talon/user). If you are on Windows, clone the repo into you home directory (in WSL) and move the talon directory into your scripts folder (/mnt/c/Users/<USERNAME>/AppData/Roaming/talon/user).

If you are new to Vim, create a file called .vimrc in your home directory and add the following line:

source ~/.talon/user/talon-tree-sitter-service/simple.vimrc

If you already have vim set up, add the following line to your Vim configuration file:

source ~/.talon/user/talon-tree-sitter-service/talon.vimrc

If you are already using Vim with Talon, you can remove vim.py and vim.talon, but keep vim_editor.py. If you're managing the title in your Vim config you might need to comment that code out.

To start the server, navigate to the backend directory and run the following commands:

npm install
npm start

If on windows, use npm start -a windows instead.

Add Elm to the list of know languages by going into knausj_talon/code/code.py and add ".elm": "elm" to the extension_lang_map.

Once you have everything set up, open Example.elm in Vim. Your window title should look something like: VIM MODE:n | (1,1) | Example.elm and you should be seeing some debug output from the server.

Note: the server is currently not being notified immediately when a change occur, so make sure to save the document after making a change so the server is in sync for the next command.

Using the System

The current version supports the following use cases:

• Dictating type signatures
• Navigating between function/type declarations or any identifier.
• Selecting and deleting function/type declarations.

Navigation and Editing

Try the following commands:

go fun update
go type update
go to sandbox

go to text
go to next
go to last

go string
go number
go import

select type person
select fun view

delete type message
save it
delete fun main

Dictation

Go into insert mode and try dictating signatures using the types declared in this file. For example:

• word person <pause> of type person -> person : Person
• word update <pause> of type message to model to model -> update : Msg -> Model -> Model

Backend Development

To build the code we use Spago (https://spacchetti.readthedocs.io/en/latest/spago.html)

npm install -g purescript@0.13.8
npm install -g purescript-spago
spago run