Merge multiple rdf:Description elements with the same rdf:about attribute

merge-multiple-rdf-descriptions.rb

#!/usr/bin/env ruby

#
# @file
#   This script takes an RDF file containing multiple rdf:Description elements
#   with the same rdf:about attribute, combines them into a single 
#   rdf:Description for each rdf:about, and outputs the resultant RDF.
#

require 'nokogiri'

unless ARGV[0]
  puts "Usage: #{__FILE__} RDF_FILENAME"
  exit
end

RDF_FILENAME=ARGV[0]

f = File.open(RDF_FILENAME)
rdf = Nokogiri::XML(f) do |config|
  config.default_xml.noblanks
end
f.close

rdf_descriptions = {}
rdf.root.xpath('./rdf:Description').each do |rdf_description|
  rdf_about = rdf_description['rdf:about']
  if rdf_descriptions.has_key?(rdf_about)
    rdf_description.children.each do |child|
      rdf_descriptions[rdf_about] << child
    end
  else
    rdf_descriptions[rdf_about] = rdf_description
  end
  rdf_description.remove
end

rdf_descriptions.each_value do |rdf_description|
  rdf.root << rdf_description
end

puts rdf.to_xml(:indent => 2)