ehhhhh.rb

require 'rubygems'
require 'sinatra'
require 'open-uri'
require 'java'
require File.dirname(__FILE__)+"/jtidy-r938.jar"

Dir.glob(File.dirname(__FILE__)+"/flyingsaucer/*.jar").each do |f|
  require f
end

java_import "com.lowagie.text.DocumentException"
java_import "org.xhtmlrenderer.pdf.ITextRenderer"
java_import "java.io.ByteArrayOutputStream"
java_import "org.w3c.tidy.Tidy"  


include Java::JavaxXmlParsers
include Java::JavaIo

get '/' do
  process_url if params[:url]
end

def process_url
  url = params[:url] 
  raw_html = open(url).read
  buf = java.lang.StringBuffer.new
  buf.append raw_html
  sbi =  java.io.StringBufferInputStream.new(buf.toString)  
  baos = java.io.ByteArrayOutputStream.new
  
  tidy = Tidy.new                                                
  #tidy.setXHTML true      
  doc = tidy.parse(sbi, baos)   
  
  builderFactory = javax.xml.parsers.DocumentBuilderFactory.newInstance
  builder = builderFactory.newDocumentBuilder
  
  doc = builder.parse java.io.StringBufferInputStream.new(baos.toString)
  renderer = ITextRenderer.new
  renderer.setDocument doc, url
  renderer.layout
  output = java.io.ByteArrayOutputStream.new
  renderer.createPDF output
  content_type 'application/pdf'
  output.toString
end