I had to build my own Markdown Editor because no tool was fast enough for me.

TL;DR

I built an open-source markdown editor called Fast Author to improve my productivity when creating written articles specific for tech tutorials involving lots screenshots.

👉 https://github.com/ExamProCo/fast-author

P.S I wrote this article in this editor.

The Pain Point

I've been working on a new version of my AWS Certified Developer course which I am publishing free-to-watch on freeCodeCamp and I had finished recording all my lecture videos with only the follow alongs remaining.

Follow Alongs (some may call them labs) is where I make a video where you follow along with me to get hands-on skills with AWS.

However, producing these videos has been a bottleneck on my production because I have to back-track as I discover things on the fly which could result in me re-recording 3-4 video sections back.

So it makes sense for me to tutorialize them along the way in written format which is easier to modify.

I have to create written versions anyway since on my paid platform we offer the written versions as plus to our the free-to-watch videos.

I am feeling the pain hard here because the Developer Associate is heavy hands-on and these follow alongs need more attention and complexity than any other Certification.

Existing Editors and My Use Case

There are many markdown editors out there, but none are designed for power-users or optimized for my use case which are:

What did I build it in and how long did it take?

I built in 3 days. 2 days building, and 1 day QA with my co-founder Bayko. Today I am using it for its intended use and I already know I am going to get a 500% increase in productivity.

Electron

I already build an open-source video game called Swap-N-Pop so it was a simple as reviewing what I had done before.

Coffeescript

Typescript would have been a better choice if I had multiple collaborators but I wanted to get this done as fast as possible and Coffeescript gives speed like no other.

This was the same path as Swap-N-Pop where when I needed test code, and more collabs I converted it from Coffeescript from Typescript.

MithrilJS

I was going to use Svelte but I wanted to get this done, so I just rolled back to leveraging Mithril where I have solved lots of javascript *hard parts and I didn't want to tack on two extra days to development.

SharpJS

I hate working with ImageMagick so I opted for SharpJS which is much easier to install, but we had considerable pain getting this work with Electron. I had to find the right version of both Electron and SharpJS.

The Editor

• The editor should use a mono-spaced font to easily align text which will render in code elements.
• Should be to quickly toggle into a publisher preview mode
• Design should be optimized for side by side preview
• Need hotkeys for custom tags for Highting, underling and marking text red.

The Images

• Should be able to drag in images into the editor
• Should be able to quickly edit images to resize, crop, border and draw rectangle and markers
• Should store the original images in the project for future reference or modifications

The Preview and Export

• Should be able to load custom css for publisher preview so I can see what it would like on DEV, Medium, freeCodeCamp, HashNode or etc
• Should rename the files in order of appearance as they are being moved around on export

Added Bonsues

Since this is an Electron app I should be able to add my Grammarly extension to better improve my writing.

A Project of Distraction or Procrastination?

So far its been worth the detour. If I worked for another company and I proposed I could try to build a tool in a few days to save weeks they probably wouldn't let me do it since most people would see it as a distraction.

I could have completed my course in the days I built this but it's so easy to focus on the short term, and knowing when to put the time, in the long run, is a skill that requires lots of attempts at failed distractions.

Interesting Code

I thought I pull out some code which was interesting:

I borrowed the online function to get the relative coordinates for a Canvas.

function relMouseCoords(event){
    var totalOffsetX = 0;
    var totalOffsetY = 0;
    var canvasX = 0;
    var canvasY = 0;
    var currentElement = this;

    do{
        totalOffsetX += currentElement.offsetLeft - currentElement.scrollLeft;
        totalOffsetY += currentElement.offsetTop - currentElement.scrollTop;
    }
    while(currentElement = currentElement.offsetParent)

    canvasX = event.pageX - totalOffsetX;
    canvasY = event.pageY - totalOffsetY;

    return {x:canvasX, y:canvasY}
}

I would overlay a canvas on an image. Then I can capture the canvas as an image using the toDataURL() and the replacing the start of the string replace(/^data:image\/png;base64,/, "")

function save(){
  console.log('saving')
  let path = "/tmp/save-drawing-overlay.png"
  const el = document.getElementById('draw')
  fs.writeFile(path, el.toDataURL().replace(/^data:image\/png;base64,/, ""), 'base64', function(err){
    console.log(err)
    ipc.send('sharp-draw',{overlay: path, source: asset.path})
  })
}

SharpJS can composite two files on top of each other which is how I am saving images.

sharp(opts.source).composite([{input: opts.overlay}]).toFile(new_asset)

I set global hotkeys and just watch on keydown.

# global hotkeys
document.addEventListener 'keydown', (e)=>
  meta =
  if os.platform() is 'darwin'
    'Meta'
  else
    'Control'
  Data.meta(true)  if e.key is meta
  Data.shift(true) if e.key is 'Shift'
  if Data.meta()
    if e.key is 'f'
      ipc.send('toggle-fullscreen')
    else if e.key is 'p'
      Data.publisher_preview !Data.publisher_preview()
      m.redraw(true)
    else if e.key is 'n'
      ipc.send('prompt-new')
    else if e.key is 's' && Data.shift()
      Data.splitview !Data.splitview()
      m.redraw(true)
    else if e.key is 'w' && Data.shift()
      Data.line_wrap !Data.line_wrap()
      m.redraw(true)
document.addEventListener 'keyup', (e)=>
  Data.meta(false)
  Data.shift(false)

All the data is stored in a Singleton. No reactive nonsense.

import stream from 'mithril/stream'

class Data
  constructor:->
    # The root directory where all the markdown files are stored
    # eg. ~/fast-author/
    @home = stream('')

    # When the current file was last saved
    @last_saved = stream('')

    # the file that shows selecte in the right hand column
    @active_file = stream(null)

    # files that appear in the right hand column
    @files  = stream([])

    # assets that appear in the right hand column
    # assets only for the current markdown file that is active
    @assets  = stream([])

    # The currently selected image in the markdown to apply editing
    @active_asset = stream null

    # the contents of the markdown file
    @document = stream('')

    # whether the meta key is being held eg. Command on Mac
    @meta = stream(false)

    # whether the shift key is behind held
    @shift = stream(false)

    # whether to wrap or not wrap lines in textarea
    @line_wrap = stream(false)
    #
    # whether to split the view (show both editor or preview, or just editor)
    @splitview = stream(true)

    # when true will hide editor and center preview.
    @publisher_preview = stream(false)

    # the start and end select for markdown textarea
    @selectionStart = stream false
    @selectionEnd   = stream false

    # current selections for infobar
    @_selectionStart = stream 0
    @_selectionEnd   = stream 0
  markdown_path:(name)=>
    path = "#{@home()}/#{name}/index.md"
    console.log path
    path
  # select can be loss after certain updates to textarea.
  # This ensures our old selection remains
  keep_selection:=>
    @selectionStart @_selectionStart()
    @selectionEnd @_selectionEnd()
  get_asset:=>
    asset = null
    for a in @assets()
      if a.path is @active_asset().replace('file://','')
        asset = a
        break
    asset
export default new Data()