Build SameTime emoticon palettes from GitHub Emoji

emoji_scrape.rb

require 'mechanize'
require 'open-uri'

EMOJI_SITE = 'http://www.emoji-cheat-sheet.com'
COLLECTIONS = %w[people nature objects places symbols]

agent = Mechanize.new
page = agent.get(EMOJI_SITE)

COLLECTIONS.each do |set|
  Dir.mkdir set unless Dir.exist? set

  builder = Nokogiri::XML::Builder.new do |xml|
    xml.palette('name' => "Emojis-#{set.capitalize}")
  end
  palette_node = builder.doc.xpath('//palette').first

  elements = page.search("#emoji-#{set} > li > div")
  elements.each do |element|
    emoji_path = element.css('.emoji').first.attributes['data-src'].value
    emoji_name = element.css('.name').first.text

    filename = emoji_path.rpartition('/').last
    File.open(File.join(set, filename), 'wb') do |saved_file|
      open(EMOJI_SITE + '/' + emoji_path, 'rb') do |read_file|
        saved_file.write(read_file.read)
      end
    end

    Nokogiri::XML::Builder.with(palette_node) do |xml|
      xml.item do |xml|
        xml.name { |xml| xml.cdata emoji_name }
        xml.imgfile { |xml| xml.text filename }
        xml.alttext { |xml| xml.cdata emoji_name }
        xml.keyboard { |xml| xml.cdata ":#{emoji_name}:" }
        xml.kbrdregexp { |xml| xml.cdata "[:]#{emoji_name}[:]" }
      end
    end

  end

  File.open("#{set}/palette.xml", 'w') { |f| f.write builder.to_xml }
end