haml--haml/bin/html2haml

#!/usr/bin/env ruby

require 'rubygems'
require 'hpricot'
require 'open-uri'

doc = open(ARGV[0]) { |f| Hpricot(f) }

def tabulate(tabs)
  '  ' * tabs
end

def parse_text(text, tabs)
  text.strip!
  if text.empty?
    String.new
  else
    text = text.gsub(/[\t ]+/, " ").gsub(/(\r|\n|\r\n)\s*/, "\n#{tabulate(tabs)}")
    "#{tabulate(tabs)}#{text}\n"
  end
end

module Hpricot::Node
  def to_haml(tabs)
    parse_text(self.to_s, tabs)
  end
end

class Hpricot::Doc
  def to_haml
    output = ''
    children.each { |child| output += child.to_haml(0) }
    output
  end
end

class Hpricot::XMLDecl
  def to_haml(tabs)
    "#{tabulate(tabs)}!!! XML\n"
  end
end

class Hpricot::DocType
  def to_haml(tabs)
    "#{tabulate(tabs)}!!!\n"
  end
end

class Hpricot::Comment
  def to_haml(tabs)
    "#{tabulate(tabs)}/\n#{parse_text(self.content, tabs + 1)}"
  end
end

class Hpricot::Elem
  def to_haml(tabs)
    output = "#{tabulate(tabs)}"
    output += "%#{name}" unless name == 'div'
      
    if attributes
      output += "##{attributes['id']}" if attributes['id']
      attributes['class'].split(' ').each { |c| output += ".#{c}" } if attributes['class']
      attributes.delete("id")
      attributes.delete("class")
      output += attributes.inspect if attributes.length > 0
    end
  
    output += "/" if children.length == 0
    output += "\n"

    self.children.each do |child|
      output += child.to_haml(tabs + 1)
    end

    output
  end
end

puts doc.to_haml