transform-markdown-mathmode

Transform LaTeX equations asynchronously in a stream of markdown

Introduction

This module takes a stream of markdown and searches for basic patterns that look like equations, being careful to exclude escaped patterns or code blocks. It doesn't do any rendering itself, but takes transformation functions that may perform some synchronous or asynchronous action and return the result. This makes it easy to render equations, store them somewhere, and insert image tags into the resulting markdown.

Installation

To install, run:

$ npm install transform-markdown-mathmode

Example

This README is generated by passing README.mdtex through the parser. The sample configuration just transforms $...$ into <...> and $$...$$ into <<...>> so it's clear that it's working (FWIW: this line started out with all $'s):

var transformMathmode = require('transform-markdown-mathmode')

process.stdin
  .pipe(
    transformMathmode({
      inline:  function(tex, cb) { cb(  '<' + tex + '>'  ) },
      display: function(tex, cb) { cb( '<<' + tex + '>>' ) }
    })
  )
  .pipe( process.stdout )

// Also, note that equations in code blocks don't get transformed,
// no escaping necessary! Hooray!
//
//   $y = x$
//

This is just invoked with:

$ cat README.mdtex | node example/transform.js > README.md

This transformation isn't particularly interesting, but it means you can perform an arbitrarily complicated asynchronous task each time you encounter an equation.

What it does

As hesitant as I was to parse a stream of markdown (with regexes? Is that even parsing?), this module uses a finite state machine to track whether a couple different types of blocks are open or closed. I suspect it's probably possible (and wayyyy easier) to do this with a couple regexes, but the current complication comes from the need to track open equations across multiple lines of streamed text and the desire to avoid ugly escaping that would make it a total pain in the butt to paste fifty lines of jQuery code into your README. That's really all it's doing through. It's pretty simple, carefully tested, and allows simple escaping for corner cases. Fundamentally though, it's a parser built on regexes, so it's not like a work of art or anything. If there are cases that fail, let me know. If you can do this in five lines of code... I'm not sure I want to know...

Goals:

• transform equations asynchronously while avoiding a bunch of ugly escaping and other workarounds.
• learn how to process streams in node.
• and learn how to write gulp plugins.
• and avoid typesetting equations in Github READMEs by hand.
• maybe use it for a static website instead of MathJax. If the baseline positioning is workable. We'll see.

Usage

require('transform-markdown-mathmode')( options )

Create a transform stream that performs the transformation.

• options: The options hash accepts two options:

  • inline: A function of format function(tex, cb){...} that receives a string of tex (with $ delimiters stripped) and executes a callback containing the transformed inline equation. See above for example.
  • display: A function of format function(tex, cb){...} that receives a string of tex (with $$ delimiters stripped) and executes a callback containing the transformed display equation. See above for example.

• Returns: The transform stream.

License

(c) 2015 Ricky Reusser. MIT License