convert resource url in a html file to data scheme uri

convert_asset_to_inline.rb

require 'open-uri'
require 'image_size'

class ConvertAssetInline
  def initialize
    @orig = nil
    @cache = {}
  end

  def convert(url)
    body = crawl(url)
    convert_img(body, url)
    convert_favicon(body, url)

    body
  end

  private
  def crawl(url, options = {})
    open(url) {|f| f.read }
  end

  def url_to_data(url, base_url = nil, content_type = nil)
    url = URI.join(base_url, url) if base_url
    url = url.to_s
    return @cache[url] if @cache[url]

    target = open(url) {|f| f.read }.freeze

    unless content_type
      target_is = ImageSize.new(target)
      content_type = "image/#{target_is.get_type.downcase}"
    end

    result = datascheme(target, content_type)
    @cache[url] = result

    result
  end

  def datascheme(binary, content_type)
    target_b64 = [binary].pack('m')
    "data:#{content_type};base64,#{target_b64}"
  end

  def convert_img(body, base_url)
    body.gsub!(/<img[^>]*?src=["'](.*?)["']/) { # don't care there are ["'] in the URL
      $&.sub($1, url_to_data($1, base_url))
    }
  end

  def convert_favicon(body, base_url)
    body.gsub!(/<link [^>]*rel="shortcut icon".*?>/) {|matched|
      matched.sub(/href=["'](.*?)["']/) {
        $&.sub($1, url_to_data($1, base_url, 'image/vnd.microsoft.icon'))
      }
    }
  end

  # def convert_stylesheets ...
end


c = ConvertAssetInline.new
puts c.convert ARGV.first

fixme :)