jch/rack-pdf_preflight.rb

## rack-pdf_preflight.rb
require 'base64'
require 'nokogiri'

module Rack
  # Rack middleware to inline css and images from underlying application so it
  # can be proccessed by PDFKit without additional network requests.
  #
  # For example, this middleware will transform this response:
  #
  #    <html>
  #      <head>
  #        <link href="/stylesheet.css" media="all" rel="stylesheet" type="text/css">
  #      </head>
  #      <body>
  #        <img src="/bananas.png" />
  #      </body>
  #    </html>
  #
  # into this response, with all the assets inlined:
  #
  #    <html>
  #      <head>
  #        <style>/* contents of /stylesheet.css */</style>
  #      </head>
  #      <body>
  #        <!-- base64 encoded version of /bananas.png -->
  #        <img src="data:image/png;base64,ZmFrZSBpbWFnZQ==%0A" />
  #      </body>
  #    </html>
  #
  # Usage:
  #
  #    Rack::Builder.app do
  #      use Rack::PDF::PDFPreflight
  #      run YourApp
  #    end
  class PDFPreflight
    def initialize(app, options={})
      @app = app
    end

    def call(env)
      @env = env
      response = Rack::Response.new(@app.call(env))

      bodies = response.body.map do |body|
        doc = Nokogiri::HTML.parse(body)
        inline_css(doc)
        inline_images(doc)
        doc.to_html
      end

      response.body = bodies
      puts response.body
      response.finish
    end

    # Internal: In place replaces `link` elements in `doc` with `style`
    # element containing the asset's styles.
    #
    # Returns nothing.
    def inline_css(doc)
      doc.css('link').each do |node|
        href = node.attr('href')
        if href =~ /css$/i
          app_response = get(href)
          if app_response.ok?
            styles = doc.create_element "style"
            styles.content = app_response.body.join("\n")
            node.replace(styles)
          end
        end
      end
    end

    # Internal: In place replace `img` elements in `doc` with base64 encoded
    # images.
    #
    # Returns nothing.
    def inline_images(doc)
      doc.css('img').each do |node|
        if src = node.attr('src')
          app_response = get(src)
          content_type = app_response.headers['Content-Type']
          asset        = app_response.body.join("")
          node['src']  = "data:#{content_type};base64,#{Base64.encode64(asset)}"
        end
      end
    end

    # Fetch a resource from the underlying Rack application
    #
    # Returns a rack response
    def get(path)
      e = @env.dup
      e['PATH'] = path
      @app.call(e)
    end
  end
end

require 'test/unit'
require 'rack/test'
class Rack::PDFPreflightTest < Test::Unit::TestCase
  include Rack::Test::Methods

  class TestApp
    HTML = <<-HTML
    <html>
      <head>
        <link href="/stylesheet.css" media="all" rel="stylesheet" type="text/css">
      </head>
      <body>
        <img src="/foo.png" />
      </body>
    </html>
    HTML

    def self.call(env)
      case env['PATH']
      when /css$/
        Rack::Response.new(["/* fake css */"], 200, "Content-Type" => "text/css")
      when /png$/
        Rack::Response.new(["fake image"], 200, "Content-Type" => "image/png")
      else
        Rack::Response.new([HTML.clone], 200, "Content-Type" => "text/html")
      end
    end
  end

  def app
    Rack::Builder.app do
      use Rack::PDFPreflight
      run TestApp
    end
  end

  def setup
    get '/'
    @doc = Nokogiri::HTML.parse(last_response.body)
  end

  def test_inline_css
    assert_equal 0, @doc.css('link').size
    assert_equal 1, @doc.css('style').size
    assert_match @doc.css('style').text, /fake css/
  end

  def test_inline_images
    assert_equal 1, @doc.css('img').size
    assert_equal "data:image/png;base64,ZmFrZSBpbWFnZQ==%0A", @doc.css('img').attr('src').to_s
  end
end
	require 'base64'
	require 'nokogiri'

	module Rack
	# Rack middleware to inline css and images from underlying application so it
	# can be proccessed by PDFKit without additional network requests.
	#
	# For example, this middleware will transform this response:
	#
	# <html>
	# <head>
	# <link href="/stylesheet.css" media="all" rel="stylesheet" type="text/css">
	# </head>
	# <body>
	# <img src="/bananas.png" />
	# </body>
	# </html>
	#
	# into this response, with all the assets inlined:
	#
	# <html>
	# <head>
	# <style>/* contents of /stylesheet.css */</style>
	# </head>
	# <body>
	# <!-- base64 encoded version of /bananas.png -->
	# <img src="data:image/png;base64,ZmFrZSBpbWFnZQ==%0A" />
	# </body>
	# </html>
	#
	# Usage:
	#
	# Rack::Builder.app do
	# use Rack::PDF::PDFPreflight
	# run YourApp
	# end
	class PDFPreflight
	def initialize(app, options={})
	@app = app
	end

	def call(env)
	@env = env
	response = Rack::Response.new(@app.call(env))

	bodies = response.body.map do \|body\|
	doc = Nokogiri::HTML.parse(body)
	inline_css(doc)
	inline_images(doc)
	doc.to_html
	end

	response.body = bodies
	puts response.body
	response.finish
	end

	# Internal: In place replaces `link` elements in `doc` with `style`
	# element containing the asset's styles.
	#
	# Returns nothing.
	def inline_css(doc)
	doc.css('link').each do \|node\|
	href = node.attr('href')
	if href =~ /css$/i
	app_response = get(href)
	if app_response.ok?
	styles = doc.create_element "style"
	styles.content = app_response.body.join("\n")
	node.replace(styles)
	end
	end
	end
	end

	# Internal: In place replace `img` elements in `doc` with base64 encoded
	# images.
	#
	# Returns nothing.
	def inline_images(doc)
	doc.css('img').each do \|node\|
	if src = node.attr('src')
	app_response = get(src)
	content_type = app_response.headers['Content-Type']
	asset = app_response.body.join("")
	node['src'] = "data:#{content_type};base64,#{Base64.encode64(asset)}"
	end
	end
	end

	# Fetch a resource from the underlying Rack application
	#
	# Returns a rack response
	def get(path)
	e = @env.dup
	e['PATH'] = path
	@app.call(e)
	end
	end
	end

	require 'test/unit'
	require 'rack/test'
	class Rack::PDFPreflightTest < Test::Unit::TestCase
	include Rack::Test::Methods

	class TestApp
	HTML = <<-HTML
	<html>
	<head>
	<link href="/stylesheet.css" media="all" rel="stylesheet" type="text/css">
	</head>
	<body>
	<img src="/foo.png" />
	</body>
	</html>
	HTML

	def self.call(env)
	case env['PATH']
	when /css$/
	Rack::Response.new(["/* fake css */"], 200, "Content-Type" => "text/css")
	when /png$/
	Rack::Response.new(["fake image"], 200, "Content-Type" => "image/png")
	else
	Rack::Response.new([HTML.clone], 200, "Content-Type" => "text/html")
	end
	end
	end

	def app
	Rack::Builder.app do
	use Rack::PDFPreflight
	run TestApp
	end
	end

	def setup
	get '/'
	@doc = Nokogiri::HTML.parse(last_response.body)
	end

	def test_inline_css
	assert_equal 0, @doc.css('link').size
	assert_equal 1, @doc.css('style').size
	assert_match @doc.css('style').text, /fake css/
	end

	def test_inline_images
	assert_equal 1, @doc.css('img').size
	assert_equal "data:image/png;base64,ZmFrZSBpbWFnZQ==%0A", @doc.css('img').attr('src').to_s
	end
	end