teste.rb

module RIndexer
  RE_TAG = %r!<([a-z\d_-]+)\s*.*?>!im

  def fetch_tags(doc_stream)
    tags = Hash.new(0)
    doc_stream.scan(RE_TAG).each do |e|
      tags[e] += 1
    end
    tags
  end
end

include RIndexer

arg = ARGV.shift
p fetch_tags(IO.read(arg))

#USAGE: ruby teste.rb document.xml (or html, etc)