gistfile1.rb

require 'rubygems'
require 'nokogiri'

class PushDocument < Nokogiri::XML::SAX::Document
  attr_reader :document, :current, :parents, :errors, :warnings

  def cdata_block(string)
    current << Nokogiri::XML::CDATA.new(document, string)
  end

  def characters(string)
    current << Nokogiri::XML::Text.new(string, document)
  end

  def comment(string)
    current << Nokogiri::XML::Comment.new(document, string)
  end

  def end_document
    puts "END DOCUMENT!!!"   # added
    # ? TODO maybe add a document callback
  end

  def end_element(name)
    puts "done with #{name}"  # added
    # TODO  check current
    pop
  end

  def end_element_namespace(name, prefix, url)
    # TODO  check current
    pop
  end

  def error(str)
    @errors << str
  end

  def start_document
    @document = Nokogiri::XML::Document.new
    @errors = []
    @warnings = []
    @current = @document
    @parents = [@document.root]
  end

  def start_element(name, attrs = [])
    puts "got #{name}"   # added
    e = Nokogiri::XML::Element.new(name, document)
#    attrs.each { |a| e.add_child a }   # removed, causes error
    push e
  end

  def start_element_namespace(name, attrs=[], prefix=nil, uri=nil, ns=[])
    start_element(name, attrs)
    # TODO apply namespace stuff to current
  end

  def warning(string)
    @warnings << string
  end

  def xmldecl(version, encoding, standalone)
    document.encoding = encoding
    # TODO  can we do anything about version and standalone?
  end

  def push(cur)
    parents.push current
    @current = cur
  end

  def pop
    new_cur = parents.pop
    new_cur.add_child current
    @current = new_cur
  end

  def to_xml(*a)
    document.to_xml(*a)
  end

  def to_html(*a)
    document.to_html(*a)
  end

  def done?
    p current   # added
    p document  # added
    return false unless current && document
#    p document.root  # added
    current == document.root
  end
end

class EasyPush < Nokogiri::XML::SAX::PushParser
  def initialize(doc = PushDocument.new, *args)
    super
  end

  def done?
    document.done?
  end
end

BEGIN { require 'rubygems' if __FILE__ == $0 }
if __FILE__ == $0
  require 'pp'

  ep = EasyPush.new
  ep << ARGF.read(1) until ARGF.eof? # or ep.done?   # ep.done? is always false :/
  puts "done? #{ep.done?}"  # added
  puts ep.document.to_xml
  p ep.document.errors
end