Reb—regex-based Brainfuck toolkit

Reb is a CLI tool for working with Brainfuck. Right now, it supports:

• Code minification and formatting.
• Pattern optimization.
• Compilation to C.
• And evaluation of both Brainfuck and optimized code format.

The catch is: it’s all regex. The goal of the project is to use regex as much as possible. Though the evaluation is not based on regular expressions. Minification and optimization are, at least.

Getting started

You only need to clone the repo:

git clone https://github.com/aartaka/reb

and run

make all

in the checked-out directory. After that, you’ll have an executable called reb around. Use it:

Usage

There are four main commands in Reb:

m

minification.

f

formatting (as per bf.style.)

o

optimization.

r

evaluation (from “run”).

c

compilation to C.

Either of these is invoked as a prefix following the executable name. I.e., minifying file.bf into file.min.bf is:

./reb m file.bf > file.min.bf

You can see the (ab)use of shell redirection here. Reb is intended to fit the UNIX-y model of tools’ interation. Want to evaluate output of some command? Pipe it to ./reb r -- (where -- stands for stdin). Want to save the optimized code to a file? Use output redirection. Want to have a pipeline of minification+optimization+evaluation? You can do that:

./reb m file.bf | ./reb o -- | ./reb r --

(although it will be slow. So you better split the stages and run the minified/optimized file when it’s ready.)

And then, you can also compile to C and use your favorite C compiler to get near-native speed:

./reb c file.min.bf > file.c
gcc file.c -O3 -o file

Optimized Code Specification (WIP)

Monadic commands

[, ], ., ,

Like in Brainfuck

+, -, <, >

Like in Brainfuck, except they can have a numeric argument specifying how many times to repeat them.

=

Set the contents of the cell to numeric argument.

Diadic commands

offset`multiplier{ and offset`multiplier}

Copy to the left and copy to the right respectively. offset defines how far (= number of cells) to copy the data, multiplier defines how to scale this data.

step`value( and step`value)

Scan the memory (to the left and right respectively) for value cells and stop there, nulling them. step defines how many cells to skip through.

History

Reb is a sequel to Sade, my Brainfuck-to-Lisp optimizing compiler. While Sade is quite fast and nice to use… The way optimizations are implemented is quite hard to maintain. So I figured that this optimization of Brainfuck command patterns is a good use for regex. Here we are: Reb is a tool utilizing regex to minify, compile, and parse Brainfuck code. Success. Alpplause. Bring down the curtain.