omg.rb

###
# Read in XML as a stream, write out JSON as a stream.  As little information
# is kept in memory as possible.
require 'nokogiri'
require 'psych'

class JSONTranslator < Nokogiri::XML::SAX::Document
  attr_reader :emitter

  def initialize emitter
    @emitter = emitter
    super()
  end

  def start_document
    emitter.start_stream Psych::Nodes::Stream::UTF8
    emitter.start_document [], [], true
  end

  def end_document
    emitter.end_document
    emitter.end_stream
  end

  def start_element_namespace name, attrs = [], prefix = nil, uri = nil, ns = []
    start_map
    scalar name
    start_map

    unless attrs.empty?
      scalar 'attributes'
      mapping do
        attrs.each { |attr|
          scalar attr.localname
          scalar attr.value
        }
      end
    end

    scalar 'children'
    start_sequence
  end

  def characters string
    scalar string
  end

  def end_element_namespace name, prefix = nil, uri = nil
    end_sequence
    end_map
    end_map
  end

  private
  def scalar string
    emitter.scalar string, nil, nil, false, true, Psych::Nodes::Scalar::PLAIN
  end

  def start_map
    emitter.start_mapping nil, nil, false, 1
  end

  def end_map
    emitter.end_mapping
  end

  def start_sequence
    emitter.start_sequence nil, nil, false, 1
  end

  def end_sequence
    emitter.end_sequence
  end

  def mapping
    start_map
    yield
    end_map
  end
end

###
# +parser+ calls XML events on the +translator+. +translator+ calls JSON events
# on the +emitter+/
emitter    = Psych::JSON::Stream::Emitter.new $stdout
translator = JSONTranslator.new emitter
parser     = Nokogiri::XML::SAX::Parser.new translator

File.open ARGV[0], 'rb' do |f|
  parser.parse f
end

__END__
<one two="three">
  four
  <ul>
    <li>
      hi
      <div>lol</div>
      mom!
    </li>
    <li>hello world</li>
  </ul>
</one>