Woxi

An interpreter for the Wolfram Language powered by Rust.

You can find the official documentation at woxi.ad-si.com/docs.

Features

The initial focus is to implement a subset of the Wolfram Language so that it can be used for CLI scripting and notebooks. For example:

#!/usr/bin/env woxi

(* Print the square of 5 random integers between 1 and 9 *)
RandomInteger[{1, 9}, 5] // Map[#^2&] // Map[Print]

It has full support for Jupyter Notebooks including graphical output:

[!TIP] Try it out yourself in our JupyterLite instance!

Check out the CLI tests directory to see all currently supported commands and their expected output. All tests must pass with Woxi and WolframScript.

Also check out the functions.csv file for a list of all Wolfram Language functions and their implementation status.

Woxi runs faster than WolframScript as there is no overhead of starting a kernel and verifying its license.

Installation

You can easily install it with Rust's cargo:

cargo install woxi

Prebuilt Binaries

Every GitHub release ships archives of the woxi CLI and the Woxi Studio notebook editor for Linux, macOS, and Windows (x86-64 and arm64), alongside a SHA256SUMS.txt to verify them against.

The macOS build of Woxi Studio comes as a Woxi Studio.app bundle. It is not notarized, so macOS quarantines it after the download — remove the flag once after unzipping:

xattr -dr com.apple.quarantine "Woxi Studio.app"

JavaScript / Node.js

Woxi is also available on npm as woxi-wasm as a WebAssembly build with JavaScript bindings (see npm/):

npm install woxi-wasm

import { evaluate } from "woxi-wasm"
evaluate("Plus[1, 2]")  //=> "3"

Python

Woxi is also available on PyPI as a Python package with pre-built wheels:

pip install woxi

import woxi

woxi.interpret("Plus[1, 2]")  # => '3'

See woxi-py/readme.md for the full Python API.

From Source

If you want to build Woxi from source, you need to have Rust installed. You can get it from rust-lang.org.

Clone the repository, build the project, and install it:

git clone https://github.com/ad-si/Woxi
cd Woxi
make install

On macOS, make install additionally builds the Woxi Studio notebook editor and installs it as a .app bundle at /Applications/Woxi Studio.app, registered with Launch Services so that .nb, .wl, and .wls files appear under Finder's "Open With…" menu.

Usage

You can use the interpreter directly from the command line:

woxi eval "1 + 2"
# 3

woxi eval 'StringJoin["Hello", " ", "World!"]'
# Hello World!

Or you can run a script:

woxi run tests/scripts/hello_world.wls

Or start an interactive REPL, where definitions and % / Out[] history persist across inputs:

woxi repl
# In[1]:= x = 5
# Out[1]= 5
# In[2]:= x^2
# Out[2]= 25

Jupyter Notebook

You can also use Woxi in Jupyter notebooks. Install the kernel with:

woxi install-kernel

The kernelspec is embedded in the woxi binary, so this works from any directory and with any installation method. It is registered with jupyter kernelspec install when the jupyter command is available, and written to Jupyter's kernels directory directly otherwise. Use --system to install it for all users instead of only the current one.

Then start the Jupyter server:

cd examples && jupyter lab

Or simply use our JupyterLite instance. It runs fully self-contained in your browser and no data is send to the cloud.

Editor Integration

Woxi ships a Language Server Protocol server for the Wolfram Language:

woxi lsp

It speaks LSP over stdin/stdout, which is how editors start it, and provides

• Diagnostics — syntax errors, plus warnings on the symbols this interpreter does not implement, so you see what a script will not do before running it
• Hover documentation and completion for every built-in symbol, from the same registry the interpreter uses, and for the symbols the edited file defines itself
• Go to definition, find references, occurrence highlighting and the document outline, computed from the file's assignments

Configure your editor to run woxi lsp for .wl, .wls, .m and .nb files. For Neovim's built-in client that is:

vim.lsp.config.woxi = {
  cmd = { "woxi", "lsp" },
  filetypes = { "wolfram" },
  root_markers = { ".git" },
}
vim.lsp.enable("woxi")

CLI Comparison With WolframScript

Woxi	WolframScript
woxi eval "1 + 2"	wolframscript -code "1 + 2"
woxi run script.wls	wolframscript script.wls
woxi run notebook.nb	not supported (can't run .nb files directly)
woxi repl	wolframscript (no arguments)
woxi lsp	not supported (no language server)

Contributing

Contributions are very welcome! Please feel free to submit a Pull Request.

Testing

To run the test suite:

make test