scraping with casperjs and functional utility belt

casperjs_scraping

_ = require 'underscore'
casper = require('casper').create()
utils = require('utils')

curry2 = (fn) ->

  (arg2) ->

    (arg1) ->

      fn arg1, arg2

curry3 = (fn) ->

  (arg3) ->
    
    (arg2) ->

      (arg1) ->

        fn arg1, arg2, arg3

n = 0
releases = []

url = 

	beep: "http://beep.com"
	meep: "http://meep.com"

link = if type = casper.cli.get(0) then url[type] else url.beep

lower = curry2(_.invoke)('toLowerCase')

substr = _.bind(Function.prototype.call, String.prototype.substr)
truncateTo80 = curry3(substr)(80)(0)

split = _.bind(Function.prototype.call, String.prototype.split)
splitByPeriod = curry2(split)('.') 
splitByDash = curry2(split)('-')

join = _.bind(Function.prototype.call, Array.prototype.join)
joinWithDash = curry2(join)('-')

parse = ->

	removeGrp = _.compose(joinWithDash, _.initial, splitByDash)

	contains = (str) ->

		_.compose(curry2(_.contains)(str.toLowerCase()), lower, splitByPeriod, removeGrp)

	is720p = contains '720p'

	isForeign = (str) ->

		languages = ['german', 'french', 'dutch', 'swedish', 'norwegian', 'nlsubbed', 'spanish', 'flemish', 'portuguese']
		_.reduce languages, (memo, language) -> 

			memo || contains(language)(str)
		, false

	elements = @getElementsInfo('#releases div span.release')
	_.chain(elements)
	.pluck('text')
	.filter(is720p)
	.reject(isForeign)
	.map(truncateTo80)
	.value()

start = (n) ->

	suffix = if n == 0 then '' else "/#{n + 1}"
	@start link + suffix, ->

		releases = releases.concat parse.call(@)
dispatch = ->

	if releases.length <= 24 && n < 10

		start.call(@, n++)
		@run dispatch
	else

		echo = _.bind(@echo, @)
		_.chain(releases).first(24).each(echo)
		@exit()

casper.start().then ->
casper.run(dispatch)