require 'rubygems'
require 'sanitize'

html = ARGF.read
puts Sanitize.clean(html,
                    :elements => ['a', 'p',
                                  'h1', 'h2', 'h3', 'h4', 'h5', 'h6',
                                  'ul', 'li',
                                  'table', 'tr', 'th', 'td'],
                    :attributes => {'a' => ['href', 'title']})