bigbang.coffee

File        : bigbang.coffee
Maintainer  : Felix C. Stegerman <flx@obfusk.net>
Date        : 2014-07-18

Copyright   : Copyright (C) 2014  Felix C. Stegerman
Licence     : LGPLv3+
Version     : v0.2.1

graphical/evented functional programming for js/coffee

Create graphical and evented programs (like games) in coffeescript or javascript using plain mathematical functions; inspired by the 2htdp library for racket.

https://github.com/obfusk/bigbang.coffee

License: LGPLv3+.

underscore + exports

U = this._ || require 'underscore'
B = -> B.bigbang arguments...
if exports? then module.exports = B else this.bigbang = B

requestAnimationFrame & polyfill

requestAnimationFrame polyfill; the default delay is 17 milliseconds (ca. 60 fps)

B.polyRequestAnimationFrame = polyRequestAnimationFrame =       # {{{1
  (opts = {}) ->
    console.warn 'polyfilling *RequestAnimationFrame ...' if opts.warn
    delay = opts.delay || 17; next = 0
    (cb) ->
      cur   = +new Date
      dt    = Math.max 0, delay - (cur - next)
      next  = cur + dt
      window.setTimeout (-> cb +new Date), dt

NB: we use the prefixed versions of requestAnimationFrame b/c requestAnimationFrame itself uses relative timestamps; also good to know: webkitRequestAnimationFrame uses floating point timestamps.

B.requestAnimationFrame = requestAnimationFrame =
  window?.webkitRequestAnimationFrame ||
  window?.mozRequestAnimationFrame ||
  polyRequestAnimationFrame warn: true

bigbang & stop_with

big bang: start a new universe whose behaviour is specified by the options and handler functions designated

bigbang
  world:        object,
  canvas:       element|object,
  to_draw:      ((world) -> scene),
  on_tick:      ((world, time) -> new_world),
  fps:          int,
  queue:        boolean|int,
  animate:      requestAnimationFrame-like-function,
  on_key:       ((world, key) -> new_world),
  on_click:     ((world, x, y) -> new_world),
  on:           { foo: ((world, ...) -> new_world), ... },
  stop_when:    ((world) -> boolean),
  last_draw:    ((world) -> scene),
  setup:        ((canvas, handlers) -> setup_value),
  teardown:     ((canvas, handlers, setup_value) ->
                  teardown_value),
  on_stop:      ((world, teardown_value) -> ...)
-> {get_world,get_done}

Options:

• world is the object representing the initial world
• canvas is the HTML5 canvas (or equivalent) to draw on

• to_draw is called every time a new world needs to be drawn

• (optional) on_tick

  • when passed, the universe is in “clock mode”: all changes are queued (unless queue is false) and drawing happens every actual clock tick (usually approx. 60 fps); on_tick is called every “virtual” clock tick (as determined by fps)
  • otherwise, events are processed immediately (by updating the world and re-drawing the scene)
• (optional) fps is the requested frame rate; defaults to 60
• (optional) queue when set to false, disables queueing in “clock mode”; when set to an int, restricts the queue size to that number of items (e.g. to draw every actual tick and ignore all but the last world, set this to 1)
• (optional) animate the requestAnimationFrame-like function to use as the clock in “clock mode”
• (optional) on_key is called every time a key is pressed to update the world
• (optional) on_click is called every time the mouse is clicked inside the canvas to update the world
• (optional) on contains handlers for user-defined events; setup and teardown are used to connect handlers to elements and events; when the user-defined event is triggered, the appropriate function is called
• (optional) stop_when is called to determine if the universe needs to stop
• (optional) last_draw is called instead of to_draw to draw the last world
• (optional) setup is called before the universe starts; the handlers are the internal event handlers for on which setup is expected to connect to the appropriate elements and events
• (optional) teardown is called after the universe has ended, before on_stop; the handlers are the same as passed to setup and can be used to cancel the event handling
• (optional) on_stop is called after the universe has ended, after teardown

Returns:

• get_world is a function that returns the current world
• get_done is a function that returns whether the universe is stopped

The canvas need not be an actual canvas: it can be any element you wish to “draw” the world with. It can be (part of) the body of an event-driven page. In this case, you will also have a different concept of “scene”. For more on “scenes”, see mk_scene.

To stop the world from on_tick etc., return stop_with(new_world) instead of new_world.

For details on key press handling, see handle_keys. If you want to use a different key press handling library, set the handle_keys option to a function with the same api as handle_keys.

For details on mouse click handling, see handle_click.

B.bigbang = (opts) ->                                           # {{{1
  tickless      = !opts.on_tick
  queue         = opts.queue ? true
  anim          = opts.animate || requestAnimationFrame
  world         = opts.world
  delay         = 1000 / (opts.fps || 60)
  delay_margin  = delay / 17                                    #  ???
  done          = opts.stop_when?(world) || false
  next          = +new Date + delay
  changes       = []
  setup_value   = null

  if tickless && (opts.queue? || opts.fps || opts.animate)
    throw new Error 'queue, fps and animate require on_tick'

  draw = (w,d) ->
    f = if d && opts.last_draw then opts.last_draw else opts.to_draw
    draw_scene(f w) opts.canvas

  draw_changes = ->
    for {world:w,done:d} in changes
      draw w, d
    changes = []

  change = (f, args...) ->
    return if done || !U.isFunction f
    x = f world, args...
    if x instanceof _Stop
      world = x.world
      done  = true
    else
      world = x
      done  = true if opts.stop_when? world
    if !tickless && queue
      changes.shift() if changes.length == queue  # != true is OK
      changes.push {world,done}
    else
      draw world, done
      finish() if done

  s = +new Date

  tick = (t) ->
    if t >= next - delay_margin
      next += delay; change opts.on_tick, t
    draw_changes() if queue
    if done
      finish() if queue
    else
      anim tick

  key   = (k) -> change opts.on_key, k
  click = (x,y) -> change opts.on_click, x, y

  handlers = {}
  for k, v of opts.on || {}
    do (k,v) -> handlers[k] = (args...) -> change v, args...

  finish = ->
    cancel_keys?(); cancel_click?()
    tv = opts.teardown? opts.canvas, handlers, setup_value
    opts.on_stop? world, tv

  hk            = opts.handle_keys  || handle_keys
  hc            = opts.handle_click || handle_click
  cancel_keys   = opts.on_key   && hk opts.canvas, key  , opts.$
  cancel_click  = opts.on_click && hc opts.canvas, click, opts.$
  setup_value   = opts.setup? opts.canvas, handlers

  draw world, done
  anim tick unless tickless

  world: (-> world), done: (-> done)

stop the universe; see bigbang

B.stop_with = stop_with = (w) -> new _Stop w

wrapper class for stop_with

class _Stop
  constructor: (@world) ->
B._Stop = _Stop

abstract scene functions

make a scene

A “scene” is a function that takes a “canvas” and “draws” on it.

A “non-recursive scene” is a “scene” that also takes a second argument specifying whether any “lower scene” should also be drawn by it and has a property lower_scene that holds the “lower scene”. This allows draw_scene to draw multiple scenes using iteration instead of recursion.

mk_scene takes a “non-recursive (lower) scene” and a “regular scene” and turns the “regular scene” into a “non-recursive scene” that draws the “lower scene” when needed.

B.mk_scene = mk_scene = (scene, f) ->
  g = (canvas, draw_lower = true) -> scene? canvas if draw_lower; f canvas
  g.lower_scene = scene; g

draw a scene

B.draw_scene = draw_scene = (scene) -> (canvas) ->
  if (s = scene).lower_scene
    scenes = [scene]; scenes.push s while s = s.lower_scene
    scenes.reverse(); s canvas, false for s in scenes; null
  else
    scene canvas

HTML5 canvas scene functions

empty scene

B.empty_scene = empty_scene = (width, height) -> mk_scene null, (canvas) ->
  canvas.width = width; canvas.height = height

text with center at coordinates

B.place_text = place_text =                                     # {{{1
  (string, x, y, fontsize, colour, scene, $ = window.$) -> \
  mk_scene scene, (canvas) ->
    ctx = canvas.getContext '2d'
    ctx.save()
    ctx.font          = "#{fontsize} sans-serif"
    ctx.fillStyle     = colour
    ctx.textBaseline  = 'bottom'
    {w,h}             = measure_text $, string, fontsize, 'sans-serif'
    ctx.fillText string, Math.round(x - w / 2), Math.round(y + h / 2)
    ctx.restore()

image with center at coordinates

B.place_image = place_image = (image, x, y, scene) ->
  mk_scene scene, (canvas) ->
    ctx = canvas.getContext '2d'
    x_  = x - Math.round(image.width  / 2)
    y_  = y - Math.round(image.height / 2)
    ctx.drawImage image, x_, y_

… TODO: more scene and image functions …

keyboard functions

Handle key presses. Listens to keydown on elem, calls f with the key_to_string’d key; returns a function that cancels the listening. the key presses handled are a-z, 0-9, and the keys in keycodes (with the exception of SHIFT); with and without shift; when alt, ctrl, or meta is pressed, the key press is ignored.

Why only this limited set of key presses you ask? Because of browser and keyboard layout inconsistencies. For example, a German keyboard layout will result in incorrect results with chromium, and even less correct results with firefox. The key presses currently handled seem to be the ones that work correctly in both firefox and chromium, for both US English and German keyboard layouts. YMMV.

B.handle_keys = handle_keys = (elem, f, $ = window.$) ->        # {{{1
  h = (e) ->
    if !e.altKey && !e.ctrlKey && !e.metaKey &&
        e.which != keycodes.SHIFT
      k = key_to_string e.which, e.shiftKey; f k if k
      k == null
    else
      true
  $(elem).on 'keydown', h
  -> $(elem).off 'keydown', h

Turn keypress into something usable: a-z without shift is returned as “a”-“z”; A-Z with shift is returned as “A”-“Z”; 0-9 without shift is returned as “0”-“9”; 0-9 with shift is returned as “SHIFT_0”..”SHIFT_9”; the other key codes are retured as the keys in keycodes, lowercase when without shift (e.g. “left”), uppercase when with shift (e.g. “HOME”).

B.key_to_string = key_to_string = (which, shift) ->             # {{{1
  w = which; s = shift
  switch
    when keyranges.ALPHA.from <= w <= keyranges.ALPHA.to
      c = String.fromCharCode w
      if s then c else c.toLowerCase()
    when keyranges.NUM.from <= w <= keyranges.NUM.to
      c = String.fromCharCode w
      if s then "SHIFT_#{c}" else c
    else
      for k, v of keycodes
        if v == w
          return if s then k else k.toLowerCase()
      null

key codes

B.keycodes = keycodes =
  BACKSPACE: 8, COMMA: 188, DELETE: 46, DOWN: 40, END: 35, ENTER: 13,
  ESCAPE: 27, HOME: 36, LEFT: 37, PAGE_DOWN: 34, PAGE_UP: 33,
  PERIOD: 190, RIGHT: 39, SHIFT: 16, SPACE: 32, TAB: 9, UP: 38

key ranges

B.keyranges = keyranges =
  ALPHA: { from: 65, to: 90 }, NUM: { from: 48, to: 57 }, # ...

mouse functions

Handle mouse clicks. Listens to click on elem, calls f with the {x,y}; returns a function that cancels the listening.

B.handle_click = handle_click = (elem, f, $ = window.$) ->
  h = (e) -> {x,y} = mouse_position e; f x, y; false
  $(elem).on 'click', h
  -> $(elem).off 'click', h

relative mouse position; returns {x,y}

B.mouse_position = mouse_position =
  (event, elem = event.target, $ = window.$, cache = {}) ->
    e            = $(elem)
    cache.left  ?= (e.outerWidth()  - e.width() ) / 2
    cache.top   ?= (e.outerHeight() - e.height()) / 2
    x: event.offsetX - cache.left, y: event.offsetY - cache.top

miscellaneous functions

measure text height and width using a temporary hidden div; returns {w:width,h:height}

B.measure_text = measure_text = ($, text, size, family) ->      # {{{1
  c = measure_text.cache["#{size}|#{family}|#{text}"]
  return c if c
  d = $ '<div>'; d.text text
  d.css display: 'none', 'font-size': size, 'font-family': family
  $('body').append d
  w = d.width(); h = d.height()
  d.remove()
  measure_text.cache["#{size}|#{family}|#{text}"] = {w,h}
measure_text.cache = {}